1Release 3.15.0 (?? ??????? 2019) 2~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 3 43.15.0 is a feature release with many improvements and the usual collection of 5bug fixes. 6 7This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux, 8PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux, 9MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android, 10X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary 11support for X86/macOS 10.13, AMD64/macOS 10.13. 12 13* ==================== CORE CHANGES =================== 14 15* The XTree Massif output format now makes use of the information obtained 16 when specifying --read-inline-info=yes. 17 18* ================== PLATFORM CHANGES ================= 19 20 21* ==================== TOOL CHANGES ==================== 22 23* Cachegrind: 24 25 - cg_annotate has a new option, --show-percs, which prints percentages next 26 to all event counts. 27 28* Callgrind: 29 30 - callgrind_annotate has a new option, --show-percs, which prints percentages 31 next to all event counts. 32 33 - callgrind_annotate now inserts commas in call counts, and 34 sort the caller/callee lists in the call tree. 35 36* Massif: 37 - The default value for --read-inline-info is now "yes" on 38 Linux/Android/Solaris. It is still "no" on other OS. 39 40* Memcheck: 41 - The option --xtree-leak=yes (to output leak result in xtree format) 42 automatically activates the option --show-leak-kinds=all, 43 as xtree visualisation tools such as kcachegrind can in any case 44 select what kind of leak to visualise. 45 46* ==================== OTHER CHANGES ==================== 47 48* The new option --show-error-list=no|yes 49 allows to show at exit the list of detected errors and the used 50 suppressions. In the previous valgrind version, showing this 51 information could only be done using a verbosity >= 2, which is 52 producing a lot of other non necessarily useful messages. 53 The option -s is equivalent to --show-error-list=yes. 54 55* ==================== FIXED BUGS ==================== 56 57The following bugs have been fixed or resolved. Note that "n-i-bz" 58stands for "not in bugzilla" -- that is, a bug that was reported to us 59but never got a bugzilla entry. We encourage you to file bugs in 60bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 61than mailing the developers (or mailing lists) directly -- bugs that 62are not entered into bugzilla tend to get forgotten about or ignored. 63 64To see details of a given bug, visit 65 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 66where XXXXXX is the bug number as listed below. 67 68399301 Use inlined frames in Massif XTree output. 69399322 Improve callgrind_annotate output 70400490 s390x: VRs allocated as if separate from FPRs 71400491 s390x: Operand of LOCH treated as unsigned integer 72397187 z13 vector register support for vgdb gdbserver 73401277 More bugs in z13 support 74401112 LLVM 5.0 generates comparison against partially initialized data 75385411 s390x: z13 vector floating-point instructions not implemented 76401578 drd: crashes sometimes on fork() 77401627 memcheck errors with glibc avx2 optimized wcsncmp 78401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings 79402006 mark helper regs defined in final_tidyup before freeres_wrapper call 80402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace) 81402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64 82402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop) 83402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking 84402480 Do not use %rsp in clobber list 85402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64 86402515 Implement new option --show-error-list=no|yes / -s 87 88Release 3.14.0 (9 October 2018) 89~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 90 913.14.0 is a feature release with many improvements and the usual collection of 92bug fixes. 93 94This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux, 95PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux, 96MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android, 97X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary 98support for X86/macOS 10.13, AMD64/macOS 10.13. 99 100* ==================== CORE CHANGES =================== 101 102* The new option --keep-debuginfo=no|yes (default no) can be used to retain 103 debug info for unloaded code. This allows saved stack traces (e.g. for 104 memory leaks) to include file/line info for code that has been dlclose'd (or 105 similar). See the user manual for more information and known limitations. 106 107* Ability to specify suppressions based on source file name and line number. 108 109* Majorly overhauled register allocator. No end-user changes, but the JIT 110 generates code a bit more quickly now. 111 112* ================== PLATFORM CHANGES ================= 113 114* Preliminary support for macOS 10.13 has been added. 115 116* mips: support for MIPS32/MIPS64 Revision 6 has been added. 117 118* mips: support for MIPS SIMD architecture (MSA) has been added. 119 120* mips: support for MIPS N32 ABI has been added. 121 122* s390: partial support for vector instructions (integer and string) has been 123 added. 124 125* ==================== TOOL CHANGES ==================== 126 127* Helgrind: Addition of a flag 128 --delta-stacktrace=no|yes [yes on linux amd64/x86] 129 which specifies how full history stack traces should be computed. 130 Setting this to =yes can speed up Helgrind by 25% when using 131 --history-level=full. 132 133* Memcheck: reduced false positive rate for optimised code created by Clang 6 134 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code 135 blocks more carefully to determine where it can avoid expensive definedness 136 checks without loss of precision. This is controlled by the flag 137 --expensive-definedness-checks=no|auto|yes [auto]. 138 139* ==================== OTHER CHANGES ==================== 140 141* Valgrind is now buildable with link-time optimisation (LTO). A new 142 configure option --enable-lto=yes allows building Valgrind with LTO. If the 143 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%). 144 Note that if you are doing Valgrind development, --enable-lto=yes massively 145 slows down the build process. 146 147* ==================== FIXED BUGS ==================== 148 149The following bugs have been fixed or resolved. Note that "n-i-bz" 150stands for "not in bugzilla" -- that is, a bug that was reported to us 151but never got a bugzilla entry. We encourage you to file bugs in 152bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 153than mailing the developers (or mailing lists) directly -- bugs that 154are not entered into bugzilla tend to get forgotten about or ignored. 155 156To see details of a given bug, visit 157 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 158where XXXXXX is the bug number as listed below. 159 16079362 Debug info is lost for .so files when they are dlclose'd 161208052 strlcpy error when n = 0 162255603 exp-sgcheck Assertion '!already_present' failed 163338252 building valgrind with -flto (link time optimisation) fails 164345763 MIPS N32 ABI support 165368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace) 166 == 388664 unhandled arm64-linux syscall: 117 (ptrace) 167372347 Replacement problem of the additional c++14/c++17 new/delete operators 168373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+ 169376257 helgrind history full speed up using a cached stack 170379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s) 171 on macOS 10.12 172379748 Fix missing pselect syscall (OS X 10.11) 173379754 Fix missing syscall ulock_wait (OS X 10.12) 174380397 s390x: __GI_strcspn() replacemenet needed 175381162 possible array overrun in VEX register allocator 176381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support 177381274 powerpc too chatty even with --sigill-diagnostics=no 178381289 epoll_pwait can have a NULL sigmask 179381553 VEX register allocator v3 180381556 arm64: Handle feature registers access on 4.11 Linux kernel or later 181381769 Use ucontext_t instead of struct ucontext 182381805 arm32 needs ld.so index hardwire for new glibc security fixes 183382256 gz compiler flag test doesn't work for gold 184382407 vg_perf needs "--terse" command line option 185382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..] 186382563 MIPS MSA ASE support 187382998 xml-socket doesn't work 188383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed 189383723 Fix missing kevent_qos syscall (macOS 10.11) 190 == 385604 illegal hardware instruction (OpenCV cv::namedWindow) 191384096 Mention AddrCheck at Memcheck's command line option [..] 192384230 vex x86->IR: 0x67 0xE8 0xAB 0x68 193 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A 194 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program 195 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29 196 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA 197384337 performance improvements to VEX register allocator v2 and v3 198384526 reduce number of spill insns generated by VEX register allocator v3 199384584 Callee saved regs listed first for AMD64, X86, and PPC architectures 200384631 Sanitise client args as printed with -v 201384633 Add a simple progress-reporting facility 202384987 VEX regalloc: allocate caller-save registers for short lived vregs 203385055 PPC VEX temporary storage exhausted 204385182 PPC64 is missing support for the DSCR 205385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp 206385207 PPC64, generate_store_FPRF() generates too many Iops 207385208 PPC64, xxperm instruction exhausts temporary memory 208385210 PPC64, vpermr instruction could exhaust temporary memory 209385279 unhandled syscall: mach:43 (mach_generate_activity_id) 210 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..] 211 == 387045 Valgrind crashing on High Sierra when testing any newly [..] 212385334 PPC64, fix vpermr, xxperm, xxpermr mask value. 213385408 s390x: z13 vector "support" instructions not implemented 214385409 s390x: z13 vector integer instructions not implemented 215385410 s390x: z13 vector string instructions not implemented 216385412 s390x: new non-vector z13 instructions not implemented 217385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning. 218385912 none/tests/rlimit_nofile fails on newer glibc/kernel. 219385939 Optionally exit on the first error 220386318 valgrind.org/info/tools.html is missing SGCheck 221386425 running valgrind + wine on armv7l gives illegal opcode 222386397 PPC64, valgrind truncates powerpc timebase to 32-bits. 223387410 MIPSr6 support 224387664 Memcheck: make expensive-definedness-checks be the default 225387712 s390x cgijnl reports Conditional jump depends on uninitialised value 226387766 asm shifts cause false positive "Conditional jump or move depends 227 on uninitialised value" 228387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink 229388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed" 230388786 Support bpf syscall in amd64 Linux 231388862 Add replacements for wmemchr and wcsnlen on Linux 232389065 valgrind meets gcc flag -Wlogical-op 233389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented 234390471 suppression by specification of source-file line number 235390723 make xtree dump files world wide readable, similar to log files 236391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa 237391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)' 238392118 unhandled amd64-linux syscall: 332 (statx) 239392449 callgrind not clearing the number of calls properly 240393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp, 241 lxssp, stxssp and stxvl instructions. 242393023 callgrind_control risks using the wrong vgdb 243393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed" 244393099 posix_memalign() invalid write if alignment == 0 245393146 failing assert "is_DebugInfo_active(di)" 246395709 PPC64 is missing support for the xvnegsp instruction 247395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code 248 == 384727 249396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds) 250395991 arm-linux: wine's unit tests enter a signal delivery loop [..] 251396839 s390x: Trap instructions not implemented 252396887 arch_prctl should return EINVAL on unknown option 253 == 397286 crash before launching binary (Unsupported arch_prctl option) 254 == 397393 valgrind: the 'impossible' happened: (Archlinux) 255 == 397521 valgrind: the 'impossible' happened: Unsupported [..] 256396906 compile tests failure on mips32-linux: broken inline asm in tests on 257 mips32-linux 258397012 glibc ld.so uses arch_prctl on i386 259397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h 260397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT 261397424 glibc 2.27 and gdb_server tests 262398028 Assertion `cfsi_fits` failing in simple C program 263398066 s390x: cgijl dep1, 0 reports false unitialised values warning 264 265n-i-bz Fix missing workq_ops operations (macOS) 266n-i-bz fix bug in strspn replacement 267n-i-bz Add support for the Linux BLKFLSBUF ioctl 268n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls 269n-i-bz Fix possible stack trashing by semctl syscall wrapping 270n-i-bz Add support for the Linux membarrier() system call 271n-i-bz x86 front end: recognise and handle UD2 correctly 272n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is 273 correctly aligned before entering the handler. 274 275(3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae) 276(3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582) 277(3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f) 278 279 280 281Release 3.13.0 (15 June 2017) 282~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 283 2843.13.0 is a feature release with many improvements and the usual collection of 285bug fixes. 286 287This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux, 288PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux, 289MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android, 290X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. 291 292* ==================== CORE CHANGES =================== 293 294* The translation cache size has been increased to keep up with the demands of 295 large applications. The maximum number of sectors has increased from 24 to 296 48. The default number of sectors has increased from 16 to 32 on all 297 targets except Android, where the increase is from 6 to 12. 298 299* The amount of memory that Valgrind can use has been increased from 64GB to 300 128GB. In particular this means your application can allocate up to about 301 60GB when running on Memcheck. 302 303* Valgrind's default load address has been changed from 0x3800'0000 to 304 0x5800'0000, so as to make it possible to load larger executables. This 305 should make it possible to load executables of size at least 1200MB. 306 307* A massive spaceleak caused by reading compressed debuginfo files has been 308 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created 309 debuginfo. 310 311* The C++ demangler has been updated. 312 313* Support for demangling Rust symbols has been added. 314 315* A new representation of stack traces, the "XTree", has been added. An XTree 316 is a tree of stacktraces with data associated with the stacktraces. This is 317 used by various tools (Memcheck, Helgrind, Massif) to report on the heap 318 consumption of your program. Reporting is controlled by the new options 319 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>. 320 321 A report can also be produced on demand using the gdbserver monitor command 322 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind 323 format' and 'massif format. The existing visualisers for these formats (e.g. 324 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and 325 analyse these reports. 326 327 Memcheck can also produce XTree leak reports using the Callgrind file 328 format. For more details, see the user manual. 329 330* ================== PLATFORM CHANGES ================= 331 332* ppc64: support for ISA 3.0B and various fixes for existing 3.0 support 333 334* amd64: fixes for JIT failure problems on long AVX2 code blocks 335 336* amd64 and x86: support for CET prefixes has been added 337 338* arm32: a few missing ARMv8 instructions have been implemented 339 340* arm64, mips64, mips32: an alternative implementation of Load-Linked and 341 Store-Conditional instructions has been added. This is to deal with 342 processor implementations that implement the LL/SC specifications strictly 343 and as a result cause Valgrind to hang in certain situations. The 344 alternative implementation is automatically enabled at startup, as required. 345 You can use the option --sim-hints=fallback-llsc to force-enable it if you 346 want. 347 348* Support for OSX 10.12 has been improved. 349 350* On Linux, clone handling has been improved to honour CLONE_VFORK that 351 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled 352 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on 353 CLONE_VM exact semantics will (still) not work. 354 355* The TileGX/Linux port has been removed because it appears to be both unused 356 and unsupported. 357 358* ==================== TOOL CHANGES ==================== 359 360* Memcheck: 361 362 - Memcheck should give fewer false positives when running optimised 363 Clang/LLVM generated code. 364 365 - Support for --xtree-memory and 'xtmemory [<filename>]>'. 366 367 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file> 368 to produce the end of execution leak report in a xtree callgrind format 369 file. 370 371 - New option 'xtleak' in the memcheck leak_check monitor command, to produce 372 the leak report in an xtree file. 373 374* Massif: 375 376 - Support for --xtree-memory and 'xtmemory [<filename>]>'. 377 378 - For some workloads (typically, for big applications), Massif memory 379 consumption and CPU consumption has decreased significantly. 380 381* Helgrind: 382 383 - Support for --xtree-memory and 'xtmemory [<filename>]>'. 384 385 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful 386 for Ada gnat compiled applications. 387 388* ==================== OTHER CHANGES ==================== 389 390* For Valgrind developers: in an outer/inner setup, the outer Valgrind will 391 append the inner guest stacktrace to the inner host stacktrace. This helps 392 to investigate the errors reported by the outer, when they are caused by the 393 inner guest program (such as an inner regtest). See README_DEVELOPERS for 394 more info. 395 396* To allow fast detection of callgrind files by desktop environments and file 397 managers, the format was extended to have an optional first line that 398 uniquely identifies the format ("# callgrind format"). Callgrind creates 399 this line now, as does the new xtree functionality. 400 401* File name template arguments (such as --log-file, --xtree-memory-file, ...) 402 have a new %n format letter that is replaced by a sequence number. 403 404* "--version -v" now shows the SVN revision numbers from which Valgrind was 405 built. 406 407* ==================== FIXED BUGS ==================== 408 409The following bugs have been fixed or resolved. Note that "n-i-bz" 410stands for "not in bugzilla" -- that is, a bug that was reported to us 411but never got a bugzilla entry. We encourage you to file bugs in 412bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 413than mailing the developers (or mailing lists) directly -- bugs that 414are not entered into bugzilla tend to get forgotten about or ignored. 415 416To see details of a given bug, visit 417 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 418where XXXXXX is the bug number as listed below. 419 420162848 --log-file output isn't split when a program forks 421340777 Illegal instruction on mips (ar71xx) 422341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms 423342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones 424 to a different stack. 425344139 x86 stack-seg overrides, needed by the Wine people 426344524 store conditional of guest applications always fail - observed on 427 Octeon3(MIPS) 428348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT) 429352395 Please provide SVN revision info in --version -v 430352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP) 431356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid != 432 INVALID_POSIX_THREADID' failed 433358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread 434 barrier implementation 435358697 valgrind.h: Some code remains even when defining NVALGRIND 436359202 Add musl libc configure/compile 437360415 amd64 instructions ADCX and ADOX are not implemented in VEX 438 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10) 439360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1) 440362223 assertion failed when .valgrindrc is a directory instead of a file 441367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags 442367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138) 443368507 can't malloc chunks larger than about 34GB 444368529 Android arm target link error, missing atexit and pthread_atfork 445368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list) 446368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp) 447368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8) 448368917 WARNING: unhandled arm64-linux syscall: 218 (request_key) 449368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval) 450368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname) 451368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range) 452368925 WARNING: unhandled arm64-linux syscall: 130 (tkill) 453368926 WARNING: unhandled arm64-linux syscall: 97 (unshare) 454369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr) 455370028 Reduce the number of compiler warnings on MIPS platforms 456370635 arm64 missing syscall getcpu 457371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64 458371227 Clean AArch64 syscall table 459371412 Rename wrap_sys_shmat to sys_shmat like other wrappers 460371471 Valgrind complains about non legit memory leaks on placement new (C++) 461371491 handleAddrOverrides() is [incorrect] when ASO prefix is used 462371503 disInstr(arm64): unhandled instruction 0xF89F0000 463371869 support '%' in symbol Z-encoding 464371916 execution tree xtree concept 465372120 c++ demangler demangles symbols which are not c++ 466372185 Support of valgrind on ARMv8 with 32 bit executable 467372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10) 468372195 Power PC, xxsel instruction is not always recognized. 