1 Release 3.13.0 (15 June 2017) 2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 3 4 3.13.0 is a feature release with many improvements and the usual collection of 5 bug fixes. 6 7 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux, 8 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux, 9 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android, 10 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. 11 12 * ==================== CORE CHANGES =================== 13 14 * The translation cache size has been increased to keep up with the demands of 15 large applications. The maximum number of sectors has increased from 24 to 16 48. The default number of sectors has increased from 16 to 32 on all 17 targets except Android, where the increase is from 6 to 12. 18 19 * The amount of memory that Valgrind can use has been increased from 64GB to 20 128GB. In particular this means your application can allocate up to about 21 60GB when running on Memcheck. 22 23 * Valgrind's default load address has been changed from 0x3800'0000 to 24 0x5800'0000, so as to make it possible to load larger executables. This 25 should make it possible to load executables of size at least 1200MB. 26 27 * A massive spaceleak caused by reading compressed debuginfo files has been 28 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created 29 debuginfo. 30 31 * The C++ demangler has been updated. 32 33 * Support for demangling Rust symbols has been added. 34 35 * A new representation of stack traces, the "XTree", has been added. An XTree 36 is a tree of stacktraces with data associated with the stacktraces. This is 37 used by various tools (Memcheck, Helgrind, Massif) to report on the heap 38 consumption of your program. Reporting is controlled by the new options 39 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>. 40 41 A report can also be produced on demand using the gdbserver monitor command 42 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind 43 format' and 'massif format. The existing visualisers for these formats (e.g. 44 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and 45 analyse these reports. 46 47 Memcheck can also produce XTree leak reports using the Callgrind file 48 format. For more details, see the user manual. 49 50 * ================== PLATFORM CHANGES ================= 51 52 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support 53 54 * amd64: fixes for JIT failure problems on long AVX2 code blocks 55 56 * amd64 and x86: support for CET prefixes has been added 57 58 * arm32: a few missing ARMv8 instructions have been implemented 59 60 * arm64, mips64, mips32: an alternative implementation of Load-Linked and 61 Store-Conditional instructions has been added. This is to deal with 62 processor implementations that implement the LL/SC specifications strictly 63 and as a result cause Valgrind to hang in certain situations. The 64 alternative implementation is automatically enabled at startup, as required. 65 You can use the option --sim-hints=fallback-llsc to force-enable it if you 66 want. 67 68 * Support for OSX 10.12 has been improved. 69 70 * On Linux, clone handling has been improved to honour CLONE_VFORK that 71 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled 72 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on 73 CLONE_VM exact semantics will (still) not work. 74 75 * The TileGX/Linux port has been removed because it appears to be both unused 76 and unsupported. 77 78 * ==================== TOOL CHANGES ==================== 79 80 * Memcheck: 81 82 - Memcheck should give fewer false positives when running optimised 83 Clang/LLVM generated code. 84 85 - Support for --xtree-memory and 'xtmemory [<filename>]>'. 86 87 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file> 88 to produce the end of execution leak report in a xtree callgrind format 89 file. 90 91 - New option 'xtleak' in the memcheck leak_check monitor command, to produce 92 the leak report in an xtree file. 93 94 * Massif: 95 96 - Support for --xtree-memory and 'xtmemory [<filename>]>'. 97 98 - For some workloads (typically, for big applications), Massif memory 99 consumption and CPU consumption has decreased significantly. 100 101 * Helgrind: 102 103 - Support for --xtree-memory and 'xtmemory [<filename>]>'. 104 105 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful 106 for Ada gnat compiled applications. 107 108 * ==================== OTHER CHANGES ==================== 109 110 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will 111 append the inner guest stacktrace to the inner host stacktrace. This helps 112 to investigate the errors reported by the outer, when they are caused by the 113 inner guest program (such as an inner regtest). See README_DEVELOPERS for 114 more info. 115 116 * To allow fast detection of callgrind files by desktop environments and file 117 managers, the format was extended to have an optional first line that 118 uniquely identifies the format ("# callgrind format"). Callgrind creates 119 this line now, as does the new xtree functionality. 120 121 * File name template arguments (such as --log-file, --xtree-memory-file, ...) 122 have a new %n format letter that is replaced by a sequence number. 123 124 * "--version -v" now shows the SVN revision numbers from which Valgrind was 125 built. 126 127 * ==================== FIXED BUGS ==================== 128 129 The following bugs have been fixed or resolved. Note that "n-i-bz" 130 stands for "not in bugzilla" -- that is, a bug that was reported to us 131 but never got a bugzilla entry. We encourage you to file bugs in 132 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 133 than mailing the developers (or mailing lists) directly -- bugs that 134 are not entered into bugzilla tend to get forgotten about or ignored. 