2 Release 3.16.0 (?? ?????? 2019)
3 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
5 3.16.0 is a feature release with many improvements and the usual collection of
8 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
9 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
10 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
11 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
12 support for X86/macOS 10.13 and AMD64/macOS 10.13.
14 * ==================== CORE CHANGES ===================
16 * It is now possible to dynamically change the value of many command
17 line options while your program (or its children) are running under
19 To have the list of dynamically changeable options, run
20 "valgrind --help-dyn-options".
21 You can change the options from the shell by using vgdb to launch
22 the monitor command "v.clo <clo option>...".
23 The same monitor command can be used from a gdb connected
24 to the valgrind gdbserver.
25 Your program can also change the dynamically changeable options using
26 the client request VALGRIND_CLO_CHANGE(option).
28 * ==================== TOOL CHANGES ====================
35 - The command option --collect-systime has been enhanced to specify
36 the unit used to record the elapsed time spent during system calls.
37 The command option now accepts the values no|yes|msec|usec|nsec,
38 where yes is a synonym of msec. When giving the value nsec, the
39 system cpu time of system calls is also recorded.
45 - Several memcheck options are now dynamically changeable.
46 Use valgrind --help-dyn-options to list them.
48 * ==================== OTHER CHANGES ====================
50 * New and modified GDB server monitor features:
52 - The gdbserver monitor commands that require an address and an optional
53 length argument now accepts the alternate 'C like' syntax "address[length]".
54 For example, the memcheck command "monitor who_points_at 0x12345678 120"
55 can now also be given as "monitor who_points_at 0x12345678[120]".
57 * ==================== FIXED BUGS ====================
59 The following bugs have been fixed or resolved. Note that "n-i-bz"
60 stands for "not in bugzilla" -- that is, a bug that was reported to us
61 but never got a bugzilla entry. We encourage you to file bugs in
62 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
63 than mailing the developers (or mailing lists) directly -- bugs that
64 are not entered into bugzilla tend to get forgotten about or ignored.
66 To see details of a given bug, visit
67 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
68 where XXXXXX is the bug number as listed below.
70 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
71 406561 mcinfcallWSRU gdbserver_test fails on ppc64
72 406824 Unsupported baseline
73 407218 Add support for the copy_file_range syscall
74 407307 Intercept stpcpy also in ld.so for arm64
75 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
76 408009 Expose rdrand and f16c even on avx if host cpu supports them
77 408091 Missing pkey syscalls
78 408414 Add support for missing for preadv2 and pwritev2 syscalls
79 404406 s390x: z14 miscellaneous instructions not implemented
80 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
81 409141 Valgrind hangs when SIGKILLed
82 409367 exit_group() after signal to thread waiting in futex() causes hangs
83 410599 Non-deterministic behaviour of pth_self_kill_15_other test
84 411134 Allow the user to change a set of command line options during execution
85 412344 Problem setting mips flags with specific paths
86 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
87 413603 callgrind_annotate/cg_annotate truncate function names at '#'
89 n-i-bz Fix minor one time leaks in dhat.
90 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
91 n-i-bz Add support for the Linux io_uring system calls
93 Release 3.15.0 (12 April 2019)
94 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
96 3.15.0 is a feature release with many improvements and the usual collection of
99 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
100 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
101 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
102 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
103 support for X86/macOS 10.13 and AMD64/macOS 10.13.
105 * ==================== CORE CHANGES ===================
107 * The XTree Massif output format now makes use of the information obtained
108 when specifying --read-inline-info=yes.
110 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
112 * ==================== TOOL CHANGES ====================
116 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
117 it has been promoted from an experimental tool to a regular tool. Run it
118 with --tool=dhat instead of --tool=exp-dhat.
120 - DHAT now prints only minimal data when the program ends, instead writing
121 the bulk of the profiling data to a file. As a result, the --show-top-n
122 and --sort-by options have been removed.
124 - Profile results can be viewed with the new viewer, dh_view.html. When
125 a run ends, a short message is printed, explaining how to view the result.
127 - See the documentation for more details.
131 - cg_annotate has a new option, --show-percs, which prints percentages next
136 - callgrind_annotate has a new option, --show-percs, which prints percentages
137 next to all event counts.
139 - callgrind_annotate now inserts commas in call counts, and
140 sort the caller/callee lists in the call tree.
144 - The default value for --read-inline-info is now "yes" on
145 Linux/Android/Solaris. It is still "no" on other OS.
149 - The option --xtree-leak=yes (to output leak result in xtree format)
150 automatically activates the option --show-leak-kinds=all, as xtree
151 visualisation tools such as kcachegrind can in any case select what kind
152 of leak to visualise.
154 - There has been further work to avoid false positives. In particular,
155 integer equality on partially defined inputs (C == and !=) is now handled
158 * ==================== OTHER CHANGES ====================
160 * The new option --show-error-list=no|yes displays, at the end of the run, the
161 list of detected errors and the used suppressions. Prior to this change,
162 showing this information could only be done by specifying "-v -v", but that
163 also produced a lot of other possibly-non-useful messages. The option -s is
164 equivalent to --show-error-list=yes.
166 * ==================== FIXED BUGS ====================
168 The following bugs have been fixed or resolved. Note that "n-i-bz"
169 stands for "not in bugzilla" -- that is, a bug that was reported to us
170 but never got a bugzilla entry. We encourage you to file bugs in
171 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
172 than mailing the developers (or mailing lists) directly -- bugs that
173 are not entered into bugzilla tend to get forgotten about or ignored.
175 To see details of a given bug, visit
176 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
177 where XXXXXX is the bug number as listed below.
179 385411 s390x: z13 vector floating-point instructions not implemented
180 397187 z13 vector register support for vgdb gdbserver
181 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
182 398870 Please add support for instruction vcvtps2ph
183 399287 amd64 front end: Illegal Instruction vcmptrueps
184 399301 Use inlined frames in Massif XTree output.
185 399322 Improve callgrind_annotate output
186 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
187 400164 helgrind test encounters mips x-compiler warnings and assembler error
188 400490 s390x: VRs allocated as if separate from FPRs
189 400491 s390x: Operand of LOCH treated as unsigned integer
190 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
191 options, which specify a mips64 processor
192 401112 LLVM 5.0 generates comparison against partially initialized data
193 401277 More bugs in z13 support
194 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
195 401578 drd: crashes sometimes on fork()
196 401627 memcheck errors with glibc avx2 optimized wcsncmp
197 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
198 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
199 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
201 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
202 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
203 402123 invalid assembler opcodes for mips32r2
204 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
205 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
206 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
207 402351 mips64 libvexmultiarch_test fails on s390x
209 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
210 402480 Do not use %rsp in clobber list
211 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
212 402515 Implement new option --show-error-list=no|yes / -s
213 402519 POWER 3.0 addex instruction incorrectly implemented
214 402781 Redo the cache used to process indirect branch targets
215 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
216 403552 s390x: wrong facility bit checked for vector facility
217 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
218 404638 Add VG_(replaceIndexXA)
219 404843 s390x: backtrace sometimes ends prematurely
220 404888 autotools cleanup series
221 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
222 405182 Valgrind fails to build with Clang
223 405205 filter_libc: remove the line holding the futex syscall error entirely
224 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
225 the upper and lower 32-bits of the 64-bit result
226 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
227 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
228 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
230 405403 s390x disassembler cannot be used on x86
231 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
232 405458 MIPS mkFormVEC arguments swapped?
233 405716 drd: Fix an integer overflow in the stack margin calculation
234 405722 Support arm64 core dump
235 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
236 of the 64-bit destination field.
237 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
238 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
239 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
241 406256 PPC64, vector floating point instructions don't handle subnormal
242 according to VSCR[NJ] bit setting.
243 406352 cachegrind/callgrind fails ann tests because of missing a.c
244 406354 dhat is broken on x86 (32bit)
245 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
246 406357 gdbserver_tests fails because of gdb output change
247 406360 memcheck/tests/libstdc++.supp needs more supression variants
248 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
249 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
250 407340 PPC64, does not support the vlogefp, vexptefp instructions.
251 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
252 n-i-bz Fix callgrind_annotate non deterministic order for equal total
253 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
254 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
255 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
257 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
258 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
259 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
263 Release 3.14.0 (9 October 2018)
264 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
266 3.14.0 is a feature release with many improvements and the usual collection of
269 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
270 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
271 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
272 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
273 support for X86/macOS 10.13, AMD64/macOS 10.13.
275 * ==================== CORE CHANGES ===================
277 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
278 debug info for unloaded code. This allows saved stack traces (e.g. for
279 memory leaks) to include file/line info for code that has been dlclose'd (or
280 similar). See the user manual for more information and known limitations.
282 * Ability to specify suppressions based on source file name and line number.
284 * Majorly overhauled register allocator. No end-user changes, but the JIT
285 generates code a bit more quickly now.
287 * ================== PLATFORM CHANGES =================
289 * Preliminary support for macOS 10.13 has been added.
291 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
293 * mips: support for MIPS SIMD architecture (MSA) has been added.
295 * mips: support for MIPS N32 ABI has been added.
297 * s390: partial support for vector instructions (integer and string) has been
300 * ==================== TOOL CHANGES ====================
302 * Helgrind: Addition of a flag
303 --delta-stacktrace=no|yes [yes on linux amd64/x86]
304 which specifies how full history stack traces should be computed.
305 Setting this to =yes can speed up Helgrind by 25% when using
306 --history-level=full.
308 * Memcheck: reduced false positive rate for optimised code created by Clang 6
309 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
310 blocks more carefully to determine where it can avoid expensive definedness
311 checks without loss of precision. This is controlled by the flag
312 --expensive-definedness-checks=no|auto|yes [auto].
314 * ==================== OTHER CHANGES ====================
316 * Valgrind is now buildable with link-time optimisation (LTO). A new
317 configure option --enable-lto=yes allows building Valgrind with LTO. If the
318 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
319 Note that if you are doing Valgrind development, --enable-lto=yes massively
320 slows down the build process.
322 * ==================== FIXED BUGS ====================
324 The following bugs have been fixed or resolved. Note that "n-i-bz"
325 stands for "not in bugzilla" -- that is, a bug that was reported to us
326 but never got a bugzilla entry. We encourage you to file bugs in
327 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
328 than mailing the developers (or mailing lists) directly -- bugs that
329 are not entered into bugzilla tend to get forgotten about or ignored.
331 To see details of a given bug, visit
332 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
333 where XXXXXX is the bug number as listed below.
335 79362 Debug info is lost for .so files when they are dlclose'd
336 208052 strlcpy error when n = 0
337 255603 exp-sgcheck Assertion '!already_present' failed
338 338252 building valgrind with -flto (link time optimisation) fails
339 345763 MIPS N32 ABI support
340 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
341 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
342 372347 Replacement problem of the additional c++14/c++17 new/delete operators
343 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
344 376257 helgrind history full speed up using a cached stack
345 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
347 379748 Fix missing pselect syscall (OS X 10.11)
348 379754 Fix missing syscall ulock_wait (OS X 10.12)
349 380397 s390x: __GI_strcspn() replacemenet needed
350 381162 possible array overrun in VEX register allocator
351 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
352 381274 powerpc too chatty even with --sigill-diagnostics=no
353 381289 epoll_pwait can have a NULL sigmask
354 381553 VEX register allocator v3
355 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
356 381769 Use ucontext_t instead of struct ucontext
357 381805 arm32 needs ld.so index hardwire for new glibc security fixes
358 382256 gz compiler flag test doesn't work for gold
359 382407 vg_perf needs "--terse" command line option
360 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
361 382563 MIPS MSA ASE support
362 382998 xml-socket doesn't work
363 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
364 383723 Fix missing kevent_qos syscall (macOS 10.11)
365 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
366 384096 Mention AddrCheck at Memcheck's command line option [..]
367 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
368 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
369 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
370 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
371 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
372 384337 performance improvements to VEX register allocator v2 and v3
373 384526 reduce number of spill insns generated by VEX register allocator v3
374 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
375 384631 Sanitise client args as printed with -v
376 384633 Add a simple progress-reporting facility
377 384987 VEX regalloc: allocate caller-save registers for short lived vregs
378 385055 PPC VEX temporary storage exhausted
379 385182 PPC64 is missing support for the DSCR
380 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
381 385207 PPC64, generate_store_FPRF() generates too many Iops
382 385208 PPC64, xxperm instruction exhausts temporary memory
383 385210 PPC64, vpermr instruction could exhaust temporary memory
384 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
385 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
386 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
387 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
388 385408 s390x: z13 vector "support" instructions not implemented
389 385409 s390x: z13 vector integer instructions not implemented
390 385410 s390x: z13 vector string instructions not implemented
391 385412 s390x: new non-vector z13 instructions not implemented
392 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
393 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
394 385939 Optionally exit on the first error
395 386318 valgrind.org/info/tools.html is missing SGCheck
396 386425 running valgrind + wine on armv7l gives illegal opcode
397 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
398 387410 MIPSr6 support
399 387664 Memcheck: make expensive-definedness-checks be the default
400 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
401 387766 asm shifts cause false positive "Conditional jump or move depends
402 on uninitialised value"
403 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
404 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
405 388786 Support bpf syscall in amd64 Linux
406 388862 Add replacements for wmemchr and wcsnlen on Linux
407 389065 valgrind meets gcc flag -Wlogical-op
408 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
409 390471 suppression by specification of source-file line number
410 390723 make xtree dump files world wide readable, similar to log files
411 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
412 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
413 392118 unhandled amd64-linux syscall: 332 (statx)
414 392449 callgrind not clearing the number of calls properly
415 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
416 lxssp, stxssp and stxvl instructions.
417 393023 callgrind_control risks using the wrong vgdb
418 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
419 393099 posix_memalign() invalid write if alignment == 0
420 393146 failing assert "is_DebugInfo_active(di)"
421 395709 PPC64 is missing support for the xvnegsp instruction
422 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
424 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
425 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
426 396839 s390x: Trap instructions not implemented
427 396887 arch_prctl should return EINVAL on unknown option
428 == 397286 crash before launching binary (Unsupported arch_prctl option)
429 == 397393 valgrind: the 'impossible' happened: (Archlinux)
430 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
431 396906 compile tests failure on mips32-linux: broken inline asm in tests on
433 397012 glibc ld.so uses arch_prctl on i386
434 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
435 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
436 397424 glibc 2.27 and gdb_server tests
437 398028 Assertion `cfsi_fits` failing in simple C program
438 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
440 n-i-bz Fix missing workq_ops operations (macOS)
441 n-i-bz fix bug in strspn replacement
442 n-i-bz Add support for the Linux BLKFLSBUF ioctl
443 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
444 n-i-bz Fix possible stack trashing by semctl syscall wrapping
445 n-i-bz Add support for the Linux membarrier() system call
446 n-i-bz x86 front end: recognise and handle UD2 correctly
447 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
448 correctly aligned before entering the handler.
450 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
451 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
452 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
456 Release 3.13.0 (15 June 2017)
457 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
459 3.13.0 is a feature release with many improvements and the usual collection of
462 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
463 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
464 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
465 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
467 * ==================== CORE CHANGES ===================
469 * The translation cache size has been increased to keep up with the demands of
470 large applications. The maximum number of sectors has increased from 24 to
471 48. The default number of sectors has increased from 16 to 32 on all
472 targets except Android, where the increase is from 6 to 12.
474 * The amount of memory that Valgrind can use has been increased from 64GB to
475 128GB. In particular this means your application can allocate up to about
476 60GB when running on Memcheck.
478 * Valgrind's default load address has been changed from 0x3800'0000 to
479 0x5800'0000, so as to make it possible to load larger executables. This
480 should make it possible to load executables of size at least 1200MB.
482 * A massive spaceleak caused by reading compressed debuginfo files has been
483 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
486 * The C++ demangler has been updated.
488 * Support for demangling Rust symbols has been added.
490 * A new representation of stack traces, the "XTree", has been added. An XTree
491 is a tree of stacktraces with data associated with the stacktraces. This is
492 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
493 consumption of your program. Reporting is controlled by the new options
494 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
496 A report can also be produced on demand using the gdbserver monitor command
497 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
498 format' and 'massif format. The existing visualisers for these formats (e.g.
499 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
500 analyse these reports.
502 Memcheck can also produce XTree leak reports using the Callgrind file
503 format. For more details, see the user manual.
505 * ================== PLATFORM CHANGES =================
507 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
509 * amd64: fixes for JIT failure problems on long AVX2 code blocks
511 * amd64 and x86: support for CET prefixes has been added
513 * arm32: a few missing ARMv8 instructions have been implemented
515 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
516 Store-Conditional instructions has been added. This is to deal with
517 processor implementations that implement the LL/SC specifications strictly
518 and as a result cause Valgrind to hang in certain situations. The
519 alternative implementation is automatically enabled at startup, as required.
520 You can use the option --sim-hints=fallback-llsc to force-enable it if you
523 * Support for OSX 10.12 has been improved.
525 * On Linux, clone handling has been improved to honour CLONE_VFORK that
526 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
527 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
528 CLONE_VM exact semantics will (still) not work.
530 * The TileGX/Linux port has been removed because it appears to be both unused
533 * ==================== TOOL CHANGES ====================
537 - Memcheck should give fewer false positives when running optimised
538 Clang/LLVM generated code.
540 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
542 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
543 to produce the end of execution leak report in a xtree callgrind format
546 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
547 the leak report in an xtree file.