469372504 Hanging on exit_group 470372600 process loops forever when fatal signals are arriving quickly 471372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed 472373046 Stacks registered by core are never deregistered 473373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+ 474373086 Implement additional Xen hypercalls 475373192 Calling posix_spawn in glibc 2.24 completely broken 476373488 Support for fanotify API on ARM64 architecture 477 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init) 478373555 Rename BBPTR to GSPTR as it denotes guest state pointer only 479373938 const IRExpr arguments for matchIRExpr() 480374719 some spelling fixes 481374963 increase valgrind's load address to prevent mmap failure 482375514 valgrind_get_tls_addr() does not work in case of static TLS 483375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address 484 for ML_(find_rx_mapping)() 485375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24 486375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns 487 == 377159 "vex: the `impossible' happened" still present 488 == 375150 Assertion 'tres.status == VexTransOK' failed 489 == 378068 valgrind crashes on AVX2 function in FFmpeg 490376142 Segfaults on MIPS Cavium Octeon boards 491376279 disInstr(arm64): unhandled instruction 0xD50320FF 492376455 Solaris: unhandled syscall lgrpsys(180) 493376518 Solaris: unhandled fast trap getlgrp(6) 494376611 ppc64 and arm64 don't know about prlimit64 syscall 495376729 PPC64, remove R2 from the clobber list 496 == 371668 497376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses 498 to be wrongly marked as addressable 499377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with 500 PIE enabled by default 501377376 memcheck/tests/linux/getregset fails with glibc2.24 502377427 PPC64, lxv instruction failing on odd destination register 503377478 PPC64: ISA 3.0 setup fixes 504377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE 505 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET, 506 and 2 args for FUTEX_TRYLOCK_PI 507377717 Fix massive space leak when reading compressed debuginfo sections 508377891 Update Xen 4.6 domctl wrappers 509377930 fcntl syscall wrapper is missing flock structure check 510378524 libvexmultiarch_test regression on s390x and ppc64 511378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper 512378673 Update libiberty demangler 513378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support 514379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes 515379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper 516379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603] 517 (task_register_dyld_image_infos) 518379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603] 519 (task_register_dyld_shared_cache_image_info) 520379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap) 521379473 MIPS: add support for rdhwr cycle counter register 522379504 remove TileGX/Linux port 523379525 Support more x86 nop opcodes 524379838 disAMode(x86): not an addr! 525379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions 526379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4) 527379895 clock_gettime does not execute POST syscall wrapper 528379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly 529379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module) 530380200 xtree generated callgrind files refer to files without directory name 531380202 Assertion failure for cache line size (cls == 64) on aarch64. 532380397 s390x: __GI_strcspn() replacement needed 533n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7. 534 535(3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434) 536(3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443) 537(3.13.0: 14 June 2017, vex r3396, valgrind r16446) 538 539 540 541Release 3.12.0 (20 October 2016) 542~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 543 5443.12.0 is a feature release with many improvements and the usual 545collection of bug fixes. 546 547This release supports X86/Linux, AMD64/Linux, ARM32/Linux, 548ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, 549MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android, 550MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX 55110.10 and AMD64/MacOSX 10.10. There is also preliminary support for 552X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux. 553 554* ================== PLATFORM CHANGES ================= 555 556* POWER: Support for ISA 3.0 has been added 557 558* mips: support for O32 FPXX ABI has been added. 559* mips: improved recognition of different processors 560* mips: determination of page size now done at run time 561 562* amd64: Partial support for AMD FMA4 instructions. 563 564* arm, arm64: Support for v8 crypto and CRC instructions. 565 566* Improvements and robustification of the Solaris port. 567 568* Preliminary support for MacOS 10.12 (Sierra) has been added. 569 570Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we 571would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64) 572where possible. Valgrind's support for 32-bit x86 has stagnated in 573recent years and has fallen far behind that for 64-bit x86 574instructions. By contrast 64-bit x86 is well supported, up to and 575including AVX2. 576 577* ==================== TOOL CHANGES ==================== 578 579* Memcheck: 580 581 - Added meta mempool support for describing a custom allocator which: 582 - Auto-frees all chunks assuming that destroying a pool destroys all 583 objects in the pool 584 - Uses itself to allocate other memory blocks 585 586 - New flag --ignore-range-below-sp to ignore memory accesses below 587 the stack pointer, if you really have to. The related flag 588 --workaround-gcc296-bugs=yes is now deprecated. Use 589 --ignore-range-below-sp=1024-1 as a replacement. 590 591* DRD: 592 593 - Improved thread startup time significantly on non-Linux platforms. 594 595* DHAT 596 597 - Added collection of the metric "tot-blocks-allocd" 598 599* ==================== OTHER CHANGES ==================== 600 601* Replacement/wrapping of malloc/new related functions is now done not just 602 for system libraries by default, but for any globally defined malloc/new 603 related function (both in shared libraries and statically linked alternative 604 malloc implementations). The dynamic (runtime) linker is excluded, though. 605 To only intercept malloc/new related functions in 606 system libraries use --soname-synonyms=somalloc=nouserintercepts (where 607 "nouserintercepts" can be any non-existing library name). 608 This new functionality is not implemented for MacOS X. 609 610* The maximum number of callers in a suppression entry is now equal to 611 the maximum size for --num-callers (500). 612 Note that --gen-suppressions=yes|all similarly generates suppressions 613 containing up to --num-callers frames. 614 615* New and modified GDB server monitor features: 616 617 - Valgrind's gdbserver now accepts the command 'catch syscall'. 618 Note that you must have GDB >= 7.11 to use 'catch syscall' with 619 gdbserver. 620 621* New option --run-cxx-freeres=<yes|no> can be used to change whether 622 __gnu_cxx::__freeres() cleanup function is called or not. Default is 623 'yes'. 624 625* Valgrind is able to read compressed debuginfo sections in two formats: 626 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib) 627 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu) 628 629* Modest JIT-cost improvements: the cost of instrumenting code blocks 630 for the most common use case (x86_64-linux, Memcheck) has been 631 reduced by 10%-15%. 632 633* Improved performance for programs that do a lot of discarding of 634 instruction address ranges of 8KB or less. 635 636* The C++ symbol demangler has been updated. 637 638* More robustness against invalid syscall parameters on Linux. 639 640* ==================== FIXED BUGS ==================== 641 642The following bugs have been fixed or resolved. Note that "n-i-bz" 643stands for "not in bugzilla" -- that is, a bug that was reported to us 644but never got a bugzilla entry. We encourage you to file bugs in 645bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 646than mailing the developers (or mailing lists) directly -- bugs that 647are not entered into bugzilla tend to get forgotten about or ignored. 648 649To see details of a given bug, visit 650 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 651where XXXXXX is the bug number as listed below. 652 653191069 Exiting due to signal not reported in XML output 654199468 Suppressions: stack size limited to 25 655 while --num-callers allows more frames 656212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM) 657278744 cvtps2pd with redundant RexW 658303877 valgrind doesn't support compressed debuginfo sections. 659345307 Warning about "still reachable" memory when using libstdc++ from gcc 5 660348345 Assertion fails for negative lineno 661348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI 662351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1 663351692 Dumps created by valgrind are not readable by gdb (mips32 specific) 664351804 Crash on generating suppressions for "printf" call on OS X 10.10 665352197 mips: mmap2() not wrapped correctly for page size > 4096 666353083 arm64 doesn't implement various xattr system calls 667353084 arm64 doesn't support sigpending system call 668353137 www: update info for Supported Platforms 669353138 www: update "The Valgrind Developers" page 670353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine 671 == 365325 672 == 357873 673353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62) 674353398 WARNING: unhandled amd64-solaris syscall: 207 675353660 XML in auxwhat tag not escaping reserved symbols properly 676353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN 677353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72) 678353802 ELF debug info reader confused with multiple .rodata sections 679353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed 680353917 unhandled amd64-solaris syscall fchdir(120) 681353920 unhandled amd64-solaris syscall: 170 682354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3) 683354392 unhandled amd64-solaris syscall: 171 684354797 Vbit test does not include Iops for Power 8 instruction support 685354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11 686 == 361351 687 == 362920 688 == 366222 689354933 Fix documentation of --kernel-variant=android-no-hw-tls option 690355188 valgrind should intercept all malloc related global functions 691355454 do not intercept malloc related symbols from the runtime linker 692355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained 693356044 Dwarf line info reader misinterprets is_stmt register 694356112 mips: replace addi with addiu 695356393 valgrind (vex) crashes because isZeroU happened 696 == 363497 697 == 364497 698356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min) 699356678 arm64-linux: unhandled syscall 232 (mincore) 700356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND 701356823 Unsupported ARM instruction: stlex 702357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state 703357338 Unhandled instruction for SHA instructions libcrypto Boring SSL 704357673 crash if I try to run valgrind with a binary link with libcurl 705357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+ 706357871 pthread_spin_destroy not properly wrapped 707357887 Calls to VG_(fclose) do not close the file descriptor 708357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm. 709358030 support direct socket calls on x86 32bit (new in linux 4.3) 710358478 drd/tests/std_thread.cpp doesn't build with GCC6 711359133 Assertion 'eltSzB <= ddpa->poolSzB' failed 712359181 Buffer Overflow during Demangling 713359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET 714359289 s390x: popcnt (B9E1) not implemented 715359472 The Power PC vsubuqm instruction doesn't always give the correct result 716359503 Add missing syscalls for aarch64 (arm64) 717359645 "You need libc6-dbg" help message could be more helpful 718359703 s390: wire up separate socketcalls system calls 719359724 getsockname might crash - deref_UInt should call safe_to_deref 720359733 amd64 implement ld.so strchr/index override like x86 721359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5 722359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses 723 uninitialized data 724359838 arm64: Unhandled instruction 0xD5033F5F (clrex) 725359871 Incorrect mask handling in ppoll 726359952 Unrecognised PCMPESTRM variants (0x70, 0x19) 727360008 Contents of Power vr registers contents is not printed correctly when 728 the --vgdb-shadow-registers=yes option is used 729360035 POWER PC instruction bcdadd and bcdsubtract generate result with 730 non-zero shadow bits 731360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2) 732360425 arm64 unsupported instruction ldpsw 733 == 364435 734360519 none/tests/arm64/memory.vgtest might fail with newer gcc 735360571 Error about the Android Runtime reading below the stack pointer on ARM 736360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64 737360749 kludge for multiple .rodata sections on Solaris no longer needed 738360752 raise the number of reserved fds in m_main.c from 10 to 12 739361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5 740361226 s390x: risbgn (EC59) not implemented 741361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create' 742361354 ppc64[le]: wire up separate socketcalls system calls 743361615 Inconsistent termination for multithreaded process terminated by signal 744361926 Unhandled Solaris syscall: sysfs(84) 745362009 V dumps core on unimplemented functionality before threads are created 746362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5 747362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64) 748362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed 749362953 Request for an update to the Valgrind Developers page 750363680 add renameat2() support 751363705 arm64 missing syscall name_to_handle_at and open_by_handle_at 752363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at 753363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5 754364058 clarify in manual limitations of array overruns detections 755364413 pselect sycallwrapper mishandles NULL sigmask 756364728 Power PC, missing support for several HW registers in 757 get_otrack_shadow_offset_wrk() 758364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5 759365273 Invalid write to stack location reported after signal handler runs 760365912 ppc64BE segfault during jm-insns test (RELRO) 761366079 FPXX Support for MIPS32 Valgrind 762366138 Fix configure errors out when using Xcode 8 (clang 8.0.0) 763366344 Multiple unhandled instruction for Aarch64 764 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060) 765367995 Integration of memcheck with custom memory allocator 766368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer 767368412 False positive result for altivec capability check 768368416 Add tc06_two_races_xml.exp output for ppc64 769368419 Perf Events ioctls not implemented 770368461 mmapunmap test fails on ppc64 771368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target 772369000 AMD64 fma4 instructions unsupported. 773369169 ppc64 fails jm_int_isa_2_07 test 774369175 jm_vec_isa_2_07 test crashes on ppc64 775369209 valgrind loops and eats up all memory if cwd doesn't exist. 776369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr 777369359 msghdr_foreachfield can crash when handling bad iovec 778369360 Bad sigprocmask old or new sets can crash valgrind 779369361 vmsplice syscall wrapper crashes on bad iovec 780369362 Bad sigaction arguments crash valgrind 781369383 x86 sys_modify_ldt wrapper crashes on bad ptr 782369402 Bad set/get_thread_area pointer crashes valgrind 783369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers 784369446 valgrind crashes on unknown fcntl command 785369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER 786369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter) 787370265 ISA 3.0 HW cap stuff needs updating 788371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work 789372195 Power PC, xxsel instruction is not always recognized 790 791n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64 792n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap 793n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments 794n-i-bz Fix ppoll_alarm exclusion on OS X 795n-i-bz Document brk segment limitation, reference manual in limit reached msg. 796n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737] 797n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206] 798n-i-bz amd64: memcheck false positive with shr %edx 799n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]" 800n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion 801n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D} 802n-i-bz DHAT: added collection of the metric "tot-blocks-allocd" 803 804(3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094) 805(3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096) 806(3.12.0: 21 October 2016, vex r3282, valgrind r16098) 807 808 809 810Release 3.11.0 (22 September 2015) 811~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 812 8133.11.0 is a feature release with many improvements and the usual 814collection of bug fixes. 815 816This release supports X86/Linux, AMD64/Linux, ARM32/Linux, 817ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, 818MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android, 819MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX 82010.10 and AMD64/MacOSX 10.10. There is also preliminary support for 821X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux. 822 823* ================== PLATFORM CHANGES ================= 824 825* Support for Solaris/x86 and Solaris/amd64 has been added. 826 827* Preliminary support for Mac OS X 10.11 (El Capitan) has been added. 828 829* Preliminary support for the Tilera TileGX architecture has been added. 830 831* s390x: It is now required for the host to have the "long displacement" 832 facility. The oldest supported machine model is z990. 833 834* x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a 835 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3. 836 837* The JIT's register allocator is significantly faster, making the JIT 838 as a whole somewhat faster, so JIT-intensive activities, for example 839 program startup, are modestly faster, around 5%. 840 841* There have been changes to the default settings of several command 842 line flags, as detailed below. 843 844* Intel AVX2 support is more complete (64 bit targets only). On AVX2 845 capable hosts, the simulated CPUID will now indicate AVX2 support. 846 847* ==================== TOOL CHANGES ==================== 848 849* Memcheck: 850 851 - The default value for --leak-check-heuristics has been changed from 852 "none" to "all". This helps to reduce the number of possibly 853 lost blocks, in particular for C++ applications. 854 855 - The default value for --keep-stacktraces has been changed from 856 "malloc-then-free" to "malloc-and-free". This has a small cost in 857 memory (one word per malloc-ed block) but allows Memcheck to show the 858 3 stacktraces of a dangling reference: where the block was allocated, 859 where it was freed, and where it is acccessed after being freed. 860 861 - The default value for --partial-loads-ok has been changed from "no" to 862 "yes", so as to avoid false positive errors resulting from some kinds 863 of vectorised loops. 864 865 - A new monitor command 'xb <addr> <len>' shows the validity bits of 866 <len> bytes at <addr>. The monitor command 'xb' is easier to use 867 than get_vbits when you need to associate byte data value with 868 their corresponding validity bits. 869 870 - The 'block_list' monitor command has been enhanced: 871 o it can print a range of loss records 872 o it now accepts an optional argument 'limited <max_blocks>' 873 to control the number of blocks printed. 874 o if a block has been found using a heuristic, then 875 'block_list' now shows the heuristic after the block size. 876 o the loss records/blocks to print can be limited to the blocks 877 found via specified heuristics. 878 879 - The C helper functions used to instrument loads on 880 x86-{linux,solaris} and arm-linux (both 32-bit only) have been 881 replaced by handwritten assembly sequences. This gives speedups 882 in the region of 0% to 7% for those targets only. 883 884 - A new command line option, --expensive-definedness-checks=yes|no, 885 has been added. This is useful for avoiding occasional invalid 886 uninitialised-value errors in optimised code. Watch out for 887 runtime degradation, as this can be up to 25%. As always, though, 888 the slowdown is highly application specific. The default setting 889 is "no". 890 891* Massif: 892 893 - A new monitor command 'all_snapshots <filename>' dumps all 894 snapshots taken so far. 895 896* Helgrind: 897 898 - Significant memory reduction and moderate speedups for 899 --history-level=full for applications accessing a lot of memory 900 with many different stacktraces. 901 902 - The default value for --conflict-cache-size=N has been doubled to 903 2000000. Users that were not using the default value should 904 preferably also double the value they give. 905 906 The default was changed due to the changes in the "full history" 907 implementation. Doubling the value gives on average a slightly more 908 complete history and uses similar memory (or significantly less memory 909 in the worst case) than the previous implementation. 910 911 - The Helgrind monitor command 'info locks' now accepts an optional 912 argument 'lock_addr', which shows information about the lock at the 913 given address only. 914 915 - When using --history-level=full, the new Helgrind monitor command 916 'accesshistory <addr> [<len>]' will show the recorded accesses for 917 <len> (or 1) bytes at <addr>. 918 919* ==================== OTHER CHANGES ==================== 920 921* The default value for the --smc-check option has been changed from 922 "stack" to "all-non-file" on targets that provide automatic D-I 923 cache coherence (x86, amd64 and s390x). The result is to provide, 924 by default, transparent support for JIT generated and self-modifying 925 code on all targets. 926 927* Mac OS X only: the default value for the --dsymutil option has been 928 changed from "no" to "yes", since any serious usage on Mac OS X 929 always required it to be "yes". 930 931* The command line options --db-attach and --db-command have been removed. 932 They were deprecated in 3.10.0. 933 934* When a process dies due to a signal, Valgrind now shows the signal 935 and the stacktrace at default verbosity (i.e. verbosity 1). 936 937* The address description logic used by Memcheck and Helgrind now 938 describes addresses in anonymous segments, file mmap-ed segments, 939 shared memory segments and the brk data segment. 940 941* The new option --error-markers=<begin>,<end> can be used to mark the 942 begin/end of errors in textual output mode, to facilitate 943 searching/extracting errors in output files that mix valgrind errors 944 with program output. 945 946* The new option --max-threads=<number> can be used to change the number 947 of threads valgrind can handle. The default is 500 threads which 948 should be more than enough for most applications. 949 950* The new option --valgrind-stacksize=<number> can be used to change the 951 size of the private thread stacks used by Valgrind. This is useful 952 for reducing memory use or increasing the stack size if Valgrind 953 segfaults due to stack overflow. 954 955* The new option --avg-transtab-entry-size=<number> can be used to specify 956 the expected instrumented block size, either to reduce memory use or 957 to avoid excessive retranslation. 958 959* Valgrind can be built with Intel's ICC compiler, version 14.0 or later. 960 961* New and modified GDB server monitor features: 962 963 - When a signal is reported in GDB, you can now use the GDB convenience 964 variable $_siginfo to examine detailed signal information. 965 966 - Valgrind's gdbserver now allows the user to change the signal 967 to deliver to the process. So, use 'signal SIGNAL' to continue execution 968 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to 969 continue without passing the signal to the process. 970 971 - With GDB >= 7.10, the command 'target remote' 972 will automatically load the executable file of the process running 973 under Valgrind. This means you do not need to specify the executable 974 file yourself, GDB will discover it itself. See GDB documentation about 975 'qXfer:exec-file:read' packet for more info. 976 977* ==================== FIXED BUGS ==================== 978 979The following bugs have been fixed or resolved. Note that "n-i-bz" 980stands for "not in bugzilla" -- that is, a bug that was reported to us 981but never got a bugzilla entry. We encourage you to file bugs in 982bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 983than mailing the developers (or mailing lists) directly -- bugs that 984are not entered into bugzilla tend to get forgotten about or ignored. 