135 136 To see details of a given bug, visit 137 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 138 where XXXXXX is the bug number as listed below. 139 140 162848 --log-file output isn't split when a program forks 141 340777 Illegal instruction on mips (ar71xx) 142 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms 143 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones 144 to a different stack. 145 344139 x86 stack-seg overrides, needed by the Wine people 146 344524 store conditional of guest applications always fail - observed on 147 Octeon3(MIPS) 148 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT) 149 352395 Please provide SVN revision info in --version -v 150 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP) 151 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid != 152 INVALID_POSIX_THREADID' failed 153 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread 154 barrier implementation 155 358697 valgrind.h: Some code remains even when defining NVALGRIND 156 359202 Add musl libc configure/compile 157 360415 amd64 instructions ADCX and ADOX are not implemented in VEX 158 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10) 159 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1) 160 362223 assertion failed when .valgrindrc is a directory instead of a file 161 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags 162 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138) 163 368507 can't malloc chunks larger than about 34GB 164 368529 Android arm target link error, missing atexit and pthread_atfork 165 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list) 166 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp) 167 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8) 168 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key) 169 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval) 170 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname) 171 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range) 172 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill) 173 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare) 174 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr) 175 370028 Reduce the number of compiler warnings on MIPS platforms 176 370635 arm64 missing syscall getcpu 177 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64 178 371227 Clean AArch64 syscall table 179 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers 180 371471 Valgrind complains about non legit memory leaks on placement new (C++) 181 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used 182 371503 disInstr(arm64): unhandled instruction 0xF89F0000 183 371869 support '%' in symbol Z-encoding 184 371916 execution tree xtree concept 185 372120 c++ demangler demangles symbols which are not c++ 186 372185 Support of valgrind on ARMv8 with 32 bit executable 187 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10) 188 372195 Power PC, xxsel instruction is not always recognized. 189 372504 Hanging on exit_group 190 372600 process loops forever when fatal signals are arriving quickly 191 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed 192 373046 Stacks registered by core are never deregistered 193 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+ 194 373086 Implement additional Xen hypercalls 195 373192 Calling posix_spawn in glibc 2.24 completely broken 196 373488 Support for fanotify API on ARM64 architecture 197 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init) 198 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only 199 373938 const IRExpr arguments for matchIRExpr() 200 374719 some spelling fixes 201 374963 increase valgrind's load address to prevent mmap failure 202 375514 valgrind_get_tls_addr() does not work in case of static TLS 203 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address 204 for ML_(find_rx_mapping)() 205 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24 206 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns 207 == 377159 "vex: the `impossible' happened" still present 208 == 375150 Assertion 'tres.status == VexTransOK' failed 209 == 378068 valgrind crashes on AVX2 function in FFmpeg 210 376142 Segfaults on MIPS Cavium Octeon boards 211 376279 disInstr(arm64): unhandled instruction 0xD50320FF 212 376455 Solaris: unhandled syscall lgrpsys(180) 213 376518 Solaris: unhandled fast trap getlgrp(6) 214 376611 ppc64 and arm64 don't know about prlimit64 syscall 215 376729 PPC64, remove R2 from the clobber list 216 == 371668 217 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses 218 to be wrongly marked as addressable 219 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with 220 PIE enabled by default 221 377376 memcheck/tests/linux/getregset fails with glibc2.24 222 377427 PPC64, lxv instruction failing on odd destination register 223 377478 PPC64: ISA 3.0 setup fixes 224 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE 225 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET, 226 and 2 args for FUTEX_TRYLOCK_PI 227 377717 Fix massive space leak when reading compressed debuginfo sections 228 377891 Update Xen 4.6 domctl wrappers 229 377930 fcntl syscall wrapper is missing flock structure check 230 378524 libvexmultiarch_test regression on s390x and ppc64 231 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper 232 378673 Update libiberty demangler 233 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support 234 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes 235 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper 236 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603] 237 (task_register_dyld_image_infos) 238 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603] 239 (task_register_dyld_shared_cache_image_info) 240 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap) 241 379473 MIPS: add support for rdhwr cycle counter register 242 379504 remove TileGX/Linux port 243 379525 Support more x86 nop opcodes 244 379838 disAMode(x86): not an addr! 245 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions 246 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4) 247 379895 clock_gettime does not execute POST syscall wrapper 248 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly 249 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module) 250 380200 xtree generated callgrind files refer to files without directory name 251 380202 Assertion failure for cache line size (cls == 64) on aarch64. 