551 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
553 - For some workloads (typically, for big applications), Massif memory
554 consumption and CPU consumption has decreased significantly.
558 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
560 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
561 for Ada gnat compiled applications.
563 * ==================== OTHER CHANGES ====================
565 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
566 append the inner guest stacktrace to the inner host stacktrace. This helps
567 to investigate the errors reported by the outer, when they are caused by the
568 inner guest program (such as an inner regtest). See README_DEVELOPERS for
571 * To allow fast detection of callgrind files by desktop environments and file
572 managers, the format was extended to have an optional first line that
573 uniquely identifies the format ("# callgrind format"). Callgrind creates
574 this line now, as does the new xtree functionality.
576 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
577 have a new %n format letter that is replaced by a sequence number.
579 * "--version -v" now shows the SVN revision numbers from which Valgrind was
582 * ==================== FIXED BUGS ====================
584 The following bugs have been fixed or resolved. Note that "n-i-bz"
585 stands for "not in bugzilla" -- that is, a bug that was reported to us
586 but never got a bugzilla entry. We encourage you to file bugs in
587 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
588 than mailing the developers (or mailing lists) directly -- bugs that
589 are not entered into bugzilla tend to get forgotten about or ignored.
591 To see details of a given bug, visit
592 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
593 where XXXXXX is the bug number as listed below.
595 162848 --log-file output isn't split when a program forks
596 340777 Illegal instruction on mips (ar71xx)
597 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
598 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
599 to a different stack.
600 344139 x86 stack-seg overrides, needed by the Wine people
601 344524 store conditional of guest applications always fail - observed on
603 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
604 352395 Please provide SVN revision info in --version -v
605 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
606 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
607 INVALID_POSIX_THREADID' failed
608 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
609 barrier implementation
610 358697 valgrind.h: Some code remains even when defining NVALGRIND
611 359202 Add musl libc configure/compile
612 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
613 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
614 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
615 362223 assertion failed when .valgrindrc is a directory instead of a file
616 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
617 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
618 368507 can't malloc chunks larger than about 34GB
619 368529 Android arm target link error, missing atexit and pthread_atfork
620 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
621 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
622 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
623 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
624 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
625 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
626 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
627 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
628 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
629 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
630 370028 Reduce the number of compiler warnings on MIPS platforms
631 370635 arm64 missing syscall getcpu
632 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
633 371227 Clean AArch64 syscall table
634 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
635 371471 Valgrind complains about non legit memory leaks on placement new (C++)
636 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
637 371503 disInstr(arm64): unhandled instruction 0xF89F0000
638 371869 support '%' in symbol Z-encoding
639 371916 execution tree xtree concept
640 372120 c++ demangler demangles symbols which are not c++
641 372185 Support of valgrind on ARMv8 with 32 bit executable
642 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
643 372195 Power PC, xxsel instruction is not always recognized.
644 372504 Hanging on exit_group
645 372600 process loops forever when fatal signals are arriving quickly
646 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
647 373046 Stacks registered by core are never deregistered
648 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
649 373086 Implement additional Xen hypercalls
650 373192 Calling posix_spawn in glibc 2.24 completely broken
651 373488 Support for fanotify API on ARM64 architecture
652 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
653 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
654 373938 const IRExpr arguments for matchIRExpr()
655 374719 some spelling fixes
656 374963 increase valgrind's load address to prevent mmap failure
657 375514 valgrind_get_tls_addr() does not work in case of static TLS
658 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
659 for ML_(find_rx_mapping)()
660 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
661 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
662 == 377159 "vex: the `impossible' happened" still present
663 == 375150 Assertion 'tres.status == VexTransOK' failed
664 == 378068 valgrind crashes on AVX2 function in FFmpeg
665 376142 Segfaults on MIPS Cavium Octeon boards
666 376279 disInstr(arm64): unhandled instruction 0xD50320FF
667 376455 Solaris: unhandled syscall lgrpsys(180)
668 376518 Solaris: unhandled fast trap getlgrp(6)
669 376611 ppc64 and arm64 don't know about prlimit64 syscall
670 376729 PPC64, remove R2 from the clobber list
672 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
673 to be wrongly marked as addressable
674 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
675 PIE enabled by default
676 377376 memcheck/tests/linux/getregset fails with glibc2.24
677 377427 PPC64, lxv instruction failing on odd destination register
678 377478 PPC64: ISA 3.0 setup fixes
679 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
680 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
681 and 2 args for FUTEX_TRYLOCK_PI
682 377717 Fix massive space leak when reading compressed debuginfo sections
683 377891 Update Xen 4.6 domctl wrappers
684 377930 fcntl syscall wrapper is missing flock structure check
685 378524 libvexmultiarch_test regression on s390x and ppc64
686 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
687 378673 Update libiberty demangler
688 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
689 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
690 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
691 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
692 (task_register_dyld_image_infos)
693 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
694 (task_register_dyld_shared_cache_image_info)
695 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
696 379473 MIPS: add support for rdhwr cycle counter register
697 379504 remove TileGX/Linux port
698 379525 Support more x86 nop opcodes
699 379838 disAMode(x86): not an addr!
700 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
701 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
702 379895 clock_gettime does not execute POST syscall wrapper
703 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
704 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
705 380200 xtree generated callgrind files refer to files without directory name
706 380202 Assertion failure for cache line size (cls == 64) on aarch64.
707 380397 s390x: __GI_strcspn() replacement needed
708 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
710 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
711 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
712 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
716 Release 3.12.0 (20 October 2016)
717 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
719 3.12.0 is a feature release with many improvements and the usual
720 collection of bug fixes.
722 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
723 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
724 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
725 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
726 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
727 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
729 * ================== PLATFORM CHANGES =================
731 * POWER: Support for ISA 3.0 has been added
733 * mips: support for O32 FPXX ABI has been added.
734 * mips: improved recognition of different processors
735 * mips: determination of page size now done at run time
737 * amd64: Partial support for AMD FMA4 instructions.
739 * arm, arm64: Support for v8 crypto and CRC instructions.
741 * Improvements and robustification of the Solaris port.
743 * Preliminary support for MacOS 10.12 (Sierra) has been added.
745 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
746 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
747 where possible. Valgrind's support for 32-bit x86 has stagnated in
748 recent years and has fallen far behind that for 64-bit x86
749 instructions. By contrast 64-bit x86 is well supported, up to and
752 * ==================== TOOL CHANGES ====================
756 - Added meta mempool support for describing a custom allocator which:
757 - Auto-frees all chunks assuming that destroying a pool destroys all
759 - Uses itself to allocate other memory blocks
761 - New flag --ignore-range-below-sp to ignore memory accesses below
762 the stack pointer, if you really have to. The related flag
763 --workaround-gcc296-bugs=yes is now deprecated. Use
764 --ignore-range-below-sp=1024-1 as a replacement.
768 - Improved thread startup time significantly on non-Linux platforms.
772 - Added collection of the metric "tot-blocks-allocd"
774 * ==================== OTHER CHANGES ====================
776 * Replacement/wrapping of malloc/new related functions is now done not just
777 for system libraries by default, but for any globally defined malloc/new
778 related function (both in shared libraries and statically linked alternative
779 malloc implementations). The dynamic (runtime) linker is excluded, though.
780 To only intercept malloc/new related functions in
781 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
782 "nouserintercepts" can be any non-existing library name).
783 This new functionality is not implemented for MacOS X.
785 * The maximum number of callers in a suppression entry is now equal to
786 the maximum size for --num-callers (500).
787 Note that --gen-suppressions=yes|all similarly generates suppressions
788 containing up to --num-callers frames.
790 * New and modified GDB server monitor features:
792 - Valgrind's gdbserver now accepts the command 'catch syscall'.
793 Note that you must have GDB >= 7.11 to use 'catch syscall' with
796 * New option --run-cxx-freeres=<yes|no> can be used to change whether
797 __gnu_cxx::__freeres() cleanup function is called or not. Default is
800 * Valgrind is able to read compressed debuginfo sections in two formats:
801 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
802 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
804 * Modest JIT-cost improvements: the cost of instrumenting code blocks
805 for the most common use case (x86_64-linux, Memcheck) has been
808 * Improved performance for programs that do a lot of discarding of
809 instruction address ranges of 8KB or less.
811 * The C++ symbol demangler has been updated.
813 * More robustness against invalid syscall parameters on Linux.
815 * ==================== FIXED BUGS ====================
817 The following bugs have been fixed or resolved. Note that "n-i-bz"
818 stands for "not in bugzilla" -- that is, a bug that was reported to us
819 but never got a bugzilla entry. We encourage you to file bugs in
820 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
821 than mailing the developers (or mailing lists) directly -- bugs that
822 are not entered into bugzilla tend to get forgotten about or ignored.
824 To see details of a given bug, visit
825 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
826 where XXXXXX is the bug number as listed below.
828 191069 Exiting due to signal not reported in XML output
829 199468 Suppressions: stack size limited to 25
830 while --num-callers allows more frames
831 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
832 278744 cvtps2pd with redundant RexW
833 303877 valgrind doesn't support compressed debuginfo sections.
834 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
835 348345 Assertion fails for negative lineno
836 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
837 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
838 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
839 351804 Crash on generating suppressions for "printf" call on OS X 10.10
840 352197 mips: mmap2() not wrapped correctly for page size > 4096
841 353083 arm64 doesn't implement various xattr system calls
842 353084 arm64 doesn't support sigpending system call
843 353137 www: update info for Supported Platforms
844 353138 www: update "The Valgrind Developers" page
845 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
848 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
849 353398 WARNING: unhandled amd64-solaris syscall: 207
850 353660 XML in auxwhat tag not escaping reserved symbols properly
851 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
852 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
853 353802 ELF debug info reader confused with multiple .rodata sections
854 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
855 353917 unhandled amd64-solaris syscall fchdir(120)
856 353920 unhandled amd64-solaris syscall: 170
857 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
858 354392 unhandled amd64-solaris syscall: 171
859 354797 Vbit test does not include Iops for Power 8 instruction support
860 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
864 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
865 355188 valgrind should intercept all malloc related global functions
866 355454 do not intercept malloc related symbols from the runtime linker
867 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
868 356044 Dwarf line info reader misinterprets is_stmt register
869 356112 mips: replace addi with addiu
870 356393 valgrind (vex) crashes because isZeroU happened
873 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
874 356678 arm64-linux: unhandled syscall 232 (mincore)
875 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
876 356823 Unsupported ARM instruction: stlex
877 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
878 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
879 357673 crash if I try to run valgrind with a binary link with libcurl
880 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
881 357871 pthread_spin_destroy not properly wrapped
882 357887 Calls to VG_(fclose) do not close the file descriptor
883 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
884 358030 support direct socket calls on x86 32bit (new in linux 4.3)
885 358478 drd/tests/std_thread.cpp doesn't build with GCC6
886 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
887 359181 Buffer Overflow during Demangling
888 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
889 359289 s390x: popcnt (B9E1) not implemented
890 359472 The Power PC vsubuqm instruction doesn't always give the correct result
891 359503 Add missing syscalls for aarch64 (arm64)
892 359645 "You need libc6-dbg" help message could be more helpful
893 359703 s390: wire up separate socketcalls system calls
894 359724 getsockname might crash - deref_UInt should call safe_to_deref
895 359733 amd64 implement ld.so strchr/index override like x86
896 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
897 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
899 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
900 359871 Incorrect mask handling in ppoll
901 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
902 360008 Contents of Power vr registers contents is not printed correctly when
903 the --vgdb-shadow-registers=yes option is used
904 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
906 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
907 360425 arm64 unsupported instruction ldpsw
909 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
910 360571 Error about the Android Runtime reading below the stack pointer on ARM
911 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
912 360749 kludge for multiple .rodata sections on Solaris no longer needed
913 360752 raise the number of reserved fds in m_main.c from 10 to 12
914 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
915 361226 s390x: risbgn (EC59) not implemented
916 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
917 361354 ppc64[le]: wire up separate socketcalls system calls
918 361615 Inconsistent termination for multithreaded process terminated by signal
919 361926 Unhandled Solaris syscall: sysfs(84)
920 362009 V dumps core on unimplemented functionality before threads are created
921 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
922 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
923 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
924 362953 Request for an update to the Valgrind Developers page
925 363680 add renameat2() support
926 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
927 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
928 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
929 364058 clarify in manual limitations of array overruns detections
930 364413 pselect sycallwrapper mishandles NULL sigmask
931 364728 Power PC, missing support for several HW registers in
932 get_otrack_shadow_offset_wrk()
933 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
934 365273 Invalid write to stack location reported after signal handler runs
935 365912 ppc64BE segfault during jm-insns test (RELRO)
936 366079 FPXX Support for MIPS32 Valgrind
937 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
938 366344 Multiple unhandled instruction for Aarch64
939 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
940 367995 Integration of memcheck with custom memory allocator
941 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
942 368412 False positive result for altivec capability check
943 368416 Add tc06_two_races_xml.exp output for ppc64
944 368419 Perf Events ioctls not implemented
945 368461 mmapunmap test fails on ppc64
946 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
947 369000 AMD64 fma4 instructions unsupported.
948 369169 ppc64 fails jm_int_isa_2_07 test
949 369175 jm_vec_isa_2_07 test crashes on ppc64
950 369209 valgrind loops and eats up all memory if cwd doesn't exist.
951 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
952 369359 msghdr_foreachfield can crash when handling bad iovec
953 369360 Bad sigprocmask old or new sets can crash valgrind
954 369361 vmsplice syscall wrapper crashes on bad iovec
955 369362 Bad sigaction arguments crash valgrind
956 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
957 369402 Bad set/get_thread_area pointer crashes valgrind
958 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
959 369446 valgrind crashes on unknown fcntl command
960 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
961 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
962 370265 ISA 3.0 HW cap stuff needs updating
963 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
964 372195 Power PC, xxsel instruction is not always recognized
966 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
967 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
968 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
969 n-i-bz Fix ppoll_alarm exclusion on OS X
970 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
971 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
972 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
973 n-i-bz amd64: memcheck false positive with shr %edx
974 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
975 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
976 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
977 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
979 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
980 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
981 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
985 Release 3.11.0 (22 September 2015)
986 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
988 3.11.0 is a feature release with many improvements and the usual
989 collection of bug fixes.
991 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
992 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
993 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
994 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
995 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
996 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
998 * ================== PLATFORM CHANGES =================
1000 * Support for Solaris/x86 and Solaris/amd64 has been added.
1002 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1004 * Preliminary support for the Tilera TileGX architecture has been added.
1006 * s390x: It is now required for the host to have the "long displacement"
1007 facility. The oldest supported machine model is z990.
1009 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1010 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1012 * The JIT's register allocator is significantly faster, making the JIT
1013 as a whole somewhat faster, so JIT-intensive activities, for example
1014 program startup, are modestly faster, around 5%.
1016 * There have been changes to the default settings of several command
1017 line flags, as detailed below.
1019 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1020 capable hosts, the simulated CPUID will now indicate AVX2 support.
1022 * ==================== TOOL CHANGES ====================
1026 - The default value for --leak-check-heuristics has been changed from
1027 "none" to "all". This helps to reduce the number of possibly
1028 lost blocks, in particular for C++ applications.
1030 - The default value for --keep-stacktraces has been changed from
1031 "malloc-then-free" to "malloc-and-free". This has a small cost in
1032 memory (one word per malloc-ed block) but allows Memcheck to show the
1033 3 stacktraces of a dangling reference: where the block was allocated,
1034 where it was freed, and where it is acccessed after being freed.
1036 - The default value for --partial-loads-ok has been changed from "no" to
1037 "yes", so as to avoid false positive errors resulting from some kinds
1038 of vectorised loops.
1040 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1041 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1042 than get_vbits when you need to associate byte data value with
1043 their corresponding validity bits.
1045 - The 'block_list' monitor command has been enhanced:
1046 o it can print a range of loss records
1047 o it now accepts an optional argument 'limited <max_blocks>'
1048 to control the number of blocks printed.
1049 o if a block has been found using a heuristic, then
1050 'block_list' now shows the heuristic after the block size.
1051 o the loss records/blocks to print can be limited to the blocks
1052 found via specified heuristics.
1054 - The C helper functions used to instrument loads on
1055 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1056 replaced by handwritten assembly sequences. This gives speedups
1057 in the region of 0% to 7% for those targets only.
1059 - A new command line option, --expensive-definedness-checks=yes|no,
1060 has been added. This is useful for avoiding occasional invalid
1061 uninitialised-value errors in optimised code. Watch out for
1062 runtime degradation, as this can be up to 25%. As always, though,
1063 the slowdown is highly application specific. The default setting
1068 - A new monitor command 'all_snapshots <filename>' dumps all
1069 snapshots taken so far.
1073 - Significant memory reduction and moderate speedups for
1074 --history-level=full for applications accessing a lot of memory
1075 with many different stacktraces.
1077 - The default value for --conflict-cache-size=N has been doubled to
1078 2000000. Users that were not using the default value should
1079 preferably also double the value they give.
1081 The default was changed due to the changes in the "full history"
1082 implementation. Doubling the value gives on average a slightly more
1083 complete history and uses similar memory (or significantly less memory
1084 in the worst case) than the previous implementation.
1086 - The Helgrind monitor command 'info locks' now accepts an optional
1087 argument 'lock_addr', which shows information about the lock at the
1090 - When using --history-level=full, the new Helgrind monitor command
1091 'accesshistory <addr> [<len>]' will show the recorded accesses for
1092 <len> (or 1) bytes at <addr>.