985 986To see details of a given bug, visit 987 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 988where XXXXXX is the bug number as listed below. 989 990116002 VG_(printf): Problems with justification of strings and integers 991155125 avoid cutting away file:lineno after long function name 992197259 Unsupported arch_prtctl PR_SET_GS option 993201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269 994201216 Fix Valgrind does not support pthread_sigmask() on OS X 995201435 Fix Darwin: -v does not show kernel version 996208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X 997211256 Fixed an outdated comment regarding the default platform. 998211529 Incomplete call stacks for code compiled by newer versions of MSVC 999211926 Avoid compilation warnings in valgrind.h with -pedantic 1000212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X 1001 == 263119 1002226609 Crediting upstream authors in man page 1003231257 Valgrind omits path when executing script from shebang line 1004254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..] 1005294065 Improve the pdb file reader by avoiding hardwired absolute pathnames 1006269360 s390x: Fix addressing mode selection for compare-and-swap 1007302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)' 1008 == 326797 1009312989 ioctl handling needs to do POST handling on generic ioctls and [..] 1010319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X 1011324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it) 1012327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8 1013330147 libmpiwrap PMPI_Get_count returns undefined value 1014333051 mmap of huge pages fails due to incorrect alignment 1015 == 339163 1016334802 valgrind does not always explain why a given option is bad 1017335618 mov.w rN, pc/sp (ARM32) 1018335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd) 1019 == 307399 1020 == 343175 1021 == 342740 1022 == 346912 1023335907 segfault when running wine's ddrawex/tests/surface.c under valgrind 1024338602 AVX2 bit in CPUID missing 1025338606 Strange message for scripts with invalid interpreter 1026338731 ppc: Fix testuite build for toolchains not supporting -maltivec 1027338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL 1028339045 Getting valgrind to compile and run on OS X Yosemite (10.10) 1029 == 340252 1030339156 gdbsrv not called for fatal signal 1031339215 Valgrind 3.10.0 contain 2013 in copyrights notice 1032339288 support Cavium Octeon MIPS specific BBIT*32 instructions 1033339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix 1034339442 Fix testsuite build failure on OS X 10.9 1035339542 Enable compilation with Intel's ICC compiler 1036339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper 1037339688 Mac-specific ASM does not support .version directive (cpuid, 1038 tronical and pushfpopf tests) 1039339745 Valgrind crash when check Marmalade app (partial fix) 1040339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9 1041339778 Linux/TileGx platform support for Valgrind 1042339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9 1043339789 Fix none/tests/execve test on Mac OS X 10.9 1044339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9 1045339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42) 1046340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR 1047340392 Allow user to select more accurate definedness checking in memcheck 1048 to avoid invalid complaints on optimised code 1049340430 Fix some grammatical weirdness in the manual. 1050341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14) 1051341419 Signal handler ucontext_t not filled out correctly on OS X 1052341539 VG_(describe_addr) should not describe address as belonging to client 1053 segment if it is past the heap end 1054341613 Enable building of manythreads and thread-exits tests on Mac OS X 1055341615 Fix none/tests/darwin/access_extended test on Mac OS X 1056341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..] 1057341789 aarch64: shmat fails with valgrind on ARMv8 1058341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly 1059342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode 1060342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy) 1061342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests 1062342117 Hang when loading PDB file for MSVC compiled Firefox under Wine 1063342221 socket connect false positive uninit memory for unknown af family 1064342353 Allow dumping full massif output while valgrind is still running 1065342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS 1066 == 346476 1067 == 348387 1068 == 350593 1069342603 Add I2C_SMBUS ioctl support 1070342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code 1071342683 Mark memory past the initial brk limit as unaddressable 1072342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12" 1073342795 Internal glibc __GI_mempcpy call should be intercepted 1074342841 s390x: Support instructions fiebr(a) and fidbr(a) 1075343012 Unhandled syscall 319 (memfd_create) 1076343069 Patch updating v4l2 API support 1077343173 helgrind crash during stack unwind 1078343219 fix GET_STARTREGS for arm 1079343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10 1080343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option 1081343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64 1082343335 unhandled instruction 0x1E638400 (fccmp) aarch64 1083343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..] 1084343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..] 1085343597 ppc64le: incorrect use of offseof macro 1086343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..] 1087343663 OS X 10.10 Memchecj always reports a leak regardless of [..] 1088343732 Unhandled syscall 144 (setgid) on aarch64 1089343733 Unhandled syscall 187 (msgctl and related) on aarch64 1090343802 s390x: False positive "conditional jump or move depends on [..] 1091343902 --vgdb=yes doesn't break when --xml=yes is used 1092343967 Don't warn about setuid/setgid/setcap executable for directories 1093343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants 1094344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344) 1095344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait 1096344054 www - update info for Solaris/illumos 1097344416 'make regtest' does not work cleanly on OS X 1098344235 Remove duplicate include of pub_core_aspacemgr.h 1099344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled 1100344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled 1101344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40) 1102344314 callgrind_annotate ... warnings about commands containing newlines 1103344318 socketcall should wrap recvmmsg and sendmmsg 1104344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap) 1105344416 Fix 'make regtest' does not work cleanly on OS X 1106344499 Fix compilation for Linux kernel >= 4.0.0 1107344512 OS X: unhandled syscall: unix:348 (__pthread_chdir), 1108 unix:349 (__pthread_fchdir) 1109344559 Garbage collection of unused segment names in address space manager 1110344560 Fix stack traces missing penultimate frame on OS X 1111344621 Fix memcheck/tests/err_disable4 test on OS X 1112344686 Fix suppression for pthread_rwlock_init on OS X 10.10 1113344702 Fix missing libobjc suppressions on OS X 10.10 1114 == 344543 1115344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10 1116344939 Fix memcheck/tests/xml1 on OS X 10.10 1117345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes 1118345079 Fix build problems in VEX/useful/test_main.c 1119345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT 1120345177 arm64: prfm (reg) not implemented 1121345215 Performance improvements for the register allocator 1122345248 add support for Solaris OS in valgrind 1123345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux 1124345394 Fix memcheck/tests/strchr on OS X 1125345637 Fix memcheck/tests/sendmsg on OS X 1126345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2 1127345824 Fix aspacem segment mismatch: seen with none/tests/bigcode 1128345887 Fix an assertion in the address space manager 1129345928 amd64: callstack only contains current function for small stacks 1130345984 disInstr(arm): unhandled instruction: 0xEE193F1E 1131345987 MIPS64: Implement cavium LHX instruction 1132346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31) 1133346185 Fix typo saving altivec register v24 1134346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER() 1135 and LibVEX_GuestPPC64_get_CR() 1136346270 Regression tests none/tests/jm_vec/isa_2_07 and 1137 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian 1138346307 fuse filesystem syscall deadlocks 1139346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions 1140346411 MIPS: SysRes::_valEx handling is incorrect 1141346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls 1142346474 PPC64 Power 8, spr TEXASRU register not supported 1143346487 Compiler generates "note" about a future ABI change for PPC64 1144346562 MIPS64: lwl/lwr instructions are performing 64bit loads 1145 and causing spurious "invalid read of size 8" warnings 1146346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack 1147347151 Fix suppression for pthread_rwlock_init on OS X 10.8 1148347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell) 1149347322 Power PC regression test cleanup 1150347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8 1151 == 217236 1152347389 unhandled syscall: 373 (Linux ARM syncfs) 1153347686 Patch set to cleanup PPC64 regtests 1154347978 Remove bash dependencies where not needed 1155347982 OS X: undefined symbols for architecture x86_64: "_global" [..] 1156347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine) 1157 == 345929 1158348102 Patch updating v4l2 API support 1159348247 amd64 front end: jno jumps wrongly when overflow is not set 1160348269 Improve mmap MAP_HUGETLB support. 1161348334 (ppc) valgrind does not simulate dcbfl - then my program terminates 1162348345 Assertion fails for negative lineno 1163348377 Unsupported ARM instruction: yield 1164348565 Fix detection of command line option availability for clang 1165348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18) 1166348728 Fix broken check for VIDIOC_G_ENC_INDEX 1167348748 Fix redundant condition 1168348890 Fix clang warning about unsupported --param inline-unit-growth=900 1169348949 Bogus "ERROR: --ignore-ranges: suspiciously large range" 1170349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK 1171349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..] 1172349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..] 1173349626 Implemented additional Xen hypercalls 1174349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64 1175349790 Clean up of the hardware capability checking utilities. 1176349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so) 1177349874 Fix typos in source code 1178349879 memcheck: add handwritten assembly for helperc_LOADV* 1179349941 di_notify_mmap might create wrong start/size DebugInfoMapping 1180350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X 1181350202 Add limited param to 'monitor block_list' 1182350290 s390x: Support instructions fixbr(a) 1183350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X 1184350809 Fix none/tests/async-sigs for Solaris 1185350811 Remove reference to --db-attach which has been removed. 1186350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too 1187350854 hard-to-understand code in VG_(load_ELF)() 1188351140 arm64 syscalls setuid (146) and setresgid (149) not implemented 1189351386 Solaris: Cannot run ld.so.1 under Valgrind 1190351474 Fix VG_(iseqsigset) as obvious 1191351531 Typo in /include/vki/vki-xen-physdev.h header guard 1192351756 Intercept platform_memchr$VARIANT$Haswell on OS X 1193351858 ldsoexec support on Solaris 1194351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode 1195352130 helgrind reports false races for printfs using mempcpy on FILE* state 1196352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf 1197352320 arm64 crash on none/tests/nestedfs 1198352765 Vbit test fails on Power 6 1199352768 The mbar instruction is missing from the Power PC support 1200352769 Power PC program priority register (PPR) is not supported 1201n-i-bz Provide implementations of certain compiler builtins to support 1202 compilers that may not provide those 1203n-i-bz Old STABS code is still being compiled, but never used. Remove it. 1204n-i-bz Fix compilation on distros with glibc < 2.5 1205n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts 1206n-i-bz Enable rt_sigpending syscall on ppc64 linux. 1207n-i-bz mremap did not work properly on shared memory 1208n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity 1209n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml 1210 1211(3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646) 1212(3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667) 1213(3.11.0: 22 September 2015, vex r3195, valgrind r15674) 1214 1215 1216 1217Release 3.10.1 (25 November 2014) 1218~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 12193.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0 1220and backports fixes for all reported missing AArch64 ARMv8 instructions 1221and syscalls from the trunk. If you package or deliver 3.10.0 for others 1222to use, you might want to consider upgrading to 3.10.1 instead. 1223 1224The following bugs have been fixed or resolved. Note that "n-i-bz" 1225stands for "not in bugzilla" -- that is, a bug that was reported to us 1226but never got a bugzilla entry. We encourage you to file bugs in 1227bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 1228than mailing the developers (or mailing lists) directly -- bugs that 1229are not entered into bugzilla tend to get forgotten about or ignored. 1230 1231To see details of a given bug, visit 1232 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 1233where XXXXXX is the bug number as listed below. 1234 1235335440 arm64: ld1 (single structure) is not implemented 1236335713 arm64: unhanded instruction: prfm (immediate) 1237339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build 1238339182 ppc64: AvSplat ought to load destination vector register with [..] 1239339336 PPC64 store quad instruction (stq) is not supposed to change [..] 1240339433 ppc64 lxvw4x instruction uses four 32-byte loads 1241339645 Use correct tag names in sys_getdents/64 wrappers 1242339706 Fix false positive for ioctl(TIOCSIG) on linux 1243339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ... 1244339853 arm64 times syscall unknown 1245339855 arm64 unhandled getsid/setsid syscalls 1246339858 arm64 dmb sy not implemented 1247339926 Unhandled instruction 0x1E674001 (frintx) on aarm64 1248339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64 1249339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla) 1250 == 339950 1251339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch 1252340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants 1253340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid) 1254340036 arm64: Unhandled instruction ld4 (multiple structures, no offset) 1255340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat 1256340509 arm64: unhandled instruction fcvtas 1257340630 arm64: fchmod (52) and fchown (55) syscalls not recognized 1258340632 arm64: unhandled instruction fcvtas 1259340722 Resolve "UNKNOWN attrlist flags 0:0x10000000" 1260340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms 1261340788 warning: unhandled syscall: 318 (getrandom) 1262340807 disInstr(arm): unhandled instruction: 0xEE989B20 1263340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel) 1264340922 arm64: unhandled getgroups/setgroups syscalls 1265350251 Fix typo in VEX utility program (test_main.c). 1266350407 arm64: unhandled instruction ucvtf (vector, integer) 1267350809 none/tests/async-sigs breaks when run under cron on Solaris 1268350811 update README.solaris after r15445 1269350813 Use handwritten memcheck assembly helpers on x86/Solaris [..] 1270350854 strange code in VG_(load_ELF)() 1271351140 arm64 syscalls setuid (146) and setresgid (149) not implemented 1272n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM) 1273n-i-bz Add missing ]] to terminate CDATA. 1274n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO 1275n-i-bz Enable sys_fadvise64_64 on arm32. 1276n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns. 1277n-i-bz Add test cases for all known arm64 load/store instructions. 1278n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..] 1279n-i-bz Add detection of old ppc32 magic instructions from bug 278808. 1280n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size". 1281n-i-bz arm64: Implement "fcvtpu w, s". 1282n-i-bz arm64: implement ADDP and various others 1283n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt). 1284n-i-bz arm64: enable FCVT{A,N}S X,S. 1285 1286(3.10.1: 25 November 2014, vex r3026, valgrind r14785) 1287 1288 1289 1290Release 3.10.0 (10 September 2014) 1291~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1292 12933.10.0 is a feature release with many improvements and the usual 1294collection of bug fixes. 1295 1296This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux, 1297PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux, 1298MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9 1299and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is 1300significantly improved relative to the 3.9.0 release. 1301 1302* ================== PLATFORM CHANGES ================= 1303 1304* Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port 1305 is mostly complete, and is usable, but some SIMD instructions are as 1306 yet unsupported. 1307 1308* Support for little-endian variant of the 64-bit POWER architecture. 1309 1310* Support for Android on MIPS32. 1311 1312* Support for 64bit FPU on MIPS32 platforms. 1313 1314* Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9. 1315 1316* Configuration for and running on Android targets has changed. 1317 See README.android in the source tree for details. 1318 1319* ================== DEPRECATED FEATURES ================= 1320 1321* --db-attach is now deprecated and will be removed in the next 1322 valgrind feature release. The built-in GDB server capabilities are 1323 superior and should be used instead. Learn more here: 1324 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver 1325 1326* ==================== TOOL CHANGES ==================== 1327 1328* Memcheck: 1329 1330 - Client code can now selectively disable and re-enable reporting of 1331 invalid address errors in specific ranges using the new client 1332 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and 1333 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE. 1334 1335 - Leak checker: there is a new leak check heuristic called 1336 "length64". This is used to detect interior pointers pointing 8 1337 bytes inside a block, on the assumption that the first 8 bytes 1338 holds the value "block size - 8". This is used by 1339 sqlite3MemMalloc, for example. 1340 1341 - Checking of system call parameters: if a syscall parameter 1342 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has 1343 several fields not initialised, an error is now reported for each 1344 field. Previously, an error was reported only for the first 1345 uninitialised field. 1346 1347 - Mismatched alloc/free checking: a new flag 1348 --show-mismatched-frees=no|yes [yes] makes it possible to turn off 1349 such checks if necessary. 1350 1351* Helgrind: 1352 1353 - Improvements to error messages: 1354 1355 o Race condition error message involving heap allocated blocks also 1356 show the thread number that allocated the raced-on block. 1357 1358 o All locks referenced by an error message are now announced. 1359 Previously, some error messages only showed the lock addresses. 1360 1361 o The message indicating where a lock was first observed now also 1362 describes the address/location of the lock. 1363 1364 - Helgrind now understands the Ada task termination rules and 1365 creates a happens-before relationship between a terminated task 1366 and its master. This avoids some false positives and avoids a big 1367 memory leak when a lot of Ada tasks are created and terminated. 1368 The interceptions are only activated with forthcoming releases of 1369 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0. 1370 1371 - A new GDB server monitor command "info locks" giving the list of 1372 locks, their location, and their status. 1373 1374* Callgrind: 1375 1376 - callgrind_control now supports the --vgdb-prefix argument, 1377 which is needed if valgrind was started with this same argument. 1378 1379* ==================== OTHER CHANGES ==================== 1380 1381* Unwinding through inlined function calls. Stack unwinding can now 1382 make use of Dwarf3 inlined-unwind information if it is available. 1383 The practical effect is that inlined calls become visible in stack 1384 traces. The suppression matching machinery has been adjusted 1385 accordingly. This is controlled by the new option 1386 --read-inline-info=yes|no. Currently this is enabled by default 1387 only on Linux and Android targets and only for the tools Memcheck, 1388 Helgrind and DRD. 1389 1390* Valgrind can now read EXIDX unwind information on 32-bit ARM 1391 targets. If an object contains both CFI and EXIDX unwind 1392 information, Valgrind will prefer the CFI over the EXIDX. This 1393 facilitates unwinding through system libraries on arm-android 1394 targets. 1395 1396* Address description logic has been improved and is now common 1397 between Memcheck and Helgrind, resulting in better address 1398 descriptions for some kinds of error messages. 1399 1400* Error messages about dubious arguments (eg, to malloc or calloc) are 1401 output like other errors. This means that they can be suppressed 1402 and they have a stack trace. 1403 1404* The C++ demangler has been updated for better C++11 support. 1405 1406* New and modified GDB server monitor features: 1407 1408 - Thread local variables/storage (__thread) can now be displayed. 1409 1410 - The GDB server monitor command "v.info location <address>" 1411 displays information about an address. The information produced 1412 depends on the tool and on the options given to valgrind. 1413 Possibly, the following are described: global variables, local 1414 (stack) variables, allocated or freed blocks, ... 1415 1416 - The option "--vgdb-stop-at=event1,event2,..." allows the user to 1417 ask the GDB server to stop at the start of program execution, at 1418 the end of the program execution and on Valgrind internal errors. 1419 1420 - A new monitor command "v.info stats" shows various Valgrind core 1421 and tool statistics. 1422 1423 - A new monitor command "v.set hostvisibility" allows the GDB server 1424 to provide access to Valgrind internal host status/memory. 1425 1426* A new option "--aspace-minaddr=<address>" can in some situations 1427 allow the use of more memory by decreasing the address above which 1428 Valgrind maps memory. It can also be used to solve address 1429 conflicts with system libraries by increasing the default value. 1430 See user manual for details. 1431 1432* The amount of memory used by Valgrind to store debug info (unwind 1433 info, line number information and symbol data) has been 1434 significantly reduced, even though Valgrind now reads more 1435 information in order to support unwinding of inlined function calls. 1436 1437* Dwarf3 handling with --read-var-info=yes has been improved: 1438 1439 - Ada and C struct containing VLAs no longer cause a "bad DIE" error 1440 1441 - Code compiled with 1442 -ffunction-sections -fdata-sections -Wl,--gc-sections 1443 no longer causes assertion failures. 