1 Release 3.15.0 (?? ??????? 2019)
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4 3.15.0 is a feature release with many improvements and the usual collection of
7 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
8 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
9 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
10 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
11 support for X86/macOS 10.13, AMD64/macOS 10.13.
13 * ==================== CORE CHANGES ===================
15 * The XTree Massif output format now makes use of the information obtained
16 when specifying --read-inline-info=yes.
18 * ================== PLATFORM CHANGES =================
21 * ==================== TOOL CHANGES ====================
25 - cg_annotate has a new option, --show-percs, which prints percentages next
30 - callgrind_annotate has a new option, --show-percs, which prints percentages
31 next to all event counts.
33 - callgrind_annotate now inserts commas in call counts, and
34 sort the caller/callee lists in the call tree.
37 - The default value for --read-inline-info is now "yes" on
38 Linux/Android/Solaris. It is still "no" on other OS.
41 - The option --xtree-leak=yes (to output leak result in xtree format)
42 automatically activates the option --show-leak-kinds=all,
43 as xtree visualisation tools such as kcachegrind can in any case
44 select what kind of leak to visualise.
46 * ==================== OTHER CHANGES ====================
48 * The new option --show-error-list=no|yes
49 allows to show at exit the list of detected errors and the used
50 suppressions. In the previous valgrind version, showing this
51 information could only be done using a verbosity >= 2, which is
52 producing a lot of other non necessarily useful messages.
53 The option -s is equivalent to --show-error-list=yes.
55 * ==================== FIXED BUGS ====================
57 The following bugs have been fixed or resolved. Note that "n-i-bz"
58 stands for "not in bugzilla" -- that is, a bug that was reported to us
59 but never got a bugzilla entry. We encourage you to file bugs in
60 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
61 than mailing the developers (or mailing lists) directly -- bugs that
62 are not entered into bugzilla tend to get forgotten about or ignored.
64 To see details of a given bug, visit
65 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
66 where XXXXXX is the bug number as listed below.
68 399301 Use inlined frames in Massif XTree output.
69 399322 Improve callgrind_annotate output
70 400490 s390x: VRs allocated as if separate from FPRs
71 400491 s390x: Operand of LOCH treated as unsigned integer
72 397187 z13 vector register support for vgdb gdbserver
73 401277 More bugs in z13 support
74 401112 LLVM 5.0 generates comparison against partially initialized data
75 385411 s390x: z13 vector floating-point instructions not implemented
76 401578 drd: crashes sometimes on fork()
77 401627 memcheck errors with glibc avx2 optimized wcsncmp
78 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
79 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
80 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
81 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
82 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
83 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
84 402480 Do not use %rsp in clobber list
85 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
86 402515 Implement new option --show-error-list=no|yes / -s
87 402519 POWER 3.0 addex instruction incorrectly implemented
89 Release 3.14.0 (9 October 2018)
90 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
92 3.14.0 is a feature release with many improvements and the usual collection of
95 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
96 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
97 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
98 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
99 support for X86/macOS 10.13, AMD64/macOS 10.13.
101 * ==================== CORE CHANGES ===================
103 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
104 debug info for unloaded code. This allows saved stack traces (e.g. for
105 memory leaks) to include file/line info for code that has been dlclose'd (or
106 similar). See the user manual for more information and known limitations.
108 * Ability to specify suppressions based on source file name and line number.
110 * Majorly overhauled register allocator. No end-user changes, but the JIT
111 generates code a bit more quickly now.
113 * ================== PLATFORM CHANGES =================
115 * Preliminary support for macOS 10.13 has been added.
117 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
119 * mips: support for MIPS SIMD architecture (MSA) has been added.
121 * mips: support for MIPS N32 ABI has been added.
123 * s390: partial support for vector instructions (integer and string) has been
126 * ==================== TOOL CHANGES ====================
128 * Helgrind: Addition of a flag
129 --delta-stacktrace=no|yes [yes on linux amd64/x86]
130 which specifies how full history stack traces should be computed.
131 Setting this to =yes can speed up Helgrind by 25% when using
132 --history-level=full.
134 * Memcheck: reduced false positive rate for optimised code created by Clang 6
135 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
136 blocks more carefully to determine where it can avoid expensive definedness
137 checks without loss of precision. This is controlled by the flag
138 --expensive-definedness-checks=no|auto|yes [auto].
140 * ==================== OTHER CHANGES ====================
142 * Valgrind is now buildable with link-time optimisation (LTO). A new
143 configure option --enable-lto=yes allows building Valgrind with LTO. If the
144 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
145 Note that if you are doing Valgrind development, --enable-lto=yes massively
146 slows down the build process.
148 * ==================== FIXED BUGS ====================
150 The following bugs have been fixed or resolved. Note that "n-i-bz"
151 stands for "not in bugzilla" -- that is, a bug that was reported to us
152 but never got a bugzilla entry. We encourage you to file bugs in
153 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
154 than mailing the developers (or mailing lists) directly -- bugs that
155 are not entered into bugzilla tend to get forgotten about or ignored.
157 To see details of a given bug, visit
158 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
159 where XXXXXX is the bug number as listed below.
161 79362 Debug info is lost for .so files when they are dlclose'd
162 208052 strlcpy error when n = 0
163 255603 exp-sgcheck Assertion '!already_present' failed
164 338252 building valgrind with -flto (link time optimisation) fails
165 345763 MIPS N32 ABI support
166 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
167 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
168 372347 Replacement problem of the additional c++14/c++17 new/delete operators
169 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
170 376257 helgrind history full speed up using a cached stack
171 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
173 379748 Fix missing pselect syscall (OS X 10.11)
174 379754 Fix missing syscall ulock_wait (OS X 10.12)
175 380397 s390x: __GI_strcspn() replacemenet needed
176 381162 possible array overrun in VEX register allocator
177 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
178 381274 powerpc too chatty even with --sigill-diagnostics=no
179 381289 epoll_pwait can have a NULL sigmask
180 381553 VEX register allocator v3
181 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
182 381769 Use ucontext_t instead of struct ucontext
183 381805 arm32 needs ld.so index hardwire for new glibc security fixes
184 382256 gz compiler flag test doesn't work for gold
185 382407 vg_perf needs "--terse" command line option
186 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
187 382563 MIPS MSA ASE support
188 382998 xml-socket doesn't work
189 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
190 383723 Fix missing kevent_qos syscall (macOS 10.11)
191 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
192 384096 Mention AddrCheck at Memcheck's command line option [..]
193 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
194 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
195 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
196 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
197 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
198 384337 performance improvements to VEX register allocator v2 and v3
199 384526 reduce number of spill insns generated by VEX register allocator v3
200 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
201 384631 Sanitise client args as printed with -v
202 384633 Add a simple progress-reporting facility
203 384987 VEX regalloc: allocate caller-save registers for short lived vregs
204 385055 PPC VEX temporary storage exhausted
205 385182 PPC64 is missing support for the DSCR
206 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
207 385207 PPC64, generate_store_FPRF() generates too many Iops
208 385208 PPC64, xxperm instruction exhausts temporary memory
209 385210 PPC64, vpermr instruction could exhaust temporary memory
210 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
211 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
212 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
213 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
214 385408 s390x: z13 vector "support" instructions not implemented
215 385409 s390x: z13 vector integer instructions not implemented
216 385410 s390x: z13 vector string instructions not implemented
217 385412 s390x: new non-vector z13 instructions not implemented
218 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
219 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
220 385939 Optionally exit on the first error
221 386318 valgrind.org/info/tools.html is missing SGCheck
222 386425 running valgrind + wine on armv7l gives illegal opcode
223 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
224 387410 MIPSr6 support
225 387664 Memcheck: make expensive-definedness-checks be the default
226 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
227 387766 asm shifts cause false positive "Conditional jump or move depends
228 on uninitialised value"
229 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
230 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
231 388786 Support bpf syscall in amd64 Linux
232 388862 Add replacements for wmemchr and wcsnlen on Linux
233 389065 valgrind meets gcc flag -Wlogical-op
234 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
235 390471 suppression by specification of source-file line number
236 390723 make xtree dump files world wide readable, similar to log files
237 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
238 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
239 392118 unhandled amd64-linux syscall: 332 (statx)
240 392449 callgrind not clearing the number of calls properly
241 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
242 lxssp, stxssp and stxvl instructions.
243 393023 callgrind_control risks using the wrong vgdb
244 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
245 393099 posix_memalign() invalid write if alignment == 0
246 393146 failing assert "is_DebugInfo_active(di)"
247 395709 PPC64 is missing support for the xvnegsp instruction
248 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
250 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
251 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
252 396839 s390x: Trap instructions not implemented
253 396887 arch_prctl should return EINVAL on unknown option
254 == 397286 crash before launching binary (Unsupported arch_prctl option)
255 == 397393 valgrind: the 'impossible' happened: (Archlinux)
256 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
257 396906 compile tests failure on mips32-linux: broken inline asm in tests on
259 397012 glibc ld.so uses arch_prctl on i386
260 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
261 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
262 397424 glibc 2.27 and gdb_server tests
263 398028 Assertion `cfsi_fits` failing in simple C program
264 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
266 n-i-bz Fix missing workq_ops operations (macOS)
267 n-i-bz fix bug in strspn replacement
268 n-i-bz Add support for the Linux BLKFLSBUF ioctl
269 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
270 n-i-bz Fix possible stack trashing by semctl syscall wrapping
271 n-i-bz Add support for the Linux membarrier() system call
272 n-i-bz x86 front end: recognise and handle UD2 correctly
273 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
274 correctly aligned before entering the handler.
276 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
277 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
278 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
282 Release 3.13.0 (15 June 2017)
283 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
285 3.13.0 is a feature release with many improvements and the usual collection of
288 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
289 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
290 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
291 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
293 * ==================== CORE CHANGES ===================
295 * The translation cache size has been increased to keep up with the demands of
296 large applications. The maximum number of sectors has increased from 24 to
297 48. The default number of sectors has increased from 16 to 32 on all
298 targets except Android, where the increase is from 6 to 12.
300 * The amount of memory that Valgrind can use has been increased from 64GB to
301 128GB. In particular this means your application can allocate up to about
302 60GB when running on Memcheck.
304 * Valgrind's default load address has been changed from 0x3800'0000 to
305 0x5800'0000, so as to make it possible to load larger executables. This
306 should make it possible to load executables of size at least 1200MB.
308 * A massive spaceleak caused by reading compressed debuginfo files has been
309 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
312 * The C++ demangler has been updated.
314 * Support for demangling Rust symbols has been added.
316 * A new representation of stack traces, the "XTree", has been added. An XTree
317 is a tree of stacktraces with data associated with the stacktraces. This is
318 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
319 consumption of your program. Reporting is controlled by the new options
320 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
322 A report can also be produced on demand using the gdbserver monitor command
323 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
324 format' and 'massif format. The existing visualisers for these formats (e.g.
325 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
326 analyse these reports.
328 Memcheck can also produce XTree leak reports using the Callgrind file
329 format. For more details, see the user manual.
331 * ================== PLATFORM CHANGES =================
333 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
335 * amd64: fixes for JIT failure problems on long AVX2 code blocks
337 * amd64 and x86: support for CET prefixes has been added
339 * arm32: a few missing ARMv8 instructions have been implemented
341 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
342 Store-Conditional instructions has been added. This is to deal with
343 processor implementations that implement the LL/SC specifications strictly
344 and as a result cause Valgrind to hang in certain situations. The
345 alternative implementation is automatically enabled at startup, as required.
346 You can use the option --sim-hints=fallback-llsc to force-enable it if you
349 * Support for OSX 10.12 has been improved.
351 * On Linux, clone handling has been improved to honour CLONE_VFORK that
352 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
353 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
354 CLONE_VM exact semantics will (still) not work.
356 * The TileGX/Linux port has been removed because it appears to be both unused
359 * ==================== TOOL CHANGES ====================
363 - Memcheck should give fewer false positives when running optimised
364 Clang/LLVM generated code.
366 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
368 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
369 to produce the end of execution leak report in a xtree callgrind format
372 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
373 the leak report in an xtree file.
377 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
379 - For some workloads (typically, for big applications), Massif memory
380 consumption and CPU consumption has decreased significantly.
384 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
386 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
387 for Ada gnat compiled applications.
389 * ==================== OTHER CHANGES ====================
391 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
392 append the inner guest stacktrace to the inner host stacktrace. This helps
393 to investigate the errors reported by the outer, when they are caused by the
394 inner guest program (such as an inner regtest). See README_DEVELOPERS for
397 * To allow fast detection of callgrind files by desktop environments and file
398 managers, the format was extended to have an optional first line that
399 uniquely identifies the format ("# callgrind format"). Callgrind creates
400 this line now, as does the new xtree functionality.
402 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
403 have a new %n format letter that is replaced by a sequence number.
405 * "--version -v" now shows the SVN revision numbers from which Valgrind was
408 * ==================== FIXED BUGS ====================
410 The following bugs have been fixed or resolved. Note that "n-i-bz"
411 stands for "not in bugzilla" -- that is, a bug that was reported to us
412 but never got a bugzilla entry. We encourage you to file bugs in
413 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
414 than mailing the developers (or mailing lists) directly -- bugs that
415 are not entered into bugzilla tend to get forgotten about or ignored.
417 To see details of a given bug, visit
418 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
419 where XXXXXX is the bug number as listed below.
421 162848 --log-file output isn't split when a program forks
422 340777 Illegal instruction on mips (ar71xx)
423 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
424 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
425 to a different stack.
426 344139 x86 stack-seg overrides, needed by the Wine people
427 344524 store conditional of guest applications always fail - observed on
429 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
430 352395 Please provide SVN revision info in --version -v
431 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
432 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
433 INVALID_POSIX_THREADID' failed
434 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
435 barrier implementation
436 358697 valgrind.h: Some code remains even when defining NVALGRIND
437 359202 Add musl libc configure/compile
438 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
439 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
440 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
441 362223 assertion failed when .valgrindrc is a directory instead of a file
442 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
443 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
444 368507 can't malloc chunks larger than about 34GB
445 368529 Android arm target link error, missing atexit and pthread_atfork
446 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
447 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
448 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
449 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
450 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
451 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
452 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
453 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
454 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
455 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
456 370028 Reduce the number of compiler warnings on MIPS platforms
457 370635 arm64 missing syscall getcpu
458 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
459 371227 Clean AArch64 syscall table
460 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
461 371471 Valgrind complains about non legit memory leaks on placement new (C++)
462 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
463 371503 disInstr(arm64): unhandled instruction 0xF89F0000
464 371869 support '%' in symbol Z-encoding
465 371916 execution tree xtree concept
466 372120 c++ demangler demangles symbols which are not c++
467 372185 Support of valgrind on ARMv8 with 32 bit executable
468 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
469 372195 Power PC, xxsel instruction is not always recognized.
470 372504 Hanging on exit_group
471 372600 process loops forever when fatal signals are arriving quickly
472 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
473 373046 Stacks registered by core are never deregistered
474 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
475 373086 Implement additional Xen hypercalls
476 373192 Calling posix_spawn in glibc 2.24 completely broken
477 373488 Support for fanotify API on ARM64 architecture
478 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
479 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
480 373938 const IRExpr arguments for matchIRExpr()
481 374719 some spelling fixes
482 374963 increase valgrind's load address to prevent mmap failure
483 375514 valgrind_get_tls_addr() does not work in case of static TLS
484 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
485 for ML_(find_rx_mapping)()
486 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
487 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
488 == 377159 "vex: the `impossible' happened" still present
489 == 375150 Assertion 'tres.status == VexTransOK' failed
490 == 378068 valgrind crashes on AVX2 function in FFmpeg
491 376142 Segfaults on MIPS Cavium Octeon boards
492 376279 disInstr(arm64): unhandled instruction 0xD50320FF
493 376455 Solaris: unhandled syscall lgrpsys(180)
494 376518 Solaris: unhandled fast trap getlgrp(6)
495 376611 ppc64 and arm64 don't know about prlimit64 syscall
496 376729 PPC64, remove R2 from the clobber list
498 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
499 to be wrongly marked as addressable
500 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
501 PIE enabled by default
502 377376 memcheck/tests/linux/getregset fails with glibc2.24
503 377427 PPC64, lxv instruction failing on odd destination register
504 377478 PPC64: ISA 3.0 setup fixes
505 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
506 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
507 and 2 args for FUTEX_TRYLOCK_PI
508 377717 Fix massive space leak when reading compressed debuginfo sections
509 377891 Update Xen 4.6 domctl wrappers
510 377930 fcntl syscall wrapper is missing flock structure check
511 378524 libvexmultiarch_test regression on s390x and ppc64
512 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
513 378673 Update libiberty demangler
514 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
515 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
516 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
517 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
518 (task_register_dyld_image_infos)
519 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
520 (task_register_dyld_shared_cache_image_info)
521 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
522 379473 MIPS: add support for rdhwr cycle counter register
523 379504 remove TileGX/Linux port
524 379525 Support more x86 nop opcodes
525 379838 disAMode(x86): not an addr!