1094 * ==================== OTHER CHANGES ====================
1096 * The default value for the --smc-check option has been changed from
1097 "stack" to "all-non-file" on targets that provide automatic D-I
1098 cache coherence (x86, amd64 and s390x). The result is to provide,
1099 by default, transparent support for JIT generated and self-modifying
1100 code on all targets.
1102 * Mac OS X only: the default value for the --dsymutil option has been
1103 changed from "no" to "yes", since any serious usage on Mac OS X
1104 always required it to be "yes".
1106 * The command line options --db-attach and --db-command have been removed.
1107 They were deprecated in 3.10.0.
1109 * When a process dies due to a signal, Valgrind now shows the signal
1110 and the stacktrace at default verbosity (i.e. verbosity 1).
1112 * The address description logic used by Memcheck and Helgrind now
1113 describes addresses in anonymous segments, file mmap-ed segments,
1114 shared memory segments and the brk data segment.
1116 * The new option --error-markers=<begin>,<end> can be used to mark the
1117 begin/end of errors in textual output mode, to facilitate
1118 searching/extracting errors in output files that mix valgrind errors
1119 with program output.
1121 * The new option --max-threads=<number> can be used to change the number
1122 of threads valgrind can handle. The default is 500 threads which
1123 should be more than enough for most applications.
1125 * The new option --valgrind-stacksize=<number> can be used to change the
1126 size of the private thread stacks used by Valgrind. This is useful
1127 for reducing memory use or increasing the stack size if Valgrind
1128 segfaults due to stack overflow.
1130 * The new option --avg-transtab-entry-size=<number> can be used to specify
1131 the expected instrumented block size, either to reduce memory use or
1132 to avoid excessive retranslation.
1134 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1136 * New and modified GDB server monitor features:
1138 - When a signal is reported in GDB, you can now use the GDB convenience
1139 variable $_siginfo to examine detailed signal information.
1141 - Valgrind's gdbserver now allows the user to change the signal
1142 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1143 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1144 continue without passing the signal to the process.
1146 - With GDB >= 7.10, the command 'target remote'
1147 will automatically load the executable file of the process running
1148 under Valgrind. This means you do not need to specify the executable
1149 file yourself, GDB will discover it itself. See GDB documentation about
1150 'qXfer:exec-file:read' packet for more info.
1152 * ==================== FIXED BUGS ====================
1154 The following bugs have been fixed or resolved. Note that "n-i-bz"
1155 stands for "not in bugzilla" -- that is, a bug that was reported to us
1156 but never got a bugzilla entry. We encourage you to file bugs in
1157 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1158 than mailing the developers (or mailing lists) directly -- bugs that
1159 are not entered into bugzilla tend to get forgotten about or ignored.
1161 To see details of a given bug, visit
1162 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1163 where XXXXXX is the bug number as listed below.
1165 116002 VG_(printf): Problems with justification of strings and integers
1166 155125 avoid cutting away file:lineno after long function name
1167 197259 Unsupported arch_prtctl PR_SET_GS option
1168 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1169 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1170 201435 Fix Darwin: -v does not show kernel version
1171 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1172 211256 Fixed an outdated comment regarding the default platform.
1173 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1174 211926 Avoid compilation warnings in valgrind.h with -pedantic
1175 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1177 226609 Crediting upstream authors in man page
1178 231257 Valgrind omits path when executing script from shebang line
1179 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1180 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1181 269360 s390x: Fix addressing mode selection for compare-and-swap
1182 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1184 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1185 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1186 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1187 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1188 330147 libmpiwrap PMPI_Get_count returns undefined value
1189 333051 mmap of huge pages fails due to incorrect alignment
1191 334802 valgrind does not always explain why a given option is bad
1192 335618 mov.w rN, pc/sp (ARM32)
1193 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1198 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1199 338602 AVX2 bit in CPUID missing
1200 338606 Strange message for scripts with invalid interpreter
1201 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1202 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1203 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1205 339156 gdbsrv not called for fatal signal
1206 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1207 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1208 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1209 339442 Fix testsuite build failure on OS X 10.9
1210 339542 Enable compilation with Intel's ICC compiler
1211 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1212 339688 Mac-specific ASM does not support .version directive (cpuid,
1213 tronical and pushfpopf tests)
1214 339745 Valgrind crash when check Marmalade app (partial fix)
1215 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1216 339778 Linux/TileGx platform support for Valgrind
1217 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1218 339789 Fix none/tests/execve test on Mac OS X 10.9
1219 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1220 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1221 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1222 340392 Allow user to select more accurate definedness checking in memcheck
1223 to avoid invalid complaints on optimised code
1224 340430 Fix some grammatical weirdness in the manual.
1225 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1226 341419 Signal handler ucontext_t not filled out correctly on OS X
1227 341539 VG_(describe_addr) should not describe address as belonging to client
1228 segment if it is past the heap end
1229 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1230 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1231 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1232 341789 aarch64: shmat fails with valgrind on ARMv8
1233 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1234 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1235 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1236 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1237 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1238 342221 socket connect false positive uninit memory for unknown af family
1239 342353 Allow dumping full massif output while valgrind is still running
1240 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1244 342603 Add I2C_SMBUS ioctl support
1245 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1246 342683 Mark memory past the initial brk limit as unaddressable
1247 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1248 342795 Internal glibc __GI_mempcpy call should be intercepted
1249 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1250 343012 Unhandled syscall 319 (memfd_create)
1251 343069 Patch updating v4l2 API support
1252 343173 helgrind crash during stack unwind
1253 343219 fix GET_STARTREGS for arm
1254 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1255 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1256 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1257 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1258 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1259 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1260 343597 ppc64le: incorrect use of offseof macro
1261 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1262 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1263 343732 Unhandled syscall 144 (setgid) on aarch64
1264 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1265 343802 s390x: False positive "conditional jump or move depends on [..]
1266 343902 --vgdb=yes doesn't break when --xml=yes is used
1267 343967 Don't warn about setuid/setgid/setcap executable for directories
1268 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1269 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1270 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1271 344054 www - update info for Solaris/illumos
1272 344416 'make regtest' does not work cleanly on OS X
1273 344235 Remove duplicate include of pub_core_aspacemgr.h
1274 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1275 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1276 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1277 344314 callgrind_annotate ... warnings about commands containing newlines
1278 344318 socketcall should wrap recvmmsg and sendmmsg
1279 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1280 344416 Fix 'make regtest' does not work cleanly on OS X
1281 344499 Fix compilation for Linux kernel >= 4.0.0
1282 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1283 unix:349 (__pthread_fchdir)
1284 344559 Garbage collection of unused segment names in address space manager
1285 344560 Fix stack traces missing penultimate frame on OS X
1286 344621 Fix memcheck/tests/err_disable4 test on OS X
1287 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1288 344702 Fix missing libobjc suppressions on OS X 10.10
1290 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1291 344939 Fix memcheck/tests/xml1 on OS X 10.10
1292 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1293 345079 Fix build problems in VEX/useful/test_main.c
1294 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1295 345177 arm64: prfm (reg) not implemented
1296 345215 Performance improvements for the register allocator
1297 345248 add support for Solaris OS in valgrind
1298 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1299 345394 Fix memcheck/tests/strchr on OS X
1300 345637 Fix memcheck/tests/sendmsg on OS X
1301 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1302 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1303 345887 Fix an assertion in the address space manager
1304 345928 amd64: callstack only contains current function for small stacks
1305 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1306 345987 MIPS64: Implement cavium LHX instruction
1307 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1308 346185 Fix typo saving altivec register v24
1309 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1310 and LibVEX_GuestPPC64_get_CR()
1311 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1312 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1313 346307 fuse filesystem syscall deadlocks
1314 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1315 346411 MIPS: SysRes::_valEx handling is incorrect
1316 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1317 346474 PPC64 Power 8, spr TEXASRU register not supported
1318 346487 Compiler generates "note" about a future ABI change for PPC64
1319 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1320 and causing spurious "invalid read of size 8" warnings
1321 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1322 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1323 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1324 347322 Power PC regression test cleanup
1325 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1327 347389 unhandled syscall: 373 (Linux ARM syncfs)
1328 347686 Patch set to cleanup PPC64 regtests
1329 347978 Remove bash dependencies where not needed
1330 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1331 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1333 348102 Patch updating v4l2 API support
1334 348247 amd64 front end: jno jumps wrongly when overflow is not set
1335 348269 Improve mmap MAP_HUGETLB support.
1336 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1337 348345 Assertion fails for negative lineno
1338 348377 Unsupported ARM instruction: yield
1339 348565 Fix detection of command line option availability for clang
1340 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1341 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1342 348748 Fix redundant condition
1343 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1344 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1345 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1346 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1347 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1348 349626 Implemented additional Xen hypercalls
1349 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1350 349790 Clean up of the hardware capability checking utilities.
1351 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1352 349874 Fix typos in source code
1353 349879 memcheck: add handwritten assembly for helperc_LOADV*
1354 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1355 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1356 350202 Add limited param to 'monitor block_list'
1357 350290 s390x: Support instructions fixbr(a)
1358 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1359 350809 Fix none/tests/async-sigs for Solaris
1360 350811 Remove reference to --db-attach which has been removed.
1361 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1362 350854 hard-to-understand code in VG_(load_ELF)()
1363 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1364 351386 Solaris: Cannot run ld.so.1 under Valgrind
1365 351474 Fix VG_(iseqsigset) as obvious
1366 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1367 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1368 351858 ldsoexec support on Solaris
1369 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1370 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1371 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1372 352320 arm64 crash on none/tests/nestedfs
1373 352765 Vbit test fails on Power 6
1374 352768 The mbar instruction is missing from the Power PC support
1375 352769 Power PC program priority register (PPR) is not supported
1376 n-i-bz Provide implementations of certain compiler builtins to support
1377 compilers that may not provide those
1378 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1379 n-i-bz Fix compilation on distros with glibc < 2.5
1380 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1381 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1382 n-i-bz mremap did not work properly on shared memory
1383 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1384 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1386 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1387 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1388 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1392 Release 3.10.1 (25 November 2014)
1393 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1394 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1395 and backports fixes for all reported missing AArch64 ARMv8 instructions
1396 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1397 to use, you might want to consider upgrading to 3.10.1 instead.
1399 The following bugs have been fixed or resolved. Note that "n-i-bz"
1400 stands for "not in bugzilla" -- that is, a bug that was reported to us
1401 but never got a bugzilla entry. We encourage you to file bugs in
1402 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1403 than mailing the developers (or mailing lists) directly -- bugs that
1404 are not entered into bugzilla tend to get forgotten about or ignored.
1406 To see details of a given bug, visit
1407 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1408 where XXXXXX is the bug number as listed below.
1410 335440 arm64: ld1 (single structure) is not implemented
1411 335713 arm64: unhanded instruction: prfm (immediate)
1412 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1413 339182 ppc64: AvSplat ought to load destination vector register with [..]
1414 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1415 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1416 339645 Use correct tag names in sys_getdents/64 wrappers
1417 339706 Fix false positive for ioctl(TIOCSIG) on linux
1418 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1419 339853 arm64 times syscall unknown
1420 339855 arm64 unhandled getsid/setsid syscalls
1421 339858 arm64 dmb sy not implemented
1422 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1423 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1424 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1426 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1427 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1428 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1429 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1430 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1431 340509 arm64: unhandled instruction fcvtas
1432 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1433 340632 arm64: unhandled instruction fcvtas
1434 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1435 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1436 340788 warning: unhandled syscall: 318 (getrandom)
1437 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1438 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1439 340922 arm64: unhandled getgroups/setgroups syscalls
1440 350251 Fix typo in VEX utility program (test_main.c).
1441 350407 arm64: unhandled instruction ucvtf (vector, integer)
1442 350809 none/tests/async-sigs breaks when run under cron on Solaris
1443 350811 update README.solaris after r15445
1444 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1445 350854 strange code in VG_(load_ELF)()
1446 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1447 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1448 n-i-bz Add missing ]] to terminate CDATA.
1449 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1450 n-i-bz Enable sys_fadvise64_64 on arm32.
1451 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1452 n-i-bz Add test cases for all known arm64 load/store instructions.
1453 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1454 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1455 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1456 n-i-bz arm64: Implement "fcvtpu w, s".
1457 n-i-bz arm64: implement ADDP and various others
1458 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1459 n-i-bz arm64: enable FCVT{A,N}S X,S.
1461 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1465 Release 3.10.0 (10 September 2014)
1466 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1468 3.10.0 is a feature release with many improvements and the usual
1469 collection of bug fixes.
1471 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1472 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1473 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1474 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1475 significantly improved relative to the 3.9.0 release.
1477 * ================== PLATFORM CHANGES =================
1479 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1480 is mostly complete, and is usable, but some SIMD instructions are as
1483 * Support for little-endian variant of the 64-bit POWER architecture.
1485 * Support for Android on MIPS32.
1487 * Support for 64bit FPU on MIPS32 platforms.
1489 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1491 * Configuration for and running on Android targets has changed.
1492 See README.android in the source tree for details.
1494 * ================== DEPRECATED FEATURES =================
1496 * --db-attach is now deprecated and will be removed in the next
1497 valgrind feature release. The built-in GDB server capabilities are
1498 superior and should be used instead. Learn more here:
1499 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1501 * ==================== TOOL CHANGES ====================
1505 - Client code can now selectively disable and re-enable reporting of
1506 invalid address errors in specific ranges using the new client
1507 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1508 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1510 - Leak checker: there is a new leak check heuristic called
1511 "length64". This is used to detect interior pointers pointing 8
1512 bytes inside a block, on the assumption that the first 8 bytes
1513 holds the value "block size - 8". This is used by
1514 sqlite3MemMalloc, for example.
1516 - Checking of system call parameters: if a syscall parameter
1517 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1518 several fields not initialised, an error is now reported for each
1519 field. Previously, an error was reported only for the first
1520 uninitialised field.
1522 - Mismatched alloc/free checking: a new flag
1523 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1524 such checks if necessary.
1528 - Improvements to error messages:
1530 o Race condition error message involving heap allocated blocks also
1531 show the thread number that allocated the raced-on block.
1533 o All locks referenced by an error message are now announced.
1534 Previously, some error messages only showed the lock addresses.
1536 o The message indicating where a lock was first observed now also
1537 describes the address/location of the lock.
1539 - Helgrind now understands the Ada task termination rules and
1540 creates a happens-before relationship between a terminated task
1541 and its master. This avoids some false positives and avoids a big
1542 memory leak when a lot of Ada tasks are created and terminated.
1543 The interceptions are only activated with forthcoming releases of
1544 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1546 - A new GDB server monitor command "info locks" giving the list of
1547 locks, their location, and their status.
1551 - callgrind_control now supports the --vgdb-prefix argument,
1552 which is needed if valgrind was started with this same argument.
1554 * ==================== OTHER CHANGES ====================
1556 * Unwinding through inlined function calls. Stack unwinding can now
1557 make use of Dwarf3 inlined-unwind information if it is available.
1558 The practical effect is that inlined calls become visible in stack
1559 traces. The suppression matching machinery has been adjusted
1560 accordingly. This is controlled by the new option
1561 --read-inline-info=yes|no. Currently this is enabled by default
1562 only on Linux and Android targets and only for the tools Memcheck,
1565 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1566 targets. If an object contains both CFI and EXIDX unwind
1567 information, Valgrind will prefer the CFI over the EXIDX. This
1568 facilitates unwinding through system libraries on arm-android
1571 * Address description logic has been improved and is now common
1572 between Memcheck and Helgrind, resulting in better address
1573 descriptions for some kinds of error messages.
1575 * Error messages about dubious arguments (eg, to malloc or calloc) are
1576 output like other errors. This means that they can be suppressed
1577 and they have a stack trace.
1579 * The C++ demangler has been updated for better C++11 support.
1581 * New and modified GDB server monitor features:
1583 - Thread local variables/storage (__thread) can now be displayed.
1585 - The GDB server monitor command "v.info location <address>"
1586 displays information about an address. The information produced
1587 depends on the tool and on the options given to valgrind.
1588 Possibly, the following are described: global variables, local
1589 (stack) variables, allocated or freed blocks, ...
1591 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1592 ask the GDB server to stop at the start of program execution, at
1593 the end of the program execution and on Valgrind internal errors.
1595 - A new monitor command "v.info stats" shows various Valgrind core
1596 and tool statistics.
1598 - A new monitor command "v.set hostvisibility" allows the GDB server
1599 to provide access to Valgrind internal host status/memory.
1601 * A new option "--aspace-minaddr=<address>" can in some situations
1602 allow the use of more memory by decreasing the address above which
1603 Valgrind maps memory. It can also be used to solve address
1604 conflicts with system libraries by increasing the default value.
1605 See user manual for details.
1607 * The amount of memory used by Valgrind to store debug info (unwind
1608 info, line number information and symbol data) has been
1609 significantly reduced, even though Valgrind now reads more
1610 information in order to support unwinding of inlined function calls.
1612 * Dwarf3 handling with --read-var-info=yes has been improved:
1614 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1616 - Code compiled with
1617 -ffunction-sections -fdata-sections -Wl,--gc-sections
1618 no longer causes assertion failures.