1444 1445* Improved checking for the --sim-hints= and --kernel-variant= 1446 options. Unknown strings are now detected and reported to the user 1447 as a usage error. 1448 1449* The semantics of stack start/end boundaries in the valgrind.h 1450 VALGRIND_STACK_REGISTER client request has been clarified and 1451 documented. The convention is that start and end are respectively 1452 the lowest and highest addressable bytes of the stack. 1453 1454* ==================== FIXED BUGS ==================== 1455 1456The following bugs have been fixed or resolved. Note that "n-i-bz" 1457stands for "not in bugzilla" -- that is, a bug that was reported to us 1458but never got a bugzilla entry. We encourage you to file bugs in 1459bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 1460than mailing the developers (or mailing lists) directly -- bugs that 1461are not entered into bugzilla tend to get forgotten about or ignored. 1462 1463To see details of a given bug, visit 1464 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 1465where XXXXXX is the bug number as listed below. 1466 1467175819 Support for ipv6 socket reporting with --track-fds 1468232510 make distcheck fails 1469249435 Analyzing wine programs with callgrind triggers a crash 1470278972 support for inlined function calls in stacktraces and suppression 1471 == 199144 1472291310 FXSAVE instruction marks memory as undefined on amd64 1473303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped 1474308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall) 1475315199 vgcore file for threaded app does not show which thread crashed 1476315952 tun/tap ioctls are not supported 1477323178 Unhandled instruction: PLDW register (ARM) 1478323179 Unhandled instruction: PLDW immediate (ARM) 1479324050 Helgrind: SEGV because of unaligned stack when using movdqa 1480325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo. 1481325124 [MIPSEL] Compilation error 1482325477 Phase 4 support for IBM Power ISA 2.07 1483325538 cavium octeon mips64, valgrind reported "dumping core" [...] 1484325628 Phase 5 support for IBM Power ISA 2.07 1485325714 Empty vgcore but RLIMIT_CORE is big enough (too big) 1486325751 Missing the two privileged Power PC Transactional Memory Instructions 1487325816 Phase 6 support for IBM Power ISA 2.07 1488325856 Make SGCheck fail gracefully on unsupported platforms 1489326026 Iop names for count leading zeros/sign bits incorrectly imply [..] 1490326436 DRD: False positive in libstdc++ std::list::push_back 1491326444 Cavium MIPS Octeon Specific Load Indexed Instructions 1492326462 Refactor vgdb to isolate invoker stuff into separate module 1493326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E) 1494326623 DRD: false positive conflict report in a field assignment 1495326724 Valgrind does not compile on OSX 1.9 Mavericks 1496326816 Intercept for __strncpy_sse2_unaligned missing? 1497326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V 1498326983 Clear direction flag after tests on amd64. 1499327212 Do not prepend the current directory to absolute path names. 1500327223 Support for Cavium MIPS Octeon Atomic and Count Instructions 1501327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed 1502327284 s390x: Fix translation of the risbg instruction 1503327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34 1504327837 dwz compressed alternate .debug_info and .debug_str not read correctly 1505327916 DW_TAG_typedef may have no name 1506327943 s390x: add a redirection for the 'index' function 1507328100 XABORT not implemented 1508328205 Implement additional Xen hypercalls 1509328454 add support Backtraces with ARM unwind tables (EXIDX) 1510328455 s390x: SIGILL after emitting wrong register pair for ldxbr 1511328711 valgrind.1 manpage "memcheck options" section is badly generated 1512328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14 1513329612 Incorrect handling of AT_BASE for image execution 1514329694 clang warns about using uninitialized variable 1515329956 valgrind crashes when lmw/stmw instructions are used on ppc64 1516330228 mmap must align to VKI_SHMLBA on mips32 1517330257 LLVM does not support `-mno-dynamic-no-pic` option 1518330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend) 1519330459 --track-fds=yes doesn't track eventfds 1520330469 Add clock_adjtime syscall support 1521330594 Missing sysalls on PowerPC / uClibc 1522330622 Add test to regression suite for POWER instruction: dcbzl 1523330939 Support for AMD's syscall instruction on x86 1524 == 308729 1525330941 Typo in PRE(poll) syscall wrapper 1526331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch) 1527331254 Fix expected output for memcheck/tests/dw4 1528331255 Fix race condition in test none/tests/coolo_sigaction 1529331257 Fix type of jump buffer in test none/tests/faultstatus 1530331305 configure uses bash specific syntax 1531331337 s390x WARNING: unhandled syscall: 326 (dup3) 1532331380 Syscall param timer_create(evp) points to uninitialised byte(s) 1533331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64) 1534331829 Unexpected ioctl opcode sign extension 1535331830 ppc64: WARNING: unhandled syscall: 96/97 1536331839 drd/tests/sem_open specifies invalid semaphore name 1537331847 outcome of drd/tests/thread_name is nondeterministic 1538332037 Valgrind cannot handle Thumb "add pc, reg" 1539332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and 1540 consistency checks enabled 1541332263 intercepts for pthread_rwlock_timedrdlock and 1542 pthread_rwlock_timedwrlock are incorrect 1543332265 drd could do with post-rwlock_init and pre-rwlock_destroy 1544 client requests 1545332276 Implement additional Xen hypercalls 1546332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment 1547332765 Fix ms_print to create temporary files in a proper directory 1548333072 drd: Add semaphore annotations 1549333145 Tests for missaligned PC+#imm access for arm 1550333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0 1551333230 AAarch64 missing instruction encodings: dc, ic, dsb. 1552333248 WARNING: unhandled syscall: unix:443 1553333428 ldr.w pc [rD, #imm] instruction leads to assertion 1554333501 cachegrind: assertion: Cache set count is not a power of two. 1555 == 336577 1556 == 292281 1557333666 Recognize MPX instructions and bnd prefix. 1558333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch) 1559333817 Valgrind reports the memory areas written to by the SG_IO 1560 ioctl as untouched 1561334049 lzcnt fails silently (x86_32) 1562334384 Valgrind does not have support Little Endian support for 1563 IBM POWER PPC 64 1564334585 recvmmsg unhandled (+patch) (arm) 1565334705 sendmsg and recvmsg should guard against bogus msghdr fields. 1566334727 Build fails with -Werror=format-security 1567334788 clarify doc about --log-file initial program directory 1568334834 PPC64 Little Endian support, patch 2 1569334836 PPC64 Little Endian support, patch 3 testcase fixes 1570334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls 1571335034 Unhandled ioctl: HCIGETDEVLIST 1572335155 vgdb, fix error print statement. 1573335262 arm64: movi 8bit version is not supported 1574335263 arm64: dmb instruction is not implemented 1575335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind 1576335496 arm64: sbc/abc instructions are not implemented 1577335554 arm64: unhandled instruction: abs 1578335564 arm64: unhandled instruction: fcvtpu Xn, Sn 1579335735 arm64: unhandled instruction: cnt 1580335736 arm64: unhandled instruction: uaddlv 1581335848 arm64: unhandled instruction: {s,u}cvtf 1582335902 arm64: unhandled instruction: sli 1583335903 arm64: unhandled instruction: umull (vector) 1584336055 arm64: unhandled instruction: mov (element) 1585336062 arm64: unhandled instruction: shrn{,2} 1586336139 mip64: [...] valgrind hangs and spins on a single core [...] 1587336189 arm64: unhandled Instruction: mvn 1588336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU 1589336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type 1590336772 Make moans about unknown ioctls more informative 1591336957 Add a section about the Solaris/illumos port on the webpage 1592337094 ifunc wrapper is broken on ppc64 1593337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported 1594337528 leak check heuristic for block prefixed by length as 64bit number 1595337740 Implement additional Xen hypercalls 1596337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed. 1597337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230) 1598337871 deprecate --db-attach 1599338023 Add support for all V4L2/media ioctls 1600338024 inlined functions are not shown if DW_AT_ranges is used 1601338106 Add support for 'kcmp' syscall 1602338115 DRD: computed conflict set differs from actual after fork 1603338160 implement display of thread local storage in gdbsrv 1604338205 configure.ac and check for -Wno-tautological-compare 1605338300 coredumps are missing one byte of every segment 1606338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX 1607338499 --sim-hints parsing broken due to wrong order in tokens 1608338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7 1609338681 Unable to unwind through clone thread created on i386-linux 1610338698 race condition between gdbsrv and vgdb on startup 1611338703 helgrind on arm-linux gets false positives in dynamic loader 1612338791 alt dwz files can be relative of debug/main file 1613338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed 1614338932 build V-trunk with gcc-trunk 1615338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390 1616345079 Fix build problems in VEX/useful/test_main.c 1617n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling 1618n-i-bz s390x: Fix memory corruption for multithreaded applications 1619n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases 1620n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked 1621n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014) 1622n-i-bz drd manual: Document how to C++11 programs that use class "std::thread" 1623n-i-bz Add command-line option --default-suppressions 1624n-i-bz Add support for BLKDISCARDZEROES ioctl 1625n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions 1626n-i-bz Add support for sys_pivot_root and sys_unshare 1627 1628(3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428) 1629(3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503) 1630(3.10.0: 10 September 2014, vex r2950, valgrind r14514) 1631 1632 1633 1634Release 3.9.0 (31 October 2013) 1635~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 16363.9.0 is a feature release with many improvements and the usual 1637collection of bug fixes. 1638 1639This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, 1640PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android, 1641X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for 1642MacOSX 10.8 is significantly improved relative to the 3.8.0 release. 1643 1644* ================== PLATFORM CHANGES ================= 1645 1646* Support for MIPS64 LE and BE running Linux. Valgrind has been 1647 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions. 1648 1649* Support for MIPS DSP ASE on MIPS32 platforms. 1650 1651* Support for s390x Decimal Floating Point instructions on hosts that 1652 have the DFP facility installed. 1653 1654* Support for POWER8 (Power ISA 2.07) instructions 1655 1656* Support for Intel AVX2 instructions. This is available only on 64 1657 bit code. 1658 1659* Initial support for Intel Transactional Synchronization Extensions, 1660 both RTM and HLE. 1661 1662* Initial support for Hardware Transactional Memory on POWER. 1663 1664* Improved support for MacOSX 10.8 (64-bit only). Memcheck can now 1665 run large GUI apps tolerably well. 1666 1667* ==================== TOOL CHANGES ==================== 1668 1669* Memcheck: 1670 1671 - Improvements in handling of vectorised code, leading to 1672 significantly fewer false error reports. You need to use the flag 1673 --partial-loads-ok=yes to get the benefits of these changes. 1674 1675 - Better control over the leak checker. It is now possible to 1676 specify which leak kinds (definite/indirect/possible/reachable) 1677 should be displayed, which should be regarded as errors, and which 1678 should be suppressed by a given leak suppression. This is done 1679 using the options --show-leak-kinds=kind1,kind2,.., 1680 --errors-for-leak-kinds=kind1,kind2,.. and an optional 1681 "match-leak-kinds:" line in suppression entries, respectively. 1682 1683 Note that generated leak suppressions contain this new line and 1684 are therefore more specific than in previous releases. To get the 1685 same behaviour as previous releases, remove the "match-leak-kinds:" 1686 line from generated suppressions before using them. 1687 1688 - Reduced "possible leak" reports from the leak checker by the use 1689 of better heuristics. The available heuristics provide detection 1690 of valid interior pointers to std::stdstring, to new[] allocated 1691 arrays with elements having destructors and to interior pointers 1692 pointing to an inner part of a C++ object using multiple 1693 inheritance. They can be selected individually using the 1694 option --leak-check-heuristics=heur1,heur2,... 1695 1696 - Better control of stacktrace acquisition for heap-allocated 1697 blocks. Using the --keep-stacktraces option, it is possible to 1698 control independently whether a stack trace is acquired for each 1699 allocation and deallocation. This can be used to create better 1700 "use after free" errors or to decrease Valgrind's resource 1701 consumption by recording less information. 1702 1703 - Better reporting of leak suppression usage. The list of used 1704 suppressions (shown when the -v option is given) now shows, for 1705 each leak suppressions, how many blocks and bytes it suppressed 1706 during the last leak search. 1707 1708* Helgrind: 1709 1710 - False errors resulting from the use of statically initialised 1711 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc) 1712 have been removed. 1713 1714 - False errors resulting from the use of pthread_cond_waits that 1715 timeout, have been removed. 1716 1717* ==================== OTHER CHANGES ==================== 1718 1719* Some attempt to tune Valgrind's space requirements to the expected 1720 capabilities of the target: 1721 1722 - The default size of the translation cache has been reduced from 8 1723 sectors to 6 on Android platforms, since each sector occupies 1724 about 40MB when using Memcheck. 1725 1726 - The default size of the translation cache has been increased to 16 1727 sectors on all other platforms, reflecting the fact that large 1728 applications require instrumentation and storage of huge amounts 1729 of code. For similar reasons, the number of memory mapped 1730 segments that can be tracked has been increased by a factor of 6. 1731 1732 - In all cases, the maximum number of sectors in the translation 1733 cache can be controlled by the new flag --num-transtab-sectors. 1734 1735* Changes in how debug info (line numbers, etc) is read: 1736 1737 - Valgrind no longer temporarily mmaps the entire object to read 1738 from it. Instead, reading is done through a small fixed sized 1739 buffer. This avoids virtual memory usage spikes when Valgrind 1740 reads debuginfo from large shared objects. 1741 1742 - A new experimental remote debug info server. Valgrind can read 1743 debug info from a different machine (typically, a build host) 1744 where debuginfo objects are stored. This can save a lot of time 1745 and hassle when running Valgrind on resource-constrained targets 1746 (phones, tablets) when the full debuginfo objects are stored 1747 somewhere else. This is enabled by the --debuginfo-server= 1748 option. 1749 1750 - Consistency checking between main and debug objects can be 1751 disabled using the --allow-mismatched-debuginfo option. 1752 1753* Stack unwinding by stack scanning, on ARM. Unwinding by stack 1754 scanning can recover stack traces in some cases when the normal 1755 unwind mechanisms fail. Stack scanning is best described as "a 1756 nasty, dangerous and misleading hack" and so is disabled by default. 1757 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable 1758 and control it. 1759 1760* Detection and merging of recursive stack frame cycles. When your 1761 program has recursive algorithms, this limits the memory used by 1762 Valgrind for recorded stack traces and avoids recording 1763 uninteresting repeated calls. This is controlled by the command 1764 line option --merge-recursive-frame and by the monitor command 1765 "v.set merge-recursive-frames". 1766 1767* File name and line numbers for used suppressions. The list of used 1768 suppressions (shown when the -v option is given) now shows, for each 1769 used suppression, the file name and line number where the suppression 1770 is defined. 1771 1772* New and modified GDB server monitor features: 1773 1774 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND, 1775 that can be used to execute gdbserver monitor commands from the 1776 client program. 1777 1778 - A new monitor command, "v.info open_fds", that gives the list of 1779 open file descriptors and additional details. 1780 1781 - An optional message in the "v.info n_errs_found" monitor command, 1782 for example "v.info n_errs_found test 1234 finished", allowing a 1783 comment string to be added to the process output, perhaps for the 1784 purpose of separating errors of different tests or test phases. 1785 1786 - A new monitor command "v.info execontext" that shows information 1787 about the stack traces recorded by Valgrind. 1788 1789 - A new monitor command "v.do expensive_sanity_check_general" to run 1790 some internal consistency checks. 1791 1792* New flag --sigill-diagnostics to control whether a diagnostic 1793 message is printed when the JIT encounters an instruction it can't 1794 translate. The actual behavior -- delivery of SIGILL to the 1795 application -- is unchanged. 1796 1797* The maximum amount of memory that Valgrind can use on 64 bit targets 1798 has been increased from 32GB to 64GB. This should make it possible 1799 to run applications on Memcheck that natively require up to about 35GB. 1800 1801* ==================== FIXED BUGS ==================== 1802 1803The following bugs have been fixed or resolved. Note that "n-i-bz" 1804stands for "not in bugzilla" -- that is, a bug that was reported to us 1805but never got a bugzilla entry. We encourage you to file bugs in 1806bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 1807than mailing the developers (or mailing lists) directly -- bugs that 1808are not entered into bugzilla tend to get forgotten about or ignored. 1809 1810To see details of a given bug, visit 1811 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 1812where XXXXXX is the bug number as listed below. 1813 1814123837 system call: 4th argument is optional, depending on cmd 1815135425 memcheck should tell you where Freed blocks were Mallocd 1816164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small 1817207815 Adds some of the drm ioctls to syswrap-linux.c 1818251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP) 1819252955 Impossible to compile with ccache 1820253519 Memcheck reports auxv pointer accesses as invalid reads. 1821263034 Crash when loading some PPC64 binaries 1822269599 Increase deepest backtrace 1823274695 s390x: Support "compare to/from logical" instructions (z196) 1824275800 s390x: Autodetect cache info (part 2) 1825280271 Valgrind reports possible memory leaks on still-reachable std::string 1826284540 Memcheck shouldn't count suppressions matching still-reachable [..] 1827289578 Backtraces with ARM unwind tables (stack scan flags) 1828296311 Wrong stack traces due to -fomit-frame-pointer (x86) 1829304832 ppc32: build failure 1830305431 Use find_buildid shdr fallback for separate .debug files 1831305728 Add support for AVX2 instructions 1832305948 ppc64: code generation for ShlD64 / ShrD64 asserts 1833306035 s390x: Fix IR generation for LAAG and friends 1834306054 s390x: Condition code computation for convert-to-int/logical 1835306098 s390x: alternate opcode form for convert to/from fixed 1836306587 Fix cache line detection from auxiliary vector for PPC. 1837306783 Mips unhandled syscall : 4025 / 4079 / 4182 1838307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al) 1839307082 HG false positive: pthread_cond_destroy: destruction of unknown CV 1840307101 sys_capget second argument can be NULL 1841307103 sys_openat: If pathname is absolute, then dirfd is ignored. 1842307106 amd64->IR: f0 0f c0 02 (lock xadd byte) 1843307113 s390x: DFP support 1844307141 valgrind does't work in mips-linux system 1845307155 filter_gdb should filter out syscall-template.S T_PSEUDO 1846307285 x86_amd64 feature test for avx in test suite is wrong 1847307290 memcheck overlap testcase needs memcpy version filter 1848307463 Please add "&limit=0" to the "all open bugs" link 1849307465 --show-possibly-lost=no should reduce the error count / exit code 1850307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..] 1851307729 pkgconfig support broken valgrind.pc 1852307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr 1853307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang 1854308089 Unhandled syscall on ppc64: prctl 1855308135 PPC32 MPC8xx has 16 bytes cache size 1856308321 testsuite memcheck filter interferes with gdb_filter 1857308333 == 307106 1858308341 vgdb should report process exit (or fatal signal) 1859308427 s390 memcheck reports tsearch cjump/cmove depends on uninit 1860308495 Remove build dependency on installed Xen headers 1861308573 Internal error on 64-bit instruction executed in 32-bit mode 1862308626 == 308627 1863308627 pmovmskb validity bit propagation is imprecise 1864308644 vgdb command for having the info for the track-fds option 1865308711 give more info about aspacemgr and arenas in out_of_memory 1866308717 ARM: implement fixed-point VCVT.F64.[SU]32 1867308718 ARM implement SMLALBB family of instructions 1868308886 Missing support for PTRACE_SET/GETREGSET 1869308930 syscall name_to_handle_at (303 on amd64) not handled 1870309229 V-bit tester does not report number of tests generated 1871309323 print unrecognized instuction on MIPS 1872309425 Provide a --sigill-diagnostics flag to suppress illegal [..] 1873309427 SSE optimized stpncpy trigger uninitialised value [..] errors 1874309430 Self hosting ppc64 encounters a vassert error on operand type 1875309600 valgrind is a bit confused about 0-sized sections 1876309823 Generate errors for still reachable blocks 1877309921 PCMPISTRI validity bit propagation is imprecise 1878309922 none/tests/ppc64/test_dfp5 sometimes fails 1879310169 The Iop_CmpORD class of Iops is not supported by the vbit checker. 1880310424 --read-var-info does not properly describe static variables 1881310792 search additional path for debug symbols 1882310931 s390x: Message-security assist (MSA) instruction extension [..] 1883311100 PPC DFP implementation of the integer operands is inconsistent [..] 1884311318 ARM: "128-bit constant is not implemented" error message 1885311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..] 1886311690 V crashes because it redirects branches inside of a redirected function 1887311880 x86_64: make regtest hangs at shell_valid1 1888311922 WARNING: unhandled syscall: 170 1889311933 == 251569 1890312171 ppc: insn selection for DFP 1891312571 Rounding mode call wrong for the DFP Iops [..] 1892312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..] 