526 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
527 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
528 379895 clock_gettime does not execute POST syscall wrapper
529 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
530 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
531 380200 xtree generated callgrind files refer to files without directory name
532 380202 Assertion failure for cache line size (cls == 64) on aarch64.
533 380397 s390x: __GI_strcspn() replacement needed
534 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
536 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
537 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
538 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
542 Release 3.12.0 (20 October 2016)
543 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
545 3.12.0 is a feature release with many improvements and the usual
546 collection of bug fixes.
548 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
549 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
550 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
551 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
552 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
553 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
555 * ================== PLATFORM CHANGES =================
557 * POWER: Support for ISA 3.0 has been added
559 * mips: support for O32 FPXX ABI has been added.
560 * mips: improved recognition of different processors
561 * mips: determination of page size now done at run time
563 * amd64: Partial support for AMD FMA4 instructions.
565 * arm, arm64: Support for v8 crypto and CRC instructions.
567 * Improvements and robustification of the Solaris port.
569 * Preliminary support for MacOS 10.12 (Sierra) has been added.
571 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
572 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
573 where possible. Valgrind's support for 32-bit x86 has stagnated in
574 recent years and has fallen far behind that for 64-bit x86
575 instructions. By contrast 64-bit x86 is well supported, up to and
578 * ==================== TOOL CHANGES ====================
582 - Added meta mempool support for describing a custom allocator which:
583 - Auto-frees all chunks assuming that destroying a pool destroys all
585 - Uses itself to allocate other memory blocks
587 - New flag --ignore-range-below-sp to ignore memory accesses below
588 the stack pointer, if you really have to. The related flag
589 --workaround-gcc296-bugs=yes is now deprecated. Use
590 --ignore-range-below-sp=1024-1 as a replacement.
594 - Improved thread startup time significantly on non-Linux platforms.
598 - Added collection of the metric "tot-blocks-allocd"
600 * ==================== OTHER CHANGES ====================
602 * Replacement/wrapping of malloc/new related functions is now done not just
603 for system libraries by default, but for any globally defined malloc/new
604 related function (both in shared libraries and statically linked alternative
605 malloc implementations). The dynamic (runtime) linker is excluded, though.
606 To only intercept malloc/new related functions in
607 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
608 "nouserintercepts" can be any non-existing library name).
609 This new functionality is not implemented for MacOS X.
611 * The maximum number of callers in a suppression entry is now equal to
612 the maximum size for --num-callers (500).
613 Note that --gen-suppressions=yes|all similarly generates suppressions
614 containing up to --num-callers frames.
616 * New and modified GDB server monitor features:
618 - Valgrind's gdbserver now accepts the command 'catch syscall'.
619 Note that you must have GDB >= 7.11 to use 'catch syscall' with
622 * New option --run-cxx-freeres=<yes|no> can be used to change whether
623 __gnu_cxx::__freeres() cleanup function is called or not. Default is
626 * Valgrind is able to read compressed debuginfo sections in two formats:
627 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
628 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
630 * Modest JIT-cost improvements: the cost of instrumenting code blocks
631 for the most common use case (x86_64-linux, Memcheck) has been
634 * Improved performance for programs that do a lot of discarding of
635 instruction address ranges of 8KB or less.
637 * The C++ symbol demangler has been updated.
639 * More robustness against invalid syscall parameters on Linux.
641 * ==================== FIXED BUGS ====================
643 The following bugs have been fixed or resolved. Note that "n-i-bz"
644 stands for "not in bugzilla" -- that is, a bug that was reported to us
645 but never got a bugzilla entry. We encourage you to file bugs in
646 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
647 than mailing the developers (or mailing lists) directly -- bugs that
648 are not entered into bugzilla tend to get forgotten about or ignored.
650 To see details of a given bug, visit
651 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
652 where XXXXXX is the bug number as listed below.
654 191069 Exiting due to signal not reported in XML output
655 199468 Suppressions: stack size limited to 25
656 while --num-callers allows more frames
657 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
658 278744 cvtps2pd with redundant RexW
659 303877 valgrind doesn't support compressed debuginfo sections.
660 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
661 348345 Assertion fails for negative lineno
662 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
663 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
664 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
665 351804 Crash on generating suppressions for "printf" call on OS X 10.10
666 352197 mips: mmap2() not wrapped correctly for page size > 4096
667 353083 arm64 doesn't implement various xattr system calls
668 353084 arm64 doesn't support sigpending system call
669 353137 www: update info for Supported Platforms
670 353138 www: update "The Valgrind Developers" page
671 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
674 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
675 353398 WARNING: unhandled amd64-solaris syscall: 207
676 353660 XML in auxwhat tag not escaping reserved symbols properly
677 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
678 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
679 353802 ELF debug info reader confused with multiple .rodata sections
680 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
681 353917 unhandled amd64-solaris syscall fchdir(120)
682 353920 unhandled amd64-solaris syscall: 170
683 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
684 354392 unhandled amd64-solaris syscall: 171
685 354797 Vbit test does not include Iops for Power 8 instruction support
686 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
690 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
691 355188 valgrind should intercept all malloc related global functions
692 355454 do not intercept malloc related symbols from the runtime linker
693 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
694 356044 Dwarf line info reader misinterprets is_stmt register
695 356112 mips: replace addi with addiu
696 356393 valgrind (vex) crashes because isZeroU happened
699 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
700 356678 arm64-linux: unhandled syscall 232 (mincore)
701 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
702 356823 Unsupported ARM instruction: stlex
703 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
704 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
705 357673 crash if I try to run valgrind with a binary link with libcurl
706 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
707 357871 pthread_spin_destroy not properly wrapped
708 357887 Calls to VG_(fclose) do not close the file descriptor
709 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
710 358030 support direct socket calls on x86 32bit (new in linux 4.3)
711 358478 drd/tests/std_thread.cpp doesn't build with GCC6
712 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
713 359181 Buffer Overflow during Demangling
714 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
715 359289 s390x: popcnt (B9E1) not implemented
716 359472 The Power PC vsubuqm instruction doesn't always give the correct result
717 359503 Add missing syscalls for aarch64 (arm64)
718 359645 "You need libc6-dbg" help message could be more helpful
719 359703 s390: wire up separate socketcalls system calls
720 359724 getsockname might crash - deref_UInt should call safe_to_deref
721 359733 amd64 implement ld.so strchr/index override like x86
722 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
723 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
725 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
726 359871 Incorrect mask handling in ppoll
727 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
728 360008 Contents of Power vr registers contents is not printed correctly when
729 the --vgdb-shadow-registers=yes option is used
730 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
732 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
733 360425 arm64 unsupported instruction ldpsw
735 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
736 360571 Error about the Android Runtime reading below the stack pointer on ARM
737 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
738 360749 kludge for multiple .rodata sections on Solaris no longer needed
739 360752 raise the number of reserved fds in m_main.c from 10 to 12
740 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
741 361226 s390x: risbgn (EC59) not implemented
742 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
743 361354 ppc64[le]: wire up separate socketcalls system calls
744 361615 Inconsistent termination for multithreaded process terminated by signal
745 361926 Unhandled Solaris syscall: sysfs(84)
746 362009 V dumps core on unimplemented functionality before threads are created
747 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
748 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
749 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
750 362953 Request for an update to the Valgrind Developers page
751 363680 add renameat2() support
752 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
753 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
754 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
755 364058 clarify in manual limitations of array overruns detections
756 364413 pselect sycallwrapper mishandles NULL sigmask
757 364728 Power PC, missing support for several HW registers in
758 get_otrack_shadow_offset_wrk()
759 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
760 365273 Invalid write to stack location reported after signal handler runs
761 365912 ppc64BE segfault during jm-insns test (RELRO)
762 366079 FPXX Support for MIPS32 Valgrind
763 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
764 366344 Multiple unhandled instruction for Aarch64
765 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
766 367995 Integration of memcheck with custom memory allocator
767 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
768 368412 False positive result for altivec capability check
769 368416 Add tc06_two_races_xml.exp output for ppc64
770 368419 Perf Events ioctls not implemented
771 368461 mmapunmap test fails on ppc64
772 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
773 369000 AMD64 fma4 instructions unsupported.
774 369169 ppc64 fails jm_int_isa_2_07 test
775 369175 jm_vec_isa_2_07 test crashes on ppc64
776 369209 valgrind loops and eats up all memory if cwd doesn't exist.
777 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
778 369359 msghdr_foreachfield can crash when handling bad iovec
779 369360 Bad sigprocmask old or new sets can crash valgrind
780 369361 vmsplice syscall wrapper crashes on bad iovec
781 369362 Bad sigaction arguments crash valgrind
782 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
783 369402 Bad set/get_thread_area pointer crashes valgrind
784 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
785 369446 valgrind crashes on unknown fcntl command
786 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
787 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
788 370265 ISA 3.0 HW cap stuff needs updating
789 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
790 372195 Power PC, xxsel instruction is not always recognized
792 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
793 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
794 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
795 n-i-bz Fix ppoll_alarm exclusion on OS X
796 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
797 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
798 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
799 n-i-bz amd64: memcheck false positive with shr %edx
800 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
801 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
802 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
803 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
805 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
806 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
807 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
811 Release 3.11.0 (22 September 2015)
812 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
814 3.11.0 is a feature release with many improvements and the usual
815 collection of bug fixes.
817 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
818 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
819 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
820 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
821 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
822 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
824 * ================== PLATFORM CHANGES =================
826 * Support for Solaris/x86 and Solaris/amd64 has been added.
828 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
830 * Preliminary support for the Tilera TileGX architecture has been added.
832 * s390x: It is now required for the host to have the "long displacement"
833 facility. The oldest supported machine model is z990.
835 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
836 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
838 * The JIT's register allocator is significantly faster, making the JIT
839 as a whole somewhat faster, so JIT-intensive activities, for example
840 program startup, are modestly faster, around 5%.
842 * There have been changes to the default settings of several command
843 line flags, as detailed below.
845 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
846 capable hosts, the simulated CPUID will now indicate AVX2 support.
848 * ==================== TOOL CHANGES ====================
852 - The default value for --leak-check-heuristics has been changed from
853 "none" to "all". This helps to reduce the number of possibly
854 lost blocks, in particular for C++ applications.
856 - The default value for --keep-stacktraces has been changed from
857 "malloc-then-free" to "malloc-and-free". This has a small cost in
858 memory (one word per malloc-ed block) but allows Memcheck to show the
859 3 stacktraces of a dangling reference: where the block was allocated,
860 where it was freed, and where it is acccessed after being freed.
862 - The default value for --partial-loads-ok has been changed from "no" to
863 "yes", so as to avoid false positive errors resulting from some kinds
866 - A new monitor command 'xb <addr> <len>' shows the validity bits of
867 <len> bytes at <addr>. The monitor command 'xb' is easier to use
868 than get_vbits when you need to associate byte data value with
869 their corresponding validity bits.
871 - The 'block_list' monitor command has been enhanced:
872 o it can print a range of loss records
873 o it now accepts an optional argument 'limited <max_blocks>'
874 to control the number of blocks printed.
875 o if a block has been found using a heuristic, then
876 'block_list' now shows the heuristic after the block size.
877 o the loss records/blocks to print can be limited to the blocks
878 found via specified heuristics.
880 - The C helper functions used to instrument loads on
881 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
882 replaced by handwritten assembly sequences. This gives speedups
883 in the region of 0% to 7% for those targets only.
885 - A new command line option, --expensive-definedness-checks=yes|no,
886 has been added. This is useful for avoiding occasional invalid
887 uninitialised-value errors in optimised code. Watch out for
888 runtime degradation, as this can be up to 25%. As always, though,
889 the slowdown is highly application specific. The default setting
894 - A new monitor command 'all_snapshots <filename>' dumps all
895 snapshots taken so far.
899 - Significant memory reduction and moderate speedups for
900 --history-level=full for applications accessing a lot of memory
901 with many different stacktraces.
903 - The default value for --conflict-cache-size=N has been doubled to
904 2000000. Users that were not using the default value should
905 preferably also double the value they give.
907 The default was changed due to the changes in the "full history"
908 implementation. Doubling the value gives on average a slightly more
909 complete history and uses similar memory (or significantly less memory
910 in the worst case) than the previous implementation.
912 - The Helgrind monitor command 'info locks' now accepts an optional
913 argument 'lock_addr', which shows information about the lock at the
916 - When using --history-level=full, the new Helgrind monitor command
917 'accesshistory <addr> [<len>]' will show the recorded accesses for
918 <len> (or 1) bytes at <addr>.
920 * ==================== OTHER CHANGES ====================
922 * The default value for the --smc-check option has been changed from
923 "stack" to "all-non-file" on targets that provide automatic D-I
924 cache coherence (x86, amd64 and s390x). The result is to provide,
925 by default, transparent support for JIT generated and self-modifying
928 * Mac OS X only: the default value for the --dsymutil option has been
929 changed from "no" to "yes", since any serious usage on Mac OS X
930 always required it to be "yes".
932 * The command line options --db-attach and --db-command have been removed.
933 They were deprecated in 3.10.0.
935 * When a process dies due to a signal, Valgrind now shows the signal
936 and the stacktrace at default verbosity (i.e. verbosity 1).
938 * The address description logic used by Memcheck and Helgrind now
939 describes addresses in anonymous segments, file mmap-ed segments,
940 shared memory segments and the brk data segment.
942 * The new option --error-markers=<begin>,<end> can be used to mark the
943 begin/end of errors in textual output mode, to facilitate
944 searching/extracting errors in output files that mix valgrind errors
947 * The new option --max-threads=<number> can be used to change the number
948 of threads valgrind can handle. The default is 500 threads which
949 should be more than enough for most applications.
951 * The new option --valgrind-stacksize=<number> can be used to change the
952 size of the private thread stacks used by Valgrind. This is useful
953 for reducing memory use or increasing the stack size if Valgrind
954 segfaults due to stack overflow.
956 * The new option --avg-transtab-entry-size=<number> can be used to specify
957 the expected instrumented block size, either to reduce memory use or
958 to avoid excessive retranslation.
960 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
962 * New and modified GDB server monitor features:
964 - When a signal is reported in GDB, you can now use the GDB convenience
965 variable $_siginfo to examine detailed signal information.
967 - Valgrind's gdbserver now allows the user to change the signal
968 to deliver to the process. So, use 'signal SIGNAL' to continue execution
969 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
970 continue without passing the signal to the process.
972 - With GDB >= 7.10, the command 'target remote'
973 will automatically load the executable file of the process running
974 under Valgrind. This means you do not need to specify the executable
975 file yourself, GDB will discover it itself. See GDB documentation about
976 'qXfer:exec-file:read' packet for more info.
978 * ==================== FIXED BUGS ====================
980 The following bugs have been fixed or resolved. Note that "n-i-bz"
981 stands for "not in bugzilla" -- that is, a bug that was reported to us
982 but never got a bugzilla entry. We encourage you to file bugs in
983 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
984 than mailing the developers (or mailing lists) directly -- bugs that
985 are not entered into bugzilla tend to get forgotten about or ignored.
987 To see details of a given bug, visit
988 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
989 where XXXXXX is the bug number as listed below.
991 116002 VG_(printf): Problems with justification of strings and integers
992 155125 avoid cutting away file:lineno after long function name
993 197259 Unsupported arch_prtctl PR_SET_GS option
994 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
995 201216 Fix Valgrind does not support pthread_sigmask() on OS X
996 201435 Fix Darwin: -v does not show kernel version
997 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
998 211256 Fixed an outdated comment regarding the default platform.
999 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1000 211926 Avoid compilation warnings in valgrind.h with -pedantic
1001 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1003 226609 Crediting upstream authors in man page
1004 231257 Valgrind omits path when executing script from shebang line
1005 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1006 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1007 269360 s390x: Fix addressing mode selection for compare-and-swap
1008 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1010 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1011 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1012 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1013 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1014 330147 libmpiwrap PMPI_Get_count returns undefined value
1015 333051 mmap of huge pages fails due to incorrect alignment
1017 334802 valgrind does not always explain why a given option is bad
1018 335618 mov.w rN, pc/sp (ARM32)
1019 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1024 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1025 338602 AVX2 bit in CPUID missing
1026 338606 Strange message for scripts with invalid interpreter
1027 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1028 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1029 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1031 339156 gdbsrv not called for fatal signal
1032 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1033 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1034 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1035 339442 Fix testsuite build failure on OS X 10.9
1036 339542 Enable compilation with Intel's ICC compiler
1037 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1038 339688 Mac-specific ASM does not support .version directive (cpuid,
1039 tronical and pushfpopf tests)
1040 339745 Valgrind crash when check Marmalade app (partial fix)
1041 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1042 339778 Linux/TileGx platform support for Valgrind
1043 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1044 339789 Fix none/tests/execve test on Mac OS X 10.9
1045 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1046 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1047 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1048 340392 Allow user to select more accurate definedness checking in memcheck
1049 to avoid invalid complaints on optimised code
1050 340430 Fix some grammatical weirdness in the manual.