1620 * Improved checking for the --sim-hints= and --kernel-variant=
1621 options. Unknown strings are now detected and reported to the user
1624 * The semantics of stack start/end boundaries in the valgrind.h
1625 VALGRIND_STACK_REGISTER client request has been clarified and
1626 documented. The convention is that start and end are respectively
1627 the lowest and highest addressable bytes of the stack.
1629 * ==================== FIXED BUGS ====================
1631 The following bugs have been fixed or resolved. Note that "n-i-bz"
1632 stands for "not in bugzilla" -- that is, a bug that was reported to us
1633 but never got a bugzilla entry. We encourage you to file bugs in
1634 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1635 than mailing the developers (or mailing lists) directly -- bugs that
1636 are not entered into bugzilla tend to get forgotten about or ignored.
1638 To see details of a given bug, visit
1639 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1640 where XXXXXX is the bug number as listed below.
1642 175819 Support for ipv6 socket reporting with --track-fds
1643 232510 make distcheck fails
1644 249435 Analyzing wine programs with callgrind triggers a crash
1645 278972 support for inlined function calls in stacktraces and suppression
1647 291310 FXSAVE instruction marks memory as undefined on amd64
1648 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1649 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1650 315199 vgcore file for threaded app does not show which thread crashed
1651 315952 tun/tap ioctls are not supported
1652 323178 Unhandled instruction: PLDW register (ARM)
1653 323179 Unhandled instruction: PLDW immediate (ARM)
1654 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1655 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1656 325124 [MIPSEL] Compilation error
1657 325477 Phase 4 support for IBM Power ISA 2.07
1658 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1659 325628 Phase 5 support for IBM Power ISA 2.07
1660 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1661 325751 Missing the two privileged Power PC Transactional Memory Instructions
1662 325816 Phase 6 support for IBM Power ISA 2.07
1663 325856 Make SGCheck fail gracefully on unsupported platforms
1664 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1665 326436 DRD: False positive in libstdc++ std::list::push_back
1666 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1667 326462 Refactor vgdb to isolate invoker stuff into separate module
1668 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1669 326623 DRD: false positive conflict report in a field assignment
1670 326724 Valgrind does not compile on OSX 1.9 Mavericks
1671 326816 Intercept for __strncpy_sse2_unaligned missing?
1672 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1673 326983 Clear direction flag after tests on amd64.
1674 327212 Do not prepend the current directory to absolute path names.
1675 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1676 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1677 327284 s390x: Fix translation of the risbg instruction
1678 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1679 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1680 327916 DW_TAG_typedef may have no name
1681 327943 s390x: add a redirection for the 'index' function
1682 328100 XABORT not implemented
1683 328205 Implement additional Xen hypercalls
1684 328454 add support Backtraces with ARM unwind tables (EXIDX)
1685 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1686 328711 valgrind.1 manpage "memcheck options" section is badly generated
1687 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1688 329612 Incorrect handling of AT_BASE for image execution
1689 329694 clang warns about using uninitialized variable
1690 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1691 330228 mmap must align to VKI_SHMLBA on mips32
1692 330257 LLVM does not support `-mno-dynamic-no-pic` option
1693 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1694 330459 --track-fds=yes doesn't track eventfds
1695 330469 Add clock_adjtime syscall support
1696 330594 Missing sysalls on PowerPC / uClibc
1697 330622 Add test to regression suite for POWER instruction: dcbzl
1698 330939 Support for AMD's syscall instruction on x86
1700 330941 Typo in PRE(poll) syscall wrapper
1701 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
1702 331254 Fix expected output for memcheck/tests/dw4
1703 331255 Fix race condition in test none/tests/coolo_sigaction
1704 331257 Fix type of jump buffer in test none/tests/faultstatus
1705 331305 configure uses bash specific syntax
1706 331337 s390x WARNING: unhandled syscall: 326 (dup3)
1707 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
1708 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
1709 331829 Unexpected ioctl opcode sign extension
1710 331830 ppc64: WARNING: unhandled syscall: 96/97
1711 331839 drd/tests/sem_open specifies invalid semaphore name
1712 331847 outcome of drd/tests/thread_name is nondeterministic
1713 332037 Valgrind cannot handle Thumb "add pc, reg"
1714 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
1715 consistency checks enabled
1716 332263 intercepts for pthread_rwlock_timedrdlock and
1717 pthread_rwlock_timedwrlock are incorrect
1718 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
1720 332276 Implement additional Xen hypercalls
1721 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
1722 332765 Fix ms_print to create temporary files in a proper directory
1723 333072 drd: Add semaphore annotations
1724 333145 Tests for missaligned PC+#imm access for arm
1725 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
1726 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
1727 333248 WARNING: unhandled syscall: unix:443
1728 333428 ldr.w pc [rD, #imm] instruction leads to assertion
1729 333501 cachegrind: assertion: Cache set count is not a power of two.
1732 333666 Recognize MPX instructions and bnd prefix.
1733 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
1734 333817 Valgrind reports the memory areas written to by the SG_IO
1736 334049 lzcnt fails silently (x86_32)
1737 334384 Valgrind does not have support Little Endian support for
1739 334585 recvmmsg unhandled (+patch) (arm)
1740 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
1741 334727 Build fails with -Werror=format-security
1742 334788 clarify doc about --log-file initial program directory
1743 334834 PPC64 Little Endian support, patch 2
1744 334836 PPC64 Little Endian support, patch 3 testcase fixes
1745 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
1746 335034 Unhandled ioctl: HCIGETDEVLIST
1747 335155 vgdb, fix error print statement.
1748 335262 arm64: movi 8bit version is not supported
1749 335263 arm64: dmb instruction is not implemented
1750 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
1751 335496 arm64: sbc/abc instructions are not implemented
1752 335554 arm64: unhandled instruction: abs
1753 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
1754 335735 arm64: unhandled instruction: cnt
1755 335736 arm64: unhandled instruction: uaddlv
1756 335848 arm64: unhandled instruction: {s,u}cvtf
1757 335902 arm64: unhandled instruction: sli
1758 335903 arm64: unhandled instruction: umull (vector)
1759 336055 arm64: unhandled instruction: mov (element)
1760 336062 arm64: unhandled instruction: shrn{,2}
1761 336139 mip64: [...] valgrind hangs and spins on a single core [...]
1762 336189 arm64: unhandled Instruction: mvn
1763 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
1764 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
1765 336772 Make moans about unknown ioctls more informative
1766 336957 Add a section about the Solaris/illumos port on the webpage
1767 337094 ifunc wrapper is broken on ppc64
1768 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
1769 337528 leak check heuristic for block prefixed by length as 64bit number
1770 337740 Implement additional Xen hypercalls
1771 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
1772 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
1773 337871 deprecate --db-attach
1774 338023 Add support for all V4L2/media ioctls
1775 338024 inlined functions are not shown if DW_AT_ranges is used
1776 338106 Add support for 'kcmp' syscall
1777 338115 DRD: computed conflict set differs from actual after fork
1778 338160 implement display of thread local storage in gdbsrv
1779 338205 configure.ac and check for -Wno-tautological-compare
1780 338300 coredumps are missing one byte of every segment
1781 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
1782 338499 --sim-hints parsing broken due to wrong order in tokens
1783 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
1784 338681 Unable to unwind through clone thread created on i386-linux
1785 338698 race condition between gdbsrv and vgdb on startup
1786 338703 helgrind on arm-linux gets false positives in dynamic loader
1787 338791 alt dwz files can be relative of debug/main file
1788 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
1789 338932 build V-trunk with gcc-trunk
1790 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
1791 345079 Fix build problems in VEX/useful/test_main.c
1792 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
1793 n-i-bz s390x: Fix memory corruption for multithreaded applications
1794 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
1795 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
1796 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
1797 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
1798 n-i-bz Add command-line option --default-suppressions
1799 n-i-bz Add support for BLKDISCARDZEROES ioctl
1800 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
1801 n-i-bz Add support for sys_pivot_root and sys_unshare
1803 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
1804 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
1805 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
1809 Release 3.9.0 (31 October 2013)
1810 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1811 3.9.0 is a feature release with many improvements and the usual
1812 collection of bug fixes.
1814 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1815 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
1816 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
1817 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
1819 * ================== PLATFORM CHANGES =================
1821 * Support for MIPS64 LE and BE running Linux. Valgrind has been
1822 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
1824 * Support for MIPS DSP ASE on MIPS32 platforms.
1826 * Support for s390x Decimal Floating Point instructions on hosts that
1827 have the DFP facility installed.
1829 * Support for POWER8 (Power ISA 2.07) instructions
1831 * Support for Intel AVX2 instructions. This is available only on 64
1834 * Initial support for Intel Transactional Synchronization Extensions,
1837 * Initial support for Hardware Transactional Memory on POWER.
1839 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
1840 run large GUI apps tolerably well.
1842 * ==================== TOOL CHANGES ====================
1846 - Improvements in handling of vectorised code, leading to
1847 significantly fewer false error reports. You need to use the flag
1848 --partial-loads-ok=yes to get the benefits of these changes.
1850 - Better control over the leak checker. It is now possible to
1851 specify which leak kinds (definite/indirect/possible/reachable)
1852 should be displayed, which should be regarded as errors, and which
1853 should be suppressed by a given leak suppression. This is done
1854 using the options --show-leak-kinds=kind1,kind2,..,
1855 --errors-for-leak-kinds=kind1,kind2,.. and an optional
1856 "match-leak-kinds:" line in suppression entries, respectively.
1858 Note that generated leak suppressions contain this new line and
1859 are therefore more specific than in previous releases. To get the
1860 same behaviour as previous releases, remove the "match-leak-kinds:"
1861 line from generated suppressions before using them.
1863 - Reduced "possible leak" reports from the leak checker by the use
1864 of better heuristics. The available heuristics provide detection
1865 of valid interior pointers to std::stdstring, to new[] allocated
1866 arrays with elements having destructors and to interior pointers
1867 pointing to an inner part of a C++ object using multiple
1868 inheritance. They can be selected individually using the
1869 option --leak-check-heuristics=heur1,heur2,...
1871 - Better control of stacktrace acquisition for heap-allocated
1872 blocks. Using the --keep-stacktraces option, it is possible to
1873 control independently whether a stack trace is acquired for each
1874 allocation and deallocation. This can be used to create better
1875 "use after free" errors or to decrease Valgrind's resource
1876 consumption by recording less information.
1878 - Better reporting of leak suppression usage. The list of used
1879 suppressions (shown when the -v option is given) now shows, for
1880 each leak suppressions, how many blocks and bytes it suppressed
1881 during the last leak search.
1885 - False errors resulting from the use of statically initialised
1886 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
1889 - False errors resulting from the use of pthread_cond_waits that
1890 timeout, have been removed.
1892 * ==================== OTHER CHANGES ====================
1894 * Some attempt to tune Valgrind's space requirements to the expected
1895 capabilities of the target:
1897 - The default size of the translation cache has been reduced from 8
1898 sectors to 6 on Android platforms, since each sector occupies
1899 about 40MB when using Memcheck.
1901 - The default size of the translation cache has been increased to 16
1902 sectors on all other platforms, reflecting the fact that large
1903 applications require instrumentation and storage of huge amounts
1904 of code. For similar reasons, the number of memory mapped
1905 segments that can be tracked has been increased by a factor of 6.
1907 - In all cases, the maximum number of sectors in the translation
1908 cache can be controlled by the new flag --num-transtab-sectors.
1910 * Changes in how debug info (line numbers, etc) is read:
1912 - Valgrind no longer temporarily mmaps the entire object to read
1913 from it. Instead, reading is done through a small fixed sized
1914 buffer. This avoids virtual memory usage spikes when Valgrind
1915 reads debuginfo from large shared objects.
1917 - A new experimental remote debug info server. Valgrind can read
1918 debug info from a different machine (typically, a build host)
1919 where debuginfo objects are stored. This can save a lot of time
1920 and hassle when running Valgrind on resource-constrained targets
1921 (phones, tablets) when the full debuginfo objects are stored
1922 somewhere else. This is enabled by the --debuginfo-server=
1925 - Consistency checking between main and debug objects can be
1926 disabled using the --allow-mismatched-debuginfo option.
1928 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
1929 scanning can recover stack traces in some cases when the normal
1930 unwind mechanisms fail. Stack scanning is best described as "a
1931 nasty, dangerous and misleading hack" and so is disabled by default.
1932 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
1935 * Detection and merging of recursive stack frame cycles. When your
1936 program has recursive algorithms, this limits the memory used by
1937 Valgrind for recorded stack traces and avoids recording
1938 uninteresting repeated calls. This is controlled by the command
1939 line option --merge-recursive-frame and by the monitor command
1940 "v.set merge-recursive-frames".
1942 * File name and line numbers for used suppressions. The list of used
1943 suppressions (shown when the -v option is given) now shows, for each
1944 used suppression, the file name and line number where the suppression
1947 * New and modified GDB server monitor features:
1949 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
1950 that can be used to execute gdbserver monitor commands from the
1953 - A new monitor command, "v.info open_fds", that gives the list of
1954 open file descriptors and additional details.
1956 - An optional message in the "v.info n_errs_found" monitor command,
1957 for example "v.info n_errs_found test 1234 finished", allowing a
1958 comment string to be added to the process output, perhaps for the
1959 purpose of separating errors of different tests or test phases.
1961 - A new monitor command "v.info execontext" that shows information
1962 about the stack traces recorded by Valgrind.
1964 - A new monitor command "v.do expensive_sanity_check_general" to run
1965 some internal consistency checks.
1967 * New flag --sigill-diagnostics to control whether a diagnostic
1968 message is printed when the JIT encounters an instruction it can't
1969 translate. The actual behavior -- delivery of SIGILL to the
1970 application -- is unchanged.
1972 * The maximum amount of memory that Valgrind can use on 64 bit targets
1973 has been increased from 32GB to 64GB. This should make it possible
1974 to run applications on Memcheck that natively require up to about 35GB.
1976 * ==================== FIXED BUGS ====================
1978 The following bugs have been fixed or resolved. Note that "n-i-bz"
1979 stands for "not in bugzilla" -- that is, a bug that was reported to us
1980 but never got a bugzilla entry. We encourage you to file bugs in
1981 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1982 than mailing the developers (or mailing lists) directly -- bugs that
1983 are not entered into bugzilla tend to get forgotten about or ignored.
1985 To see details of a given bug, visit
1986 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1987 where XXXXXX is the bug number as listed below.
1989 123837 system call: 4th argument is optional, depending on cmd
1990 135425 memcheck should tell you where Freed blocks were Mallocd
1991 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
1992 207815 Adds some of the drm ioctls to syswrap-linux.c
1993 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
1994 252955 Impossible to compile with ccache
1995 253519 Memcheck reports auxv pointer accesses as invalid reads.
1996 263034 Crash when loading some PPC64 binaries
1997 269599 Increase deepest backtrace
1998 274695 s390x: Support "compare to/from logical" instructions (z196)
1999 275800 s390x: Autodetect cache info (part 2)
2000 280271 Valgrind reports possible memory leaks on still-reachable std::string
2001 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2002 289578 Backtraces with ARM unwind tables (stack scan flags)
2003 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2004 304832 ppc32: build failure
2005 305431 Use find_buildid shdr fallback for separate .debug files
2006 305728 Add support for AVX2 instructions
2007 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2008 306035 s390x: Fix IR generation for LAAG and friends
2009 306054 s390x: Condition code computation for convert-to-int/logical
2010 306098 s390x: alternate opcode form for convert to/from fixed
2011 306587 Fix cache line detection from auxiliary vector for PPC.
2012 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2013 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2014 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2015 307101 sys_capget second argument can be NULL
2016 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2017 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2018 307113 s390x: DFP support
2019 307141 valgrind does't work in mips-linux system
2020 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2021 307285 x86_amd64 feature test for avx in test suite is wrong
2022 307290 memcheck overlap testcase needs memcpy version filter
2023 307463 Please add "&limit=0" to the "all open bugs" link
2024 307465 --show-possibly-lost=no should reduce the error count / exit code
2025 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2026 307729 pkgconfig support broken valgrind.pc
2027 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2028 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2029 308089 Unhandled syscall on ppc64: prctl
2030 308135 PPC32 MPC8xx has 16 bytes cache size
2031 308321 testsuite memcheck filter interferes with gdb_filter
2033 308341 vgdb should report process exit (or fatal signal)
2034 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2035 308495 Remove build dependency on installed Xen headers
2036 308573 Internal error on 64-bit instruction executed in 32-bit mode
2038 308627 pmovmskb validity bit propagation is imprecise
2039 308644 vgdb command for having the info for the track-fds option
2040 308711 give more info about aspacemgr and arenas in out_of_memory
2041 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2042 308718 ARM implement SMLALBB family of instructions
2043 308886 Missing support for PTRACE_SET/GETREGSET
2044 308930 syscall name_to_handle_at (303 on amd64) not handled
2045 309229 V-bit tester does not report number of tests generated
2046 309323 print unrecognized instuction on MIPS
2047 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2048 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2049 309430 Self hosting ppc64 encounters a vassert error on operand type
2050 309600 valgrind is a bit confused about 0-sized sections
2051 309823 Generate errors for still reachable blocks
2052 309921 PCMPISTRI validity bit propagation is imprecise
2053 309922 none/tests/ppc64/test_dfp5 sometimes fails
2054 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2055 310424 --read-var-info does not properly describe static variables
2056 310792 search additional path for debug symbols
2057 310931 s390x: Message-security assist (MSA) instruction extension [..]