1893312913 Dangling pointers error should also report the alloc stack trace 1894312980 Building on Mountain Lion generates some compiler warnings 1895313267 Adding MIPS64/Linux port to Valgrind 1896313348 == 251569 1897313354 == 251569 1898313811 Buffer overflow in assert_fail 1899314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix 1900314269 ppc: dead code in insn selection 1901314718 ARM: implement integer divide instruction (sdiv and udiv) 1902315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi= 1903315441 sendmsg syscall should ignore unset msghdr msg_flags 1904315534 msgrcv inside a thread causes valgrind to hang (block) 1905315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed 1906315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT) 1907315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64) 1908315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section 1909316144 valgrind.1 manpage contains unknown ??? strings [..] 1910316145 callgrind command line options in manpage reference (unknown) [..] 1911316145 callgrind command line options in manpage reference [..] 1912316181 drd: Fixed a 4x slowdown for certain applications 1913316503 Valgrind does not support SSE4 "movntdqa" instruction 1914316535 Use of |signed int| instead of |size_t| in valgrind messages 1915316696 fluidanimate program of parsec 2.1 stuck 1916316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled 1917317091 Use -Wl,-Ttext-segment when static linking if possible [..] 1918317186 "Impossible happens" when occurs VCVT instruction on ARM 1919317318 Support for Threading Building Blocks "scalable_malloc" 1920317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps) 1921317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs 1922317463 bmi testcase IR SANITY CHECK FAILURE 1923317506 memcheck/tests/vbit-test fails with unknown opcode after [..] 1924318050 libmpiwrap fails to compile with out-of-source build 1925318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER 1926318643 annotate_trace_memory tests infinite loop on arm and ppc [..] 1927318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F 1928318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16) 1929318932 Add missing PPC64 and PPC32 system call support 1930319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled 1931319395 Crash with unhandled instruction on STRT (Thumb) instructions 1932319494 VEX Makefile-gcc standalone build update after r2702 1933319505 [MIPSEL] Crash: unhandled UNRAY operator. 1934319858 disInstr(thumb): unhandled instruction on instruction STRBT 1935319932 disInstr(thumb): unhandled instruction on instruction STRHT 1936320057 Problems when we try to mmap more than 12 memory pages on MIPS32 1937320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised 1938320083 disInstr(thumb): unhandled instruction on instruction LDRBT 1939320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding 1940320131 WARNING: unhandled syscall: 369 on ARM (prlimit64) 1941320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR 1942320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname' 1943320895 add fanotify support (patch included) 1944320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction 1945321065 Valgrind updates for Xen 4.3 1946321148 Unhandled instruction: PLI (Thumb 1, 2, 3) 1947321363 Unhandled instruction: SSAX (ARM + Thumb) 1948321364 Unhandled instruction: SXTAB16 (ARM + Thumb) 1949321466 Unhandled instruction: SHASX (ARM + Thumb) 1950321467 Unhandled instruction: SHSAX (ARM + Thumb) 1951321468 Unhandled instruction: SHSUB16 (ARM + Thumb) 1952321619 Unhandled instruction: SHSUB8 (ARM + Thumb) 1953321620 Unhandled instruction: UASX (ARM + Thumb) 1954321621 Unhandled instruction: USAX (ARM + Thumb) 1955321692 Unhandled instruction: UQADD16 (ARM + Thumb) 1956321693 Unhandled instruction: LDRSBT (Thumb) 1957321694 Unhandled instruction: UQASX (ARM + Thumb) 1958321696 Unhandled instruction: UQSAX (Thumb + ARM) 1959321697 Unhandled instruction: UHASX (ARM + Thumb) 1960321703 Unhandled instruction: UHSAX (ARM + Thumb) 1961321704 Unhandled instruction: REVSH (ARM + Thumb) 1962321730 Add cg_diff and cg_merge man pages 1963321738 Add vgdb and valgrind-listener man pages 1964321814 == 315545 1965321891 Unhandled instruction: LDRHT (Thumb) 1966321960 pthread_create() then alloca() causing invalid stack write errors 1967321969 ppc32 and ppc64 don't support [lf]setxattr 1968322254 Show threadname together with tid if set by application 1969322294 Add initial support for IBM Power ISA 2.07 1970322368 Assertion failure in wqthread_hijack under OS X 10.8 1971322563 vex mips->IR: 0x70 0x83 0xF0 0x3A 1972322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr 1973322851 0bXXX binary literal syntax is not standard 1974323035 Unhandled instruction: LDRSHT(Thumb) 1975323036 Unhandled instruction: SMMLS (ARM and Thumb) 1976323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..] 1977323175 Unhandled instruction: SMLALD (ARM + Thumb) 1978323177 Unhandled instruction: SMLSLD (ARM + Thumb) 1979323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..] 1980323437 Phase 2 support for IBM Power ISA 2.07 1981323713 Support mmxext (integer sse) subset on i386 (athlon) 1982323803 Transactional memory instructions are not supported for Power 1983323893 SSE3 not available on amd cpus in valgrind 1984323905 Probable false positive from Valgrind/drd on close() 1985323912 valgrind.h header isn't compatible for mingw64 1986324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions 1987324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..] 1988324181 mmap does not handle MAP_32BIT 1989324227 memcheck false positive leak when a thread calls exit+block [..] 1990324421 Support for fanotify API on ARM architecture 1991324514 gdbserver monitor cmd output behaviour consistency [..] 1992324518 ppc64: Emulation of dcbt instructions does not handle [..] 1993324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64 1994324582 When access is made to freed memory, report both allocation [..] 1995324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo. 1996324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc 1997324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs 1998324834 Unhandled instructions in Microsoft C run-time for x86_64 1999324894 Phase 3 support for IBM Power ISA 2.07 2000326091 drd: Avoid false race reports from optimized strlen() impls 2001326113 valgrind libvex hwcaps error on AMD64 2002n-i-bz Some wrong command line options could be ignored 2003n-i-bz patch to allow fair-sched on android 2004n-i-bz report error for vgdb snapshot requested before execution 2005n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android 2006 2007(3.9.0: 31 October 2013, vex r2796, valgrind r13708) 2008 2009 2010 2011Release 3.8.1 (19 September 2012) 2012~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 20133.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0 2014that occur moderately frequently in real use cases, adds support for 2015some missing instructions on ARM, and fixes a deadlock condition on 2016MacOSX. If you package or deliver 3.8.0 for others to use, you might 2017want to consider upgrading to 3.8.1 instead. 2018 2019The following bugs have been fixed or resolved. Note that "n-i-bz" 2020stands for "not in bugzilla" -- that is, a bug that was reported to us 2021but never got a bugzilla entry. We encourage you to file bugs in 2022bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 2023than mailing the developers (or mailing lists) directly -- bugs that 2024are not entered into bugzilla tend to get forgotten about or ignored. 2025 2026To see details of a given bug, visit 2027 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 2028where XXXXXX is the bug number as listed below. 2029 2030284004 == 301281 2031289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS) 2032295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT) 2033298281 wcslen causes false(?) uninitialised value warnings 2034301281 valgrind hangs on OS X when the process calls system() 2035304035 disInstr(arm): unhandled instruction 0xE1023053 2036304867 implement MOVBE instruction in x86 mode 2037304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping 2038305042 amd64: implement 0F 7F encoding of movq between two registers 2039305199 ARM: implement QDADD and QDSUB 2040305321 amd64->IR: 0xF 0xD 0xC (prefetchw) 2041305513 killed by fatal signal: SIGSEGV 2042305690 DRD reporting invalid semaphore when sem_trywait fails 2043305926 Invalid alignment checks for some AVX instructions 2044306297 disInstr(thumb): unhandled instruction 0xE883 0x000C 2045306310 3.8.0 release tarball missing some files 2046306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns 2047306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF 2048n-i-bz shmat of a segment > 4Gb does not work 2049n-i-bz simulate_control_c script wrong USR1 signal number on mips 2050n-i-bz vgdb ptrace calls wrong on mips [...] 2051n-i-bz Fixes for more MPI false positives 2052n-i-bz exp-sgcheck's memcpy causes programs to segfault 2053n-i-bz OSX build w/ clang: asserts at startup 2054n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB 2055n-i-bz fix a couple of union tag-vs-field mixups 2056n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll 2057 2058The following bugs were fixed in 3.8.0 but not listed in this NEWS 2059file at the time: 2060 2061254088 Valgrind should know about UD2 instruction 2062301280 == 254088 2063301902 == 254088 2064304754 NEWS blows TeX's little mind 2065 2066(3.8.1: 19 September 2012, vex r2537, valgrind r12996) 2067 2068 2069 2070Release 3.8.0 (10 August 2012) 2071~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 20723.8.0 is a feature release with many improvements and the usual 2073collection of bug fixes. 2074 2075This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, 2076PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android, 2077X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent 2078distros and toolchain components (glibc 2.16, gcc 4.7) has been added. 2079There is initial support for MacOSX 10.8, but it is not usable for 2080serious work at present. 2081 2082* ================== PLATFORM CHANGES ================= 2083 2084* Support for MIPS32 platforms running Linux. Valgrind has been 2085 tested on MIPS32 and MIPS32r2 platforms running different Debian 2086 Squeeze and MeeGo distributions. Both little-endian and big-endian 2087 cores are supported. The tools Memcheck, Massif and Lackey have 2088 been tested and are known to work. See README.mips for more details. 2089 2090* Preliminary support for Android running on x86. 2091 2092* Preliminary (as-yet largely unusable) support for MacOSX 10.8. 2093 2094* Support for Intel AVX instructions and for AES instructions. This 2095 support is available only for 64 bit code. 2096 2097* Support for POWER Decimal Floating Point instructions. 2098 2099* ==================== TOOL CHANGES ==================== 2100 2101* Non-libc malloc implementations are now supported. This is useful 2102 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind). 2103 Using the new option --soname-synonyms, such tools can be informed 2104 that the malloc implementation is either linked statically into the 2105 executable, or is present in some other shared library different 2106 from libc.so. This makes it possible to process statically linked 2107 programs, and programs using other malloc libraries, for example 2108 TCMalloc or JEMalloc. 2109 2110* For tools that provide their own replacement for malloc et al, the 2111 option --redzone-size=<number> allows users to specify the size of 2112 the padding blocks (redzones) added before and after each client 2113 allocated block. Smaller redzones decrease the memory needed by 2114 Valgrind. Bigger redzones increase the chance to detect blocks 2115 overrun or underrun. Prior to this change, the redzone size was 2116 hardwired to 16 bytes in Memcheck. 2117 2118* Memcheck: 2119 2120 - The leak_check GDB server monitor command now can 2121 control the maximum nr of loss records to output. 2122 2123 - Reduction of memory use for applications allocating 2124 many blocks and/or having many partially defined bytes. 2125 2126 - Addition of GDB server monitor command 'block_list' that lists 2127 the addresses/sizes of the blocks of a leak search loss record. 2128 2129 - Addition of GDB server monitor command 'who_points_at' that lists 2130 the locations pointing at a block. 2131 2132 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will 2133 detect an invalid access of these redzones, by marking them 2134 noaccess. Similarly, if a redzone size is given for a memory 2135 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access. 2136 This still allows to find some bugs if the user has forgotten to 2137 mark the pool superblock noaccess. 2138 2139 - Performance of memory leak check has been improved, especially in 2140 cases where there are many leaked blocks and/or many suppression 2141 rules used to suppress leak reports. 2142 2143 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to 2144 more precise analysis, which is important for LLVM/Clang 2145 generated code. This is at the cost of somewhat reduced 2146 performance. Note there is no change to analysis precision or 2147 costs on Linux targets. 2148 2149* DRD: 2150 2151 - Added even more facilities that can help finding the cause of a data 2152 race, namely the command-line option --ptrace-addr and the macro 2153 DRD_STOP_TRACING_VAR(x). More information can be found in the manual. 2154 2155 - Fixed a subtle bug that could cause false positive data race reports. 2156 2157* ==================== OTHER CHANGES ==================== 2158 2159* The C++ demangler has been updated so as to work well with C++ 2160 compiled by up to at least g++ 4.6. 2161 2162* Tool developers can make replacement/wrapping more flexible thanks 2163 to the new option --soname-synonyms. This was reported above, but 2164 in fact is very general and applies to all function 2165 replacement/wrapping, not just to malloc-family functions. 2166 2167* Round-robin scheduling of threads can be selected, using the new 2168 option --fair-sched= yes. Prior to this change, the pipe-based 2169 thread serialisation mechanism (which is still the default) could 2170 give very unfair scheduling. --fair-sched=yes improves 2171 responsiveness of interactive multithreaded applications, and 2172 improves repeatability of results from the thread checkers Helgrind 2173 and DRD. 2174 2175* For tool developers: support to run Valgrind on Valgrind has been 2176 improved. We can now routinely Valgrind on Helgrind or Memcheck. 2177 2178* gdbserver now shows the float shadow registers as integer 2179 rather than float values, as the shadow values are mostly 2180 used as bit patterns. 2181 2182* Increased limit for the --num-callers command line flag to 500. 2183 2184* Performance improvements for error matching when there are many 2185 suppression records in use. 2186 2187* Improved support for DWARF4 debugging information (bug 284184). 2188 2189* Initial support for DWZ compressed Dwarf debug info. 2190 2191* Improved control over the IR optimiser's handling of the tradeoff 2192 between performance and precision of exceptions. Specifically, 2193 --vex-iropt-precise-memory-exns has been removed and replaced by 2194 --vex-iropt-register-updates, with extended functionality. This 2195 allows the Valgrind gdbserver to always show up to date register 2196 values to GDB. 2197 2198* Modest performance gains through the use of translation chaining for 2199 JIT-generated code. 2200 2201* ==================== FIXED BUGS ==================== 2202 2203The following bugs have been fixed or resolved. Note that "n-i-bz" 2204stands for "not in bugzilla" -- that is, a bug that was reported to us 2205but never got a bugzilla entry. We encourage you to file bugs in 2206bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 2207than mailing the developers (or mailing lists) directly -- bugs that 2208are not entered into bugzilla tend to get forgotten about or ignored. 2209 2210To see details of a given bug, visit 2211 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 2212where XXXXXX is the bug number as listed below. 2213 2214197914 Building valgrind from svn now requires automake-1.10 2215203877 increase to 16Mb maximum allowed alignment for memalign et al 2216219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc) 2217247386 make perf does not run all performance tests 2218270006 Valgrind scheduler unfair 2219270777 Adding MIPS/Linux port to Valgrind 2220270796 s390x: Removed broken support for the TS insn 2221271438 Fix configure for proper SSE4.2 detection 2222273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions 2223273475 Add support for AVX instructions 2224274078 improved configure logic for mpicc 2225276993 fix mremap 'no thrash checks' 2226278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag 2227281482 memcheck incorrect byte allocation count in realloc() for silly argument 2228282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit 2229283413 Fix wrong sanity check 2230283671 Robustize alignment computation in LibVEX_Alloc 2231283961 Adding support for some HCI IOCTLs 2232284124 parse_type_DIE: confused by: DWARF 4 2233284864 == 273475 (Add support for AVX instructions) 2234285219 Too-restrictive constraints for Thumb2 "SP plus/minus register" 2235285662 (MacOSX): Memcheck needs to replace memcpy/memmove 2236285725 == 273475 (Add support for AVX instructions) 2237286261 add wrapper for linux I2C_RDWR ioctl 2238286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings 2239286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails 2240286384 configure fails "checking for a supported version of gcc" 2241286497 == 273475 (Add support for AVX instructions) 2242286596 == 273475 (Add support for AVX instructions) 2243286917 disInstr(arm): unhandled instruction: QADD (also QSUB) 2244287175 ARM: scalar VFP fixed-point VCVT instructions not handled 2245287260 Incorrect conditional jump or move depends on uninitialised value(s) 2246287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW) 2247287307 == 273475 (Add support for AVX instructions) 2248287858 VG_(strerror): unknown error 2249288298 (MacOSX) unhandled syscall shm_unlink 2250288995 == 273475 (Add support for AVX instructions) 2251289470 Loading of large Mach-O thin binaries fails. 2252289656 == 273475 (Add support for AVX instructions) 2253289699 vgdb connection in relay mode erroneously closed due to buffer overrun 2254289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters) 2255289839 s390x: Provide support for unicode conversion instructions 2256289939 monitor cmd 'leak_check' with details about leaked or reachable blocks 2257290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm" 2258290655 Add support for AESKEYGENASSIST instruction 2259290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage 2260290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM 2261291253 ES register not initialised in valgrind simulation 2262291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch) 2263291865 s390x: Support the "Compare Double and Swap" family of instructions 2264292300 == 273475 (Add support for AVX instructions) 2265292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid 2266292493 == 273475 (Add support for AVX instructions) 2267292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support 2268292627 Missing support for some SCSI ioctls 2269292628 none/tests/x86/bug125959-x86.c triggers undefined behavior 2270292841 == 273475 (Add support for AVX instructions) 2271292993 implement the getcpu syscall on amd64-linux 2272292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2 2273293088 Add some VEX sanity checks for ppc64 unhandled instructions 2274293751 == 290655 (Add support for AESKEYGENASSIST instruction) 2275293754 PCMPxSTRx not implemented for 16-bit characters 2276293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters) 2277293808 CLFLUSH not supported by latest VEX for amd64 2278294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...) 2279294048 MPSADBW instruction not implemented 2280294055 regtest none/tests/shell fails when locale is not set to C 2281294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM 2282294190 --vgdb-error=xxx can be out of sync with errors shown to the user 2283294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions 2284294260 disInstr_AMD64: disInstr miscalculated next %rip 2285294523 --partial-loads-ok=yes causes false negatives 2286294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A 2287294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83 2288294812 patch allowing to run (on x86 at least) helgrind/drd on tool. 2289295089 can not annotate source for both helgrind and drd 2290295221 POWER Processor decimal floating point instruction support missing 2291295427 building for i386 with clang on darwin11 requires "-new_linker linker" 2292295428 coregrind/m_main.c has incorrect x86 assembly for darwin 2293295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed 2294295617 ARM - Add some missing syscalls 2295295799 Missing \n with get_vbits in gdbserver when line is % 80 [...] 2296296229 Linux user input device ioctls missing wrappers 2297296318 ELF Debug info improvements (more than one rx/rw mapping) 2298296422 Add translation chaining support 2299296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES) 2300296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper 2301296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains 2302297078 gdbserver signal handling problems [..] 2303297147 drd false positives on newly allocated memory 2304297329 disallow decoding of IBM Power DFP insns on some machines 2305297497 POWER Processor decimal floating point instruction support missing 2306297701 Another alias for strncasecmp_l in libc-2.13.so 2307297911 'invalid write' not reported when using APIs for custom mem allocators. 2308297976 s390x: revisit EX implementation 2309297991 Valgrind interferes with mmap()+ftell() 2310297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux) 2311297993 Fix compilation of valgrind with gcc -g3. 2312298080 POWER Processor DFP support missing, part 3 2313298227 == 273475 (Add support for AVX instructions) 2314298335 == 273475 (Add support for AVX instructions) 2315298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq) 2316298394 s390x: Don't bail out on an unknown machine model. [..] 2317298421 accept4() syscall (366) support is missing for ARM 2318298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45 2319298732 valgrind installation problem in ubuntu with kernel version 3.x 2320298862 POWER Processor DFP instruction support missing, part 4 2321298864 DWARF reader mis-parses DW_FORM_ref_addr 2322298943 massif asserts with --pages-as-heap=yes when brk is changing [..] 2323299053 Support DWARF4 DW_AT_high_pc constant form 2324299104 == 273475 (Add support for AVX instructions) 2325299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed. 2326299629 dup3() syscall (358) support is missing for ARM 2327299694 POWER Processor DFP instruction support missing, part 5 2328299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests 2329299803 == 273475 (Add support for AVX instructions) 2330299804 == 273475 (Add support for AVX instructions) 2331299805 == 273475 (Add support for AVX instructions) 2332300140 ARM - Missing (T1) SMMUL 2333300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping)) 2334300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed. 2335300414 FCOM and FCOMP unimplemented for amd64 guest 2336301204 infinite loop in canonicaliseSymtab with ifunc symbol 2337301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc) 2338301265 add x86 support to Android build 2339301984 configure script doesn't detect certain versions of clang 2340302205 Fix compiler warnings for POWER VEX code and POWER test cases 2341302287 Unhandled movbe instruction on Atom processors 2342302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result 2343302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0. 