1051 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1052 341419 Signal handler ucontext_t not filled out correctly on OS X
1053 341539 VG_(describe_addr) should not describe address as belonging to client
1054 segment if it is past the heap end
1055 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1056 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1057 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1058 341789 aarch64: shmat fails with valgrind on ARMv8
1059 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1060 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1061 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1062 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1063 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1064 342221 socket connect false positive uninit memory for unknown af family
1065 342353 Allow dumping full massif output while valgrind is still running
1066 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1070 342603 Add I2C_SMBUS ioctl support
1071 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1072 342683 Mark memory past the initial brk limit as unaddressable
1073 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1074 342795 Internal glibc __GI_mempcpy call should be intercepted
1075 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1076 343012 Unhandled syscall 319 (memfd_create)
1077 343069 Patch updating v4l2 API support
1078 343173 helgrind crash during stack unwind
1079 343219 fix GET_STARTREGS for arm
1080 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1081 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1082 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1083 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1084 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1085 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1086 343597 ppc64le: incorrect use of offseof macro
1087 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1088 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1089 343732 Unhandled syscall 144 (setgid) on aarch64
1090 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1091 343802 s390x: False positive "conditional jump or move depends on [..]
1092 343902 --vgdb=yes doesn't break when --xml=yes is used
1093 343967 Don't warn about setuid/setgid/setcap executable for directories
1094 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1095 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1096 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1097 344054 www - update info for Solaris/illumos
1098 344416 'make regtest' does not work cleanly on OS X
1099 344235 Remove duplicate include of pub_core_aspacemgr.h
1100 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1101 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1102 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1103 344314 callgrind_annotate ... warnings about commands containing newlines
1104 344318 socketcall should wrap recvmmsg and sendmmsg
1105 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1106 344416 Fix 'make regtest' does not work cleanly on OS X
1107 344499 Fix compilation for Linux kernel >= 4.0.0
1108 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1109 unix:349 (__pthread_fchdir)
1110 344559 Garbage collection of unused segment names in address space manager
1111 344560 Fix stack traces missing penultimate frame on OS X
1112 344621 Fix memcheck/tests/err_disable4 test on OS X
1113 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1114 344702 Fix missing libobjc suppressions on OS X 10.10
1116 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1117 344939 Fix memcheck/tests/xml1 on OS X 10.10
1118 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1119 345079 Fix build problems in VEX/useful/test_main.c
1120 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1121 345177 arm64: prfm (reg) not implemented
1122 345215 Performance improvements for the register allocator
1123 345248 add support for Solaris OS in valgrind
1124 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1125 345394 Fix memcheck/tests/strchr on OS X
1126 345637 Fix memcheck/tests/sendmsg on OS X
1127 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1128 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1129 345887 Fix an assertion in the address space manager
1130 345928 amd64: callstack only contains current function for small stacks
1131 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1132 345987 MIPS64: Implement cavium LHX instruction
1133 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1134 346185 Fix typo saving altivec register v24
1135 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1136 and LibVEX_GuestPPC64_get_CR()
1137 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1138 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1139 346307 fuse filesystem syscall deadlocks
1140 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1141 346411 MIPS: SysRes::_valEx handling is incorrect
1142 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1143 346474 PPC64 Power 8, spr TEXASRU register not supported
1144 346487 Compiler generates "note" about a future ABI change for PPC64
1145 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1146 and causing spurious "invalid read of size 8" warnings
1147 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1148 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1149 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1150 347322 Power PC regression test cleanup
1151 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1153 347389 unhandled syscall: 373 (Linux ARM syncfs)
1154 347686 Patch set to cleanup PPC64 regtests
1155 347978 Remove bash dependencies where not needed
1156 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1157 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1159 348102 Patch updating v4l2 API support
1160 348247 amd64 front end: jno jumps wrongly when overflow is not set
1161 348269 Improve mmap MAP_HUGETLB support.
1162 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1163 348345 Assertion fails for negative lineno
1164 348377 Unsupported ARM instruction: yield
1165 348565 Fix detection of command line option availability for clang
1166 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1167 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1168 348748 Fix redundant condition
1169 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1170 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1171 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1172 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1173 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1174 349626 Implemented additional Xen hypercalls
1175 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1176 349790 Clean up of the hardware capability checking utilities.
1177 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1178 349874 Fix typos in source code
1179 349879 memcheck: add handwritten assembly for helperc_LOADV*
1180 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1181 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1182 350202 Add limited param to 'monitor block_list'
1183 350290 s390x: Support instructions fixbr(a)
1184 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1185 350809 Fix none/tests/async-sigs for Solaris
1186 350811 Remove reference to --db-attach which has been removed.
1187 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1188 350854 hard-to-understand code in VG_(load_ELF)()
1189 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1190 351386 Solaris: Cannot run ld.so.1 under Valgrind
1191 351474 Fix VG_(iseqsigset) as obvious
1192 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1193 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1194 351858 ldsoexec support on Solaris
1195 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1196 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1197 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1198 352320 arm64 crash on none/tests/nestedfs
1199 352765 Vbit test fails on Power 6
1200 352768 The mbar instruction is missing from the Power PC support
1201 352769 Power PC program priority register (PPR) is not supported
1202 n-i-bz Provide implementations of certain compiler builtins to support
1203 compilers that may not provide those
1204 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1205 n-i-bz Fix compilation on distros with glibc < 2.5
1206 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1207 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1208 n-i-bz mremap did not work properly on shared memory
1209 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1210 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1212 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1213 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1214 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1218 Release 3.10.1 (25 November 2014)
1219 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1220 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1221 and backports fixes for all reported missing AArch64 ARMv8 instructions
1222 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1223 to use, you might want to consider upgrading to 3.10.1 instead.
1225 The following bugs have been fixed or resolved. Note that "n-i-bz"
1226 stands for "not in bugzilla" -- that is, a bug that was reported to us
1227 but never got a bugzilla entry. We encourage you to file bugs in
1228 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1229 than mailing the developers (or mailing lists) directly -- bugs that
1230 are not entered into bugzilla tend to get forgotten about or ignored.
1232 To see details of a given bug, visit
1233 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1234 where XXXXXX is the bug number as listed below.
1236 335440 arm64: ld1 (single structure) is not implemented
1237 335713 arm64: unhanded instruction: prfm (immediate)
1238 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1239 339182 ppc64: AvSplat ought to load destination vector register with [..]
1240 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1241 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1242 339645 Use correct tag names in sys_getdents/64 wrappers
1243 339706 Fix false positive for ioctl(TIOCSIG) on linux
1244 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1245 339853 arm64 times syscall unknown
1246 339855 arm64 unhandled getsid/setsid syscalls
1247 339858 arm64 dmb sy not implemented
1248 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1249 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1250 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1252 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1253 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1254 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1255 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1256 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1257 340509 arm64: unhandled instruction fcvtas
1258 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1259 340632 arm64: unhandled instruction fcvtas
1260 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1261 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1262 340788 warning: unhandled syscall: 318 (getrandom)
1263 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1264 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1265 340922 arm64: unhandled getgroups/setgroups syscalls
1266 350251 Fix typo in VEX utility program (test_main.c).
1267 350407 arm64: unhandled instruction ucvtf (vector, integer)
1268 350809 none/tests/async-sigs breaks when run under cron on Solaris
1269 350811 update README.solaris after r15445
1270 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1271 350854 strange code in VG_(load_ELF)()
1272 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1273 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1274 n-i-bz Add missing ]] to terminate CDATA.
1275 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1276 n-i-bz Enable sys_fadvise64_64 on arm32.
1277 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1278 n-i-bz Add test cases for all known arm64 load/store instructions.
1279 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1280 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1281 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1282 n-i-bz arm64: Implement "fcvtpu w, s".
1283 n-i-bz arm64: implement ADDP and various others
1284 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1285 n-i-bz arm64: enable FCVT{A,N}S X,S.
1287 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1291 Release 3.10.0 (10 September 2014)
1292 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1294 3.10.0 is a feature release with many improvements and the usual
1295 collection of bug fixes.
1297 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1298 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1299 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1300 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1301 significantly improved relative to the 3.9.0 release.
1303 * ================== PLATFORM CHANGES =================
1305 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1306 is mostly complete, and is usable, but some SIMD instructions are as
1309 * Support for little-endian variant of the 64-bit POWER architecture.
1311 * Support for Android on MIPS32.
1313 * Support for 64bit FPU on MIPS32 platforms.
1315 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1317 * Configuration for and running on Android targets has changed.
1318 See README.android in the source tree for details.
1320 * ================== DEPRECATED FEATURES =================
1322 * --db-attach is now deprecated and will be removed in the next
1323 valgrind feature release. The built-in GDB server capabilities are
1324 superior and should be used instead. Learn more here:
1325 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1327 * ==================== TOOL CHANGES ====================
1331 - Client code can now selectively disable and re-enable reporting of
1332 invalid address errors in specific ranges using the new client
1333 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1334 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1336 - Leak checker: there is a new leak check heuristic called
1337 "length64". This is used to detect interior pointers pointing 8
1338 bytes inside a block, on the assumption that the first 8 bytes
1339 holds the value "block size - 8". This is used by
1340 sqlite3MemMalloc, for example.
1342 - Checking of system call parameters: if a syscall parameter
1343 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1344 several fields not initialised, an error is now reported for each
1345 field. Previously, an error was reported only for the first
1346 uninitialised field.
1348 - Mismatched alloc/free checking: a new flag
1349 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1350 such checks if necessary.
1354 - Improvements to error messages:
1356 o Race condition error message involving heap allocated blocks also
1357 show the thread number that allocated the raced-on block.
1359 o All locks referenced by an error message are now announced.
1360 Previously, some error messages only showed the lock addresses.
1362 o The message indicating where a lock was first observed now also
1363 describes the address/location of the lock.
1365 - Helgrind now understands the Ada task termination rules and
1366 creates a happens-before relationship between a terminated task
1367 and its master. This avoids some false positives and avoids a big
1368 memory leak when a lot of Ada tasks are created and terminated.
1369 The interceptions are only activated with forthcoming releases of
1370 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1372 - A new GDB server monitor command "info locks" giving the list of
1373 locks, their location, and their status.
1377 - callgrind_control now supports the --vgdb-prefix argument,
1378 which is needed if valgrind was started with this same argument.
1380 * ==================== OTHER CHANGES ====================
1382 * Unwinding through inlined function calls. Stack unwinding can now
1383 make use of Dwarf3 inlined-unwind information if it is available.
1384 The practical effect is that inlined calls become visible in stack
1385 traces. The suppression matching machinery has been adjusted
1386 accordingly. This is controlled by the new option
1387 --read-inline-info=yes|no. Currently this is enabled by default
1388 only on Linux and Android targets and only for the tools Memcheck,
1391 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1392 targets. If an object contains both CFI and EXIDX unwind
1393 information, Valgrind will prefer the CFI over the EXIDX. This
1394 facilitates unwinding through system libraries on arm-android
1397 * Address description logic has been improved and is now common
1398 between Memcheck and Helgrind, resulting in better address
1399 descriptions for some kinds of error messages.
1401 * Error messages about dubious arguments (eg, to malloc or calloc) are
1402 output like other errors. This means that they can be suppressed
1403 and they have a stack trace.
1405 * The C++ demangler has been updated for better C++11 support.
1407 * New and modified GDB server monitor features:
1409 - Thread local variables/storage (__thread) can now be displayed.
1411 - The GDB server monitor command "v.info location <address>"
1412 displays information about an address. The information produced
1413 depends on the tool and on the options given to valgrind.
1414 Possibly, the following are described: global variables, local
1415 (stack) variables, allocated or freed blocks, ...
1417 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1418 ask the GDB server to stop at the start of program execution, at
1419 the end of the program execution and on Valgrind internal errors.
1421 - A new monitor command "v.info stats" shows various Valgrind core
1422 and tool statistics.
1424 - A new monitor command "v.set hostvisibility" allows the GDB server
1425 to provide access to Valgrind internal host status/memory.
1427 * A new option "--aspace-minaddr=<address>" can in some situations
1428 allow the use of more memory by decreasing the address above which
1429 Valgrind maps memory. It can also be used to solve address
1430 conflicts with system libraries by increasing the default value.
1431 See user manual for details.
1433 * The amount of memory used by Valgrind to store debug info (unwind
1434 info, line number information and symbol data) has been
1435 significantly reduced, even though Valgrind now reads more
1436 information in order to support unwinding of inlined function calls.
1438 * Dwarf3 handling with --read-var-info=yes has been improved:
1440 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1442 - Code compiled with
1443 -ffunction-sections -fdata-sections -Wl,--gc-sections
1444 no longer causes assertion failures.
1446 * Improved checking for the --sim-hints= and --kernel-variant=
1447 options. Unknown strings are now detected and reported to the user
1450 * The semantics of stack start/end boundaries in the valgrind.h
1451 VALGRIND_STACK_REGISTER client request has been clarified and
1452 documented. The convention is that start and end are respectively
1453 the lowest and highest addressable bytes of the stack.
1455 * ==================== FIXED BUGS ====================
1457 The following bugs have been fixed or resolved. Note that "n-i-bz"
1458 stands for "not in bugzilla" -- that is, a bug that was reported to us
1459 but never got a bugzilla entry. We encourage you to file bugs in
1460 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1461 than mailing the developers (or mailing lists) directly -- bugs that
1462 are not entered into bugzilla tend to get forgotten about or ignored.
1464 To see details of a given bug, visit
1465 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1466 where XXXXXX is the bug number as listed below.
1468 175819 Support for ipv6 socket reporting with --track-fds
1469 232510 make distcheck fails
1470 249435 Analyzing wine programs with callgrind triggers a crash
1471 278972 support for inlined function calls in stacktraces and suppression
1473 291310 FXSAVE instruction marks memory as undefined on amd64
1474 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1475 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1476 315199 vgcore file for threaded app does not show which thread crashed
1477 315952 tun/tap ioctls are not supported
1478 323178 Unhandled instruction: PLDW register (ARM)
1479 323179 Unhandled instruction: PLDW immediate (ARM)
1480 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1481 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1482 325124 [MIPSEL] Compilation error
1483 325477 Phase 4 support for IBM Power ISA 2.07
1484 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1485 325628 Phase 5 support for IBM Power ISA 2.07
1486 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1487 325751 Missing the two privileged Power PC Transactional Memory Instructions
1488 325816 Phase 6 support for IBM Power ISA 2.07
1489 325856 Make SGCheck fail gracefully on unsupported platforms
1490 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1491 326436 DRD: False positive in libstdc++ std::list::push_back
1492 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1493 326462 Refactor vgdb to isolate invoker stuff into separate module
1494 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1495 326623 DRD: false positive conflict report in a field assignment
1496 326724 Valgrind does not compile on OSX 1.9 Mavericks
1497 326816 Intercept for __strncpy_sse2_unaligned missing?
1498 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1499 326983 Clear direction flag after tests on amd64.
1500 327212 Do not prepend the current directory to absolute path names.
1501 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1502 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1503 327284 s390x: Fix translation of the risbg instruction
1504 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1505 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1506 327916 DW_TAG_typedef may have no name
1507 327943 s390x: add a redirection for the 'index' function
1508 328100 XABORT not implemented
1509 328205 Implement additional Xen hypercalls
1510 328454 add support Backtraces with ARM unwind tables (EXIDX)
1511 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1512 328711 valgrind.1 manpage "memcheck options" section is badly generated
1513 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1514 329612 Incorrect handling of AT_BASE for image execution
1515 329694 clang warns about using uninitialized variable
1516 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1517 330228 mmap must align to VKI_SHMLBA on mips32
1518 330257 LLVM does not support `-mno-dynamic-no-pic` option
1519 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1520 330459 --track-fds=yes doesn't track eventfds
1521 330469 Add clock_adjtime syscall support
1522 330594 Missing sysalls on PowerPC / uClibc
1523 330622 Add test to regression suite for POWER instruction: dcbzl
1524 330939 Support for AMD's syscall instruction on x86
1526 330941 Typo in PRE(poll) syscall wrapper
1527 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
1528 331254 Fix expected output for memcheck/tests/dw4
1529 331255 Fix race condition in test none/tests/coolo_sigaction
1530 331257 Fix type of jump buffer in test none/tests/faultstatus
1531 331305 configure uses bash specific syntax
1532 331337 s390x WARNING: unhandled syscall: 326 (dup3)
1533 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
1534 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
1535 331829 Unexpected ioctl opcode sign extension
1536 331830 ppc64: WARNING: unhandled syscall: 96/97
1537 331839 drd/tests/sem_open specifies invalid semaphore name
1538 331847 outcome of drd/tests/thread_name is nondeterministic
1539 332037 Valgrind cannot handle Thumb "add pc, reg"
1540 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
1541 consistency checks enabled
1542 332263 intercepts for pthread_rwlock_timedrdlock and
1543 pthread_rwlock_timedwrlock are incorrect
1544 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
1546 332276 Implement additional Xen hypercalls
1547 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
1548 332765 Fix ms_print to create temporary files in a proper directory
1549 333072 drd: Add semaphore annotations
1550 333145 Tests for missaligned PC+#imm access for arm
1551 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
1552 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
1553 333248 WARNING: unhandled syscall: unix:443
1554 333428 ldr.w pc [rD, #imm] instruction leads to assertion
1555 333501 cachegrind: assertion: Cache set count is not a power of two.