2058 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2059 311318 ARM: "128-bit constant is not implemented" error message
2060 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2061 311690 V crashes because it redirects branches inside of a redirected function
2062 311880 x86_64: make regtest hangs at shell_valid1
2063 311922 WARNING: unhandled syscall: 170
2065 312171 ppc: insn selection for DFP
2066 312571 Rounding mode call wrong for the DFP Iops [..]
2067 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2068 312913 Dangling pointers error should also report the alloc stack trace
2069 312980 Building on Mountain Lion generates some compiler warnings
2070 313267 Adding MIPS64/Linux port to Valgrind
2073 313811 Buffer overflow in assert_fail
2074 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2075 314269 ppc: dead code in insn selection
2076 314718 ARM: implement integer divide instruction (sdiv and udiv)
2077 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2078 315441 sendmsg syscall should ignore unset msghdr msg_flags
2079 315534 msgrcv inside a thread causes valgrind to hang (block)
2080 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2081 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2082 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2083 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2084 316144 valgrind.1 manpage contains unknown ??? strings [..]
2085 316145 callgrind command line options in manpage reference (unknown) [..]
2086 316145 callgrind command line options in manpage reference [..]
2087 316181 drd: Fixed a 4x slowdown for certain applications
2088 316503 Valgrind does not support SSE4 "movntdqa" instruction
2089 316535 Use of |signed int| instead of |size_t| in valgrind messages
2090 316696 fluidanimate program of parsec 2.1 stuck
2091 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2092 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2093 317186 "Impossible happens" when occurs VCVT instruction on ARM
2094 317318 Support for Threading Building Blocks "scalable_malloc"
2095 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2096 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2097 317463 bmi testcase IR SANITY CHECK FAILURE
2098 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2099 318050 libmpiwrap fails to compile with out-of-source build
2100 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2101 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2102 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2103 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2104 318932 Add missing PPC64 and PPC32 system call support
2105 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2106 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2107 319494 VEX Makefile-gcc standalone build update after r2702
2108 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2109 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2110 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2111 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2112 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2113 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2114 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2115 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2116 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2117 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2118 320895 add fanotify support (patch included)
2119 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2120 321065 Valgrind updates for Xen 4.3
2121 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2122 321363 Unhandled instruction: SSAX (ARM + Thumb)
2123 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2124 321466 Unhandled instruction: SHASX (ARM + Thumb)
2125 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2126 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2127 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2128 321620 Unhandled instruction: UASX (ARM + Thumb)
2129 321621 Unhandled instruction: USAX (ARM + Thumb)
2130 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2131 321693 Unhandled instruction: LDRSBT (Thumb)
2132 321694 Unhandled instruction: UQASX (ARM + Thumb)
2133 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2134 321697 Unhandled instruction: UHASX (ARM + Thumb)
2135 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2136 321704 Unhandled instruction: REVSH (ARM + Thumb)
2137 321730 Add cg_diff and cg_merge man pages
2138 321738 Add vgdb and valgrind-listener man pages
2140 321891 Unhandled instruction: LDRHT (Thumb)
2141 321960 pthread_create() then alloca() causing invalid stack write errors
2142 321969 ppc32 and ppc64 don't support [lf]setxattr
2143 322254 Show threadname together with tid if set by application
2144 322294 Add initial support for IBM Power ISA 2.07
2145 322368 Assertion failure in wqthread_hijack under OS X 10.8
2146 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2147 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2148 322851 0bXXX binary literal syntax is not standard
2149 323035 Unhandled instruction: LDRSHT(Thumb)
2150 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2151 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2152 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2153 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2154 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2155 323437 Phase 2 support for IBM Power ISA 2.07
2156 323713 Support mmxext (integer sse) subset on i386 (athlon)
2157 323803 Transactional memory instructions are not supported for Power
2158 323893 SSE3 not available on amd cpus in valgrind
2159 323905 Probable false positive from Valgrind/drd on close()
2160 323912 valgrind.h header isn't compatible for mingw64
2161 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2162 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2163 324181 mmap does not handle MAP_32BIT
2164 324227 memcheck false positive leak when a thread calls exit+block [..]
2165 324421 Support for fanotify API on ARM architecture
2166 324514 gdbserver monitor cmd output behaviour consistency [..]
2167 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2168 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2169 324582 When access is made to freed memory, report both allocation [..]
2170 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2171 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2172 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2173 324834 Unhandled instructions in Microsoft C run-time for x86_64
2174 324894 Phase 3 support for IBM Power ISA 2.07
2175 326091 drd: Avoid false race reports from optimized strlen() impls
2176 326113 valgrind libvex hwcaps error on AMD64
2177 n-i-bz Some wrong command line options could be ignored
2178 n-i-bz patch to allow fair-sched on android
2179 n-i-bz report error for vgdb snapshot requested before execution
2180 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2182 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2186 Release 3.8.1 (19 September 2012)
2187 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2188 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2189 that occur moderately frequently in real use cases, adds support for
2190 some missing instructions on ARM, and fixes a deadlock condition on
2191 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2192 want to consider upgrading to 3.8.1 instead.
2194 The following bugs have been fixed or resolved. Note that "n-i-bz"
2195 stands for "not in bugzilla" -- that is, a bug that was reported to us
2196 but never got a bugzilla entry. We encourage you to file bugs in
2197 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2198 than mailing the developers (or mailing lists) directly -- bugs that
2199 are not entered into bugzilla tend to get forgotten about or ignored.
2201 To see details of a given bug, visit
2202 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2203 where XXXXXX is the bug number as listed below.
2206 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2207 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2208 298281 wcslen causes false(?) uninitialised value warnings
2209 301281 valgrind hangs on OS X when the process calls system()
2210 304035 disInstr(arm): unhandled instruction 0xE1023053
2211 304867 implement MOVBE instruction in x86 mode
2212 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2213 305042 amd64: implement 0F 7F encoding of movq between two registers
2214 305199 ARM: implement QDADD and QDSUB
2215 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2216 305513 killed by fatal signal: SIGSEGV
2217 305690 DRD reporting invalid semaphore when sem_trywait fails
2218 305926 Invalid alignment checks for some AVX instructions
2219 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2220 306310 3.8.0 release tarball missing some files
2221 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2222 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2223 n-i-bz shmat of a segment > 4Gb does not work
2224 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2225 n-i-bz vgdb ptrace calls wrong on mips [...]
2226 n-i-bz Fixes for more MPI false positives
2227 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2228 n-i-bz OSX build w/ clang: asserts at startup
2229 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2230 n-i-bz fix a couple of union tag-vs-field mixups
2231 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2233 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2236 254088 Valgrind should know about UD2 instruction
2239 304754 NEWS blows TeX's little mind
2241 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2245 Release 3.8.0 (10 August 2012)
2246 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2247 3.8.0 is a feature release with many improvements and the usual
2248 collection of bug fixes.
2250 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2251 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2252 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2253 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2254 There is initial support for MacOSX 10.8, but it is not usable for
2255 serious work at present.
2257 * ================== PLATFORM CHANGES =================
2259 * Support for MIPS32 platforms running Linux. Valgrind has been
2260 tested on MIPS32 and MIPS32r2 platforms running different Debian
2261 Squeeze and MeeGo distributions. Both little-endian and big-endian
2262 cores are supported. The tools Memcheck, Massif and Lackey have
2263 been tested and are known to work. See README.mips for more details.
2265 * Preliminary support for Android running on x86.
2267 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2269 * Support for Intel AVX instructions and for AES instructions. This
2270 support is available only for 64 bit code.
2272 * Support for POWER Decimal Floating Point instructions.
2274 * ==================== TOOL CHANGES ====================
2276 * Non-libc malloc implementations are now supported. This is useful
2277 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2278 Using the new option --soname-synonyms, such tools can be informed
2279 that the malloc implementation is either linked statically into the
2280 executable, or is present in some other shared library different
2281 from libc.so. This makes it possible to process statically linked
2282 programs, and programs using other malloc libraries, for example
2283 TCMalloc or JEMalloc.
2285 * For tools that provide their own replacement for malloc et al, the
2286 option --redzone-size=<number> allows users to specify the size of
2287 the padding blocks (redzones) added before and after each client
2288 allocated block. Smaller redzones decrease the memory needed by
2289 Valgrind. Bigger redzones increase the chance to detect blocks
2290 overrun or underrun. Prior to this change, the redzone size was
2291 hardwired to 16 bytes in Memcheck.
2295 - The leak_check GDB server monitor command now can
2296 control the maximum nr of loss records to output.
2298 - Reduction of memory use for applications allocating
2299 many blocks and/or having many partially defined bytes.
2301 - Addition of GDB server monitor command 'block_list' that lists
2302 the addresses/sizes of the blocks of a leak search loss record.
2304 - Addition of GDB server monitor command 'who_points_at' that lists
2305 the locations pointing at a block.
2307 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2308 detect an invalid access of these redzones, by marking them
2309 noaccess. Similarly, if a redzone size is given for a memory
2310 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2311 This still allows to find some bugs if the user has forgotten to
2312 mark the pool superblock noaccess.
2314 - Performance of memory leak check has been improved, especially in
2315 cases where there are many leaked blocks and/or many suppression
2316 rules used to suppress leak reports.
2318 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2319 more precise analysis, which is important for LLVM/Clang
2320 generated code. This is at the cost of somewhat reduced
2321 performance. Note there is no change to analysis precision or
2322 costs on Linux targets.
2326 - Added even more facilities that can help finding the cause of a data
2327 race, namely the command-line option --ptrace-addr and the macro
2328 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2330 - Fixed a subtle bug that could cause false positive data race reports.
2332 * ==================== OTHER CHANGES ====================
2334 * The C++ demangler has been updated so as to work well with C++
2335 compiled by up to at least g++ 4.6.
2337 * Tool developers can make replacement/wrapping more flexible thanks
2338 to the new option --soname-synonyms. This was reported above, but
2339 in fact is very general and applies to all function
2340 replacement/wrapping, not just to malloc-family functions.
2342 * Round-robin scheduling of threads can be selected, using the new
2343 option --fair-sched= yes. Prior to this change, the pipe-based
2344 thread serialisation mechanism (which is still the default) could
2345 give very unfair scheduling. --fair-sched=yes improves
2346 responsiveness of interactive multithreaded applications, and
2347 improves repeatability of results from the thread checkers Helgrind
2350 * For tool developers: support to run Valgrind on Valgrind has been
2351 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2353 * gdbserver now shows the float shadow registers as integer
2354 rather than float values, as the shadow values are mostly
2355 used as bit patterns.
2357 * Increased limit for the --num-callers command line flag to 500.
2359 * Performance improvements for error matching when there are many
2360 suppression records in use.
2362 * Improved support for DWARF4 debugging information (bug 284184).
2364 * Initial support for DWZ compressed Dwarf debug info.
2366 * Improved control over the IR optimiser's handling of the tradeoff
2367 between performance and precision of exceptions. Specifically,
2368 --vex-iropt-precise-memory-exns has been removed and replaced by
2369 --vex-iropt-register-updates, with extended functionality. This
2370 allows the Valgrind gdbserver to always show up to date register
2373 * Modest performance gains through the use of translation chaining for
2376 * ==================== FIXED BUGS ====================
2378 The following bugs have been fixed or resolved. Note that "n-i-bz"
2379 stands for "not in bugzilla" -- that is, a bug that was reported to us
2380 but never got a bugzilla entry. We encourage you to file bugs in
2381 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2382 than mailing the developers (or mailing lists) directly -- bugs that
2383 are not entered into bugzilla tend to get forgotten about or ignored.
2385 To see details of a given bug, visit
2386 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2387 where XXXXXX is the bug number as listed below.
2389 197914 Building valgrind from svn now requires automake-1.10
2390 203877 increase to 16Mb maximum allowed alignment for memalign et al
2391 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2392 247386 make perf does not run all performance tests
2393 270006 Valgrind scheduler unfair
2394 270777 Adding MIPS/Linux port to Valgrind
2395 270796 s390x: Removed broken support for the TS insn
2396 271438 Fix configure for proper SSE4.2 detection
2397 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2398 273475 Add support for AVX instructions
2399 274078 improved configure logic for mpicc
2400 276993 fix mremap 'no thrash checks'
2401 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2402 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2403 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2404 283413 Fix wrong sanity check
2405 283671 Robustize alignment computation in LibVEX_Alloc
2406 283961 Adding support for some HCI IOCTLs
2407 284124 parse_type_DIE: confused by: DWARF 4
2408 284864 == 273475 (Add support for AVX instructions)
2409 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2410 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2411 285725 == 273475 (Add support for AVX instructions)
2412 286261 add wrapper for linux I2C_RDWR ioctl
2413 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2414 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2415 286384 configure fails "checking for a supported version of gcc"
2416 286497 == 273475 (Add support for AVX instructions)
2417 286596 == 273475 (Add support for AVX instructions)
2418 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2419 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2420 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2421 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2422 287307 == 273475 (Add support for AVX instructions)
2423 287858 VG_(strerror): unknown error
2424 288298 (MacOSX) unhandled syscall shm_unlink
2425 288995 == 273475 (Add support for AVX instructions)
2426 289470 Loading of large Mach-O thin binaries fails.
2427 289656 == 273475 (Add support for AVX instructions)
2428 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2429 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2430 289839 s390x: Provide support for unicode conversion instructions
2431 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2432 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2433 290655 Add support for AESKEYGENASSIST instruction
2434 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2435 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2436 291253 ES register not initialised in valgrind simulation
2437 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2438 291865 s390x: Support the "Compare Double and Swap" family of instructions
2439 292300 == 273475 (Add support for AVX instructions)
2440 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2441 292493 == 273475 (Add support for AVX instructions)
2442 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2443 292627 Missing support for some SCSI ioctls
2444 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2445 292841 == 273475 (Add support for AVX instructions)
2446 292993 implement the getcpu syscall on amd64-linux
2447 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2448 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2449 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2450 293754 PCMPxSTRx not implemented for 16-bit characters
2451 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2452 293808 CLFLUSH not supported by latest VEX for amd64
2453 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2454 294048 MPSADBW instruction not implemented
2455 294055 regtest none/tests/shell fails when locale is not set to C
2456 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2457 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2458 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2459 294260 disInstr_AMD64: disInstr miscalculated next %rip
2460 294523 --partial-loads-ok=yes causes false negatives
2461 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2462 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2463 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2464 295089 can not annotate source for both helgrind and drd
2465 295221 POWER Processor decimal floating point instruction support missing
2466 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2467 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2468 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2469 295617 ARM - Add some missing syscalls
2470 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2471 296229 Linux user input device ioctls missing wrappers
2472 296318 ELF Debug info improvements (more than one rx/rw mapping)
2473 296422 Add translation chaining support
2474 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2475 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2476 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2477 297078 gdbserver signal handling problems [..]
2478 297147 drd false positives on newly allocated memory
2479 297329 disallow decoding of IBM Power DFP insns on some machines
2480 297497 POWER Processor decimal floating point instruction support missing
2481 297701 Another alias for strncasecmp_l in libc-2.13.so
2482 297911 'invalid write' not reported when using APIs for custom mem allocators.
2483 297976 s390x: revisit EX implementation
2484 297991 Valgrind interferes with mmap()+ftell()
2485 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2486 297993 Fix compilation of valgrind with gcc -g3.
2487 298080 POWER Processor DFP support missing, part 3
2488 298227 == 273475 (Add support for AVX instructions)
2489 298335 == 273475 (Add support for AVX instructions)
2490 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2491 298394 s390x: Don't bail out on an unknown machine model. [..]
2492 298421 accept4() syscall (366) support is missing for ARM
2493 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2494 298732 valgrind installation problem in ubuntu with kernel version 3.x
2495 298862 POWER Processor DFP instruction support missing, part 4
2496 298864 DWARF reader mis-parses DW_FORM_ref_addr
2497 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2498 299053 Support DWARF4 DW_AT_high_pc constant form
2499 299104 == 273475 (Add support for AVX instructions)
2500 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2501 299629 dup3() syscall (358) support is missing for ARM
2502 299694 POWER Processor DFP instruction support missing, part 5
2503 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2504 299803 == 273475 (Add support for AVX instructions)
2505 299804 == 273475 (Add support for AVX instructions)
2506 299805 == 273475 (Add support for AVX instructions)
2507 300140 ARM - Missing (T1) SMMUL
2508 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2509 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2510 300414 FCOM and FCOMP unimplemented for amd64 guest
2511 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2512 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2513 301265 add x86 support to Android build
2514 301984 configure script doesn't detect certain versions of clang
2515 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2516 302287 Unhandled movbe instruction on Atom processors
2517 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2518 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2519 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2520 302656 == 273475 (Add support for AVX instructions)
2521 302709 valgrind for ARM needs extra tls support for android emulator [..]
2522 302827 add wrapper for CDROM_GET_CAPABILITY
2523 302901 Valgrind crashes with dwz optimized debuginfo
2524 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2525 303116 Add support for the POWER instruction popcntb
2526 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2527 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2528 303466 == 273475 (Add support for AVX instructions)
2529 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2530 303963 strstr() function produces wrong results under valgrind callgrind
2531 304054 CALL_FN_xx macros need to enforce stack alignment
2532 304561 tee system call not supported
2533 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2534 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2535 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2536 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2537 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2538 n-i-bz s390x: Shadow registers can now be examined using vgdb
2540 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2541 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2545 Release 3.7.0 (5 November 2011)
2546 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2547 3.7.0 is a feature release with many significant improvements and the
2548 usual collection of bug fixes.