2344302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess 2345302656 == 273475 (Add support for AVX instructions) 2346302709 valgrind for ARM needs extra tls support for android emulator [..] 2347302827 add wrapper for CDROM_GET_CAPABILITY 2348302901 Valgrind crashes with dwz optimized debuginfo 2349302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite 2350303116 Add support for the POWER instruction popcntb 2351303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions. 2352303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code 2353303466 == 273475 (Add support for AVX instructions) 2354303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP) 2355303963 strstr() function produces wrong results under valgrind callgrind 2356304054 CALL_FN_xx macros need to enforce stack alignment 2357304561 tee system call not supported 2358715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#) 2359n-i-bz Add missing gdbserver xml files for shadow registers for ppc32 2360n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts 2361n-i-bz Fix assert in gdbserver for watchpoints watching the same address 2362n-i-bz Fix false positive in sys_clone on amd64 when optional args [..] 2363n-i-bz s390x: Shadow registers can now be examined using vgdb 2364 2365(3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865) 2366(3.8.0: 10 August 2012, vex r2465, valgrind r12866) 2367 2368 2369 2370Release 3.7.0 (5 November 2011) 2371~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 23723.7.0 is a feature release with many significant improvements and the 2373usual collection of bug fixes. 2374 2375This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, 2376PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin. 2377Support for recent distros and toolchain components (glibc 2.14, gcc 23784.6, MacOSX 10.7) has been added. 2379 2380* ================== PLATFORM CHANGES ================= 2381 2382* Support for IBM z/Architecture (s390x) running Linux. Valgrind can 2383 analyse 64-bit programs running on z/Architecture. Most user space 2384 instructions up to and including z10 are supported. Valgrind has 2385 been tested extensively on z9, z10, and z196 machines running SLES 2386 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are 2387 known to work well. Callgrind, Helgrind, and DRD work reasonably 2388 well on z9 and later models. See README.s390 for more details. 2389 2390* Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and 2391 64-bit processes are supported. Some complex threaded applications 2392 (Firefox) are observed to hang when run as 32 bit applications, 2393 whereas 64-bit versions run OK. The cause is unknown. Memcheck 2394 will likely report some false errors. In general, expect some rough 2395 spots. This release also supports MacOSX 10.6, but drops support 2396 for 10.5. 2397 2398* Preliminary support for Android (on ARM). Valgrind can now run 2399 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See 2400 README.android for more details, plus instructions on how to get 2401 started. 2402 2403* Support for the IBM Power ISA 2.06 (Power7 instructions) 2404 2405* General correctness and performance improvements for ARM/Linux, and, 2406 by extension, ARM/Android. 2407 2408* Further solidification of support for SSE 4.2 in 64-bit mode. AVX 2409 instruction set support is under development but is not available in 2410 this release. 2411 2412* Support for AIX5 has been removed. 2413 2414* ==================== TOOL CHANGES ==================== 2415 2416* Memcheck: some incremental changes: 2417 2418 - reduction of memory use in some circumstances 2419 2420 - improved handling of freed memory, which in some circumstances 2421 can cause detection of use-after-free that would previously have 2422 been missed 2423 2424 - fix of a longstanding bug that could cause false negatives (missed 2425 errors) in programs doing vector saturated narrowing instructions. 2426 2427* Helgrind: performance improvements and major memory use reductions, 2428 particularly for large, long running applications which perform many 2429 synchronisation (lock, unlock, etc) events. Plus many smaller 2430 changes: 2431 2432 - display of locksets for both threads involved in a race 2433 2434 - general improvements in formatting/clarity of error messages 2435 2436 - addition of facilities and documentation regarding annotation 2437 of thread safe reference counted C++ classes 2438 2439 - new flag --check-stack-refs=no|yes [yes], to disable race checking 2440 on thread stacks (a performance hack) 2441 2442 - new flag --free-is-write=no|yes [no], to enable detection of races 2443 where one thread accesses heap memory but another one frees it, 2444 without any coordinating synchronisation event 2445 2446* DRD: enabled XML output; added support for delayed thread deletion 2447 in order to detect races that occur close to the end of a thread 2448 (--join-list-vol); fixed a memory leak triggered by repeated client 2449 memory allocatation and deallocation; improved Darwin support. 2450 2451* exp-ptrcheck: this tool has been renamed to exp-sgcheck 2452 2453* exp-sgcheck: this tool has been reduced in scope so as to improve 2454 performance and remove checking that Memcheck does better. 2455 Specifically, the ability to check for overruns for stack and global 2456 arrays is unchanged, but the ability to check for overruns of heap 2457 blocks has been removed. The tool has accordingly been renamed to 2458 exp-sgcheck ("Stack and Global Array Checking"). 2459 2460* ==================== OTHER CHANGES ==================== 2461 2462* GDB server: Valgrind now has an embedded GDB server. That means it 2463 is possible to control a Valgrind run from GDB, doing all the usual 2464 things that GDB can do (single stepping, breakpoints, examining 2465 data, etc). Tool-specific functionality is also available. For 2466 example, it is possible to query the definedness state of variables 2467 or memory from within GDB when running Memcheck; arbitrarily large 2468 memory watchpoints are supported, etc. To use the GDB server, start 2469 Valgrind with the flag --vgdb-error=0 and follow the on-screen 2470 instructions. 2471 2472* Improved support for unfriendly self-modifying code: a new option 2473 --smc-check=all-non-file is available. This adds the relevant 2474 consistency checks only to code that originates in non-file-backed 2475 mappings. In effect this confines the consistency checking only to 2476 code that is or might be JIT generated, and avoids checks on code 2477 that must have been compiled ahead of time. This significantly 2478 improves performance on applications that generate code at run time. 2479 2480* It is now possible to build a working Valgrind using Clang-2.9 on 2481 Linux. 2482 2483* new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING. 2484 These enable and disable error reporting on a per-thread, and 2485 nestable, basis. This is useful for hiding errors in particularly 2486 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c) 2487 now uses this facility. 2488 2489* Added the --mod-funcname option to cg_diff. 2490 2491* ==================== FIXED BUGS ==================== 2492 2493The following bugs have been fixed or resolved. Note that "n-i-bz" 2494stands for "not in bugzilla" -- that is, a bug that was reported to us 2495but never got a bugzilla entry. We encourage you to file bugs in 2496bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than 2497mailing the developers (or mailing lists) directly -- bugs that are 2498not entered into bugzilla tend to get forgotten about or ignored. 2499 2500To see details of a given bug, visit 2501https://bugs.kde.org/show_bug.cgi?id=XXXXXX 2502where XXXXXX is the bug number as listed below. 2503 2504 79311 malloc silly arg warning does not give stack trace 2505210935 port valgrind.h (not valgrind) to win32 to support client requests 2506214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10 2507243404 Port to zSeries 2508243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER() 2509247223 non-x86: Suppress warning: 'regparm' attribute directive ignored 2510250101 huge "free" memory usage due to m_mallocfree.c fragmentation 2511253206 Some fixes for the faultstatus testcase 2512255223 capget testcase fails when running as root 2513256703 xlc_dbl_u32.c testcase broken 2514256726 Helgrind tests have broken inline asm 2515259977 == 214223 (Valgrind segfaults doing __builtin_longjmp) 2516264800 testcase compile failure on zseries 2517265762 make public VEX headers compilable by G++ 3.x 2518265771 assertion in jumps.c (r11523) fails with glibc-2.3 2519266753 configure script does not give the user the option to not use QtCore 2520266931 gen_insn_test.pl is broken 2521266961 ld-linux.so.2 i?86-linux strlen issues 2522266990 setns instruction causes false positive 2523267020 Make directory for temporary files configurable at run-time. 2524267342 == 267997 (segmentation fault on Mac OS 10.6) 2525267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed 2526267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed. 2527267488 regtest: darwin support for 64-bit build 2528267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools 2529267630 Add support for IBM Power ISA 2.06 -- stage 1 2530267769 == 267997 (Darwin: memcheck triggers segmentation fault) 2531267819 Add client request for informing the core about reallocation 2532267925 laog data structure quadratic for a single sequence of lock 2533267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed 2534267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1 2535268513 missed optimizations in fold_Expr 2536268619 s390x: fpr - gpr transfer facility 2537268620 s390x: reconsider "long displacement" requirement 2538268621 s390x: improve IR generation for XC 2539268715 s390x: FLOGR is not universally available 2540268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4) 2541268930 s390x: MHY is not universally available 2542269078 arm->IR: unhandled instruction SUB (SP minus immediate/register) 2543269079 Support ptrace system call on ARM 2544269144 missing "Bad option" error message 2545269209 conditional load and store facility (z196) 2546269354 Shift by zero on x86 can incorrectly clobber CC_NDEP 2547269641 == 267997 (valgrind segfaults immediately (segmentation fault)) 2548269736 s390x: minor code generation tweaks 2549269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..) 2550269863 s390x: remove unused function parameters 2551269864 s390x: tweak s390_emit_load_cc 2552269884 == 250101 (overhead for huge blocks exhausts space too soon) 2553270082 s390x: Make sure to point the PSW address to the next address on SIGILL 2554270115 s390x: rewrite some testcases 2555270309 == 267997 (valgrind crash on startup) 2556270320 add support for Linux FIOQSIZE ioctl() call 2557270326 segfault while trying to sanitize the environment passed to execle 2558270794 IBM POWER7 support patch causes regression in none/tests 2559270851 IBM POWER7 fcfidus instruction causes memcheck to fail 2560270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app 2561270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix 2562270959 s390x: invalid use of R0 as base register 2563271042 VSX configure check fails when it should not 2564271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21 2565271259 s390x: fix code confusion 2566271337 == 267997 (Valgrind segfaults on MacOS X) 2567271385 s390x: Implement Ist_MBE 2568271501 s390x: misc cleanups 2569271504 s390x: promote likely and unlikely 2570271579 ppc: using wrong enum type 2571271615 unhandled instruction "popcnt" (arch=amd10h) 2572271730 Fix bug when checking ioctls: duplicate check 2573271776 s390x: provide STFLE instruction support 2574271779 s390x: provide clock instructions like STCK 2575271799 Darwin: ioctls without an arg report a memory error 2576271820 arm: fix type confusion 2577271917 pthread_cond_timedwait failure leads to not-locked false positive 2578272067 s390x: fix DISP20 macro 2579272615 A typo in debug output in mc_leakcheck.c 2580272661 callgrind_annotate chokes when run from paths containing regex chars 2581272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup) 2582272955 Unhandled syscall error for pwrite64 on ppc64 arch 2583272967 make documentation build-system more robust 2584272986 Fix gcc-4.6 warnings with valgrind.h 2585273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case) 2586273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 2587273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039) 2588273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...' 2589273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer' 2590273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169) 2591273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction) 2592273778 exp-ptrcheck: unhandled sysno == 259 2593274089 exp-ptrcheck: unhandled sysno == 208 2594274378 s390x: Various dispatcher tweaks 2595274447 WARNING: unhandled syscall: 340 2596274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66 2597274784 == 267997 (valgrind ls -l results in Segmentation Fault) 2598274926 valgrind does not build against linux-3 2599275148 configure FAIL with glibc-2.14 2600275151 Fedora 15 / glibc-2.14 'make regtest' FAIL 2601275168 Make Valgrind work for MacOSX 10.7 Lion 2602275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al) 2603275278 valgrind does not build on Linux kernel 3.0.* due to silly 2604275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64 2605275308 Fix implementation for ppc64 fres instruc 2606275339 s390x: fix testcase compile warnings 2607275517 s390x: Provide support for CKSM instruction 2608275710 s390x: get rid of redundant address mode calculation 2609275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall) 2610275852 == 250101 (valgrind uses all swap space and is killed) 2611276784 Add support for IBM Power ISA 2.06 -- stage 3 2612276987 gdbsrv: fix tests following recent commits 2613277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a 2614277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink 2615277471 Unhandled syscall: 340 2616277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...) 2617277653 ARM: support Thumb2 PLD instruction 2618277663 ARM: NEON float VMUL by scalar incorrect 2619277689 ARM: tests for VSTn with register post-index are broken 2620277694 ARM: BLX LR instruction broken in ARM mode 2621277780 ARM: VMOV.F32 (immediate) instruction is broken 2622278057 fuse filesystem syscall deadlocks 2623278078 Unimplemented syscall 280 on ppc32 2624278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands 2625278454 VALGRIND_STACK_DEREGISTER has wrong output type 2626278502 == 275284 (Valgrind confuses memcpy() and memmove()) 2627278892 gdbsrv: factorize gdb version handling, fix doc and typos 2628279027 Support for MVCL and CLCL instruction 2629279027 s390x: Provide support for CLCL and MVCL instructions 2630279062 Remove a redundant check in the insn selector for ppc. 2631279071 JDK creates PTEST with redundant REX.W prefix 2632279212 gdbsrv: add monitor cmd v.info scheduler. 2633279378 exp-ptrcheck: the 'impossible' happened on mkfifo call 2634279698 memcheck discards valid-bits for packuswb 2635279795 memcheck reports uninitialised values for mincore on amd64 2636279994 Add support for IBM Power ISA 2.06 -- stage 3 2637280083 mempolicy syscall check errors 2638280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F 2639280710 s390x: config files for nightly builds 2640280757 /tmp dir still used by valgrind even if TMPDIR is specified 2641280965 Valgrind breaks fcntl locks when program does mmap 2642281138 WARNING: unhandled syscall: 340 2643281241 == 275168 (valgrind useless on Macos 10.7.1 Lion) 2644281304 == 275168 (Darwin: dyld "cannot load inserted library") 2645281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1) 2646281468 s390x: handle do_clone and gcc clones in call traces 2647281488 ARM: VFP register corruption 2648281828 == 275284 (false memmove warning: "Source and destination overlap") 2649281883 s390x: Fix system call wrapper for "clone". 2650282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock 2651282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp) 2652282238 SLES10: make check fails 2653282979 strcasestr needs replacement with recent(>=2.12) glibc 2654283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF 2655283243 Regression in ppc64 memcheck tests 2656283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0) 2657283427 re-connect epoll_pwait syscall on ARM linux 2658283600 gdbsrv: android: port vgdb.c 2659283709 none/tests/faultstatus needs to account for page size 2660284305 filter_gdb needs enhancement to work on ppc64 2661284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h 2662284472 Thumb2 ROR.W encoding T2 not implemented 2663284621 XML-escape process command line in XML output 2664n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs 2665 that have non-power-of-2 sizes (also AMDs) 2666n-i-bz don't be spooked by libraries mashed by elfhack 2667n-i-bz don't be spooked by libxul.so linked with gold 2668n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED 2669 2670(3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245) 2671(3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257) 2672(3.7.0: 5 November 2011, vex r2231, valgrind r12258) 2673 2674 2675 2676Release 3.6.1 (16 February 2011) 2677~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 26783.6.1 is a bug fix release. It adds support for some SSE4 2679instructions that were omitted in 3.6.0 due to lack of time. Initial 2680support for glibc-2.13 has been added. A number of bugs causing 2681crashing or assertion failures have been fixed. 2682 2683The following bugs have been fixed or resolved. Note that "n-i-bz" 2684stands for "not in bugzilla" -- that is, a bug that was reported to us 2685but never got a bugzilla entry. We encourage you to file bugs in 2686bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than 2687mailing the developers (or mailing lists) directly -- bugs that are 2688not entered into bugzilla tend to get forgotten about or ignored. 2689 2690To see details of a given bug, visit 2691https://bugs.kde.org/show_bug.cgi?id=XXXXXX 2692where XXXXXX is the bug number as listed below. 2693 2694188572 Valgrind on Mac should suppress setenv() mem leak 2695194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support) 2696210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ) 2697246152 callgrind internal error after pthread_cancel on 32 Bit Linux 2698250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest 2699254420 memory pool tracking broken 2700254957 Test code failing to compile due to changes in memcheck.h 2701255009 helgrind/drd: crash on chmod with invalid parameter 2702255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types 2703255355 helgrind/drd: crash on threaded programs doing fork 2704255358 == 255355 2705255418 (SSE4.x) rint call compiled with ICC 2706255822 --gen-suppressions can create invalid files: "too many callers [...]" 2707255888 closing valgrindoutput tag outputted to log-stream on error 2708255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD) 2709255966 Slowness when using mempool annotations 2710256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM) 2711256600 super-optimized strcasecmp() false positive 2712256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64 2713256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx) 2714257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW) 2715257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS) 2716257276 Missing case in memcheck --track-origins=yes 2717258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction 2718261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L}) 2719262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX 2720262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0) 2721263099 callgrind_annotate counts Ir improperly [...] 2722263877 undefined coprocessor instruction on ARMv7 2723265964 configure FAIL with glibc-2.13 2724n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c 2725n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al 2726n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes 2727n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler 2728n-i-bz DRD: disable --free-is-write due to implementation difficulties 2729 2730(3.6.1: 16 February 2011, vex r2103, valgrind r11561). 2731 2732 2733 2734Release 3.6.0 (21 October 2010) 2735~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 27363.6.0 is a feature release with many significant improvements and the 2737usual collection of bug fixes. 2738 2739This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, 2740PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros 2741and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added. 2742 2743 ------------------------- 2744 2745Here are some highlights. Details are shown further down: 2746 2747* Support for ARM/Linux. 2748 2749* Support for recent Linux distros: Ubuntu 10.10 and Fedora 14. 2750 2751* Support for Mac OS X 10.6, both 32- and 64-bit executables. 2752 2753* Support for the SSE4.2 instruction set. 2754 2755* Enhancements to the Callgrind profiler, including the ability to 2756 handle CPUs with three levels of cache. 2757 2758* A new experimental heap profiler, DHAT. 2759 2760* A huge number of bug fixes and small enhancements. 2761 2762 ------------------------- 2763 2764Here are details of the above changes, together with descriptions of 2765many other changes, and a list of fixed bugs. 2766 2767* ================== PLATFORM CHANGES ================= 2768 2769* Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs 2770 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10, 2771 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want. 2772 2773 This requires a CPU capable of running the ARMv7-A instruction set 2774 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage 2775 of the user space instruction set, including ARM and Thumb integer 2776 code, VFPv3, NEON and V6 media instructions. The Memcheck, 2777 Cachegrind and Massif tools work properly; other tools work to 2778 varying degrees. 2779 2780* Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along 2781 with support for recent releases of the underlying toolchain 2782 components, notably gcc-4.5 and glibc-2.12. 2783 2784* Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit 2785 support also works much better on OS X 10.5, and is as solid as 2786 32-bit support now. 2787 2788* Support for the SSE4.2 instruction set. SSE4.2 is supported in 2789 64-bit mode. In 32-bit mode, support is only available up to and 2790 including SSSE3. Some exceptions: SSE4.2 AES instructions are not 2791 supported in 64-bit mode, and 32-bit mode does in fact support the 2792 bare minimum SSE4 instructions to needed to run programs on Mac OS X 2793 10.6 on 32-bit targets. 2794 2795* Support for IBM POWER6 cpus has been improved. The Power ISA up to 2796 and including version 2.05 is supported. 2797 2798* ==================== TOOL CHANGES ==================== 2799 2800* Cachegrind has a new processing script, cg_diff, which finds the 2801 difference between two profiles. It's very useful for evaluating 2802 the performance effects of a change in a program. 2803 2804 Related to this change, the meaning of cg_annotate's (rarely-used) 2805 --threshold option has changed; this is unlikely to affect many 2806 people, if you do use it please see the user manual for details. 2807 2808* Callgrind now can do branch prediction simulation, similar to 2809 Cachegrind. In addition, it optionally can count the number of 2810 executed global bus events. Both can be used for a better 2811 approximation of a "Cycle Estimation" as derived event (you need to 2812 update the event formula in KCachegrind yourself). 