1558 333666 Recognize MPX instructions and bnd prefix.
1559 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
1560 333817 Valgrind reports the memory areas written to by the SG_IO
1562 334049 lzcnt fails silently (x86_32)
1563 334384 Valgrind does not have support Little Endian support for
1565 334585 recvmmsg unhandled (+patch) (arm)
1566 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
1567 334727 Build fails with -Werror=format-security
1568 334788 clarify doc about --log-file initial program directory
1569 334834 PPC64 Little Endian support, patch 2
1570 334836 PPC64 Little Endian support, patch 3 testcase fixes
1571 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
1572 335034 Unhandled ioctl: HCIGETDEVLIST
1573 335155 vgdb, fix error print statement.
1574 335262 arm64: movi 8bit version is not supported
1575 335263 arm64: dmb instruction is not implemented
1576 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
1577 335496 arm64: sbc/abc instructions are not implemented
1578 335554 arm64: unhandled instruction: abs
1579 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
1580 335735 arm64: unhandled instruction: cnt
1581 335736 arm64: unhandled instruction: uaddlv
1582 335848 arm64: unhandled instruction: {s,u}cvtf
1583 335902 arm64: unhandled instruction: sli
1584 335903 arm64: unhandled instruction: umull (vector)
1585 336055 arm64: unhandled instruction: mov (element)
1586 336062 arm64: unhandled instruction: shrn{,2}
1587 336139 mip64: [...] valgrind hangs and spins on a single core [...]
1588 336189 arm64: unhandled Instruction: mvn
1589 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
1590 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
1591 336772 Make moans about unknown ioctls more informative
1592 336957 Add a section about the Solaris/illumos port on the webpage
1593 337094 ifunc wrapper is broken on ppc64
1594 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
1595 337528 leak check heuristic for block prefixed by length as 64bit number
1596 337740 Implement additional Xen hypercalls
1597 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
1598 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
1599 337871 deprecate --db-attach
1600 338023 Add support for all V4L2/media ioctls
1601 338024 inlined functions are not shown if DW_AT_ranges is used
1602 338106 Add support for 'kcmp' syscall
1603 338115 DRD: computed conflict set differs from actual after fork
1604 338160 implement display of thread local storage in gdbsrv
1605 338205 configure.ac and check for -Wno-tautological-compare
1606 338300 coredumps are missing one byte of every segment
1607 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
1608 338499 --sim-hints parsing broken due to wrong order in tokens
1609 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
1610 338681 Unable to unwind through clone thread created on i386-linux
1611 338698 race condition between gdbsrv and vgdb on startup
1612 338703 helgrind on arm-linux gets false positives in dynamic loader
1613 338791 alt dwz files can be relative of debug/main file
1614 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
1615 338932 build V-trunk with gcc-trunk
1616 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
1617 345079 Fix build problems in VEX/useful/test_main.c
1618 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
1619 n-i-bz s390x: Fix memory corruption for multithreaded applications
1620 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
1621 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
1622 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
1623 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
1624 n-i-bz Add command-line option --default-suppressions
1625 n-i-bz Add support for BLKDISCARDZEROES ioctl
1626 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
1627 n-i-bz Add support for sys_pivot_root and sys_unshare
1629 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
1630 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
1631 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
1635 Release 3.9.0 (31 October 2013)
1636 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1637 3.9.0 is a feature release with many improvements and the usual
1638 collection of bug fixes.
1640 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1641 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
1642 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
1643 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
1645 * ================== PLATFORM CHANGES =================
1647 * Support for MIPS64 LE and BE running Linux. Valgrind has been
1648 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
1650 * Support for MIPS DSP ASE on MIPS32 platforms.
1652 * Support for s390x Decimal Floating Point instructions on hosts that
1653 have the DFP facility installed.
1655 * Support for POWER8 (Power ISA 2.07) instructions
1657 * Support for Intel AVX2 instructions. This is available only on 64
1660 * Initial support for Intel Transactional Synchronization Extensions,
1663 * Initial support for Hardware Transactional Memory on POWER.
1665 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
1666 run large GUI apps tolerably well.
1668 * ==================== TOOL CHANGES ====================
1672 - Improvements in handling of vectorised code, leading to
1673 significantly fewer false error reports. You need to use the flag
1674 --partial-loads-ok=yes to get the benefits of these changes.
1676 - Better control over the leak checker. It is now possible to
1677 specify which leak kinds (definite/indirect/possible/reachable)
1678 should be displayed, which should be regarded as errors, and which
1679 should be suppressed by a given leak suppression. This is done
1680 using the options --show-leak-kinds=kind1,kind2,..,
1681 --errors-for-leak-kinds=kind1,kind2,.. and an optional
1682 "match-leak-kinds:" line in suppression entries, respectively.
1684 Note that generated leak suppressions contain this new line and
1685 are therefore more specific than in previous releases. To get the
1686 same behaviour as previous releases, remove the "match-leak-kinds:"
1687 line from generated suppressions before using them.
1689 - Reduced "possible leak" reports from the leak checker by the use
1690 of better heuristics. The available heuristics provide detection
1691 of valid interior pointers to std::stdstring, to new[] allocated
1692 arrays with elements having destructors and to interior pointers
1693 pointing to an inner part of a C++ object using multiple
1694 inheritance. They can be selected individually using the
1695 option --leak-check-heuristics=heur1,heur2,...
1697 - Better control of stacktrace acquisition for heap-allocated
1698 blocks. Using the --keep-stacktraces option, it is possible to
1699 control independently whether a stack trace is acquired for each
1700 allocation and deallocation. This can be used to create better
1701 "use after free" errors or to decrease Valgrind's resource
1702 consumption by recording less information.
1704 - Better reporting of leak suppression usage. The list of used
1705 suppressions (shown when the -v option is given) now shows, for
1706 each leak suppressions, how many blocks and bytes it suppressed
1707 during the last leak search.
1711 - False errors resulting from the use of statically initialised
1712 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
1715 - False errors resulting from the use of pthread_cond_waits that
1716 timeout, have been removed.
1718 * ==================== OTHER CHANGES ====================
1720 * Some attempt to tune Valgrind's space requirements to the expected
1721 capabilities of the target:
1723 - The default size of the translation cache has been reduced from 8
1724 sectors to 6 on Android platforms, since each sector occupies
1725 about 40MB when using Memcheck.
1727 - The default size of the translation cache has been increased to 16
1728 sectors on all other platforms, reflecting the fact that large
1729 applications require instrumentation and storage of huge amounts
1730 of code. For similar reasons, the number of memory mapped
1731 segments that can be tracked has been increased by a factor of 6.
1733 - In all cases, the maximum number of sectors in the translation
1734 cache can be controlled by the new flag --num-transtab-sectors.
1736 * Changes in how debug info (line numbers, etc) is read:
1738 - Valgrind no longer temporarily mmaps the entire object to read
1739 from it. Instead, reading is done through a small fixed sized
1740 buffer. This avoids virtual memory usage spikes when Valgrind
1741 reads debuginfo from large shared objects.
1743 - A new experimental remote debug info server. Valgrind can read
1744 debug info from a different machine (typically, a build host)
1745 where debuginfo objects are stored. This can save a lot of time
1746 and hassle when running Valgrind on resource-constrained targets
1747 (phones, tablets) when the full debuginfo objects are stored
1748 somewhere else. This is enabled by the --debuginfo-server=
1751 - Consistency checking between main and debug objects can be
1752 disabled using the --allow-mismatched-debuginfo option.
1754 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
1755 scanning can recover stack traces in some cases when the normal
1756 unwind mechanisms fail. Stack scanning is best described as "a
1757 nasty, dangerous and misleading hack" and so is disabled by default.
1758 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
1761 * Detection and merging of recursive stack frame cycles. When your
1762 program has recursive algorithms, this limits the memory used by
1763 Valgrind for recorded stack traces and avoids recording
1764 uninteresting repeated calls. This is controlled by the command
1765 line option --merge-recursive-frame and by the monitor command
1766 "v.set merge-recursive-frames".
1768 * File name and line numbers for used suppressions. The list of used
1769 suppressions (shown when the -v option is given) now shows, for each
1770 used suppression, the file name and line number where the suppression
1773 * New and modified GDB server monitor features:
1775 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
1776 that can be used to execute gdbserver monitor commands from the
1779 - A new monitor command, "v.info open_fds", that gives the list of
1780 open file descriptors and additional details.
1782 - An optional message in the "v.info n_errs_found" monitor command,
1783 for example "v.info n_errs_found test 1234 finished", allowing a
1784 comment string to be added to the process output, perhaps for the
1785 purpose of separating errors of different tests or test phases.
1787 - A new monitor command "v.info execontext" that shows information
1788 about the stack traces recorded by Valgrind.
1790 - A new monitor command "v.do expensive_sanity_check_general" to run
1791 some internal consistency checks.
1793 * New flag --sigill-diagnostics to control whether a diagnostic
1794 message is printed when the JIT encounters an instruction it can't
1795 translate. The actual behavior -- delivery of SIGILL to the
1796 application -- is unchanged.
1798 * The maximum amount of memory that Valgrind can use on 64 bit targets
1799 has been increased from 32GB to 64GB. This should make it possible
1800 to run applications on Memcheck that natively require up to about 35GB.
1802 * ==================== FIXED BUGS ====================
1804 The following bugs have been fixed or resolved. Note that "n-i-bz"
1805 stands for "not in bugzilla" -- that is, a bug that was reported to us
1806 but never got a bugzilla entry. We encourage you to file bugs in
1807 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1808 than mailing the developers (or mailing lists) directly -- bugs that
1809 are not entered into bugzilla tend to get forgotten about or ignored.
1811 To see details of a given bug, visit
1812 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1813 where XXXXXX is the bug number as listed below.
1815 123837 system call: 4th argument is optional, depending on cmd
1816 135425 memcheck should tell you where Freed blocks were Mallocd
1817 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
1818 207815 Adds some of the drm ioctls to syswrap-linux.c
1819 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
1820 252955 Impossible to compile with ccache
1821 253519 Memcheck reports auxv pointer accesses as invalid reads.
1822 263034 Crash when loading some PPC64 binaries
1823 269599 Increase deepest backtrace
1824 274695 s390x: Support "compare to/from logical" instructions (z196)
1825 275800 s390x: Autodetect cache info (part 2)
1826 280271 Valgrind reports possible memory leaks on still-reachable std::string
1827 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
1828 289578 Backtraces with ARM unwind tables (stack scan flags)
1829 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
1830 304832 ppc32: build failure
1831 305431 Use find_buildid shdr fallback for separate .debug files
1832 305728 Add support for AVX2 instructions
1833 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
1834 306035 s390x: Fix IR generation for LAAG and friends
1835 306054 s390x: Condition code computation for convert-to-int/logical
1836 306098 s390x: alternate opcode form for convert to/from fixed
1837 306587 Fix cache line detection from auxiliary vector for PPC.
1838 306783 Mips unhandled syscall : 4025 / 4079 / 4182
1839 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
1840 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
1841 307101 sys_capget second argument can be NULL
1842 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
1843 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
1844 307113 s390x: DFP support
1845 307141 valgrind does't work in mips-linux system
1846 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
1847 307285 x86_amd64 feature test for avx in test suite is wrong
1848 307290 memcheck overlap testcase needs memcpy version filter
1849 307463 Please add "&limit=0" to the "all open bugs" link
1850 307465 --show-possibly-lost=no should reduce the error count / exit code
1851 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
1852 307729 pkgconfig support broken valgrind.pc
1853 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
1854 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
1855 308089 Unhandled syscall on ppc64: prctl
1856 308135 PPC32 MPC8xx has 16 bytes cache size
1857 308321 testsuite memcheck filter interferes with gdb_filter
1859 308341 vgdb should report process exit (or fatal signal)
1860 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
1861 308495 Remove build dependency on installed Xen headers
1862 308573 Internal error on 64-bit instruction executed in 32-bit mode
1864 308627 pmovmskb validity bit propagation is imprecise
1865 308644 vgdb command for having the info for the track-fds option
1866 308711 give more info about aspacemgr and arenas in out_of_memory
1867 308717 ARM: implement fixed-point VCVT.F64.[SU]32
1868 308718 ARM implement SMLALBB family of instructions
1869 308886 Missing support for PTRACE_SET/GETREGSET
1870 308930 syscall name_to_handle_at (303 on amd64) not handled
1871 309229 V-bit tester does not report number of tests generated
1872 309323 print unrecognized instuction on MIPS
1873 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
1874 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
1875 309430 Self hosting ppc64 encounters a vassert error on operand type
1876 309600 valgrind is a bit confused about 0-sized sections
1877 309823 Generate errors for still reachable blocks
1878 309921 PCMPISTRI validity bit propagation is imprecise
1879 309922 none/tests/ppc64/test_dfp5 sometimes fails
1880 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
1881 310424 --read-var-info does not properly describe static variables
1882 310792 search additional path for debug symbols
1883 310931 s390x: Message-security assist (MSA) instruction extension [..]
1884 311100 PPC DFP implementation of the integer operands is inconsistent [..]
1885 311318 ARM: "128-bit constant is not implemented" error message
1886 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
1887 311690 V crashes because it redirects branches inside of a redirected function
1888 311880 x86_64: make regtest hangs at shell_valid1
1889 311922 WARNING: unhandled syscall: 170
1891 312171 ppc: insn selection for DFP
1892 312571 Rounding mode call wrong for the DFP Iops [..]
1893 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
1894 312913 Dangling pointers error should also report the alloc stack trace
1895 312980 Building on Mountain Lion generates some compiler warnings
1896 313267 Adding MIPS64/Linux port to Valgrind
1899 313811 Buffer overflow in assert_fail
1900 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
1901 314269 ppc: dead code in insn selection
1902 314718 ARM: implement integer divide instruction (sdiv and udiv)
1903 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
1904 315441 sendmsg syscall should ignore unset msghdr msg_flags
1905 315534 msgrcv inside a thread causes valgrind to hang (block)
1906 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
1907 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
1908 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
1909 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
1910 316144 valgrind.1 manpage contains unknown ??? strings [..]
1911 316145 callgrind command line options in manpage reference (unknown) [..]
1912 316145 callgrind command line options in manpage reference [..]
1913 316181 drd: Fixed a 4x slowdown for certain applications
1914 316503 Valgrind does not support SSE4 "movntdqa" instruction
1915 316535 Use of |signed int| instead of |size_t| in valgrind messages
1916 316696 fluidanimate program of parsec 2.1 stuck
1917 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
1918 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
1919 317186 "Impossible happens" when occurs VCVT instruction on ARM
1920 317318 Support for Threading Building Blocks "scalable_malloc"
1921 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
1922 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
1923 317463 bmi testcase IR SANITY CHECK FAILURE
1924 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
1925 318050 libmpiwrap fails to compile with out-of-source build
1926 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
1927 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
1928 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
1929 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
1930 318932 Add missing PPC64 and PPC32 system call support
1931 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
1932 319395 Crash with unhandled instruction on STRT (Thumb) instructions
1933 319494 VEX Makefile-gcc standalone build update after r2702
1934 319505 [MIPSEL] Crash: unhandled UNRAY operator.