2550 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2551 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2552 Support for recent distros and toolchain components (glibc 2.14, gcc
2553 4.6, MacOSX 10.7) has been added.
2555 * ================== PLATFORM CHANGES =================
2557 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2558 analyse 64-bit programs running on z/Architecture. Most user space
2559 instructions up to and including z10 are supported. Valgrind has
2560 been tested extensively on z9, z10, and z196 machines running SLES
2561 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2562 known to work well. Callgrind, Helgrind, and DRD work reasonably
2563 well on z9 and later models. See README.s390 for more details.
2565 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2566 64-bit processes are supported. Some complex threaded applications
2567 (Firefox) are observed to hang when run as 32 bit applications,
2568 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2569 will likely report some false errors. In general, expect some rough
2570 spots. This release also supports MacOSX 10.6, but drops support
2573 * Preliminary support for Android (on ARM). Valgrind can now run
2574 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2575 README.android for more details, plus instructions on how to get
2578 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2580 * General correctness and performance improvements for ARM/Linux, and,
2581 by extension, ARM/Android.
2583 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2584 instruction set support is under development but is not available in
2587 * Support for AIX5 has been removed.
2589 * ==================== TOOL CHANGES ====================
2591 * Memcheck: some incremental changes:
2593 - reduction of memory use in some circumstances
2595 - improved handling of freed memory, which in some circumstances
2596 can cause detection of use-after-free that would previously have
2599 - fix of a longstanding bug that could cause false negatives (missed
2600 errors) in programs doing vector saturated narrowing instructions.
2602 * Helgrind: performance improvements and major memory use reductions,
2603 particularly for large, long running applications which perform many
2604 synchronisation (lock, unlock, etc) events. Plus many smaller
2607 - display of locksets for both threads involved in a race
2609 - general improvements in formatting/clarity of error messages
2611 - addition of facilities and documentation regarding annotation
2612 of thread safe reference counted C++ classes
2614 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2615 on thread stacks (a performance hack)
2617 - new flag --free-is-write=no|yes [no], to enable detection of races
2618 where one thread accesses heap memory but another one frees it,
2619 without any coordinating synchronisation event
2621 * DRD: enabled XML output; added support for delayed thread deletion
2622 in order to detect races that occur close to the end of a thread
2623 (--join-list-vol); fixed a memory leak triggered by repeated client
2624 memory allocatation and deallocation; improved Darwin support.
2626 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2628 * exp-sgcheck: this tool has been reduced in scope so as to improve
2629 performance and remove checking that Memcheck does better.
2630 Specifically, the ability to check for overruns for stack and global
2631 arrays is unchanged, but the ability to check for overruns of heap
2632 blocks has been removed. The tool has accordingly been renamed to
2633 exp-sgcheck ("Stack and Global Array Checking").
2635 * ==================== OTHER CHANGES ====================
2637 * GDB server: Valgrind now has an embedded GDB server. That means it
2638 is possible to control a Valgrind run from GDB, doing all the usual
2639 things that GDB can do (single stepping, breakpoints, examining
2640 data, etc). Tool-specific functionality is also available. For
2641 example, it is possible to query the definedness state of variables
2642 or memory from within GDB when running Memcheck; arbitrarily large
2643 memory watchpoints are supported, etc. To use the GDB server, start
2644 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2647 * Improved support for unfriendly self-modifying code: a new option
2648 --smc-check=all-non-file is available. This adds the relevant
2649 consistency checks only to code that originates in non-file-backed
2650 mappings. In effect this confines the consistency checking only to
2651 code that is or might be JIT generated, and avoids checks on code
2652 that must have been compiled ahead of time. This significantly
2653 improves performance on applications that generate code at run time.
2655 * It is now possible to build a working Valgrind using Clang-2.9 on
2658 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2659 These enable and disable error reporting on a per-thread, and
2660 nestable, basis. This is useful for hiding errors in particularly
2661 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2662 now uses this facility.
2664 * Added the --mod-funcname option to cg_diff.
2666 * ==================== FIXED BUGS ====================
2668 The following bugs have been fixed or resolved. Note that "n-i-bz"
2669 stands for "not in bugzilla" -- that is, a bug that was reported to us
2670 but never got a bugzilla entry. We encourage you to file bugs in
2671 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2672 mailing the developers (or mailing lists) directly -- bugs that are
2673 not entered into bugzilla tend to get forgotten about or ignored.
2675 To see details of a given bug, visit
2676 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2677 where XXXXXX is the bug number as listed below.
2679 79311 malloc silly arg warning does not give stack trace
2680 210935 port valgrind.h (not valgrind) to win32 to support client requests
2681 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2682 243404 Port to zSeries
2683 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2684 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2685 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2686 253206 Some fixes for the faultstatus testcase
2687 255223 capget testcase fails when running as root
2688 256703 xlc_dbl_u32.c testcase broken
2689 256726 Helgrind tests have broken inline asm
2690 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2691 264800 testcase compile failure on zseries
2692 265762 make public VEX headers compilable by G++ 3.x
2693 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2694 266753 configure script does not give the user the option to not use QtCore
2695 266931 gen_insn_test.pl is broken
2696 266961 ld-linux.so.2 i?86-linux strlen issues
2697 266990 setns instruction causes false positive
2698 267020 Make directory for temporary files configurable at run-time.
2699 267342 == 267997 (segmentation fault on Mac OS 10.6)
2700 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
2701 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
2702 267488 regtest: darwin support for 64-bit build
2703 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
2704 267630 Add support for IBM Power ISA 2.06 -- stage 1
2705 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
2706 267819 Add client request for informing the core about reallocation
2707 267925 laog data structure quadratic for a single sequence of lock
2708 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
2709 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
2710 268513 missed optimizations in fold_Expr
2711 268619 s390x: fpr - gpr transfer facility
2712 268620 s390x: reconsider "long displacement" requirement
2713 268621 s390x: improve IR generation for XC
2714 268715 s390x: FLOGR is not universally available
2715 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
2716 268930 s390x: MHY is not universally available
2717 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
2718 269079 Support ptrace system call on ARM
2719 269144 missing "Bad option" error message
2720 269209 conditional load and store facility (z196)
2721 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
2722 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
2723 269736 s390x: minor code generation tweaks
2724 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
2725 269863 s390x: remove unused function parameters
2726 269864 s390x: tweak s390_emit_load_cc
2727 269884 == 250101 (overhead for huge blocks exhausts space too soon)
2728 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
2729 270115 s390x: rewrite some testcases
2730 270309 == 267997 (valgrind crash on startup)
2731 270320 add support for Linux FIOQSIZE ioctl() call
2732 270326 segfault while trying to sanitize the environment passed to execle
2733 270794 IBM POWER7 support patch causes regression in none/tests
2734 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
2735 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
2736 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
2737 270959 s390x: invalid use of R0 as base register
2738 271042 VSX configure check fails when it should not
2739 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
2740 271259 s390x: fix code confusion
2741 271337 == 267997 (Valgrind segfaults on MacOS X)
2742 271385 s390x: Implement Ist_MBE
2743 271501 s390x: misc cleanups
2744 271504 s390x: promote likely and unlikely
2745 271579 ppc: using wrong enum type
2746 271615 unhandled instruction "popcnt" (arch=amd10h)
2747 271730 Fix bug when checking ioctls: duplicate check
2748 271776 s390x: provide STFLE instruction support
2749 271779 s390x: provide clock instructions like STCK
2750 271799 Darwin: ioctls without an arg report a memory error
2751 271820 arm: fix type confusion
2752 271917 pthread_cond_timedwait failure leads to not-locked false positive
2753 272067 s390x: fix DISP20 macro
2754 272615 A typo in debug output in mc_leakcheck.c
2755 272661 callgrind_annotate chokes when run from paths containing regex chars
2756 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
2757 272955 Unhandled syscall error for pwrite64 on ppc64 arch
2758 272967 make documentation build-system more robust
2759 272986 Fix gcc-4.6 warnings with valgrind.h
2760 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
2761 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
2762 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
2763 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
2764 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
2765 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
2766 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
2767 273778 exp-ptrcheck: unhandled sysno == 259
2768 274089 exp-ptrcheck: unhandled sysno == 208
2769 274378 s390x: Various dispatcher tweaks
2770 274447 WARNING: unhandled syscall: 340
2771 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
2772 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
2773 274926 valgrind does not build against linux-3
2774 275148 configure FAIL with glibc-2.14
2775 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
2776 275168 Make Valgrind work for MacOSX 10.7 Lion
2777 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
2778 275278 valgrind does not build on Linux kernel 3.0.* due to silly
2779 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
2780 275308 Fix implementation for ppc64 fres instruc
2781 275339 s390x: fix testcase compile warnings
2782 275517 s390x: Provide support for CKSM instruction
2783 275710 s390x: get rid of redundant address mode calculation
2784 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
2785 275852 == 250101 (valgrind uses all swap space and is killed)
2786 276784 Add support for IBM Power ISA 2.06 -- stage 3
2787 276987 gdbsrv: fix tests following recent commits
2788 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
2789 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
2790 277471 Unhandled syscall: 340
2791 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
2792 277653 ARM: support Thumb2 PLD instruction
2793 277663 ARM: NEON float VMUL by scalar incorrect
2794 277689 ARM: tests for VSTn with register post-index are broken
2795 277694 ARM: BLX LR instruction broken in ARM mode
2796 277780 ARM: VMOV.F32 (immediate) instruction is broken
2797 278057 fuse filesystem syscall deadlocks
2798 278078 Unimplemented syscall 280 on ppc32
2799 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
2800 278454 VALGRIND_STACK_DEREGISTER has wrong output type
2801 278502 == 275284 (Valgrind confuses memcpy() and memmove())
2802 278892 gdbsrv: factorize gdb version handling, fix doc and typos
2803 279027 Support for MVCL and CLCL instruction
2804 279027 s390x: Provide support for CLCL and MVCL instructions
2805 279062 Remove a redundant check in the insn selector for ppc.
2806 279071 JDK creates PTEST with redundant REX.W prefix
2807 279212 gdbsrv: add monitor cmd v.info scheduler.
2808 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
2809 279698 memcheck discards valid-bits for packuswb
2810 279795 memcheck reports uninitialised values for mincore on amd64
2811 279994 Add support for IBM Power ISA 2.06 -- stage 3
2812 280083 mempolicy syscall check errors
2813 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
2814 280710 s390x: config files for nightly builds
2815 280757 /tmp dir still used by valgrind even if TMPDIR is specified
2816 280965 Valgrind breaks fcntl locks when program does mmap
2817 281138 WARNING: unhandled syscall: 340
2818 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
2819 281304 == 275168 (Darwin: dyld "cannot load inserted library")
2820 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
2821 281468 s390x: handle do_clone and gcc clones in call traces
2822 281488 ARM: VFP register corruption
2823 281828 == 275284 (false memmove warning: "Source and destination overlap")
2824 281883 s390x: Fix system call wrapper for "clone".
2825 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
2826 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
2827 282238 SLES10: make check fails
2828 282979 strcasestr needs replacement with recent(>=2.12) glibc
2829 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
2830 283243 Regression in ppc64 memcheck tests
2831 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
2832 283427 re-connect epoll_pwait syscall on ARM linux
2833 283600 gdbsrv: android: port vgdb.c
2834 283709 none/tests/faultstatus needs to account for page size
2835 284305 filter_gdb needs enhancement to work on ppc64
2836 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
2837 284472 Thumb2 ROR.W encoding T2 not implemented
2838 284621 XML-escape process command line in XML output
2839 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
2840 that have non-power-of-2 sizes (also AMDs)
2841 n-i-bz don't be spooked by libraries mashed by elfhack
2842 n-i-bz don't be spooked by libxul.so linked with gold
2843 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
2845 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
2846 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
2847 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
2851 Release 3.6.1 (16 February 2011)
2852 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2853 3.6.1 is a bug fix release. It adds support for some SSE4
2854 instructions that were omitted in 3.6.0 due to lack of time. Initial
2855 support for glibc-2.13 has been added. A number of bugs causing
2856 crashing or assertion failures have been fixed.
2858 The following bugs have been fixed or resolved. Note that "n-i-bz"
2859 stands for "not in bugzilla" -- that is, a bug that was reported to us
2860 but never got a bugzilla entry. We encourage you to file bugs in
2861 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2862 mailing the developers (or mailing lists) directly -- bugs that are
2863 not entered into bugzilla tend to get forgotten about or ignored.
2865 To see details of a given bug, visit
2866 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2867 where XXXXXX is the bug number as listed below.
2869 188572 Valgrind on Mac should suppress setenv() mem leak
2870 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
2871 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
2872 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2873 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
2874 254420 memory pool tracking broken
2875 254957 Test code failing to compile due to changes in memcheck.h
2876 255009 helgrind/drd: crash on chmod with invalid parameter
2877 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
2878 255355 helgrind/drd: crash on threaded programs doing fork
2880 255418 (SSE4.x) rint call compiled with ICC
2881 255822 --gen-suppressions can create invalid files: "too many callers [...]"
2882 255888 closing valgrindoutput tag outputted to log-stream on error
2883 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
2884 255966 Slowness when using mempool annotations
2885 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
2886 256600 super-optimized strcasecmp() false positive
2887 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
2888 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
2889 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
2890 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
2891 257276 Missing case in memcheck --track-origins=yes
2892 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
2893 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
2894 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
2895 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
2896 263099 callgrind_annotate counts Ir improperly [...]
2897 263877 undefined coprocessor instruction on ARMv7
2898 265964 configure FAIL with glibc-2.13
2899 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
2900 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
2901 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
2902 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
2903 n-i-bz DRD: disable --free-is-write due to implementation difficulties
2905 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
2909 Release 3.6.0 (21 October 2010)
2910 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2911 3.6.0 is a feature release with many significant improvements and the
2912 usual collection of bug fixes.
2914 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2915 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
2916 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
2918 -------------------------
2920 Here are some highlights. Details are shown further down:
2922 * Support for ARM/Linux.
2924 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
2926 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
2928 * Support for the SSE4.2 instruction set.
2930 * Enhancements to the Callgrind profiler, including the ability to
2931 handle CPUs with three levels of cache.
2933 * A new experimental heap profiler, DHAT.
2935 * A huge number of bug fixes and small enhancements.
2937 -------------------------
2939 Here are details of the above changes, together with descriptions of
2940 many other changes, and a list of fixed bugs.
2942 * ================== PLATFORM CHANGES =================
2944 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
2945 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
2946 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
2948 This requires a CPU capable of running the ARMv7-A instruction set
2949 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
2950 of the user space instruction set, including ARM and Thumb integer
2951 code, VFPv3, NEON and V6 media instructions. The Memcheck,
2952 Cachegrind and Massif tools work properly; other tools work to
2955 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
2956 with support for recent releases of the underlying toolchain
2957 components, notably gcc-4.5 and glibc-2.12.
2959 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
2960 support also works much better on OS X 10.5, and is as solid as
2963 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
2964 64-bit mode. In 32-bit mode, support is only available up to and
2965 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
2966 supported in 64-bit mode, and 32-bit mode does in fact support the
2967 bare minimum SSE4 instructions to needed to run programs on Mac OS X
2968 10.6 on 32-bit targets.
2970 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
2971 and including version 2.05 is supported.
2973 * ==================== TOOL CHANGES ====================
2975 * Cachegrind has a new processing script, cg_diff, which finds the
2976 difference between two profiles. It's very useful for evaluating
2977 the performance effects of a change in a program.
2979 Related to this change, the meaning of cg_annotate's (rarely-used)
2980 --threshold option has changed; this is unlikely to affect many
2981 people, if you do use it please see the user manual for details.
2983 * Callgrind now can do branch prediction simulation, similar to
2984 Cachegrind. In addition, it optionally can count the number of
2985 executed global bus events. Both can be used for a better
2986 approximation of a "Cycle Estimation" as derived event (you need to
2987 update the event formula in KCachegrind yourself).
2989 * Cachegrind and Callgrind now refer to the LL (last-level) cache
2990 rather than the L2 cache. This is to accommodate machines with
2991 three levels of caches -- if Cachegrind/Callgrind auto-detects the
2992 cache configuration of such a machine it will run the simulation as
2993 if the L2 cache isn't present. This means the results are less
2994 likely to match the true result for the machine, but
2995 Cachegrind/Callgrind's results are already only approximate, and
2996 should not be considered authoritative. The results are still
2997 useful for giving a general idea about a program's locality.
2999 * Massif has a new option, --pages-as-heap, which is disabled by
3000 default. When enabled, instead of tracking allocations at the level
3001 of heap blocks (as allocated with malloc/new/new[]), it instead
3002 tracks memory allocations at the level of memory pages (as mapped by
3003 mmap, brk, etc). Each mapped page is treated as its own block.
3004 Interpreting the page-level output is harder than the heap-level
3005 output, but this option is useful if you want to account for every
3006 byte of memory used by a program.
3008 * DRD has two new command-line options: --free-is-write and
3009 --trace-alloc. The former allows to detect reading from already freed
3010 memory, and the latter allows tracing of all memory allocations and
3013 * DRD has several new annotations. Custom barrier implementations can
3014 now be annotated, as well as benign races on static variables.