2813 2814* Cachegrind and Callgrind now refer to the LL (last-level) cache 2815 rather than the L2 cache. This is to accommodate machines with 2816 three levels of caches -- if Cachegrind/Callgrind auto-detects the 2817 cache configuration of such a machine it will run the simulation as 2818 if the L2 cache isn't present. This means the results are less 2819 likely to match the true result for the machine, but 2820 Cachegrind/Callgrind's results are already only approximate, and 2821 should not be considered authoritative. The results are still 2822 useful for giving a general idea about a program's locality. 2823 2824* Massif has a new option, --pages-as-heap, which is disabled by 2825 default. When enabled, instead of tracking allocations at the level 2826 of heap blocks (as allocated with malloc/new/new[]), it instead 2827 tracks memory allocations at the level of memory pages (as mapped by 2828 mmap, brk, etc). Each mapped page is treated as its own block. 2829 Interpreting the page-level output is harder than the heap-level 2830 output, but this option is useful if you want to account for every 2831 byte of memory used by a program. 2832 2833* DRD has two new command-line options: --free-is-write and 2834 --trace-alloc. The former allows to detect reading from already freed 2835 memory, and the latter allows tracing of all memory allocations and 2836 deallocations. 2837 2838* DRD has several new annotations. Custom barrier implementations can 2839 now be annotated, as well as benign races on static variables. 2840 2841* DRD's happens before / happens after annotations have been made more 2842 powerful, so that they can now also be used to annotate e.g. a smart 2843 pointer implementation. 2844 2845* Helgrind's annotation set has also been drastically improved, so as 2846 to provide to users a general set of annotations to describe locks, 2847 semaphores, barriers and condition variables. Annotations to 2848 describe thread-safe reference counted heap objects have also been 2849 added. 2850 2851* Memcheck has a new command-line option, --show-possibly-lost, which 2852 is enabled by default. When disabled, the leak detector will not 2853 show possibly-lost blocks. 2854 2855* A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool), 2856 has been added. DHAT keeps track of allocated heap blocks, and also 2857 inspects every memory reference to see which block (if any) is being 2858 accessed. This gives a lot of insight into block lifetimes, 2859 utilisation, turnover, liveness, and the location of hot and cold 2860 fields. You can use DHAT to do hot-field profiling. 2861 2862* ==================== OTHER CHANGES ==================== 2863 2864* Improved support for unfriendly self-modifying code: the extra 2865 overhead incurred by --smc-check=all has been reduced by 2866 approximately a factor of 5 as compared with 3.5.0. 2867 2868* Ability to show directory names for source files in error messages. 2869 This is combined with a flexible mechanism for specifying which 2870 parts of the paths should be shown. This is enabled by the new flag 2871 --fullpath-after. 2872 2873* A new flag, --require-text-symbol, which will stop the run if a 2874 specified symbol is not found it a given shared object when it is 2875 loaded into the process. This makes advanced working with function 2876 intercepting and wrapping safer and more reliable. 2877 2878* Improved support for the Valkyrie GUI, version 2.0.0. GUI output 2879 and control of Valgrind is now available for the tools Memcheck and 2880 Helgrind. XML output from Valgrind is available for Memcheck, 2881 Helgrind and exp-Ptrcheck. 2882 2883* More reliable stack unwinding on amd64-linux, particularly in the 2884 presence of function wrappers, and with gcc-4.5 compiled code. 2885 2886* Modest scalability (performance improvements) for massive 2887 long-running applications, particularly for those with huge amounts 2888 of code. 2889 2890* Support for analyzing programs running under Wine with has been 2891 improved. The header files <valgrind/valgrind.h>, 2892 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in 2893 Windows-programs compiled with MinGW or one of the Microsoft Visual 2894 Studio compilers. 2895 2896* A rare but serious error in the 64-bit x86 CPU simulation was fixed. 2897 The 32-bit simulator was not affected. This did not occur often, 2898 but when it did would usually crash the program under test. 2899 Bug 245925. 2900 2901* A large number of bugs were fixed. These are shown below. 2902 2903* A number of bugs were investigated, and were candidates for fixing, 2904 but are not fixed in 3.6.0, due to lack of developer time. They may 2905 get fixed in later releases. They are: 2906 2907 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64) 2908 212419 false positive "lock order violated" (A+B vs A) 2909 213685 Undefined value propagates past dependency breaking instruction 2910 216837 Incorrect instrumentation of NSOperationQueue on Darwin 2911 237920 valgrind segfault on fork failure 2912 242137 support for code compiled by LLVM-2.8 2913 242423 Another unknown Intel cache config value 2914 243232 Inconsistent Lock Orderings report with trylock 2915 243483 ppc: callgrind triggers VEX assertion failure 2916 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong 2917 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion 2918 'thr' failed. 2919 246152 callgrind internal error after pthread_cancel on 32 Bit Linux 2920 249435 Analyzing wine programs with callgrind triggers a crash 2921 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest 2922 250065 Handling large allocations 2923 250101 huge "free" memory usage due to m_mallocfree.c 2924 "superblocks fragmentation" 2925 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP) 2926 252091 Callgrind on ARM does not detect function returns correctly 2927 252600 [PATCH] Allow lhs to be a pointer for shl/shr 2928 254420 memory pool tracking broken 2929 n-i-bz support for adding symbols for JIT generated code 2930 2931 2932The following bugs have been fixed or resolved. Note that "n-i-bz" 2933stands for "not in bugzilla" -- that is, a bug that was reported to us 2934but never got a bugzilla entry. We encourage you to file bugs in 2935bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than 2936mailing the developers (or mailing lists) directly -- bugs that are 2937not entered into bugzilla tend to get forgotten about or ignored. 2938 2939To see details of a given bug, visit 2940https://bugs.kde.org/show_bug.cgi?id=XXXXXX 2941where XXXXXX is the bug number as listed below. 2942 2943135264 dcbzl instruction missing 2944142688 == 250799 2945153699 Valgrind should report unaligned reads with movdqa 2946180217 == 212335 2947190429 Valgrind reports lost of errors in ld.so 2948 with x86_64 2.9.90 glibc 2949197266 valgrind appears to choke on the xmms instruction 2950 "roundsd" on x86_64 2951197988 Crash when demangling very large symbol names 2952202315 unhandled syscall: 332 (inotify_init1) 2953203256 Add page-level profiling to Massif 2954205093 dsymutil=yes needs quotes, locking (partial fix) 2955205241 Snow Leopard 10.6 support (partial fix) 2956206600 Leak checker fails to upgrade indirect blocks when their 2957 parent becomes reachable 2958210935 port valgrind.h (not valgrind) to win32 so apps run under 2959 wine can make client requests 2960211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89 2961 within Linux ip-stack checksum functions 2962212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0 2963 (lzcnt %eax,%eax) 2964213685 Undefined value propagates past dependency breaking instruction 2965 (partial fix) 2966215914 Valgrind inserts bogus empty environment variable 2967217863 == 197988 2968219538 adjtimex syscall wrapper wrong in readonly adjtime mode 2969222545 shmat fails under valgind on some arm targets 2970222560 ARM NEON support 2971230407 == 202315 2972231076 == 202315 2973232509 Docs build fails with formatting inside <title></title> elements 2974232793 == 202315 2975235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls 2976236546 vex x86->IR: 0x66 0xF 0x3A 0xA 2977237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B 2978237371 better support for VALGRIND_MALLOCLIKE_BLOCK 2979237485 symlink (syscall 57) is not supported on Mac OS 2980237723 sysno == 101 exp-ptrcheck: the 'impossible' happened: 2981 unhandled syscall 2982238208 is_just_below_ESP doesn't take into account red-zone 2983238345 valgrind passes wrong $0 when executing a shell script 2984238679 mq_timedreceive syscall doesn't flag the reception buffer 2985 as "defined" 2986238696 fcntl command F_DUPFD_CLOEXEC not supported 2987238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6 2988238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6 2989238745 3.5.0 Make fails on PPC Altivec opcodes, though configure 2990 says "Altivec off" 2991239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48 2992240488 == 197988 2993240639 == 212335 2994241377 == 236546 2995241903 == 202315 2996241920 == 212335 2997242606 unhandled syscall: setegid (in Ptrcheck) 2998242814 Helgrind "Impossible has happened" during 2999 QApplication::initInstance(); 3000243064 Valgrind attempting to read debug information from iso 3001243270 Make stack unwinding in Valgrind wrappers more reliable 3002243884 exp-ptrcheck: the 'impossible happened: unhandled syscall 3003 sysno = 277 (mq_open) 3004244009 exp-ptrcheck unknown syscalls in analyzing lighttpd 3005244493 ARM VFP d16-d31 registers support 3006244670 add support for audit_session_self syscall on Mac OS 10.6 3007244921 The xml report of helgrind tool is not well format 3008244923 In the xml report file, the <preamble> not escape the 3009 xml char, eg '<','&','>' 3010245535 print full path names in plain text reports 3011245925 x86-64 red zone handling problem 3012246258 Valgrind not catching integer underruns + new [] s 3013246311 reg/reg cmpxchg doesn't work on amd64 3014246549 unhandled syscall unix:277 while testing 32-bit Darwin app 3015246888 Improve Makefile.vex.am 3016247510 [OS X 10.6] Memcheck reports unaddressable bytes passed 3017 to [f]chmod_extended 3018247526 IBM POWER6 (ISA 2.05) support is incomplete 3019247561 Some leak testcases fails due to reachable addresses in 3020 caller save regs 3021247875 sizeofIRType to handle Ity_I128 3022247894 [PATCH] unhandled syscall sys_readahead 3023247980 Doesn't honor CFLAGS passed to configure 3024248373 darwin10.supp is empty in the trunk 3025248822 Linux FIBMAP ioctl has int parameter instead of long 3026248893 [PATCH] make readdwarf.c big endianess safe to enable 3027 unwinding on big endian systems 3028249224 Syscall 336 not supported (SYS_proc_info) 3029249359 == 245535 3030249775 Incorrect scheme for detecting NEON capabilities of host CPU 3031249943 jni JVM init fails when using valgrind 3032249991 Valgrind incorrectly declares AESKEYGENASSIST support 3033 since VEX r2011 3034249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64) 3035250799 frexp$fenv_access_off function generates SIGILL 3036250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E 3037251251 support pclmulqdq insn 3038251362 valgrind: ARM: attach to debugger either fails or provokes 3039 kernel oops 3040251674 Unhandled syscall 294 3041251818 == 254550 3042 3043254257 Add support for debugfiles found by build-id 3044254550 [PATCH] Implement DW_ATE_UTF (DWARF4) 3045254646 Wrapped functions cause stack misalignment on OS X 3046 (and possibly Linux) 3047254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0 3048 3049(3.6.0: 21 October 2010, vex r2068, valgrind r11471). 3050 3051 3052 3053Release 3.5.0 (19 August 2009) 3054~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 30553.5.0 is a feature release with many significant improvements and the 3056usual collection of bug fixes. The main improvement is that Valgrind 3057now works on Mac OS X. 3058 3059This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux 3060and X86/Darwin. Support for recent distros and toolchain components 3061(glibc 2.10, gcc 4.5) has been added. 3062 3063 ------------------------- 3064 3065Here is a short summary of the changes. Details are shown further 3066down: 3067 3068* Support for Mac OS X (10.5.x). 3069 3070* Improvements and simplifications to Memcheck's leak checker. 3071 3072* Clarification and simplifications in various aspects of Valgrind's 3073 text output. 3074 3075* XML output for Helgrind and Ptrcheck. 3076 3077* Performance and stability improvements for Helgrind and DRD. 3078 3079* Genuinely atomic support for x86/amd64/ppc atomic instructions. 3080 3081* A new experimental tool, BBV, useful for computer architecture 3082 research. 3083 3084* Improved Wine support, including ability to read Windows PDB 3085 debuginfo. 3086 3087 ------------------------- 3088 3089Here are details of the above changes, followed by descriptions of 3090many other minor changes, and a list of fixed bugs. 3091 3092 3093* Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes 3094 called "Darwin" because that is the name of the OS core, which is the 3095 level that Valgrind works at.) 3096 3097 Supported systems: 3098 3099 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned 3100 because it would require work and 10.4 is only becoming less common. 3101 3102 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported 3103 fairly well. For 10.5.x, 32-bit programs are the default even on 3104 64-bit machines, so it handles most current programs. 3105 3106 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not 3107 officially supported, but simple programs at least will probably work. 3108 However, start-up is slow. 3109 3110 - PowerPC machines are not supported. 3111 3112 Things that don't work: 3113 3114 - The Ptrcheck tool. 3115 3116 - Objective-C garbage collection. 3117 3118 - --db-attach=yes. 3119 3120 - If you have Rogue Amoeba's "Instant Hijack" program installed, 3121 Valgrind will fail with a SIGTRAP at start-up. See 3122 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a 3123 simple work-around. 3124 3125 Usage notes: 3126 3127 - You will likely find --dsymutil=yes a useful option, as error 3128 messages may be imprecise without it. 3129 3130 - Mac OS X support is new and therefore will be less robust than the 3131 Linux support. Please report any bugs you find. 3132 3133 - Threaded programs may run more slowly than on Linux. 3134 3135 Many thanks to Greg Parker for developing this port over several years. 3136 3137 3138* Memcheck's leak checker has been improved. 3139 3140 - The results for --leak-check=summary now match the summary results 3141 for --leak-check=full. Previously they could differ because 3142 --leak-check=summary counted "indirectly lost" blocks and 3143 "suppressed" blocks as "definitely lost". 3144 3145 - Blocks that are only reachable via at least one interior-pointer, 3146 but are directly pointed to by a start-pointer, were previously 3147 marked as "still reachable". They are now correctly marked as 3148 "possibly lost". 3149 3150 - The default value for the --leak-resolution option has been 3151 changed from "low" to "high". In general, this means that more 3152 leak reports will be produced, but each leak report will describe 3153 fewer leaked blocks. 3154 3155 - With --leak-check=full, "definitely lost" and "possibly lost" 3156 leaks are now considered as proper errors, ie. they are counted 3157 for the "ERROR SUMMARY" and affect the behaviour of 3158 --error-exitcode. These leaks are not counted as errors if 3159 --leak-check=summary is specified, however. 3160 3161 - Documentation for the leak checker has been improved. 3162 3163 3164* Various aspects of Valgrind's text output have changed. 3165 3166 - Valgrind's start-up message has changed. It is shorter but also 3167 includes the command being run, which makes it easier to use 3168 --trace-children=yes. An example: 3169 3170 - Valgrind's shut-down messages have also changed. This is most 3171 noticeable with Memcheck, where the leak summary now occurs before 3172 the error summary. This change was necessary to allow leaks to be 3173 counted as proper errors (see the description of the leak checker 3174 changes above for more details). This was also necessary to fix a 3175 longstanding bug in which uses of suppressions against leaks were 3176 not "counted", leading to difficulties in maintaining suppression 3177 files (see https://bugs.kde.org/show_bug.cgi?id=186790). 3178 3179 - Behavior of -v has changed. In previous versions, -v printed out 3180 a mixture of marginally-user-useful information, and tool/core 3181 statistics. The statistics printing has now been moved to its own 3182 flag, --stats=yes. This means -v is less verbose and more likely 3183 to convey useful end-user information. 3184 3185 - The format of some (non-XML) stack trace entries has changed a 3186 little. Previously there were six possible forms: 3187 3188 0x80483BF: really (a.c:20) 3189 0x80483BF: really (in /foo/a.out) 3190 0x80483BF: really 3191 0x80483BF: (within /foo/a.out) 3192 0x80483BF: ??? (a.c:20) 3193 0x80483BF: ??? 3194 3195 The third and fourth of these forms have been made more consistent 3196 with the others. The six possible forms are now: 3197 3198 0x80483BF: really (a.c:20) 3199 0x80483BF: really (in /foo/a.out) 3200 0x80483BF: really (in ???) 3201 0x80483BF: ??? (in /foo/a.out) 3202 0x80483BF: ??? (a.c:20) 3203 0x80483BF: ??? 3204 3205 Stack traces produced when --xml=yes is specified are different 3206 and unchanged. 3207 3208 3209* Helgrind and Ptrcheck now support XML output, so they can be used 3210 from GUI tools. Also, the XML output mechanism has been 3211 overhauled. 3212 3213 - The XML format has been overhauled and generalised, so it is more 3214 suitable for error reporting tools in general. The Memcheck 3215 specific aspects of it have been removed. The new format, which 3216 is an evolution of the old format, is described in 3217 docs/internals/xml-output-protocol4.txt. 3218 3219 - Memcheck has been updated to use the new format. 3220 3221 - Helgrind and Ptrcheck are now able to emit output in this format. 3222 3223 - The XML output mechanism has been overhauled. XML is now output 3224 to its own file descriptor, which means that: 3225 3226 * Valgrind can output text and XML independently. 3227 3228 * The longstanding problem of XML output being corrupted by 3229 unexpected un-tagged text messages is solved. 3230 3231 As before, the destination for text output is specified using 3232 --log-file=, --log-fd= or --log-socket=. 3233 3234 As before, XML output for a tool is enabled using --xml=yes. 3235 3236 Because there's a new XML output channel, the XML output 3237 destination is now specified by --xml-file=, --xml-fd= or 3238 --xml-socket=. 3239 3240 Initial feedback has shown this causes some confusion. To 3241 clarify, the two envisaged usage scenarios are: 3242 3243 (1) Normal text output. In this case, do not specify --xml=yes 3244 nor any of --xml-file=, --xml-fd= or --xml-socket=. 3245 3246 (2) XML output. In this case, specify --xml=yes, and one of 3247 --xml-file=, --xml-fd= or --xml-socket= to select the XML 3248 destination, one of --log-file=, --log-fd= or --log-socket= 3249 to select the destination for any remaining text messages, 3250 and, importantly, -q. 3251 3252 -q makes Valgrind completely silent on the text channel, 3253 except in the case of critical failures, such as Valgrind 3254 itself segfaulting, or failing to read debugging information. 3255 Hence, in this scenario, it suffices to check whether or not 3256 any output appeared on the text channel. If yes, then it is 3257 likely to be a critical error which should be brought to the 3258 attention of the user. If no (the text channel produced no 3259 output) then it can be assumed that the run was successful. 3260 3261 This allows GUIs to make the critical distinction they need to 3262 make (did the run fail or not?) without having to search or 3263 filter the text output channel in any way. 3264 3265 It is also recommended to use --child-silent-after-fork=yes in 3266 scenario (2). 3267 3268 3269* Improvements and changes in Helgrind: 3270 3271 - XML output, as described above 3272 3273 - Checks for consistent association between pthread condition 3274 variables and their associated mutexes are now performed. 3275 3276 - pthread_spinlock functions are supported. 3277 3278 - Modest performance improvements. 3279 3280 - Initial (skeletal) support for describing the behaviour of 3281 non-POSIX synchronisation objects through ThreadSanitizer 3282 compatible ANNOTATE_* macros. 3283 3284 - More controllable tradeoffs between performance and the level of 3285 detail of "previous" accesses in a race. There are now three 3286 settings: 3287 3288 * --history-level=full. This is the default, and was also the 3289 default in 3.4.x. It shows both stacks involved in a race, but 3290 requires a lot of memory and can be very slow in programs that 3291 do many inter-thread synchronisation events. 3292 3293 * --history-level=none. This only shows the later stack involved 3294 in a race. This can be much faster than --history-level=full, 3295 but makes it much more difficult to find the other access 3296 involved in the race. 3297 3298 The new intermediate setting is 3299 3300 * --history-level=approx 3301 3302 For the earlier (other) access, two stacks are presented. The 3303 earlier access is guaranteed to be somewhere in between the two 3304 program points denoted by those stacks. This is not as useful 3305 as showing the exact stack for the previous access (as per 3306 --history-level=full), but it is better than nothing, and it's 3307 almost as fast as --history-level=none. 3308 3309 3310* New features and improvements in DRD: 3311 3312 - The error messages printed by DRD are now easier to interpret. 3313 Instead of using two different numbers to identify each thread 3314 (Valgrind thread ID and DRD thread ID), DRD does now identify 3315 threads via a single number (the DRD thread ID). Furthermore 3316 "first observed at" information is now printed for all error 3317 messages related to synchronization objects. 3318 3319 - Added support for named semaphores (sem_open() and sem_close()). 3320 3321 - Race conditions between pthread_barrier_wait() and 3322 pthread_barrier_destroy() calls are now reported. 3323 3324 - Added support for custom allocators through the macros 3325 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in 3326 in <valgrind/valgrind.h>). An alternative for these two macros is 3327 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in 3328 <valgrind/drd.h>). 3329 3330 - Added support for annotating non-POSIX synchronization objects 3331 through several new ANNOTATE_*() macros. 3332 3333 - OpenMP: added support for the OpenMP runtime (libgomp) included 3334 with gcc versions 4.3.0 and 4.4.0. 3335 3336 - Faster operation. 3337 3338 - Added two new command-line options (--first-race-only and 3339 --segment-merging-interval). 3340 3341 3342* Genuinely atomic support for x86/amd64/ppc atomic instructions 3343 3344 Valgrind will now preserve (memory-access) atomicity of LOCK- 3345 prefixed x86/amd64 instructions, and any others implying a global 3346 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions. 3347 3348 This means that Valgrinded processes will "play nicely" in 3349 situations where communication with other processes, or the kernel, 3350 is done through shared memory and coordinated with such atomic 3351 instructions. Prior to this change, such arrangements usually 3352 resulted in hangs, races or other synchronisation failures, because 3353 Valgrind did not honour atomicity of such instructions. 