1935 319858 disInstr(thumb): unhandled instruction on instruction STRBT
1936 319932 disInstr(thumb): unhandled instruction on instruction STRHT
1937 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
1938 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
1939 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
1940 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
1941 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
1942 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
1943 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
1944 320895 add fanotify support (patch included)
1945 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
1946 321065 Valgrind updates for Xen 4.3
1947 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
1948 321363 Unhandled instruction: SSAX (ARM + Thumb)
1949 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
1950 321466 Unhandled instruction: SHASX (ARM + Thumb)
1951 321467 Unhandled instruction: SHSAX (ARM + Thumb)
1952 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
1953 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
1954 321620 Unhandled instruction: UASX (ARM + Thumb)
1955 321621 Unhandled instruction: USAX (ARM + Thumb)
1956 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
1957 321693 Unhandled instruction: LDRSBT (Thumb)
1958 321694 Unhandled instruction: UQASX (ARM + Thumb)
1959 321696 Unhandled instruction: UQSAX (Thumb + ARM)
1960 321697 Unhandled instruction: UHASX (ARM + Thumb)
1961 321703 Unhandled instruction: UHSAX (ARM + Thumb)
1962 321704 Unhandled instruction: REVSH (ARM + Thumb)
1963 321730 Add cg_diff and cg_merge man pages
1964 321738 Add vgdb and valgrind-listener man pages
1966 321891 Unhandled instruction: LDRHT (Thumb)
1967 321960 pthread_create() then alloca() causing invalid stack write errors
1968 321969 ppc32 and ppc64 don't support [lf]setxattr
1969 322254 Show threadname together with tid if set by application
1970 322294 Add initial support for IBM Power ISA 2.07
1971 322368 Assertion failure in wqthread_hijack under OS X 10.8
1972 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
1973 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
1974 322851 0bXXX binary literal syntax is not standard
1975 323035 Unhandled instruction: LDRSHT(Thumb)
1976 323036 Unhandled instruction: SMMLS (ARM and Thumb)
1977 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
1978 323175 Unhandled instruction: SMLALD (ARM + Thumb)
1979 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
1980 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
1981 323437 Phase 2 support for IBM Power ISA 2.07
1982 323713 Support mmxext (integer sse) subset on i386 (athlon)
1983 323803 Transactional memory instructions are not supported for Power
1984 323893 SSE3 not available on amd cpus in valgrind
1985 323905 Probable false positive from Valgrind/drd on close()
1986 323912 valgrind.h header isn't compatible for mingw64
1987 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
1988 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
1989 324181 mmap does not handle MAP_32BIT
1990 324227 memcheck false positive leak when a thread calls exit+block [..]
1991 324421 Support for fanotify API on ARM architecture
1992 324514 gdbserver monitor cmd output behaviour consistency [..]
1993 324518 ppc64: Emulation of dcbt instructions does not handle [..]
1994 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
1995 324582 When access is made to freed memory, report both allocation [..]
1996 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
1997 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
1998 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
1999 324834 Unhandled instructions in Microsoft C run-time for x86_64
2000 324894 Phase 3 support for IBM Power ISA 2.07
2001 326091 drd: Avoid false race reports from optimized strlen() impls
2002 326113 valgrind libvex hwcaps error on AMD64
2003 n-i-bz Some wrong command line options could be ignored
2004 n-i-bz patch to allow fair-sched on android
2005 n-i-bz report error for vgdb snapshot requested before execution
2006 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2008 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2012 Release 3.8.1 (19 September 2012)
2013 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2014 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2015 that occur moderately frequently in real use cases, adds support for
2016 some missing instructions on ARM, and fixes a deadlock condition on
2017 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2018 want to consider upgrading to 3.8.1 instead.
2020 The following bugs have been fixed or resolved. Note that "n-i-bz"
2021 stands for "not in bugzilla" -- that is, a bug that was reported to us
2022 but never got a bugzilla entry. We encourage you to file bugs in
2023 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2024 than mailing the developers (or mailing lists) directly -- bugs that
2025 are not entered into bugzilla tend to get forgotten about or ignored.
2027 To see details of a given bug, visit
2028 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2029 where XXXXXX is the bug number as listed below.
2032 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2033 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2034 298281 wcslen causes false(?) uninitialised value warnings
2035 301281 valgrind hangs on OS X when the process calls system()
2036 304035 disInstr(arm): unhandled instruction 0xE1023053
2037 304867 implement MOVBE instruction in x86 mode
2038 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2039 305042 amd64: implement 0F 7F encoding of movq between two registers
2040 305199 ARM: implement QDADD and QDSUB
2041 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2042 305513 killed by fatal signal: SIGSEGV
2043 305690 DRD reporting invalid semaphore when sem_trywait fails
2044 305926 Invalid alignment checks for some AVX instructions
2045 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2046 306310 3.8.0 release tarball missing some files
2047 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2048 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2049 n-i-bz shmat of a segment > 4Gb does not work
2050 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2051 n-i-bz vgdb ptrace calls wrong on mips [...]
2052 n-i-bz Fixes for more MPI false positives
2053 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2054 n-i-bz OSX build w/ clang: asserts at startup
2055 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2056 n-i-bz fix a couple of union tag-vs-field mixups
2057 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2059 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2062 254088 Valgrind should know about UD2 instruction
2065 304754 NEWS blows TeX's little mind
2067 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2071 Release 3.8.0 (10 August 2012)
2072 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2073 3.8.0 is a feature release with many improvements and the usual
2074 collection of bug fixes.
2076 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2077 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2078 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2079 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2080 There is initial support for MacOSX 10.8, but it is not usable for
2081 serious work at present.
2083 * ================== PLATFORM CHANGES =================
2085 * Support for MIPS32 platforms running Linux. Valgrind has been
2086 tested on MIPS32 and MIPS32r2 platforms running different Debian
2087 Squeeze and MeeGo distributions. Both little-endian and big-endian
2088 cores are supported. The tools Memcheck, Massif and Lackey have
2089 been tested and are known to work. See README.mips for more details.
2091 * Preliminary support for Android running on x86.
2093 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2095 * Support for Intel AVX instructions and for AES instructions. This
2096 support is available only for 64 bit code.
2098 * Support for POWER Decimal Floating Point instructions.
2100 * ==================== TOOL CHANGES ====================
2102 * Non-libc malloc implementations are now supported. This is useful
2103 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2104 Using the new option --soname-synonyms, such tools can be informed
2105 that the malloc implementation is either linked statically into the
2106 executable, or is present in some other shared library different
2107 from libc.so. This makes it possible to process statically linked
2108 programs, and programs using other malloc libraries, for example
2109 TCMalloc or JEMalloc.
2111 * For tools that provide their own replacement for malloc et al, the
2112 option --redzone-size=<number> allows users to specify the size of
2113 the padding blocks (redzones) added before and after each client
2114 allocated block. Smaller redzones decrease the memory needed by
2115 Valgrind. Bigger redzones increase the chance to detect blocks
2116 overrun or underrun. Prior to this change, the redzone size was
2117 hardwired to 16 bytes in Memcheck.
2121 - The leak_check GDB server monitor command now can
2122 control the maximum nr of loss records to output.
2124 - Reduction of memory use for applications allocating
2125 many blocks and/or having many partially defined bytes.
2127 - Addition of GDB server monitor command 'block_list' that lists
2128 the addresses/sizes of the blocks of a leak search loss record.
2130 - Addition of GDB server monitor command 'who_points_at' that lists
2131 the locations pointing at a block.
2133 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2134 detect an invalid access of these redzones, by marking them
2135 noaccess. Similarly, if a redzone size is given for a memory
2136 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2137 This still allows to find some bugs if the user has forgotten to
2138 mark the pool superblock noaccess.
2140 - Performance of memory leak check has been improved, especially in
2141 cases where there are many leaked blocks and/or many suppression
2142 rules used to suppress leak reports.
2144 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2145 more precise analysis, which is important for LLVM/Clang
2146 generated code. This is at the cost of somewhat reduced
2147 performance. Note there is no change to analysis precision or
2148 costs on Linux targets.
2152 - Added even more facilities that can help finding the cause of a data
2153 race, namely the command-line option --ptrace-addr and the macro
2154 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2156 - Fixed a subtle bug that could cause false positive data race reports.
2158 * ==================== OTHER CHANGES ====================
2160 * The C++ demangler has been updated so as to work well with C++
2161 compiled by up to at least g++ 4.6.
2163 * Tool developers can make replacement/wrapping more flexible thanks
2164 to the new option --soname-synonyms. This was reported above, but
2165 in fact is very general and applies to all function
2166 replacement/wrapping, not just to malloc-family functions.
2168 * Round-robin scheduling of threads can be selected, using the new
2169 option --fair-sched= yes. Prior to this change, the pipe-based
2170 thread serialisation mechanism (which is still the default) could
2171 give very unfair scheduling. --fair-sched=yes improves
2172 responsiveness of interactive multithreaded applications, and
2173 improves repeatability of results from the thread checkers Helgrind
2176 * For tool developers: support to run Valgrind on Valgrind has been
2177 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2179 * gdbserver now shows the float shadow registers as integer
2180 rather than float values, as the shadow values are mostly
2181 used as bit patterns.
2183 * Increased limit for the --num-callers command line flag to 500.
2185 * Performance improvements for error matching when there are many
2186 suppression records in use.
2188 * Improved support for DWARF4 debugging information (bug 284184).
2190 * Initial support for DWZ compressed Dwarf debug info.
2192 * Improved control over the IR optimiser's handling of the tradeoff
2193 between performance and precision of exceptions. Specifically,
2194 --vex-iropt-precise-memory-exns has been removed and replaced by
2195 --vex-iropt-register-updates, with extended functionality. This
2196 allows the Valgrind gdbserver to always show up to date register
2199 * Modest performance gains through the use of translation chaining for
2202 * ==================== FIXED BUGS ====================
2204 The following bugs have been fixed or resolved. Note that "n-i-bz"
2205 stands for "not in bugzilla" -- that is, a bug that was reported to us
2206 but never got a bugzilla entry. We encourage you to file bugs in
2207 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2208 than mailing the developers (or mailing lists) directly -- bugs that
2209 are not entered into bugzilla tend to get forgotten about or ignored.
2211 To see details of a given bug, visit
2212 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2213 where XXXXXX is the bug number as listed below.
2215 197914 Building valgrind from svn now requires automake-1.10
2216 203877 increase to 16Mb maximum allowed alignment for memalign et al
2217 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2218 247386 make perf does not run all performance tests
2219 270006 Valgrind scheduler unfair
2220 270777 Adding MIPS/Linux port to Valgrind
2221 270796 s390x: Removed broken support for the TS insn
2222 271438 Fix configure for proper SSE4.2 detection
2223 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2224 273475 Add support for AVX instructions
2225 274078 improved configure logic for mpicc
2226 276993 fix mremap 'no thrash checks'
2227 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2228 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2229 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2230 283413 Fix wrong sanity check
2231 283671 Robustize alignment computation in LibVEX_Alloc
2232 283961 Adding support for some HCI IOCTLs
2233 284124 parse_type_DIE: confused by: DWARF 4
2234 284864 == 273475 (Add support for AVX instructions)
2235 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2236 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2237 285725 == 273475 (Add support for AVX instructions)
2238 286261 add wrapper for linux I2C_RDWR ioctl
2239 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2240 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2241 286384 configure fails "checking for a supported version of gcc"
2242 286497 == 273475 (Add support for AVX instructions)
2243 286596 == 273475 (Add support for AVX instructions)
2244 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2245 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2246 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2247 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2248 287307 == 273475 (Add support for AVX instructions)
2249 287858 VG_(strerror): unknown error
2250 288298 (MacOSX) unhandled syscall shm_unlink
2251 288995 == 273475 (Add support for AVX instructions)
2252 289470 Loading of large Mach-O thin binaries fails.
2253 289656 == 273475 (Add support for AVX instructions)
2254 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2255 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2256 289839 s390x: Provide support for unicode conversion instructions
2257 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2258 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2259 290655 Add support for AESKEYGENASSIST instruction
2260 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2261 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2262 291253 ES register not initialised in valgrind simulation
2263 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2264 291865 s390x: Support the "Compare Double and Swap" family of instructions
2265 292300 == 273475 (Add support for AVX instructions)
2266 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2267 292493 == 273475 (Add support for AVX instructions)
2268 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2269 292627 Missing support for some SCSI ioctls
2270 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2271 292841 == 273475 (Add support for AVX instructions)
2272 292993 implement the getcpu syscall on amd64-linux
2273 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2274 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2275 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2276 293754 PCMPxSTRx not implemented for 16-bit characters
2277 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2278 293808 CLFLUSH not supported by latest VEX for amd64
2279 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2280 294048 MPSADBW instruction not implemented
2281 294055 regtest none/tests/shell fails when locale is not set to C
2282 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2283 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2284 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2285 294260 disInstr_AMD64: disInstr miscalculated next %rip
2286 294523 --partial-loads-ok=yes causes false negatives
2287 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2288 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2289 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2290 295089 can not annotate source for both helgrind and drd
2291 295221 POWER Processor decimal floating point instruction support missing
2292 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2293 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2294 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2295 295617 ARM - Add some missing syscalls
2296 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2297 296229 Linux user input device ioctls missing wrappers
2298 296318 ELF Debug info improvements (more than one rx/rw mapping)
2299 296422 Add translation chaining support
2300 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2301 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2302 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2303 297078 gdbserver signal handling problems [..]
2304 297147 drd false positives on newly allocated memory
2305 297329 disallow decoding of IBM Power DFP insns on some machines
2306 297497 POWER Processor decimal floating point instruction support missing
2307 297701 Another alias for strncasecmp_l in libc-2.13.so
2308 297911 'invalid write' not reported when using APIs for custom mem allocators.
2309 297976 s390x: revisit EX implementation
2310 297991 Valgrind interferes with mmap()+ftell()
2311 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2312 297993 Fix compilation of valgrind with gcc -g3.
2313 298080 POWER Processor DFP support missing, part 3
2314 298227 == 273475 (Add support for AVX instructions)
2315 298335 == 273475 (Add support for AVX instructions)
2316 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2317 298394 s390x: Don't bail out on an unknown machine model. [..]
2318 298421 accept4() syscall (366) support is missing for ARM
2319 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2320 298732 valgrind installation problem in ubuntu with kernel version 3.x
2321 298862 POWER Processor DFP instruction support missing, part 4
2322 298864 DWARF reader mis-parses DW_FORM_ref_addr
2323 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2324 299053 Support DWARF4 DW_AT_high_pc constant form
2325 299104 == 273475 (Add support for AVX instructions)
2326 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2327 299629 dup3() syscall (358) support is missing for ARM
2328 299694 POWER Processor DFP instruction support missing, part 5
2329 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2330 299803 == 273475 (Add support for AVX instructions)
2331 299804 == 273475 (Add support for AVX instructions)
2332 299805 == 273475 (Add support for AVX instructions)
2333 300140 ARM - Missing (T1) SMMUL
2334 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2335 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2336 300414 FCOM and FCOMP unimplemented for amd64 guest
2337 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2338 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2339 301265 add x86 support to Android build
2340 301984 configure script doesn't detect certain versions of clang
2341 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2342 302287 Unhandled movbe instruction on Atom processors
2343 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2344 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2345 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2346 302656 == 273475 (Add support for AVX instructions)
2347 302709 valgrind for ARM needs extra tls support for android emulator [..]
2348 302827 add wrapper for CDROM_GET_CAPABILITY
2349 302901 Valgrind crashes with dwz optimized debuginfo
2350 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2351 303116 Add support for the POWER instruction popcntb
2352 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2353 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2354 303466 == 273475 (Add support for AVX instructions)
2355 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2356 303963 strstr() function produces wrong results under valgrind callgrind
2357 304054 CALL_FN_xx macros need to enforce stack alignment
2358 304561 tee system call not supported
2359 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2360 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2361 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2362 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2363 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2364 n-i-bz s390x: Shadow registers can now be examined using vgdb
2366 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2367 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2371 Release 3.7.0 (5 November 2011)
2372 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2373 3.7.0 is a feature release with many significant improvements and the
2374 usual collection of bug fixes.
2376 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2377 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2378 Support for recent distros and toolchain components (glibc 2.14, gcc
2379 4.6, MacOSX 10.7) has been added.
2381 * ================== PLATFORM CHANGES =================
2383 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2384 analyse 64-bit programs running on z/Architecture. Most user space
2385 instructions up to and including z10 are supported. Valgrind has
2386 been tested extensively on z9, z10, and z196 machines running SLES
2387 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2388 known to work well. Callgrind, Helgrind, and DRD work reasonably
2389 well on z9 and later models. See README.s390 for more details.
2391 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2392 64-bit processes are supported. Some complex threaded applications
2393 (Firefox) are observed to hang when run as 32 bit applications,
2394 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2395 will likely report some false errors. In general, expect some rough
2396 spots. This release also supports MacOSX 10.6, but drops support
2399 * Preliminary support for Android (on ARM). Valgrind can now run
2400 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2401 README.android for more details, plus instructions on how to get
2404 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2406 * General correctness and performance improvements for ARM/Linux, and,
2407 by extension, ARM/Android.
2409 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2410 instruction set support is under development but is not available in
2413 * Support for AIX5 has been removed.
2415 * ==================== TOOL CHANGES ====================
2417 * Memcheck: some incremental changes:
2419 - reduction of memory use in some circumstances
2421 - improved handling of freed memory, which in some circumstances
2422 can cause detection of use-after-free that would previously have
2425 - fix of a longstanding bug that could cause false negatives (missed
2426 errors) in programs doing vector saturated narrowing instructions.