3016 * DRD's happens before / happens after annotations have been made more
3017 powerful, so that they can now also be used to annotate e.g. a smart
3018 pointer implementation.
3020 * Helgrind's annotation set has also been drastically improved, so as
3021 to provide to users a general set of annotations to describe locks,
3022 semaphores, barriers and condition variables. Annotations to
3023 describe thread-safe reference counted heap objects have also been
3026 * Memcheck has a new command-line option, --show-possibly-lost, which
3027 is enabled by default. When disabled, the leak detector will not
3028 show possibly-lost blocks.
3030 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3031 has been added. DHAT keeps track of allocated heap blocks, and also
3032 inspects every memory reference to see which block (if any) is being
3033 accessed. This gives a lot of insight into block lifetimes,
3034 utilisation, turnover, liveness, and the location of hot and cold
3035 fields. You can use DHAT to do hot-field profiling.
3037 * ==================== OTHER CHANGES ====================
3039 * Improved support for unfriendly self-modifying code: the extra
3040 overhead incurred by --smc-check=all has been reduced by
3041 approximately a factor of 5 as compared with 3.5.0.
3043 * Ability to show directory names for source files in error messages.
3044 This is combined with a flexible mechanism for specifying which
3045 parts of the paths should be shown. This is enabled by the new flag
3048 * A new flag, --require-text-symbol, which will stop the run if a
3049 specified symbol is not found it a given shared object when it is
3050 loaded into the process. This makes advanced working with function
3051 intercepting and wrapping safer and more reliable.
3053 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3054 and control of Valgrind is now available for the tools Memcheck and
3055 Helgrind. XML output from Valgrind is available for Memcheck,
3056 Helgrind and exp-Ptrcheck.
3058 * More reliable stack unwinding on amd64-linux, particularly in the
3059 presence of function wrappers, and with gcc-4.5 compiled code.
3061 * Modest scalability (performance improvements) for massive
3062 long-running applications, particularly for those with huge amounts
3065 * Support for analyzing programs running under Wine with has been
3066 improved. The header files <valgrind/valgrind.h>,
3067 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3068 Windows-programs compiled with MinGW or one of the Microsoft Visual
3071 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3072 The 32-bit simulator was not affected. This did not occur often,
3073 but when it did would usually crash the program under test.
3076 * A large number of bugs were fixed. These are shown below.
3078 * A number of bugs were investigated, and were candidates for fixing,
3079 but are not fixed in 3.6.0, due to lack of developer time. They may
3080 get fixed in later releases. They are:
3082 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3083 212419 false positive "lock order violated" (A+B vs A)
3084 213685 Undefined value propagates past dependency breaking instruction
3085 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3086 237920 valgrind segfault on fork failure
3087 242137 support for code compiled by LLVM-2.8
3088 242423 Another unknown Intel cache config value
3089 243232 Inconsistent Lock Orderings report with trylock
3090 243483 ppc: callgrind triggers VEX assertion failure
3091 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3092 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3094 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3095 249435 Analyzing wine programs with callgrind triggers a crash
3096 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3097 250065 Handling large allocations
3098 250101 huge "free" memory usage due to m_mallocfree.c
3099 "superblocks fragmentation"
3100 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3101 252091 Callgrind on ARM does not detect function returns correctly
3102 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3103 254420 memory pool tracking broken
3104 n-i-bz support for adding symbols for JIT generated code
3107 The following bugs have been fixed or resolved. Note that "n-i-bz"
3108 stands for "not in bugzilla" -- that is, a bug that was reported to us
3109 but never got a bugzilla entry. We encourage you to file bugs in
3110 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3111 mailing the developers (or mailing lists) directly -- bugs that are
3112 not entered into bugzilla tend to get forgotten about or ignored.
3114 To see details of a given bug, visit
3115 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3116 where XXXXXX is the bug number as listed below.
3118 135264 dcbzl instruction missing
3120 153699 Valgrind should report unaligned reads with movdqa
3122 190429 Valgrind reports lost of errors in ld.so
3123 with x86_64 2.9.90 glibc
3124 197266 valgrind appears to choke on the xmms instruction
3126 197988 Crash when demangling very large symbol names
3127 202315 unhandled syscall: 332 (inotify_init1)
3128 203256 Add page-level profiling to Massif
3129 205093 dsymutil=yes needs quotes, locking (partial fix)
3130 205241 Snow Leopard 10.6 support (partial fix)
3131 206600 Leak checker fails to upgrade indirect blocks when their
3132 parent becomes reachable
3133 210935 port valgrind.h (not valgrind) to win32 so apps run under
3134 wine can make client requests
3135 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3136 within Linux ip-stack checksum functions
3137 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3139 213685 Undefined value propagates past dependency breaking instruction
3141 215914 Valgrind inserts bogus empty environment variable
3143 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3144 222545 shmat fails under valgind on some arm targets
3145 222560 ARM NEON support
3148 232509 Docs build fails with formatting inside <title></title> elements
3150 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3151 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3152 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3153 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3154 237485 symlink (syscall 57) is not supported on Mac OS
3155 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3157 238208 is_just_below_ESP doesn't take into account red-zone
3158 238345 valgrind passes wrong $0 when executing a shell script
3159 238679 mq_timedreceive syscall doesn't flag the reception buffer
3161 238696 fcntl command F_DUPFD_CLOEXEC not supported
3162 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3163 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3164 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3166 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3172 242606 unhandled syscall: setegid (in Ptrcheck)
3173 242814 Helgrind "Impossible has happened" during
3174 QApplication::initInstance();
3175 243064 Valgrind attempting to read debug information from iso
3176 243270 Make stack unwinding in Valgrind wrappers more reliable
3177 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3178 sysno = 277 (mq_open)
3179 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3180 244493 ARM VFP d16-d31 registers support
3181 244670 add support for audit_session_self syscall on Mac OS 10.6
3182 244921 The xml report of helgrind tool is not well format
3183 244923 In the xml report file, the <preamble> not escape the
3184 xml char, eg '<','&','>'
3185 245535 print full path names in plain text reports
3186 245925 x86-64 red zone handling problem
3187 246258 Valgrind not catching integer underruns + new [] s
3188 246311 reg/reg cmpxchg doesn't work on amd64
3189 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3190 246888 Improve Makefile.vex.am
3191 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3192 to [f]chmod_extended
3193 247526 IBM POWER6 (ISA 2.05) support is incomplete
3194 247561 Some leak testcases fails due to reachable addresses in
3196 247875 sizeofIRType to handle Ity_I128
3197 247894 [PATCH] unhandled syscall sys_readahead
3198 247980 Doesn't honor CFLAGS passed to configure
3199 248373 darwin10.supp is empty in the trunk
3200 248822 Linux FIBMAP ioctl has int parameter instead of long
3201 248893 [PATCH] make readdwarf.c big endianess safe to enable
3202 unwinding on big endian systems
3203 249224 Syscall 336 not supported (SYS_proc_info)
3205 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3206 249943 jni JVM init fails when using valgrind
3207 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3209 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3210 250799 frexp$fenv_access_off function generates SIGILL
3211 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3212 251251 support pclmulqdq insn
3213 251362 valgrind: ARM: attach to debugger either fails or provokes
3215 251674 Unhandled syscall 294
3218 254257 Add support for debugfiles found by build-id
3219 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3220 254646 Wrapped functions cause stack misalignment on OS X
3221 (and possibly Linux)
3222 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3224 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3228 Release 3.5.0 (19 August 2009)
3229 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3230 3.5.0 is a feature release with many significant improvements and the
3231 usual collection of bug fixes. The main improvement is that Valgrind
3232 now works on Mac OS X.
3234 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3235 and X86/Darwin. Support for recent distros and toolchain components
3236 (glibc 2.10, gcc 4.5) has been added.
3238 -------------------------
3240 Here is a short summary of the changes. Details are shown further
3243 * Support for Mac OS X (10.5.x).
3245 * Improvements and simplifications to Memcheck's leak checker.
3247 * Clarification and simplifications in various aspects of Valgrind's
3250 * XML output for Helgrind and Ptrcheck.
3252 * Performance and stability improvements for Helgrind and DRD.
3254 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3256 * A new experimental tool, BBV, useful for computer architecture
3259 * Improved Wine support, including ability to read Windows PDB
3262 -------------------------
3264 Here are details of the above changes, followed by descriptions of
3265 many other minor changes, and a list of fixed bugs.
3268 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3269 called "Darwin" because that is the name of the OS core, which is the
3270 level that Valgrind works at.)
3274 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3275 because it would require work and 10.4 is only becoming less common.
3277 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3278 fairly well. For 10.5.x, 32-bit programs are the default even on
3279 64-bit machines, so it handles most current programs.
3281 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3282 officially supported, but simple programs at least will probably work.
3283 However, start-up is slow.
3285 - PowerPC machines are not supported.
3287 Things that don't work:
3289 - The Ptrcheck tool.
3291 - Objective-C garbage collection.
3295 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3296 Valgrind will fail with a SIGTRAP at start-up. See
3297 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3302 - You will likely find --dsymutil=yes a useful option, as error
3303 messages may be imprecise without it.
3305 - Mac OS X support is new and therefore will be less robust than the
3306 Linux support. Please report any bugs you find.
3308 - Threaded programs may run more slowly than on Linux.
3310 Many thanks to Greg Parker for developing this port over several years.
3313 * Memcheck's leak checker has been improved.
3315 - The results for --leak-check=summary now match the summary results
3316 for --leak-check=full. Previously they could differ because
3317 --leak-check=summary counted "indirectly lost" blocks and
3318 "suppressed" blocks as "definitely lost".
3320 - Blocks that are only reachable via at least one interior-pointer,
3321 but are directly pointed to by a start-pointer, were previously
3322 marked as "still reachable". They are now correctly marked as
3325 - The default value for the --leak-resolution option has been
3326 changed from "low" to "high". In general, this means that more
3327 leak reports will be produced, but each leak report will describe
3328 fewer leaked blocks.
3330 - With --leak-check=full, "definitely lost" and "possibly lost"
3331 leaks are now considered as proper errors, ie. they are counted
3332 for the "ERROR SUMMARY" and affect the behaviour of
3333 --error-exitcode. These leaks are not counted as errors if
3334 --leak-check=summary is specified, however.
3336 - Documentation for the leak checker has been improved.
3339 * Various aspects of Valgrind's text output have changed.
3341 - Valgrind's start-up message has changed. It is shorter but also
3342 includes the command being run, which makes it easier to use
3343 --trace-children=yes. An example:
3345 - Valgrind's shut-down messages have also changed. This is most
3346 noticeable with Memcheck, where the leak summary now occurs before
3347 the error summary. This change was necessary to allow leaks to be
3348 counted as proper errors (see the description of the leak checker
3349 changes above for more details). This was also necessary to fix a
3350 longstanding bug in which uses of suppressions against leaks were
3351 not "counted", leading to difficulties in maintaining suppression
3352 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3354 - Behavior of -v has changed. In previous versions, -v printed out
3355 a mixture of marginally-user-useful information, and tool/core
3356 statistics. The statistics printing has now been moved to its own
3357 flag, --stats=yes. This means -v is less verbose and more likely
3358 to convey useful end-user information.
3360 - The format of some (non-XML) stack trace entries has changed a
3361 little. Previously there were six possible forms:
3363 0x80483BF: really (a.c:20)
3364 0x80483BF: really (in /foo/a.out)
3366 0x80483BF: (within /foo/a.out)
3367 0x80483BF: ??? (a.c:20)
3370 The third and fourth of these forms have been made more consistent
3371 with the others. The six possible forms are now:
3373 0x80483BF: really (a.c:20)
3374 0x80483BF: really (in /foo/a.out)
3375 0x80483BF: really (in ???)
3376 0x80483BF: ??? (in /foo/a.out)
3377 0x80483BF: ??? (a.c:20)
3380 Stack traces produced when --xml=yes is specified are different
3384 * Helgrind and Ptrcheck now support XML output, so they can be used
3385 from GUI tools. Also, the XML output mechanism has been
3388 - The XML format has been overhauled and generalised, so it is more
3389 suitable for error reporting tools in general. The Memcheck
3390 specific aspects of it have been removed. The new format, which
3391 is an evolution of the old format, is described in
3392 docs/internals/xml-output-protocol4.txt.
3394 - Memcheck has been updated to use the new format.
3396 - Helgrind and Ptrcheck are now able to emit output in this format.
3398 - The XML output mechanism has been overhauled. XML is now output
3399 to its own file descriptor, which means that:
3401 * Valgrind can output text and XML independently.
3403 * The longstanding problem of XML output being corrupted by
3404 unexpected un-tagged text messages is solved.
3406 As before, the destination for text output is specified using
3407 --log-file=, --log-fd= or --log-socket=.
3409 As before, XML output for a tool is enabled using --xml=yes.
3411 Because there's a new XML output channel, the XML output
3412 destination is now specified by --xml-file=, --xml-fd= or
3415 Initial feedback has shown this causes some confusion. To
3416 clarify, the two envisaged usage scenarios are:
3418 (1) Normal text output. In this case, do not specify --xml=yes
3419 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3421 (2) XML output. In this case, specify --xml=yes, and one of
3422 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3423 destination, one of --log-file=, --log-fd= or --log-socket=
3424 to select the destination for any remaining text messages,
3425 and, importantly, -q.
3427 -q makes Valgrind completely silent on the text channel,
3428 except in the case of critical failures, such as Valgrind
3429 itself segfaulting, or failing to read debugging information.
3430 Hence, in this scenario, it suffices to check whether or not
3431 any output appeared on the text channel. If yes, then it is
3432 likely to be a critical error which should be brought to the
3433 attention of the user. If no (the text channel produced no
3434 output) then it can be assumed that the run was successful.
3436 This allows GUIs to make the critical distinction they need to
3437 make (did the run fail or not?) without having to search or
3438 filter the text output channel in any way.
3440 It is also recommended to use --child-silent-after-fork=yes in
3444 * Improvements and changes in Helgrind:
3446 - XML output, as described above
3448 - Checks for consistent association between pthread condition
3449 variables and their associated mutexes are now performed.
3451 - pthread_spinlock functions are supported.
3453 - Modest performance improvements.
3455 - Initial (skeletal) support for describing the behaviour of
3456 non-POSIX synchronisation objects through ThreadSanitizer
3457 compatible ANNOTATE_* macros.
3459 - More controllable tradeoffs between performance and the level of
3460 detail of "previous" accesses in a race. There are now three
3463 * --history-level=full. This is the default, and was also the
3464 default in 3.4.x. It shows both stacks involved in a race, but
3465 requires a lot of memory and can be very slow in programs that
3466 do many inter-thread synchronisation events.
3468 * --history-level=none. This only shows the later stack involved
3469 in a race. This can be much faster than --history-level=full,
3470 but makes it much more difficult to find the other access
3471 involved in the race.
3473 The new intermediate setting is
3475 * --history-level=approx
3477 For the earlier (other) access, two stacks are presented. The
3478 earlier access is guaranteed to be somewhere in between the two
3479 program points denoted by those stacks. This is not as useful
3480 as showing the exact stack for the previous access (as per
3481 --history-level=full), but it is better than nothing, and it's
3482 almost as fast as --history-level=none.
3485 * New features and improvements in DRD:
3487 - The error messages printed by DRD are now easier to interpret.
3488 Instead of using two different numbers to identify each thread
3489 (Valgrind thread ID and DRD thread ID), DRD does now identify
3490 threads via a single number (the DRD thread ID). Furthermore
3491 "first observed at" information is now printed for all error
3492 messages related to synchronization objects.
3494 - Added support for named semaphores (sem_open() and sem_close()).
3496 - Race conditions between pthread_barrier_wait() and
3497 pthread_barrier_destroy() calls are now reported.
3499 - Added support for custom allocators through the macros
3500 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3501 in <valgrind/valgrind.h>). An alternative for these two macros is
3502 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3505 - Added support for annotating non-POSIX synchronization objects
3506 through several new ANNOTATE_*() macros.
3508 - OpenMP: added support for the OpenMP runtime (libgomp) included
3509 with gcc versions 4.3.0 and 4.4.0.
3513 - Added two new command-line options (--first-race-only and
3514 --segment-merging-interval).
3517 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3519 Valgrind will now preserve (memory-access) atomicity of LOCK-
3520 prefixed x86/amd64 instructions, and any others implying a global
3521 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3523 This means that Valgrinded processes will "play nicely" in
3524 situations where communication with other processes, or the kernel,
3525 is done through shared memory and coordinated with such atomic
3526 instructions. Prior to this change, such arrangements usually
3527 resulted in hangs, races or other synchronisation failures, because
3528 Valgrind did not honour atomicity of such instructions.
3531 * A new experimental tool, BBV, has been added. BBV generates basic
3532 block vectors for use with the SimPoint analysis tool, which allows
3533 a program's overall behaviour to be approximated by running only a
3534 fraction of it. This is useful for computer architecture
3535 researchers. You can run BBV by specifying --tool=exp-bbv (the
3536 "exp-" prefix is short for "experimental"). BBV was written by
3540 * Modestly improved support for running Windows applications under
3541 Wine. In particular, initial support for reading Windows .PDB debug
3542 information has been added.
3545 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3546 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3550 * The Valgrind client requests VALGRIND_PRINTF and
3551 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3552 the string was always printed immediately on its own line. Now, the
3553 string will be added to a buffer but not printed until a newline is
3554 encountered, or other Valgrind output is printed (note that for
3555 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3556 "other Valgrind output"). This allows you to use multiple
3557 VALGRIND_PRINTF calls to build up a single output line, and also to
3558 print multiple output lines with a single request (by embedding
3559 multiple newlines in the string).