3354 3355 3356* A new experimental tool, BBV, has been added. BBV generates basic 3357 block vectors for use with the SimPoint analysis tool, which allows 3358 a program's overall behaviour to be approximated by running only a 3359 fraction of it. This is useful for computer architecture 3360 researchers. You can run BBV by specifying --tool=exp-bbv (the 3361 "exp-" prefix is short for "experimental"). BBV was written by 3362 Vince Weaver. 3363 3364 3365* Modestly improved support for running Windows applications under 3366 Wine. In particular, initial support for reading Windows .PDB debug 3367 information has been added. 3368 3369 3370* A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been 3371 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks 3372 instead of bytes. 3373 3374 3375* The Valgrind client requests VALGRIND_PRINTF and 3376 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously, 3377 the string was always printed immediately on its own line. Now, the 3378 string will be added to a buffer but not printed until a newline is 3379 encountered, or other Valgrind output is printed (note that for 3380 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered 3381 "other Valgrind output"). This allows you to use multiple 3382 VALGRIND_PRINTF calls to build up a single output line, and also to 3383 print multiple output lines with a single request (by embedding 3384 multiple newlines in the string). 3385 3386 3387* The graphs drawn by Massif's ms_print program have changed slightly: 3388 3389 - The half-height chars '.' and ',' are no longer drawn, because 3390 they are confusing. The --y option can be used if the default 3391 y-resolution is not high enough. 3392 3393 - Horizontal lines are now drawn after the top of a snapshot if 3394 there is a gap until the next snapshot. This makes it clear that 3395 the memory usage has not dropped to zero between snapshots. 3396 3397 3398* Something that happened in 3.4.0, but wasn't clearly announced: the 3399 option --read-var-info=yes can be used by some tools (Memcheck, 3400 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3 3401 variable type and location information. This makes those tools 3402 start up more slowly and increases memory consumption, but 3403 descriptions of data addresses in error messages become more 3404 detailed. 3405 3406 3407* exp-Omega, an experimental instantaneous leak-detecting tool, was 3408 disabled in 3.4.0 due to a lack of interest and maintenance, 3409 although the source code was still in the distribution. The source 3410 code has now been removed from the distribution. For anyone 3411 interested, the removal occurred in SVN revision r10247. 3412 3413 3414* Some changes have been made to the build system. 3415 3416 - VEX/ is now integrated properly into the build system. This means 3417 that dependency tracking within VEX/ now works properly, "make 3418 install" will work without requiring "make" before it, and 3419 parallel builds (ie. 'make -j') now work (previously a 3420 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j' 3421 was effectively ignored). 3422 3423 - The --with-vex configure option has been removed. It was of 3424 little use and removing it simplified the build system. 3425 3426 - The location of some install files has changed. This should not 3427 affect most users. Those who might be affected: 3428 3429 * For people who use Valgrind with MPI programs, the installed 3430 libmpiwrap.so library has moved from 3431 $(INSTALL)/<platform>/libmpiwrap.so to 3432 $(INSTALL)/libmpiwrap-<platform>.so. 3433 3434 * For people who distribute standalone Valgrind tools, the 3435 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a 3436 have moved to $(INSTALL)/libcoregrind-<platform>.a. 3437 3438 These changes simplify the build system. 3439 3440 - Previously, all the distributed suppression (*.supp) files were 3441 installed. Now, only default.supp is installed. This should not 3442 affect users as the other installed suppression files were not 3443 read; the fact that they were installed was a mistake. 3444 3445 3446* KNOWN LIMITATIONS: 3447 3448 - Memcheck is unusable with the Intel compiler suite version 11.1, 3449 when it generates code for SSE2-and-above capable targets. This 3450 is because of icc's use of highly optimised inlined strlen 3451 implementations. It causes Memcheck to report huge numbers of 3452 false errors even in simple programs. Helgrind and DRD may also 3453 have problems. 3454 3455 Versions 11.0 and earlier may be OK, but this has not been 3456 properly tested. 3457 3458 3459The following bugs have been fixed or resolved. Note that "n-i-bz" 3460stands for "not in bugzilla" -- that is, a bug that was reported to us 3461but never got a bugzilla entry. We encourage you to file bugs in 3462bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than 3463mailing the developers (or mailing lists) directly -- bugs that are 3464not entered into bugzilla tend to get forgotten about or ignored. 3465 3466To see details of a given bug, visit 3467https://bugs.kde.org/show_bug.cgi?id=XXXXXX 3468where XXXXXX is the bug number as listed below. 3469 347084303 How about a LockCheck tool? 347191633 dereference of null ptr in vgPlain_st_basetype 347297452 Valgrind doesn't report any pthreads problems 3473100628 leak-check gets assertion failure when using 3474 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory 3475108528 NPTL pthread cleanup handlers not called 3476110126 Valgrind 2.4.1 configure.in tramples CFLAGS 3477110128 mallinfo is not implemented... 3478110770 VEX: Generated files not always updated when making valgrind 3479111102 Memcheck: problems with large (memory footprint) applications 3480115673 Vex's decoder should never assert 3481117564 False positive: Syscall param clone(child_tidptr) contains 3482 uninitialised byte(s) 3483119404 executing ssh from inside valgrind fails 3484133679 Callgrind does not write path names to sources with dwarf debug 3485 info 3486135847 configure.in problem with non gnu compilers (and possible fix) 3487136154 threads.c:273 (vgCallgrind_post_signal): Assertion 3488 '*(vgCallgrind_current_fn_stack.top) == 0' failed. 3489136230 memcheck reports "possibly lost", should be "still reachable" 3490137073 NULL arg to MALLOCLIKE_BLOCK causes crash 3491137904 Valgrind reports a memory leak when using POSIX threads, 3492 while it shouldn't 3493139076 valgrind VT_GETSTATE error 3494142228 complaint of elf_dynamic_do_rela in trivial usage 3495145347 spurious warning with USBDEVFS_REAPURB 3496148441 (wine) can't find memory leak in Wine, win32 binary 3497 executable file. 3498148742 Leak-check fails assert on exit 3499149878 add (proper) check for calloc integer overflow 3500150606 Call graph is broken when using callgrind control 3501152393 leak errors produce an exit code of 0. I need some way to 3502 cause leak errors to result in a nonzero exit code. 3503157154 documentation (leak-resolution doc speaks about num-callers 3504 def=4) + what is a loss record 3505159501 incorrect handling of ALSA ioctls 3506162020 Valgrinding an empty/zero-byte file crashes valgrind 3507162482 ppc: Valgrind crashes while reading stabs information 3508162718 x86: avoid segment selector 0 in sys_set_thread_area() 3509163253 (wine) canonicaliseSymtab forgot some fields in DiSym 3510163560 VEX/test_main.c is missing from valgrind-3.3.1 3511164353 malloc_usable_size() doesn't return a usable size 3512165468 Inconsistent formatting in memcheck manual -- please fix 3513169505 main.c:286 (endOfInstr): 3514 Assertion 'ii->cost_offset == *cost_offset' failed 3515177206 Generate default.supp during compile instead of configure 3516177209 Configure valt_load_address based on arch+os 3517177305 eventfd / syscall 323 patch lost 3518179731 Tests fail to build because of inlining of non-local asm labels 3519181394 helgrind: libhb_core.c:3762 (msm_write): Assertion 3520 'ordxx == POrd_EQ || ordxx == POrd_LT' failed. 3521181594 Bogus warning for empty text segment 3522181707 dwarf doesn't require enumerations to have name 3523185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64 3524185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree): 3525 Assertion '!already_present' failed. 3526185359 exp-ptrcheck: unhandled syscall getresuid() 3527185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64 3528185816 Valgrind is unable to handle debug info for files with split 3529 debug info that are prelinked afterwards 3530185980 [darwin] unhandled syscall: sem_open 3531186238 bbToIR_AMD64: disInstr miscalculated next %rip 3532186507 exp-ptrcheck unhandled syscalls prctl, etc. 3533186790 Suppression pattern used for leaks are not reported 3534186796 Symbols with length>200 in suppression files are ignored 3535187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation 3536187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid} 3537188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened 3538188046 bashisms in the configure script 3539188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA 3540188161 memcheck: --track-origins=yes asserts "mc_machine.c:672 3541 (get_otrack_shadow_offset_wrk): the 'impossible' happened." 3542188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock, 3543 assertion fail "!lock->heldBy" 3544188427 Add support for epoll_create1 (with patch) 3545188530 Support for SIOCGSTAMPNS 3546188560 Include valgrind.spec in the tarball 3547188572 Valgrind on Mac should suppress setenv() mem leak 3548189054 Valgrind fails to build because of duplicate non-local asm labels 3549189737 vex amd64->IR: unhandled instruction bytes: 0xAC 3550189762 epoll_create syscall not handled (--tool=exp-ptrcheck) 3551189763 drd assertion failure: s_threadinfo[tid].is_recording 3552190219 unhandled syscall: 328 (x86-linux) 3553190391 dup of 181394; see above 3554190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc 3555190820 No debug information on powerpc-linux 3556191095 PATCH: Improve usbdevfs ioctl handling 3557191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks 3558 or big nr of errors 3559191189 --xml=yes should obey --gen-suppressions=all 3560191192 syslog() needs a suppression on macosx 3561191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265 3562191761 getrlimit on MacOSX 3563191992 multiple --fn-skip only works sometimes; dependent on order 3564192634 V. reports "aspacem sync_check_mapping_callback: 3565 segment mismatch" on Darwin 3566192954 __extension__ missing on 2 client requests 3567194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29 3568194474 "INSTALL" file has different build instructions than "README" 3569194671 Unhandled syscall (sem_wait?) from mac valgrind 3570195069 memcheck: reports leak (memory still reachable) for 3571 printf("%d', x) 3572195169 drd: (vgDrd_barrier_post_wait): 3573 Assertion 'r->sg[p->post_iteration]' failed. 3574195268 valgrind --log-file doesn't accept ~/... 3575195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate 3576195860 WARNING: unhandled syscall: unix:223 3577196528 need a error suppression for pthread_rwlock_init under os x? 3578197227 Support aio_* syscalls on Darwin 3579197456 valgrind should reject --suppressions=(directory) 3580197512 DWARF2 CFI reader: unhandled CFI instruction 0:10 3581197591 unhandled syscall 27 (mincore) 3582197793 Merge DCAS branch to the trunk == 85756, 142103 3583197794 Avoid duplicate filenames in Vex 3584197898 make check fails on current SVN 3585197901 make check fails also under exp-ptrcheck in current SVN 3586197929 Make --leak-resolution=high the default 3587197930 Reduce spacing between leak reports 3588197933 Print command line of client at start-up, and shorten preamble 3589197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck) 3590198395 add BBV to the distribution as an experimental tool 3591198624 Missing syscalls on Darwin: 82, 167, 281, 347 3592198649 callgrind_annotate doesn't cumulate counters 3593199338 callgrind_annotate sorting/thresholds are broken for all but Ir 3594199977 Valgrind complains about an unrecognized instruction in the 3595 atomic_incs test program 3596200029 valgrind isn't able to read Fedora 12 debuginfo 3597200760 darwin unhandled syscall: unix:284 3598200827 DRD doesn't work on Mac OS X 3599200990 VG_(read_millisecond_timer)() does not work correctly 3600201016 Valgrind does not support pthread_kill() on Mac OS 3601201169 Document --read-var-info 3602201323 Pre-3.5.0 performance sanity checking 3603201384 Review user manual for the 3.5.0 release 3604201585 mfpvr not implemented on ppc 3605201708 tests failing because x86 direction flag is left set 3606201757 Valgrind doesn't handle any recent sys_futex additions 3607204377 64-bit valgrind can not start a shell script 3608 (with #!/path/to/shell) if the shell is a 32-bit executable 3609n-i-bz drd: fixed assertion failure triggered by mutex reinitialization. 3610n-i-bz drd: fixed a bug that caused incorrect messages to be printed 3611 about memory allocation events with memory access tracing enabled 3612n-i-bz drd: fixed a memory leak triggered by vector clock deallocation 3613 3614(3.5.0: 19 Aug 2009, vex r1913, valgrind r10846). 3615 3616 3617 3618Release 3.4.1 (28 February 2009) 3619~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 36203.4.1 is a bug-fix release that fixes some regressions and assertion 3621failures in debug info reading in 3.4.0, most notably incorrect stack 3622traces on amd64-linux on older (glibc-2.3 based) systems. Various 3623other debug info problems are also fixed. A number of bugs in the 3624exp-ptrcheck tool introduced in 3.4.0 have been fixed. 3625 3626In view of the fact that 3.4.0 contains user-visible regressions 3627relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are 3628encouraged to ship 3.4.1 in preference to 3.4.0. 3629 3630The fixed bugs are as follows. Note that "n-i-bz" stands for "not in 3631bugzilla" -- that is, a bug that was reported to us but never got a 3632bugzilla entry. We encourage you to file bugs in bugzilla 3633(http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the 3634developers (or mailing lists) directly -- bugs that are not entered 3635into bugzilla tend to get forgotten about or ignored. 3636 3637n-i-bz Fix various bugs reading icc-11 generated debug info 3638n-i-bz Fix various bugs reading gcc-4.4 generated debug info 3639n-i-bz Preliminary support for glibc-2.10 / Fedora 11 3640n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes, 3641 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2. 3642179618 exp-ptrcheck crashed / exit prematurely 3643179624 helgrind: false positive races with pthread_create and 3644 recv/open/close/read 3645134207 pkg-config output contains @VG_PLATFORM@ 3646176926 floating point exception at valgrind startup with PPC 440EPX 3647181594 Bogus warning for empty text segment 3648173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes) 3649181707 Dwarf3 doesn't require enumerations to have name 3650185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64 3651185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree): 3652 Assertion '!already_present' failed. 3653185359 exp-ptrcheck unhandled syscall getresuid() 3654 3655(3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253). 3656(3.4.1: 28 Feb 2008, vex r1884, valgrind r9293). 3657 3658 3659 3660Release 3.4.0 (2 January 2009) 3661~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 36623.4.0 is a feature release with many significant improvements and the 3663usual collection of bug fixes. This release supports X86/Linux, 3664AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros 3665(using gcc 4.4, glibc 2.8 and 2.9) has been added. 3666 36673.4.0 brings some significant tool improvements. Memcheck can now 3668report the origin of uninitialised values, the thread checkers 3669Helgrind and DRD are much improved, and we have a new experimental 3670tool, exp-Ptrcheck, which is able to detect overruns of stack and 3671global arrays. In detail: 3672 3673* Memcheck is now able to track the origin of uninitialised values. 3674 When it reports an uninitialised value error, it will try to show 3675 the origin of the value, as either a heap or stack allocation. 3676 Origin tracking is expensive and so is not enabled by default. To 3677 use it, specify --track-origins=yes. Memcheck's speed will be 3678 essentially halved, and memory usage will be significantly 3679 increased. Nevertheless it can drastically reduce the effort 3680 required to identify the root cause of uninitialised value errors, 3681 and so is often a programmer productivity win, despite running more 3682 slowly. 3683 3684* A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in 3685 3.4.0, will be released shortly. 3686 3687* Helgrind's race detection algorithm has been completely redesigned 3688 and reimplemented, to address usability and scalability concerns: 3689 3690 - The new algorithm has a lower false-error rate: it is much less 3691 likely to report races that do not really exist. 3692 3693 - Helgrind will display full call stacks for both accesses involved 3694 in a race. This makes it easier to identify the root causes of 3695 races. 3696 3697 - Limitations on the size of program that can run have been removed. 3698 3699 - Performance has been modestly improved, although that is very 3700 workload-dependent. 3701 3702 - Direct support for Qt4 threading has been added. 3703 3704 - pthread_barriers are now directly supported. 3705 3706 - Helgrind works well on all supported Linux targets. 3707 3708* The DRD thread debugging tool has seen major improvements: 3709 3710 - Greatly improved performance and significantly reduced memory 3711 usage. 3712 3713 - Support for several major threading libraries (Boost.Thread, Qt4, 3714 glib, OpenMP) has been added. 3715 3716 - Support for atomic instructions, POSIX semaphores, barriers and 3717 reader-writer locks has been added. 3718 3719 - Works now on PowerPC CPUs too. 3720 3721 - Added support for printing thread stack usage at thread exit time. 3722 3723 - Added support for debugging lock contention. 3724 3725 - Added a manual for Drd. 3726 3727* A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck 3728 checks for misuses of pointers. In that sense it is a bit like 3729 Memcheck. However, Ptrcheck can do things Memcheck can't: it can 3730 detect overruns of stack and global arrays, it can detect 3731 arbitrarily far out-of-bounds accesses to heap blocks, and it can 3732 detect accesses heap blocks that have been freed a very long time 3733 ago (millions of blocks in the past). 3734 3735 Ptrcheck currently works only on x86-linux and amd64-linux. To use 3736 it, use --tool=exp-ptrcheck. A simple manual is provided, as part 3737 of the main Valgrind documentation. As this is an experimental 3738 tool, we would be particularly interested in hearing about your 3739 experiences with it. 3740 3741* exp-Omega, an experimental instantaneous leak-detecting tool, is no 3742 longer built by default, although the code remains in the repository 3743 and the tarball. This is due to three factors: a perceived lack of 3744 users, a lack of maintenance, and concerns that it may not be 3745 possible to achieve reliable operation using the existing design. 3746 3747* As usual, support for the latest Linux distros and toolchain 3748 components has been added. It should work well on Fedora Core 10, 3749 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release 3750 state) is supported, as is glibc-2.9. The C++ demangler has been 3751 updated so as to work well with C++ compiled by even the most recent 3752 g++'s. 3753 3754* You can now use frame-level wildcards in suppressions. This was a 3755 frequently-requested enhancement. A line "..." in a suppression now 3756 matches zero or more frames. This makes it easier to write 3757 suppressions which are precise yet insensitive to changes in 3758 inlining behaviour. 3759 3760* 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set. 3761 3762* Very basic support for IBM Power6 has been added (64-bit processes only). 3763 3764* Valgrind is now cross-compilable. For example, it is possible to 3765 cross compile Valgrind on an x86/amd64-linux host, so that it runs 3766 on a ppc32/64-linux target. 3767 3768* You can set the main thread's stack size at startup using the 3769 new --main-stacksize= flag (subject of course to ulimit settings). 3770 This is useful for running apps that need a lot of stack space. 3771 3772* The limitation that you can't use --trace-children=yes together 3773 with --db-attach=yes has been removed. 3774 3775* The following bugs have been fixed. Note that "n-i-bz" stands for 3776 "not in bugzilla" -- that is, a bug that was reported to us but 3777 never got a bugzilla entry. We encourage you to file bugs in 3778 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than 3779 mailing the developers (or mailing lists) directly. 3780 3781 n-i-bz Make return types for some client requests 64-bit clean 3782 n-i-bz glibc 2.9 support 3783 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865) 3784 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults 3785 n-i-bz Building in an env without gdb gives bogus gdb attach 3786 92456 Tracing the origin of uninitialised memory 3787 106497 Valgrind does not demangle some C++ template symbols 3788 162222 ==106497 3789 151612 Suppression with "..." (frame-level wildcards in .supp files) 3790 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit) 3791 159285 unhandled syscall:25 (stime, on x86-linux) 3792 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig" 3793 160954 ppc build of valgrind crashes with illegal instruction (isel) 3794 160956 mallinfo implementation, w/ patch 3795 162092 Valgrind fails to start gnome-system-monitor 3796 162819 malloc_free_fill test doesn't pass on glibc2.8 x86 3797 163794 assertion failure with "--track-origins=yes" 3798 163933 sigcontext.err and .trapno must be set together 3799 163955 remove constraint !(--db-attach=yes && --trace-children=yes) 3800 164476 Missing kernel module loading system calls 3801 164669 SVN regression: mmap() drops posix file locks 3802 166581 Callgrind output corruption when program forks 3803 167288 Patch file for missing system calls on Cell BE 3804 168943 unsupported scas instruction pentium 3805 171645 Unrecognised instruction (MOVSD, non-binutils encoding) 3806 172417 x86->IR: 0x82 ... 3807 172563 amd64->IR: 0xD9 0xF5 - fprem1 3808 173099 .lds linker script generation error 3809 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl) 3810 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes) 3811 174532 == 173751 3812 174908 --log-file value not expanded correctly for core file 3813 175044 Add lookup_dcookie for amd64 3814 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding) 3815 3816Developer-visible changes: 3817 3818* Valgrind's debug-info reading machinery has been majorly overhauled. 3819 It can now correctly establish the addresses for ELF data symbols, 3820 which is something that has never worked properly before now. 3821 3822 Also, Valgrind can now read DWARF3 type and location information for 3823 stack and global variables. This makes it possible to use the 3824 framework to build tools that rely on knowing the type and locations 3825 of stack and global variables, for example exp-Ptrcheck. 3826 3827 Reading of such information is disabled by default, because most 3828 tools don't need it, and because it is expensive in space and time. 3829 However, you can force Valgrind to read it, using the 3830 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to 3831 make use of such information, if present, to provide source-level 3832 descriptions of data addresses in the error messages they create. 3833 3834(3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882). 3835(3.4.0: 3 Jan 2009, vex r1878, valgrind r8899). 3836