2428 * Helgrind: performance improvements and major memory use reductions,
2429 particularly for large, long running applications which perform many
2430 synchronisation (lock, unlock, etc) events. Plus many smaller
2433 - display of locksets for both threads involved in a race
2435 - general improvements in formatting/clarity of error messages
2437 - addition of facilities and documentation regarding annotation
2438 of thread safe reference counted C++ classes
2440 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2441 on thread stacks (a performance hack)
2443 - new flag --free-is-write=no|yes [no], to enable detection of races
2444 where one thread accesses heap memory but another one frees it,
2445 without any coordinating synchronisation event
2447 * DRD: enabled XML output; added support for delayed thread deletion
2448 in order to detect races that occur close to the end of a thread
2449 (--join-list-vol); fixed a memory leak triggered by repeated client
2450 memory allocatation and deallocation; improved Darwin support.
2452 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2454 * exp-sgcheck: this tool has been reduced in scope so as to improve
2455 performance and remove checking that Memcheck does better.
2456 Specifically, the ability to check for overruns for stack and global
2457 arrays is unchanged, but the ability to check for overruns of heap
2458 blocks has been removed. The tool has accordingly been renamed to
2459 exp-sgcheck ("Stack and Global Array Checking").
2461 * ==================== OTHER CHANGES ====================
2463 * GDB server: Valgrind now has an embedded GDB server. That means it
2464 is possible to control a Valgrind run from GDB, doing all the usual
2465 things that GDB can do (single stepping, breakpoints, examining
2466 data, etc). Tool-specific functionality is also available. For
2467 example, it is possible to query the definedness state of variables
2468 or memory from within GDB when running Memcheck; arbitrarily large
2469 memory watchpoints are supported, etc. To use the GDB server, start
2470 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2473 * Improved support for unfriendly self-modifying code: a new option
2474 --smc-check=all-non-file is available. This adds the relevant
2475 consistency checks only to code that originates in non-file-backed
2476 mappings. In effect this confines the consistency checking only to
2477 code that is or might be JIT generated, and avoids checks on code
2478 that must have been compiled ahead of time. This significantly
2479 improves performance on applications that generate code at run time.
2481 * It is now possible to build a working Valgrind using Clang-2.9 on
2484 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2485 These enable and disable error reporting on a per-thread, and
2486 nestable, basis. This is useful for hiding errors in particularly
2487 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2488 now uses this facility.
2490 * Added the --mod-funcname option to cg_diff.
2492 * ==================== FIXED BUGS ====================
2494 The following bugs have been fixed or resolved. Note that "n-i-bz"
2495 stands for "not in bugzilla" -- that is, a bug that was reported to us
2496 but never got a bugzilla entry. We encourage you to file bugs in
2497 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2498 mailing the developers (or mailing lists) directly -- bugs that are
2499 not entered into bugzilla tend to get forgotten about or ignored.
2501 To see details of a given bug, visit
2502 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2503 where XXXXXX is the bug number as listed below.
2505 79311 malloc silly arg warning does not give stack trace
2506 210935 port valgrind.h (not valgrind) to win32 to support client requests
2507 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2508 243404 Port to zSeries
2509 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2510 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2511 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2512 253206 Some fixes for the faultstatus testcase
2513 255223 capget testcase fails when running as root
2514 256703 xlc_dbl_u32.c testcase broken
2515 256726 Helgrind tests have broken inline asm
2516 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2517 264800 testcase compile failure on zseries
2518 265762 make public VEX headers compilable by G++ 3.x
2519 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2520 266753 configure script does not give the user the option to not use QtCore
2521 266931 gen_insn_test.pl is broken
2522 266961 ld-linux.so.2 i?86-linux strlen issues
2523 266990 setns instruction causes false positive
2524 267020 Make directory for temporary files configurable at run-time.
2525 267342 == 267997 (segmentation fault on Mac OS 10.6)
2526 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
2527 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
2528 267488 regtest: darwin support for 64-bit build
2529 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
2530 267630 Add support for IBM Power ISA 2.06 -- stage 1
2531 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
2532 267819 Add client request for informing the core about reallocation
2533 267925 laog data structure quadratic for a single sequence of lock
2534 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
2535 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
2536 268513 missed optimizations in fold_Expr
2537 268619 s390x: fpr - gpr transfer facility
2538 268620 s390x: reconsider "long displacement" requirement
2539 268621 s390x: improve IR generation for XC
2540 268715 s390x: FLOGR is not universally available
2541 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
2542 268930 s390x: MHY is not universally available
2543 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
2544 269079 Support ptrace system call on ARM
2545 269144 missing "Bad option" error message
2546 269209 conditional load and store facility (z196)
2547 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
2548 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
2549 269736 s390x: minor code generation tweaks
2550 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
2551 269863 s390x: remove unused function parameters
2552 269864 s390x: tweak s390_emit_load_cc
2553 269884 == 250101 (overhead for huge blocks exhausts space too soon)
2554 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
2555 270115 s390x: rewrite some testcases
2556 270309 == 267997 (valgrind crash on startup)
2557 270320 add support for Linux FIOQSIZE ioctl() call
2558 270326 segfault while trying to sanitize the environment passed to execle
2559 270794 IBM POWER7 support patch causes regression in none/tests
2560 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
2561 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
2562 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
2563 270959 s390x: invalid use of R0 as base register
2564 271042 VSX configure check fails when it should not
2565 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
2566 271259 s390x: fix code confusion
2567 271337 == 267997 (Valgrind segfaults on MacOS X)
2568 271385 s390x: Implement Ist_MBE
2569 271501 s390x: misc cleanups
2570 271504 s390x: promote likely and unlikely
2571 271579 ppc: using wrong enum type
2572 271615 unhandled instruction "popcnt" (arch=amd10h)
2573 271730 Fix bug when checking ioctls: duplicate check
2574 271776 s390x: provide STFLE instruction support
2575 271779 s390x: provide clock instructions like STCK
2576 271799 Darwin: ioctls without an arg report a memory error
2577 271820 arm: fix type confusion
2578 271917 pthread_cond_timedwait failure leads to not-locked false positive
2579 272067 s390x: fix DISP20 macro
2580 272615 A typo in debug output in mc_leakcheck.c
2581 272661 callgrind_annotate chokes when run from paths containing regex chars
2582 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
2583 272955 Unhandled syscall error for pwrite64 on ppc64 arch
2584 272967 make documentation build-system more robust
2585 272986 Fix gcc-4.6 warnings with valgrind.h
2586 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
2587 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
2588 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
2589 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
2590 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
2591 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
2592 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
2593 273778 exp-ptrcheck: unhandled sysno == 259
2594 274089 exp-ptrcheck: unhandled sysno == 208
2595 274378 s390x: Various dispatcher tweaks
2596 274447 WARNING: unhandled syscall: 340
2597 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
2598 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
2599 274926 valgrind does not build against linux-3
2600 275148 configure FAIL with glibc-2.14
2601 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
2602 275168 Make Valgrind work for MacOSX 10.7 Lion
2603 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
2604 275278 valgrind does not build on Linux kernel 3.0.* due to silly
2605 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
2606 275308 Fix implementation for ppc64 fres instruc
2607 275339 s390x: fix testcase compile warnings
2608 275517 s390x: Provide support for CKSM instruction
2609 275710 s390x: get rid of redundant address mode calculation
2610 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
2611 275852 == 250101 (valgrind uses all swap space and is killed)
2612 276784 Add support for IBM Power ISA 2.06 -- stage 3
2613 276987 gdbsrv: fix tests following recent commits
2614 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
2615 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
2616 277471 Unhandled syscall: 340
2617 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
2618 277653 ARM: support Thumb2 PLD instruction
2619 277663 ARM: NEON float VMUL by scalar incorrect
2620 277689 ARM: tests for VSTn with register post-index are broken
2621 277694 ARM: BLX LR instruction broken in ARM mode
2622 277780 ARM: VMOV.F32 (immediate) instruction is broken
2623 278057 fuse filesystem syscall deadlocks
2624 278078 Unimplemented syscall 280 on ppc32
2625 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
2626 278454 VALGRIND_STACK_DEREGISTER has wrong output type
2627 278502 == 275284 (Valgrind confuses memcpy() and memmove())
2628 278892 gdbsrv: factorize gdb version handling, fix doc and typos
2629 279027 Support for MVCL and CLCL instruction
2630 279027 s390x: Provide support for CLCL and MVCL instructions
2631 279062 Remove a redundant check in the insn selector for ppc.
2632 279071 JDK creates PTEST with redundant REX.W prefix
2633 279212 gdbsrv: add monitor cmd v.info scheduler.
2634 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
2635 279698 memcheck discards valid-bits for packuswb
2636 279795 memcheck reports uninitialised values for mincore on amd64
2637 279994 Add support for IBM Power ISA 2.06 -- stage 3
2638 280083 mempolicy syscall check errors
2639 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
2640 280710 s390x: config files for nightly builds
2641 280757 /tmp dir still used by valgrind even if TMPDIR is specified
2642 280965 Valgrind breaks fcntl locks when program does mmap
2643 281138 WARNING: unhandled syscall: 340
2644 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
2645 281304 == 275168 (Darwin: dyld "cannot load inserted library")
2646 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
2647 281468 s390x: handle do_clone and gcc clones in call traces
2648 281488 ARM: VFP register corruption
2649 281828 == 275284 (false memmove warning: "Source and destination overlap")
2650 281883 s390x: Fix system call wrapper for "clone".
2651 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
2652 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
2653 282238 SLES10: make check fails
2654 282979 strcasestr needs replacement with recent(>=2.12) glibc
2655 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
2656 283243 Regression in ppc64 memcheck tests
2657 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
2658 283427 re-connect epoll_pwait syscall on ARM linux
2659 283600 gdbsrv: android: port vgdb.c
2660 283709 none/tests/faultstatus needs to account for page size
2661 284305 filter_gdb needs enhancement to work on ppc64
2662 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
2663 284472 Thumb2 ROR.W encoding T2 not implemented
2664 284621 XML-escape process command line in XML output
2665 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
2666 that have non-power-of-2 sizes (also AMDs)
2667 n-i-bz don't be spooked by libraries mashed by elfhack
2668 n-i-bz don't be spooked by libxul.so linked with gold
2669 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
2671 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
2672 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
2673 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
2677 Release 3.6.1 (16 February 2011)
2678 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2679 3.6.1 is a bug fix release. It adds support for some SSE4
2680 instructions that were omitted in 3.6.0 due to lack of time. Initial
2681 support for glibc-2.13 has been added. A number of bugs causing
2682 crashing or assertion failures have been fixed.
2684 The following bugs have been fixed or resolved. Note that "n-i-bz"
2685 stands for "not in bugzilla" -- that is, a bug that was reported to us
2686 but never got a bugzilla entry. We encourage you to file bugs in
2687 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2688 mailing the developers (or mailing lists) directly -- bugs that are
2689 not entered into bugzilla tend to get forgotten about or ignored.
2691 To see details of a given bug, visit
2692 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2693 where XXXXXX is the bug number as listed below.
2695 188572 Valgrind on Mac should suppress setenv() mem leak
2696 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
2697 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
2698 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2699 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
2700 254420 memory pool tracking broken
2701 254957 Test code failing to compile due to changes in memcheck.h
2702 255009 helgrind/drd: crash on chmod with invalid parameter
2703 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
2704 255355 helgrind/drd: crash on threaded programs doing fork
2706 255418 (SSE4.x) rint call compiled with ICC
2707 255822 --gen-suppressions can create invalid files: "too many callers [...]"
2708 255888 closing valgrindoutput tag outputted to log-stream on error
2709 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
2710 255966 Slowness when using mempool annotations
2711 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
2712 256600 super-optimized strcasecmp() false positive
2713 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
2714 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
2715 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
2716 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
2717 257276 Missing case in memcheck --track-origins=yes
2718 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
2719 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
2720 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
2721 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
2722 263099 callgrind_annotate counts Ir improperly [...]
2723 263877 undefined coprocessor instruction on ARMv7
2724 265964 configure FAIL with glibc-2.13
2725 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
2726 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
2727 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
2728 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
2729 n-i-bz DRD: disable --free-is-write due to implementation difficulties
2731 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
2735 Release 3.6.0 (21 October 2010)
2736 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2737 3.6.0 is a feature release with many significant improvements and the
2738 usual collection of bug fixes.
2740 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2741 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
2742 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
2744 -------------------------
2746 Here are some highlights. Details are shown further down:
2748 * Support for ARM/Linux.
2750 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
2752 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
2754 * Support for the SSE4.2 instruction set.
2756 * Enhancements to the Callgrind profiler, including the ability to
2757 handle CPUs with three levels of cache.
2759 * A new experimental heap profiler, DHAT.
2761 * A huge number of bug fixes and small enhancements.
2763 -------------------------
2765 Here are details of the above changes, together with descriptions of
2766 many other changes, and a list of fixed bugs.
2768 * ================== PLATFORM CHANGES =================
2770 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
2771 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
2772 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
2774 This requires a CPU capable of running the ARMv7-A instruction set
2775 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
2776 of the user space instruction set, including ARM and Thumb integer
2777 code, VFPv3, NEON and V6 media instructions. The Memcheck,
2778 Cachegrind and Massif tools work properly; other tools work to
2781 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
2782 with support for recent releases of the underlying toolchain
2783 components, notably gcc-4.5 and glibc-2.12.
2785 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
2786 support also works much better on OS X 10.5, and is as solid as
2789 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
2790 64-bit mode. In 32-bit mode, support is only available up to and
2791 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
2792 supported in 64-bit mode, and 32-bit mode does in fact support the
2793 bare minimum SSE4 instructions to needed to run programs on Mac OS X
2794 10.6 on 32-bit targets.
2796 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
2797 and including version 2.05 is supported.
2799 * ==================== TOOL CHANGES ====================
2801 * Cachegrind has a new processing script, cg_diff, which finds the
2802 difference between two profiles. It's very useful for evaluating
2803 the performance effects of a change in a program.
2805 Related to this change, the meaning of cg_annotate's (rarely-used)
2806 --threshold option has changed; this is unlikely to affect many
2807 people, if you do use it please see the user manual for details.
2809 * Callgrind now can do branch prediction simulation, similar to
2810 Cachegrind. In addition, it optionally can count the number of
2811 executed global bus events. Both can be used for a better
2812 approximation of a "Cycle Estimation" as derived event (you need to
2813 update the event formula in KCachegrind yourself).
2815 * Cachegrind and Callgrind now refer to the LL (last-level) cache
2816 rather than the L2 cache. This is to accommodate machines with
2817 three levels of caches -- if Cachegrind/Callgrind auto-detects the
2818 cache configuration of such a machine it will run the simulation as
2819 if the L2 cache isn't present. This means the results are less
2820 likely to match the true result for the machine, but
2821 Cachegrind/Callgrind's results are already only approximate, and
2822 should not be considered authoritative. The results are still
2823 useful for giving a general idea about a program's locality.
2825 * Massif has a new option, --pages-as-heap, which is disabled by
2826 default. When enabled, instead of tracking allocations at the level
2827 of heap blocks (as allocated with malloc/new/new[]), it instead
2828 tracks memory allocations at the level of memory pages (as mapped by
2829 mmap, brk, etc). Each mapped page is treated as its own block.
2830 Interpreting the page-level output is harder than the heap-level
2831 output, but this option is useful if you want to account for every
2832 byte of memory used by a program.
2834 * DRD has two new command-line options: --free-is-write and
2835 --trace-alloc. The former allows to detect reading from already freed
2836 memory, and the latter allows tracing of all memory allocations and
2839 * DRD has several new annotations. Custom barrier implementations can
2840 now be annotated, as well as benign races on static variables.
2842 * DRD's happens before / happens after annotations have been made more
2843 powerful, so that they can now also be used to annotate e.g. a smart
2844 pointer implementation.
2846 * Helgrind's annotation set has also been drastically improved, so as
2847 to provide to users a general set of annotations to describe locks,
2848 semaphores, barriers and condition variables. Annotations to
2849 describe thread-safe reference counted heap objects have also been
2852 * Memcheck has a new command-line option, --show-possibly-lost, which
2853 is enabled by default. When disabled, the leak detector will not
2854 show possibly-lost blocks.
2856 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
2857 has been added. DHAT keeps track of allocated heap blocks, and also
2858 inspects every memory reference to see which block (if any) is being
2859 accessed. This gives a lot of insight into block lifetimes,
2860 utilisation, turnover, liveness, and the location of hot and cold
2861 fields. You can use DHAT to do hot-field profiling.
2863 * ==================== OTHER CHANGES ====================
2865 * Improved support for unfriendly self-modifying code: the extra
2866 overhead incurred by --smc-check=all has been reduced by
2867 approximately a factor of 5 as compared with 3.5.0.
2869 * Ability to show directory names for source files in error messages.