3562 * The graphs drawn by Massif's ms_print program have changed slightly:
3564 - The half-height chars '.' and ',' are no longer drawn, because
3565 they are confusing. The --y option can be used if the default
3566 y-resolution is not high enough.
3568 - Horizontal lines are now drawn after the top of a snapshot if
3569 there is a gap until the next snapshot. This makes it clear that
3570 the memory usage has not dropped to zero between snapshots.
3573 * Something that happened in 3.4.0, but wasn't clearly announced: the
3574 option --read-var-info=yes can be used by some tools (Memcheck,
3575 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3576 variable type and location information. This makes those tools
3577 start up more slowly and increases memory consumption, but
3578 descriptions of data addresses in error messages become more
3582 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3583 disabled in 3.4.0 due to a lack of interest and maintenance,
3584 although the source code was still in the distribution. The source
3585 code has now been removed from the distribution. For anyone
3586 interested, the removal occurred in SVN revision r10247.
3589 * Some changes have been made to the build system.
3591 - VEX/ is now integrated properly into the build system. This means
3592 that dependency tracking within VEX/ now works properly, "make
3593 install" will work without requiring "make" before it, and
3594 parallel builds (ie. 'make -j') now work (previously a
3595 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3596 was effectively ignored).
3598 - The --with-vex configure option has been removed. It was of
3599 little use and removing it simplified the build system.
3601 - The location of some install files has changed. This should not
3602 affect most users. Those who might be affected:
3604 * For people who use Valgrind with MPI programs, the installed
3605 libmpiwrap.so library has moved from
3606 $(INSTALL)/<platform>/libmpiwrap.so to
3607 $(INSTALL)/libmpiwrap-<platform>.so.
3609 * For people who distribute standalone Valgrind tools, the
3610 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3611 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3613 These changes simplify the build system.
3615 - Previously, all the distributed suppression (*.supp) files were
3616 installed. Now, only default.supp is installed. This should not
3617 affect users as the other installed suppression files were not
3618 read; the fact that they were installed was a mistake.
3621 * KNOWN LIMITATIONS:
3623 - Memcheck is unusable with the Intel compiler suite version 11.1,
3624 when it generates code for SSE2-and-above capable targets. This
3625 is because of icc's use of highly optimised inlined strlen
3626 implementations. It causes Memcheck to report huge numbers of
3627 false errors even in simple programs. Helgrind and DRD may also
3630 Versions 11.0 and earlier may be OK, but this has not been
3634 The following bugs have been fixed or resolved. Note that "n-i-bz"
3635 stands for "not in bugzilla" -- that is, a bug that was reported to us
3636 but never got a bugzilla entry. We encourage you to file bugs in
3637 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3638 mailing the developers (or mailing lists) directly -- bugs that are
3639 not entered into bugzilla tend to get forgotten about or ignored.
3641 To see details of a given bug, visit
3642 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3643 where XXXXXX is the bug number as listed below.
3645 84303 How about a LockCheck tool?
3646 91633 dereference of null ptr in vgPlain_st_basetype
3647 97452 Valgrind doesn't report any pthreads problems
3648 100628 leak-check gets assertion failure when using
3649 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3650 108528 NPTL pthread cleanup handlers not called
3651 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3652 110128 mallinfo is not implemented...
3653 110770 VEX: Generated files not always updated when making valgrind
3654 111102 Memcheck: problems with large (memory footprint) applications
3655 115673 Vex's decoder should never assert
3656 117564 False positive: Syscall param clone(child_tidptr) contains
3657 uninitialised byte(s)
3658 119404 executing ssh from inside valgrind fails
3659 133679 Callgrind does not write path names to sources with dwarf debug
3661 135847 configure.in problem with non gnu compilers (and possible fix)
3662 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3663 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3664 136230 memcheck reports "possibly lost", should be "still reachable"
3665 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3666 137904 Valgrind reports a memory leak when using POSIX threads,
3668 139076 valgrind VT_GETSTATE error
3669 142228 complaint of elf_dynamic_do_rela in trivial usage
3670 145347 spurious warning with USBDEVFS_REAPURB
3671 148441 (wine) can't find memory leak in Wine, win32 binary
3673 148742 Leak-check fails assert on exit
3674 149878 add (proper) check for calloc integer overflow
3675 150606 Call graph is broken when using callgrind control
3676 152393 leak errors produce an exit code of 0. I need some way to
3677 cause leak errors to result in a nonzero exit code.
3678 157154 documentation (leak-resolution doc speaks about num-callers
3679 def=4) + what is a loss record
3680 159501 incorrect handling of ALSA ioctls
3681 162020 Valgrinding an empty/zero-byte file crashes valgrind
3682 162482 ppc: Valgrind crashes while reading stabs information
3683 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3684 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3685 163560 VEX/test_main.c is missing from valgrind-3.3.1
3686 164353 malloc_usable_size() doesn't return a usable size
3687 165468 Inconsistent formatting in memcheck manual -- please fix
3688 169505 main.c:286 (endOfInstr):
3689 Assertion 'ii->cost_offset == *cost_offset' failed
3690 177206 Generate default.supp during compile instead of configure
3691 177209 Configure valt_load_address based on arch+os
3692 177305 eventfd / syscall 323 patch lost
3693 179731 Tests fail to build because of inlining of non-local asm labels
3694 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3695 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3696 181594 Bogus warning for empty text segment
3697 181707 dwarf doesn't require enumerations to have name
3698 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3699 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3700 Assertion '!already_present' failed.
3701 185359 exp-ptrcheck: unhandled syscall getresuid()
3702 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
3703 185816 Valgrind is unable to handle debug info for files with split
3704 debug info that are prelinked afterwards
3705 185980 [darwin] unhandled syscall: sem_open
3706 186238 bbToIR_AMD64: disInstr miscalculated next %rip
3707 186507 exp-ptrcheck unhandled syscalls prctl, etc.
3708 186790 Suppression pattern used for leaks are not reported
3709 186796 Symbols with length>200 in suppression files are ignored
3710 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
3711 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
3712 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
3713 188046 bashisms in the configure script
3714 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
3715 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
3716 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
3717 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
3718 assertion fail "!lock->heldBy"
3719 188427 Add support for epoll_create1 (with patch)
3720 188530 Support for SIOCGSTAMPNS
3721 188560 Include valgrind.spec in the tarball
3722 188572 Valgrind on Mac should suppress setenv() mem leak
3723 189054 Valgrind fails to build because of duplicate non-local asm labels
3724 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
3725 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
3726 189763 drd assertion failure: s_threadinfo[tid].is_recording
3727 190219 unhandled syscall: 328 (x86-linux)
3728 190391 dup of 181394; see above
3729 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
3730 190820 No debug information on powerpc-linux
3731 191095 PATCH: Improve usbdevfs ioctl handling
3732 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
3734 191189 --xml=yes should obey --gen-suppressions=all
3735 191192 syslog() needs a suppression on macosx
3736 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
3737 191761 getrlimit on MacOSX
3738 191992 multiple --fn-skip only works sometimes; dependent on order
3739 192634 V. reports "aspacem sync_check_mapping_callback:
3740 segment mismatch" on Darwin
3741 192954 __extension__ missing on 2 client requests
3742 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
3743 194474 "INSTALL" file has different build instructions than "README"
3744 194671 Unhandled syscall (sem_wait?) from mac valgrind
3745 195069 memcheck: reports leak (memory still reachable) for
3747 195169 drd: (vgDrd_barrier_post_wait):
3748 Assertion 'r->sg[p->post_iteration]' failed.
3749 195268 valgrind --log-file doesn't accept ~/...
3750 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
3751 195860 WARNING: unhandled syscall: unix:223
3752 196528 need a error suppression for pthread_rwlock_init under os x?
3753 197227 Support aio_* syscalls on Darwin
3754 197456 valgrind should reject --suppressions=(directory)
3755 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
3756 197591 unhandled syscall 27 (mincore)
3757 197793 Merge DCAS branch to the trunk == 85756, 142103
3758 197794 Avoid duplicate filenames in Vex
3759 197898 make check fails on current SVN
3760 197901 make check fails also under exp-ptrcheck in current SVN
3761 197929 Make --leak-resolution=high the default
3762 197930 Reduce spacing between leak reports
3763 197933 Print command line of client at start-up, and shorten preamble
3764 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
3765 198395 add BBV to the distribution as an experimental tool
3766 198624 Missing syscalls on Darwin: 82, 167, 281, 347
3767 198649 callgrind_annotate doesn't cumulate counters
3768 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
3769 199977 Valgrind complains about an unrecognized instruction in the
3770 atomic_incs test program
3771 200029 valgrind isn't able to read Fedora 12 debuginfo
3772 200760 darwin unhandled syscall: unix:284
3773 200827 DRD doesn't work on Mac OS X
3774 200990 VG_(read_millisecond_timer)() does not work correctly
3775 201016 Valgrind does not support pthread_kill() on Mac OS
3776 201169 Document --read-var-info
3777 201323 Pre-3.5.0 performance sanity checking
3778 201384 Review user manual for the 3.5.0 release
3779 201585 mfpvr not implemented on ppc
3780 201708 tests failing because x86 direction flag is left set
3781 201757 Valgrind doesn't handle any recent sys_futex additions
3782 204377 64-bit valgrind can not start a shell script
3783 (with #!/path/to/shell) if the shell is a 32-bit executable
3784 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
3785 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
3786 about memory allocation events with memory access tracing enabled
3787 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
3789 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
3793 Release 3.4.1 (28 February 2009)
3794 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3795 3.4.1 is a bug-fix release that fixes some regressions and assertion
3796 failures in debug info reading in 3.4.0, most notably incorrect stack
3797 traces on amd64-linux on older (glibc-2.3 based) systems. Various
3798 other debug info problems are also fixed. A number of bugs in the
3799 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
3801 In view of the fact that 3.4.0 contains user-visible regressions
3802 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
3803 encouraged to ship 3.4.1 in preference to 3.4.0.
3805 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
3806 bugzilla" -- that is, a bug that was reported to us but never got a
3807 bugzilla entry. We encourage you to file bugs in bugzilla
3808 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
3809 developers (or mailing lists) directly -- bugs that are not entered
3810 into bugzilla tend to get forgotten about or ignored.
3812 n-i-bz Fix various bugs reading icc-11 generated debug info
3813 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
3814 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
3815 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
3816 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
3817 179618 exp-ptrcheck crashed / exit prematurely
3818 179624 helgrind: false positive races with pthread_create and
3819 recv/open/close/read
3820 134207 pkg-config output contains @VG_PLATFORM@
3821 176926 floating point exception at valgrind startup with PPC 440EPX
3822 181594 Bogus warning for empty text segment
3823 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
3824 181707 Dwarf3 doesn't require enumerations to have name
3825 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3826 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3827 Assertion '!already_present' failed.
3828 185359 exp-ptrcheck unhandled syscall getresuid()
3830 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
3831 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
3835 Release 3.4.0 (2 January 2009)
3836 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3837 3.4.0 is a feature release with many significant improvements and the
3838 usual collection of bug fixes. This release supports X86/Linux,
3839 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
3840 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
3842 3.4.0 brings some significant tool improvements. Memcheck can now
3843 report the origin of uninitialised values, the thread checkers
3844 Helgrind and DRD are much improved, and we have a new experimental
3845 tool, exp-Ptrcheck, which is able to detect overruns of stack and
3846 global arrays. In detail:
3848 * Memcheck is now able to track the origin of uninitialised values.
3849 When it reports an uninitialised value error, it will try to show
3850 the origin of the value, as either a heap or stack allocation.
3851 Origin tracking is expensive and so is not enabled by default. To
3852 use it, specify --track-origins=yes. Memcheck's speed will be
3853 essentially halved, and memory usage will be significantly
3854 increased. Nevertheless it can drastically reduce the effort
3855 required to identify the root cause of uninitialised value errors,
3856 and so is often a programmer productivity win, despite running more
3859 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
3860 3.4.0, will be released shortly.
3862 * Helgrind's race detection algorithm has been completely redesigned
3863 and reimplemented, to address usability and scalability concerns:
3865 - The new algorithm has a lower false-error rate: it is much less
3866 likely to report races that do not really exist.
3868 - Helgrind will display full call stacks for both accesses involved
3869 in a race. This makes it easier to identify the root causes of
3872 - Limitations on the size of program that can run have been removed.
3874 - Performance has been modestly improved, although that is very
3877 - Direct support for Qt4 threading has been added.
3879 - pthread_barriers are now directly supported.
3881 - Helgrind works well on all supported Linux targets.
3883 * The DRD thread debugging tool has seen major improvements:
3885 - Greatly improved performance and significantly reduced memory
3888 - Support for several major threading libraries (Boost.Thread, Qt4,
3889 glib, OpenMP) has been added.
3891 - Support for atomic instructions, POSIX semaphores, barriers and
3892 reader-writer locks has been added.
3894 - Works now on PowerPC CPUs too.
3896 - Added support for printing thread stack usage at thread exit time.
3898 - Added support for debugging lock contention.
3900 - Added a manual for Drd.
3902 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
3903 checks for misuses of pointers. In that sense it is a bit like
3904 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
3905 detect overruns of stack and global arrays, it can detect
3906 arbitrarily far out-of-bounds accesses to heap blocks, and it can
3907 detect accesses heap blocks that have been freed a very long time
3908 ago (millions of blocks in the past).
3910 Ptrcheck currently works only on x86-linux and amd64-linux. To use
3911 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
3912 of the main Valgrind documentation. As this is an experimental
3913 tool, we would be particularly interested in hearing about your
3914 experiences with it.
3916 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
3917 longer built by default, although the code remains in the repository
3918 and the tarball. This is due to three factors: a perceived lack of
3919 users, a lack of maintenance, and concerns that it may not be
3920 possible to achieve reliable operation using the existing design.
3922 * As usual, support for the latest Linux distros and toolchain
3923 components has been added. It should work well on Fedora Core 10,
3924 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
3925 state) is supported, as is glibc-2.9. The C++ demangler has been
3926 updated so as to work well with C++ compiled by even the most recent
3929 * You can now use frame-level wildcards in suppressions. This was a
3930 frequently-requested enhancement. A line "..." in a suppression now
3931 matches zero or more frames. This makes it easier to write
3932 suppressions which are precise yet insensitive to changes in
3935 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
3937 * Very basic support for IBM Power6 has been added (64-bit processes only).
3939 * Valgrind is now cross-compilable. For example, it is possible to
3940 cross compile Valgrind on an x86/amd64-linux host, so that it runs
3941 on a ppc32/64-linux target.
3943 * You can set the main thread's stack size at startup using the
3944 new --main-stacksize= flag (subject of course to ulimit settings).
3945 This is useful for running apps that need a lot of stack space.
3947 * The limitation that you can't use --trace-children=yes together
3948 with --db-attach=yes has been removed.
3950 * The following bugs have been fixed. Note that "n-i-bz" stands for
3951 "not in bugzilla" -- that is, a bug that was reported to us but
3952 never got a bugzilla entry. We encourage you to file bugs in
3953 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3954 mailing the developers (or mailing lists) directly.
3956 n-i-bz Make return types for some client requests 64-bit clean
3957 n-i-bz glibc 2.9 support
3958 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
3959 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
3960 n-i-bz Building in an env without gdb gives bogus gdb attach
3961 92456 Tracing the origin of uninitialised memory
3962 106497 Valgrind does not demangle some C++ template symbols
3964 151612 Suppression with "..." (frame-level wildcards in .supp files)
3965 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
3966 159285 unhandled syscall:25 (stime, on x86-linux)
3967 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
3968 160954 ppc build of valgrind crashes with illegal instruction (isel)
3969 160956 mallinfo implementation, w/ patch
3970 162092 Valgrind fails to start gnome-system-monitor
3971 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
3972 163794 assertion failure with "--track-origins=yes"
3973 163933 sigcontext.err and .trapno must be set together
3974 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
3975 164476 Missing kernel module loading system calls
3976 164669 SVN regression: mmap() drops posix file locks
3977 166581 Callgrind output corruption when program forks
3978 167288 Patch file for missing system calls on Cell BE
3979 168943 unsupported scas instruction pentium
3980 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
3981 172417 x86->IR: 0x82 ...
3982 172563 amd64->IR: 0xD9 0xF5 - fprem1
3983 173099 .lds linker script generation error
3984 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
3985 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
3987 174908 --log-file value not expanded correctly for core file
3988 175044 Add lookup_dcookie for amd64
3989 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
3991 Developer-visible changes:
3993 * Valgrind's debug-info reading machinery has been majorly overhauled.
3994 It can now correctly establish the addresses for ELF data symbols,
3995 which is something that has never worked properly before now.
3997 Also, Valgrind can now read DWARF3 type and location information for
3998 stack and global variables. This makes it possible to use the
3999 framework to build tools that rely on knowing the type and locations
4000 of stack and global variables, for example exp-Ptrcheck.
4002 Reading of such information is disabled by default, because most
4003 tools don't need it, and because it is expensive in space and time.
4004 However, you can force Valgrind to read it, using the
4005 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4006 make use of such information, if present, to provide source-level
4007 descriptions of data addresses in the error messages they create.
4009 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4010 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).