2870 This is combined with a flexible mechanism for specifying which
2871 parts of the paths should be shown. This is enabled by the new flag
2874 * A new flag, --require-text-symbol, which will stop the run if a
2875 specified symbol is not found it a given shared object when it is
2876 loaded into the process. This makes advanced working with function
2877 intercepting and wrapping safer and more reliable.
2879 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
2880 and control of Valgrind is now available for the tools Memcheck and
2881 Helgrind. XML output from Valgrind is available for Memcheck,
2882 Helgrind and exp-Ptrcheck.
2884 * More reliable stack unwinding on amd64-linux, particularly in the
2885 presence of function wrappers, and with gcc-4.5 compiled code.
2887 * Modest scalability (performance improvements) for massive
2888 long-running applications, particularly for those with huge amounts
2891 * Support for analyzing programs running under Wine with has been
2892 improved. The header files <valgrind/valgrind.h>,
2893 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
2894 Windows-programs compiled with MinGW or one of the Microsoft Visual
2897 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
2898 The 32-bit simulator was not affected. This did not occur often,
2899 but when it did would usually crash the program under test.
2902 * A large number of bugs were fixed. These are shown below.
2904 * A number of bugs were investigated, and were candidates for fixing,
2905 but are not fixed in 3.6.0, due to lack of developer time. They may
2906 get fixed in later releases. They are:
2908 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
2909 212419 false positive "lock order violated" (A+B vs A)
2910 213685 Undefined value propagates past dependency breaking instruction
2911 216837 Incorrect instrumentation of NSOperationQueue on Darwin
2912 237920 valgrind segfault on fork failure
2913 242137 support for code compiled by LLVM-2.8
2914 242423 Another unknown Intel cache config value
2915 243232 Inconsistent Lock Orderings report with trylock
2916 243483 ppc: callgrind triggers VEX assertion failure
2917 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
2918 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
2920 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2921 249435 Analyzing wine programs with callgrind triggers a crash
2922 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
2923 250065 Handling large allocations
2924 250101 huge "free" memory usage due to m_mallocfree.c
2925 "superblocks fragmentation"
2926 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
2927 252091 Callgrind on ARM does not detect function returns correctly
2928 252600 [PATCH] Allow lhs to be a pointer for shl/shr
2929 254420 memory pool tracking broken
2930 n-i-bz support for adding symbols for JIT generated code
2933 The following bugs have been fixed or resolved. Note that "n-i-bz"
2934 stands for "not in bugzilla" -- that is, a bug that was reported to us
2935 but never got a bugzilla entry. We encourage you to file bugs in
2936 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2937 mailing the developers (or mailing lists) directly -- bugs that are
2938 not entered into bugzilla tend to get forgotten about or ignored.
2940 To see details of a given bug, visit
2941 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2942 where XXXXXX is the bug number as listed below.
2944 135264 dcbzl instruction missing
2946 153699 Valgrind should report unaligned reads with movdqa
2948 190429 Valgrind reports lost of errors in ld.so
2949 with x86_64 2.9.90 glibc
2950 197266 valgrind appears to choke on the xmms instruction
2952 197988 Crash when demangling very large symbol names
2953 202315 unhandled syscall: 332 (inotify_init1)
2954 203256 Add page-level profiling to Massif
2955 205093 dsymutil=yes needs quotes, locking (partial fix)
2956 205241 Snow Leopard 10.6 support (partial fix)
2957 206600 Leak checker fails to upgrade indirect blocks when their
2958 parent becomes reachable
2959 210935 port valgrind.h (not valgrind) to win32 so apps run under
2960 wine can make client requests
2961 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
2962 within Linux ip-stack checksum functions
2963 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
2965 213685 Undefined value propagates past dependency breaking instruction
2967 215914 Valgrind inserts bogus empty environment variable
2969 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
2970 222545 shmat fails under valgind on some arm targets
2971 222560 ARM NEON support
2974 232509 Docs build fails with formatting inside <title></title> elements
2976 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
2977 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
2978 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
2979 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
2980 237485 symlink (syscall 57) is not supported on Mac OS
2981 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
2983 238208 is_just_below_ESP doesn't take into account red-zone
2984 238345 valgrind passes wrong $0 when executing a shell script
2985 238679 mq_timedreceive syscall doesn't flag the reception buffer
2987 238696 fcntl command F_DUPFD_CLOEXEC not supported
2988 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
2989 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
2990 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
2992 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
2998 242606 unhandled syscall: setegid (in Ptrcheck)
2999 242814 Helgrind "Impossible has happened" during
3000 QApplication::initInstance();
3001 243064 Valgrind attempting to read debug information from iso
3002 243270 Make stack unwinding in Valgrind wrappers more reliable
3003 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3004 sysno = 277 (mq_open)
3005 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3006 244493 ARM VFP d16-d31 registers support
3007 244670 add support for audit_session_self syscall on Mac OS 10.6
3008 244921 The xml report of helgrind tool is not well format
3009 244923 In the xml report file, the <preamble> not escape the
3010 xml char, eg '<','&','>'
3011 245535 print full path names in plain text reports
3012 245925 x86-64 red zone handling problem
3013 246258 Valgrind not catching integer underruns + new [] s
3014 246311 reg/reg cmpxchg doesn't work on amd64
3015 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3016 246888 Improve Makefile.vex.am
3017 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3018 to [f]chmod_extended
3019 247526 IBM POWER6 (ISA 2.05) support is incomplete
3020 247561 Some leak testcases fails due to reachable addresses in
3022 247875 sizeofIRType to handle Ity_I128
3023 247894 [PATCH] unhandled syscall sys_readahead
3024 247980 Doesn't honor CFLAGS passed to configure
3025 248373 darwin10.supp is empty in the trunk
3026 248822 Linux FIBMAP ioctl has int parameter instead of long
3027 248893 [PATCH] make readdwarf.c big endianess safe to enable
3028 unwinding on big endian systems
3029 249224 Syscall 336 not supported (SYS_proc_info)
3031 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3032 249943 jni JVM init fails when using valgrind
3033 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3035 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3036 250799 frexp$fenv_access_off function generates SIGILL
3037 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3038 251251 support pclmulqdq insn
3039 251362 valgrind: ARM: attach to debugger either fails or provokes
3041 251674 Unhandled syscall 294
3044 254257 Add support for debugfiles found by build-id
3045 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3046 254646 Wrapped functions cause stack misalignment on OS X
3047 (and possibly Linux)
3048 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3050 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3054 Release 3.5.0 (19 August 2009)
3055 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3056 3.5.0 is a feature release with many significant improvements and the
3057 usual collection of bug fixes. The main improvement is that Valgrind
3058 now works on Mac OS X.
3060 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3061 and X86/Darwin. Support for recent distros and toolchain components
3062 (glibc 2.10, gcc 4.5) has been added.
3064 -------------------------
3066 Here is a short summary of the changes. Details are shown further
3069 * Support for Mac OS X (10.5.x).
3071 * Improvements and simplifications to Memcheck's leak checker.
3073 * Clarification and simplifications in various aspects of Valgrind's
3076 * XML output for Helgrind and Ptrcheck.
3078 * Performance and stability improvements for Helgrind and DRD.
3080 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3082 * A new experimental tool, BBV, useful for computer architecture
3085 * Improved Wine support, including ability to read Windows PDB
3088 -------------------------
3090 Here are details of the above changes, followed by descriptions of
3091 many other minor changes, and a list of fixed bugs.
3094 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3095 called "Darwin" because that is the name of the OS core, which is the
3096 level that Valgrind works at.)
3100 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3101 because it would require work and 10.4 is only becoming less common.
3103 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3104 fairly well. For 10.5.x, 32-bit programs are the default even on
3105 64-bit machines, so it handles most current programs.
3107 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3108 officially supported, but simple programs at least will probably work.
3109 However, start-up is slow.
3111 - PowerPC machines are not supported.
3113 Things that don't work:
3115 - The Ptrcheck tool.
3117 - Objective-C garbage collection.
3121 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3122 Valgrind will fail with a SIGTRAP at start-up. See
3123 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3128 - You will likely find --dsymutil=yes a useful option, as error
3129 messages may be imprecise without it.
3131 - Mac OS X support is new and therefore will be less robust than the
3132 Linux support. Please report any bugs you find.
3134 - Threaded programs may run more slowly than on Linux.
3136 Many thanks to Greg Parker for developing this port over several years.
3139 * Memcheck's leak checker has been improved.
3141 - The results for --leak-check=summary now match the summary results
3142 for --leak-check=full. Previously they could differ because
3143 --leak-check=summary counted "indirectly lost" blocks and
3144 "suppressed" blocks as "definitely lost".
3146 - Blocks that are only reachable via at least one interior-pointer,
3147 but are directly pointed to by a start-pointer, were previously
3148 marked as "still reachable". They are now correctly marked as
3151 - The default value for the --leak-resolution option has been
3152 changed from "low" to "high". In general, this means that more
3153 leak reports will be produced, but each leak report will describe
3154 fewer leaked blocks.
3156 - With --leak-check=full, "definitely lost" and "possibly lost"
3157 leaks are now considered as proper errors, ie. they are counted
3158 for the "ERROR SUMMARY" and affect the behaviour of
3159 --error-exitcode. These leaks are not counted as errors if
3160 --leak-check=summary is specified, however.
3162 - Documentation for the leak checker has been improved.
3165 * Various aspects of Valgrind's text output have changed.
3167 - Valgrind's start-up message has changed. It is shorter but also
3168 includes the command being run, which makes it easier to use
3169 --trace-children=yes. An example:
3171 - Valgrind's shut-down messages have also changed. This is most
3172 noticeable with Memcheck, where the leak summary now occurs before
3173 the error summary. This change was necessary to allow leaks to be
3174 counted as proper errors (see the description of the leak checker
3175 changes above for more details). This was also necessary to fix a
3176 longstanding bug in which uses of suppressions against leaks were
3177 not "counted", leading to difficulties in maintaining suppression
3178 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3180 - Behavior of -v has changed. In previous versions, -v printed out
3181 a mixture of marginally-user-useful information, and tool/core
3182 statistics. The statistics printing has now been moved to its own
3183 flag, --stats=yes. This means -v is less verbose and more likely
3184 to convey useful end-user information.
3186 - The format of some (non-XML) stack trace entries has changed a
3187 little. Previously there were six possible forms:
3189 0x80483BF: really (a.c:20)
3190 0x80483BF: really (in /foo/a.out)
3192 0x80483BF: (within /foo/a.out)
3193 0x80483BF: ??? (a.c:20)
3196 The third and fourth of these forms have been made more consistent
3197 with the others. The six possible forms are now:
3199 0x80483BF: really (a.c:20)
3200 0x80483BF: really (in /foo/a.out)
3201 0x80483BF: really (in ???)
3202 0x80483BF: ??? (in /foo/a.out)
3203 0x80483BF: ??? (a.c:20)
3206 Stack traces produced when --xml=yes is specified are different
3210 * Helgrind and Ptrcheck now support XML output, so they can be used
3211 from GUI tools. Also, the XML output mechanism has been
3214 - The XML format has been overhauled and generalised, so it is more
3215 suitable for error reporting tools in general. The Memcheck
3216 specific aspects of it have been removed. The new format, which
3217 is an evolution of the old format, is described in
3218 docs/internals/xml-output-protocol4.txt.
3220 - Memcheck has been updated to use the new format.
3222 - Helgrind and Ptrcheck are now able to emit output in this format.
3224 - The XML output mechanism has been overhauled. XML is now output
3225 to its own file descriptor, which means that:
3227 * Valgrind can output text and XML independently.
3229 * The longstanding problem of XML output being corrupted by
3230 unexpected un-tagged text messages is solved.
3232 As before, the destination for text output is specified using
3233 --log-file=, --log-fd= or --log-socket=.
3235 As before, XML output for a tool is enabled using --xml=yes.
3237 Because there's a new XML output channel, the XML output
3238 destination is now specified by --xml-file=, --xml-fd= or
3241 Initial feedback has shown this causes some confusion. To
3242 clarify, the two envisaged usage scenarios are:
3244 (1) Normal text output. In this case, do not specify --xml=yes
3245 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3247 (2) XML output. In this case, specify --xml=yes, and one of
3248 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3249 destination, one of --log-file=, --log-fd= or --log-socket=
3250 to select the destination for any remaining text messages,
3251 and, importantly, -q.
3253 -q makes Valgrind completely silent on the text channel,
3254 except in the case of critical failures, such as Valgrind
3255 itself segfaulting, or failing to read debugging information.
3256 Hence, in this scenario, it suffices to check whether or not
3257 any output appeared on the text channel. If yes, then it is
3258 likely to be a critical error which should be brought to the
3259 attention of the user. If no (the text channel produced no
3260 output) then it can be assumed that the run was successful.
3262 This allows GUIs to make the critical distinction they need to
3263 make (did the run fail or not?) without having to search or
3264 filter the text output channel in any way.
3266 It is also recommended to use --child-silent-after-fork=yes in
3270 * Improvements and changes in Helgrind:
3272 - XML output, as described above
3274 - Checks for consistent association between pthread condition
3275 variables and their associated mutexes are now performed.
3277 - pthread_spinlock functions are supported.
3279 - Modest performance improvements.
3281 - Initial (skeletal) support for describing the behaviour of
3282 non-POSIX synchronisation objects through ThreadSanitizer
3283 compatible ANNOTATE_* macros.
3285 - More controllable tradeoffs between performance and the level of
3286 detail of "previous" accesses in a race. There are now three
3289 * --history-level=full. This is the default, and was also the
3290 default in 3.4.x. It shows both stacks involved in a race, but
3291 requires a lot of memory and can be very slow in programs that
3292 do many inter-thread synchronisation events.
3294 * --history-level=none. This only shows the later stack involved
3295 in a race. This can be much faster than --history-level=full,
3296 but makes it much more difficult to find the other access
3297 involved in the race.
3299 The new intermediate setting is
3301 * --history-level=approx
3303 For the earlier (other) access, two stacks are presented. The
3304 earlier access is guaranteed to be somewhere in between the two
3305 program points denoted by those stacks. This is not as useful
3306 as showing the exact stack for the previous access (as per
3307 --history-level=full), but it is better than nothing, and it's
3308 almost as fast as --history-level=none.
3311 * New features and improvements in DRD:
3313 - The error messages printed by DRD are now easier to interpret.
3314 Instead of using two different numbers to identify each thread
3315 (Valgrind thread ID and DRD thread ID), DRD does now identify
3316 threads via a single number (the DRD thread ID). Furthermore
3317 "first observed at" information is now printed for all error
3318 messages related to synchronization objects.
3320 - Added support for named semaphores (sem_open() and sem_close()).
3322 - Race conditions between pthread_barrier_wait() and
3323 pthread_barrier_destroy() calls are now reported.
3325 - Added support for custom allocators through the macros
3326 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3327 in <valgrind/valgrind.h>). An alternative for these two macros is
3328 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3331 - Added support for annotating non-POSIX synchronization objects
3332 through several new ANNOTATE_*() macros.
3334 - OpenMP: added support for the OpenMP runtime (libgomp) included
3335 with gcc versions 4.3.0 and 4.4.0.
3339 - Added two new command-line options (--first-race-only and
3340 --segment-merging-interval).
3343 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3345 Valgrind will now preserve (memory-access) atomicity of LOCK-
3346 prefixed x86/amd64 instructions, and any others implying a global
3347 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3349 This means that Valgrinded processes will "play nicely" in
3350 situations where communication with other processes, or the kernel,
3351 is done through shared memory and coordinated with such atomic
3352 instructions. Prior to this change, such arrangements usually
3353 resulted in hangs, races or other synchronisation failures, because
3354 Valgrind did not honour atomicity of such instructions.
3357 * A new experimental tool, BBV, has been added. BBV generates basic
3358 block vectors for use with the SimPoint analysis tool, which allows
3359 a program's overall behaviour to be approximated by running only a
3360 fraction of it. This is useful for computer architecture
3361 researchers. You can run BBV by specifying --tool=exp-bbv (the
3362 "exp-" prefix is short for "experimental"). BBV was written by
3366 * Modestly improved support for running Windows applications under
3367 Wine. In particular, initial support for reading Windows .PDB debug
3368 information has been added.
3371 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3372 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3376 * The Valgrind client requests VALGRIND_PRINTF and
3377 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3378 the string was always printed immediately on its own line. Now, the
3379 string will be added to a buffer but not printed until a newline is
3380 encountered, or other Valgrind output is printed (note that for
3381 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3382 "other Valgrind output"). This allows you to use multiple
3383 VALGRIND_PRINTF calls to build up a single output line, and also to
3384 print multiple output lines with a single request (by embedding
3385 multiple newlines in the string).
3388 * The graphs drawn by Massif's ms_print program have changed slightly:
3390 - The half-height chars '.' and ',' are no longer drawn, because
3391 they are confusing. The --y option can be used if the default
3392 y-resolution is not high enough.
3394 - Horizontal lines are now drawn after the top of a snapshot if
3395 there is a gap until the next snapshot. This makes it clear that
3396 the memory usage has not dropped to zero between snapshots.
3399 * Something that happened in 3.4.0, but wasn't clearly announced: the
3400 option --read-var-info=yes can be used by some tools (Memcheck,
3401 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3402 variable type and location information. This makes those tools
3403 start up more slowly and increases memory consumption, but
3404 descriptions of data addresses in error messages become more
3408 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3409 disabled in 3.4.0 due to a lack of interest and maintenance,
3410 although the source code was still in the distribution. The source
3411 code has now been removed from the distribution. For anyone
3412 interested, the removal occurred in SVN revision r10247.
3415 * Some changes have been made to the build system.
3417 - VEX/ is now integrated properly into the build system. This means
3418 that dependency tracking within VEX/ now works properly, "make
3419 install" will work without requiring "make" before it, and
3420 parallel builds (ie. 'make -j') now work (previously a
3421 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3422 was effectively ignored).
3424 - The --with-vex configure option has been removed. It was of
3425 little use and removing it simplified the build system.
3427 - The location of some install files has changed. This should not
3428 affect most users. Those who might be affected:
3430 * For people who use Valgrind with MPI programs, the installed
3431 libmpiwrap.so library has moved from
3432 $(INSTALL)/<platform>/libmpiwrap.so to
3433 $(INSTALL)/libmpiwrap-<platform>.so.
3435 * For people who distribute standalone Valgrind tools, the
3436 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3437 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3439 These changes simplify the build system.
3441 - Previously, all the distributed suppression (*.supp) files were
3442 installed. Now, only default.supp is installed. This should not
3443 affect users as the other installed suppression files were not
3444 read; the fact that they were installed was a mistake.
3447 * KNOWN LIMITATIONS:
3449 - Memcheck is unusable with the Intel compiler suite version 11.1,
3450 when it generates code for SSE2-and-above capable targets. This
3451 is because of icc's use of highly optimised inlined strlen
3452 implementations. It causes Memcheck to report huge numbers of
3453 false errors even in simple programs. Helgrind and DRD may also
3456 Versions 11.0 and earlier may be OK, but this has not been
3460 The following bugs have been fixed or resolved. Note that "n-i-bz"
3461 stands for "not in bugzilla" -- that is, a bug that was reported to us
3462 but never got a bugzilla entry. We encourage you to file bugs in
3463 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3464 mailing the developers (or mailing lists) directly -- bugs that are
3465 not entered into bugzilla tend to get forgotten about or ignored.
3467 To see details of a given bug, visit
3468 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3469 where XXXXXX is the bug number as listed below.
3471 84303 How about a LockCheck tool?
3472 91633 dereference of null ptr in vgPlain_st_basetype
3473 97452 Valgrind doesn't report any pthreads problems
3474 100628 leak-check gets assertion failure when using
3475 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3476 108528 NPTL pthread cleanup handlers not called
3477 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3478 110128 mallinfo is not implemented...
3479 110770 VEX: Generated files not always updated when making valgrind
3480 111102 Memcheck: problems with large (memory footprint) applications
3481 115673 Vex's decoder should never assert
3482 117564 False positive: Syscall param clone(child_tidptr) contains
3483 uninitialised byte(s)
3484 119404 executing ssh from inside valgrind fails
3485 133679 Callgrind does not write path names to sources with dwarf debug
3487 135847 configure.in problem with non gnu compilers (and possible fix)
3488 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3489 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3490 136230 memcheck reports "possibly lost", should be "still reachable"
3491 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3492 137904 Valgrind reports a memory leak when using POSIX threads,
3494 139076 valgrind VT_GETSTATE error
3495 142228 complaint of elf_dynamic_do_rela in trivial usage
3496 145347 spurious warning with USBDEVFS_REAPURB
3497 148441 (wine) can't find memory leak in Wine, win32 binary
3499 148742 Leak-check fails assert on exit
3500 149878 add (proper) check for calloc integer overflow
3501 150606 Call graph is broken when using callgrind control
3502 152393 leak errors produce an exit code of 0. I need some way to
3503 cause leak errors to result in a nonzero exit code.
3504 157154 documentation (leak-resolution doc speaks about num-callers
3505 def=4) + what is a loss record
3506 159501 incorrect handling of ALSA ioctls
3507 162020 Valgrinding an empty/zero-byte file crashes valgrind
3508 162482 ppc: Valgrind crashes while reading stabs information
3509 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3510 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3511 163560 VEX/test_main.c is missing from valgrind-3.3.1
3512 164353 malloc_usable_size() doesn't return a usable size
3513 165468 Inconsistent formatting in memcheck manual -- please fix
3514 169505 main.c:286 (endOfInstr):
3515 Assertion 'ii->cost_offset == *cost_offset' failed
3516 177206 Generate default.supp during compile instead of configure
3517 177209 Configure valt_load_address based on arch+os
3518 177305 eventfd / syscall 323 patch lost
3519 179731 Tests fail to build because of inlining of non-local asm labels
3520 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3521 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3522 181594 Bogus warning for empty text segment
3523 181707 dwarf doesn't require enumerations to have name
3524 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3525 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3526 Assertion '!already_present' failed.
3527 185359 exp-ptrcheck: unhandled syscall getresuid()
3528 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
3529 185816 Valgrind is unable to handle debug info for files with split
3530 debug info that are prelinked afterwards
3531 185980 [darwin] unhandled syscall: sem_open
3532 186238 bbToIR_AMD64: disInstr miscalculated next %rip
3533 186507 exp-ptrcheck unhandled syscalls prctl, etc.
3534 186790 Suppression pattern used for leaks are not reported
3535 186796 Symbols with length>200 in suppression files are ignored
3536 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
3537 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
3538 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
3539 188046 bashisms in the configure script
3540 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
3541 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
3542 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
3543 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
3544 assertion fail "!lock->heldBy"
3545 188427 Add support for epoll_create1 (with patch)
3546 188530 Support for SIOCGSTAMPNS
3547 188560 Include valgrind.spec in the tarball
3548 188572 Valgrind on Mac should suppress setenv() mem leak
3549 189054 Valgrind fails to build because of duplicate non-local asm labels
3550 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
3551 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
3552 189763 drd assertion failure: s_threadinfo[tid].is_recording
3553 190219 unhandled syscall: 328 (x86-linux)
3554 190391 dup of 181394; see above
3555 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
3556 190820 No debug information on powerpc-linux
3557 191095 PATCH: Improve usbdevfs ioctl handling
3558 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
3560 191189 --xml=yes should obey --gen-suppressions=all
3561 191192 syslog() needs a suppression on macosx
3562 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
3563 191761 getrlimit on MacOSX
3564 191992 multiple --fn-skip only works sometimes; dependent on order
3565 192634 V. reports "aspacem sync_check_mapping_callback:
3566 segment mismatch" on Darwin
3567 192954 __extension__ missing on 2 client requests
3568 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
3569 194474 "INSTALL" file has different build instructions than "README"
3570 194671 Unhandled syscall (sem_wait?) from mac valgrind
3571 195069 memcheck: reports leak (memory still reachable) for
3573 195169 drd: (vgDrd_barrier_post_wait):
3574 Assertion 'r->sg[p->post_iteration]' failed.
3575 195268 valgrind --log-file doesn't accept ~/...
3576 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
3577 195860 WARNING: unhandled syscall: unix:223
3578 196528 need a error suppression for pthread_rwlock_init under os x?
3579 197227 Support aio_* syscalls on Darwin
3580 197456 valgrind should reject --suppressions=(directory)
3581 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
3582 197591 unhandled syscall 27 (mincore)
3583 197793 Merge DCAS branch to the trunk == 85756, 142103
3584 197794 Avoid duplicate filenames in Vex
3585 197898 make check fails on current SVN
3586 197901 make check fails also under exp-ptrcheck in current SVN
3587 197929 Make --leak-resolution=high the default
3588 197930 Reduce spacing between leak reports
3589 197933 Print command line of client at start-up, and shorten preamble
3590 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
3591 198395 add BBV to the distribution as an experimental tool
3592 198624 Missing syscalls on Darwin: 82, 167, 281, 347
3593 198649 callgrind_annotate doesn't cumulate counters
3594 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
3595 199977 Valgrind complains about an unrecognized instruction in the
3596 atomic_incs test program
3597 200029 valgrind isn't able to read Fedora 12 debuginfo
3598 200760 darwin unhandled syscall: unix:284
3599 200827 DRD doesn't work on Mac OS X
3600 200990 VG_(read_millisecond_timer)() does not work correctly
3601 201016 Valgrind does not support pthread_kill() on Mac OS
3602 201169 Document --read-var-info
3603 201323 Pre-3.5.0 performance sanity checking
3604 201384 Review user manual for the 3.5.0 release
3605 201585 mfpvr not implemented on ppc
3606 201708 tests failing because x86 direction flag is left set
3607 201757 Valgrind doesn't handle any recent sys_futex additions
3608 204377 64-bit valgrind can not start a shell script
3609 (with #!/path/to/shell) if the shell is a 32-bit executable
3610 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
3611 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
3612 about memory allocation events with memory access tracing enabled
3613 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
3615 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
3619 Release 3.4.1 (28 February 2009)
3620 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3621 3.4.1 is a bug-fix release that fixes some regressions and assertion
3622 failures in debug info reading in 3.4.0, most notably incorrect stack
3623 traces on amd64-linux on older (glibc-2.3 based) systems. Various
3624 other debug info problems are also fixed. A number of bugs in the
3625 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
3627 In view of the fact that 3.4.0 contains user-visible regressions
3628 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
3629 encouraged to ship 3.4.1 in preference to 3.4.0.
3631 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
3632 bugzilla" -- that is, a bug that was reported to us but never got a
3633 bugzilla entry. We encourage you to file bugs in bugzilla
3634 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
3635 developers (or mailing lists) directly -- bugs that are not entered
3636 into bugzilla tend to get forgotten about or ignored.
3638 n-i-bz Fix various bugs reading icc-11 generated debug info
3639 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
3640 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
3641 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
3642 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
3643 179618 exp-ptrcheck crashed / exit prematurely
3644 179624 helgrind: false positive races with pthread_create and
3645 recv/open/close/read
3646 134207 pkg-config output contains @VG_PLATFORM@
3647 176926 floating point exception at valgrind startup with PPC 440EPX
3648 181594 Bogus warning for empty text segment
3649 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
3650 181707 Dwarf3 doesn't require enumerations to have name
3651 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3652 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3653 Assertion '!already_present' failed.
3654 185359 exp-ptrcheck unhandled syscall getresuid()
3656 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
3657 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
3661 Release 3.4.0 (2 January 2009)
3662 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3663 3.4.0 is a feature release with many significant improvements and the
3664 usual collection of bug fixes. This release supports X86/Linux,
3665 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
3666 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
3668 3.4.0 brings some significant tool improvements. Memcheck can now
3669 report the origin of uninitialised values, the thread checkers
3670 Helgrind and DRD are much improved, and we have a new experimental
3671 tool, exp-Ptrcheck, which is able to detect overruns of stack and
3672 global arrays. In detail:
3674 * Memcheck is now able to track the origin of uninitialised values.
3675 When it reports an uninitialised value error, it will try to show
3676 the origin of the value, as either a heap or stack allocation.
3677 Origin tracking is expensive and so is not enabled by default. To
3678 use it, specify --track-origins=yes. Memcheck's speed will be
3679 essentially halved, and memory usage will be significantly
3680 increased. Nevertheless it can drastically reduce the effort
3681 required to identify the root cause of uninitialised value errors,
3682 and so is often a programmer productivity win, despite running more
3685 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
3686 3.4.0, will be released shortly.
3688 * Helgrind's race detection algorithm has been completely redesigned
3689 and reimplemented, to address usability and scalability concerns:
3691 - The new algorithm has a lower false-error rate: it is much less
3692 likely to report races that do not really exist.
3694 - Helgrind will display full call stacks for both accesses involved
3695 in a race. This makes it easier to identify the root causes of
3698 - Limitations on the size of program that can run have been removed.
3700 - Performance has been modestly improved, although that is very
3703 - Direct support for Qt4 threading has been added.
3705 - pthread_barriers are now directly supported.
3707 - Helgrind works well on all supported Linux targets.
3709 * The DRD thread debugging tool has seen major improvements:
3711 - Greatly improved performance and significantly reduced memory
3714 - Support for several major threading libraries (Boost.Thread, Qt4,
3715 glib, OpenMP) has been added.
3717 - Support for atomic instructions, POSIX semaphores, barriers and
3718 reader-writer locks has been added.
3720 - Works now on PowerPC CPUs too.
3722 - Added support for printing thread stack usage at thread exit time.
3724 - Added support for debugging lock contention.
3726 - Added a manual for Drd.
3728 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
3729 checks for misuses of pointers. In that sense it is a bit like
3730 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
3731 detect overruns of stack and global arrays, it can detect
3732 arbitrarily far out-of-bounds accesses to heap blocks, and it can
3733 detect accesses heap blocks that have been freed a very long time
3734 ago (millions of blocks in the past).
3736 Ptrcheck currently works only on x86-linux and amd64-linux. To use
3737 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
3738 of the main Valgrind documentation. As this is an experimental
3739 tool, we would be particularly interested in hearing about your
3740 experiences with it.
3742 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
3743 longer built by default, although the code remains in the repository
3744 and the tarball. This is due to three factors: a perceived lack of
3745 users, a lack of maintenance, and concerns that it may not be
3746 possible to achieve reliable operation using the existing design.
3748 * As usual, support for the latest Linux distros and toolchain
3749 components has been added. It should work well on Fedora Core 10,
3750 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
3751 state) is supported, as is glibc-2.9. The C++ demangler has been
3752 updated so as to work well with C++ compiled by even the most recent
3755 * You can now use frame-level wildcards in suppressions. This was a
3756 frequently-requested enhancement. A line "..." in a suppression now
3757 matches zero or more frames. This makes it easier to write
3758 suppressions which are precise yet insensitive to changes in
3761 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
3763 * Very basic support for IBM Power6 has been added (64-bit processes only).
3765 * Valgrind is now cross-compilable. For example, it is possible to
3766 cross compile Valgrind on an x86/amd64-linux host, so that it runs
3767 on a ppc32/64-linux target.
3769 * You can set the main thread's stack size at startup using the
3770 new --main-stacksize= flag (subject of course to ulimit settings).
3771 This is useful for running apps that need a lot of stack space.
3773 * The limitation that you can't use --trace-children=yes together
3774 with --db-attach=yes has been removed.
3776 * The following bugs have been fixed. Note that "n-i-bz" stands for
3777 "not in bugzilla" -- that is, a bug that was reported to us but
3778 never got a bugzilla entry. We encourage you to file bugs in
3779 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3780 mailing the developers (or mailing lists) directly.
3782 n-i-bz Make return types for some client requests 64-bit clean
3783 n-i-bz glibc 2.9 support
3784 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
3785 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
3786 n-i-bz Building in an env without gdb gives bogus gdb attach
3787 92456 Tracing the origin of uninitialised memory
3788 106497 Valgrind does not demangle some C++ template symbols
3790 151612 Suppression with "..." (frame-level wildcards in .supp files)
3791 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
3792 159285 unhandled syscall:25 (stime, on x86-linux)
3793 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
3794 160954 ppc build of valgrind crashes with illegal instruction (isel)
3795 160956 mallinfo implementation, w/ patch
3796 162092 Valgrind fails to start gnome-system-monitor
3797 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
3798 163794 assertion failure with "--track-origins=yes"
3799 163933 sigcontext.err and .trapno must be set together
3800 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
3801 164476 Missing kernel module loading system calls
3802 164669 SVN regression: mmap() drops posix file locks
3803 166581 Callgrind output corruption when program forks
3804 167288 Patch file for missing system calls on Cell BE
3805 168943 unsupported scas instruction pentium
3806 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
3807 172417 x86->IR: 0x82 ...
3808 172563 amd64->IR: 0xD9 0xF5 - fprem1
3809 173099 .lds linker script generation error
3810 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
3811 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
3813 174908 --log-file value not expanded correctly for core file
3814 175044 Add lookup_dcookie for amd64
3815 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
3817 Developer-visible changes:
3819 * Valgrind's debug-info reading machinery has been majorly overhauled.
3820 It can now correctly establish the addresses for ELF data symbols,
3821 which is something that has never worked properly before now.
3823 Also, Valgrind can now read DWARF3 type and location information for
3824 stack and global variables. This makes it possible to use the
3825 framework to build tools that rely on knowing the type and locations
3826 of stack and global variables, for example exp-Ptrcheck.
3828 Reading of such information is disabled by default, because most
3829 tools don't need it, and because it is expensive in space and time.
3830 However, you can force Valgrind to read it, using the
3831 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
3832 make use of such information, if present, to provide source-level
3833 descriptions of data addresses in the error messages they create.
3835 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
3836 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).