252 380397 s390x: __GI_strcspn() replacement needed 253 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7. 254 255 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434) 256 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443) 257 (3.13.0: 14 June 2017, vex r3396, valgrind r16446) 258 259 260 261 Release 3.12.0 (20 October 2016) 262 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 263 264 3.12.0 is a feature release with many improvements and the usual 265 collection of bug fixes. 266 267 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, 268 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, 269 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android, 270 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX 271 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for 272 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux. 273 274 * ================== PLATFORM CHANGES ================= 275 276 * POWER: Support for ISA 3.0 has been added 277 278 * mips: support for O32 FPXX ABI has been added. 279 * mips: improved recognition of different processors 280 * mips: determination of page size now done at run time 281 282 * amd64: Partial support for AMD FMA4 instructions. 283 284 * arm, arm64: Support for v8 crypto and CRC instructions. 285 286 * Improvements and robustification of the Solaris port. 287 288 * Preliminary support for MacOS 10.12 (Sierra) has been added. 289 290 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we 291 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64) 292 where possible. Valgrind's support for 32-bit x86 has stagnated in 293 recent years and has fallen far behind that for 64-bit x86 294 instructions. By contrast 64-bit x86 is well supported, up to and 295 including AVX2. 296 297 * ==================== TOOL CHANGES ==================== 298 299 * Memcheck: 300 301 - Added meta mempool support for describing a custom allocator which: 302 - Auto-frees all chunks assuming that destroying a pool destroys all 303 objects in the pool 304 - Uses itself to allocate other memory blocks 305 306 - New flag --ignore-range-below-sp to ignore memory accesses below 307 the stack pointer, if you really have to. The related flag 308 --workaround-gcc296-bugs=yes is now deprecated. Use 309 --ignore-range-below-sp=1024-1 as a replacement. 310 311 * DRD: 312 313 - Improved thread startup time significantly on non-Linux platforms. 314 315 * DHAT 316 317 - Added collection of the metric "tot-blocks-allocd" 318 319 * ==================== OTHER CHANGES ==================== 320 321 * Replacement/wrapping of malloc/new related functions is now done not just 322 for system libraries by default, but for any globally defined malloc/new 323 related function (both in shared libraries and statically linked alternative 324 malloc implementations). The dynamic (runtime) linker is excluded, though. 325 To only intercept malloc/new related functions in 326 system libraries use --soname-synonyms=somalloc=nouserintercepts (where 327 "nouserintercepts" can be any non-existing library name). 328 This new functionality is not implemented for MacOS X. 329 330 * The maximum number of callers in a suppression entry is now equal to 331 the maximum size for --num-callers (500). 332 Note that --gen-suppressions=yes|all similarly generates suppressions 333 containing up to --num-callers frames. 334 335 * New and modified GDB server monitor features: 336 337 - Valgrind's gdbserver now accepts the command 'catch syscall'. 338 Note that you must have GDB >= 7.11 to use 'catch syscall' with 339 gdbserver. 340 341 * New option --run-cxx-freeres=<yes|no> can be used to change whether 342 __gnu_cxx::__freeres() cleanup function is called or not. Default is 343 'yes'. 344 345 * Valgrind is able to read compressed debuginfo sections in two formats: 346 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib) 347 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu) 348 349 * Modest JIT-cost improvements: the cost of instrumenting code blocks 350 for the most common use case (x86_64-linux, Memcheck) has been 351 reduced by 10%-15%. 352 353 * Improved performance for programs that do a lot of discarding of 354 instruction address ranges of 8KB or less. 355 356 * The C++ symbol demangler has been updated. 357 358 * More robustness against invalid syscall parameters on Linux. 359 360 * ==================== FIXED BUGS ==================== 361 362 The following bugs have been fixed or resolved. Note that "n-i-bz" 363 stands for "not in bugzilla" -- that is, a bug that was reported to us 364 but never got a bugzilla entry. We encourage you to file bugs in 365 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 366 than mailing the developers (or mailing lists) directly -- bugs that 367 are not entered into bugzilla tend to get forgotten about or ignored. 368 369 To see details of a given bug, visit 370 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 371 where XXXXXX is the bug number as listed below. 372 373 191069 Exiting due to signal not reported in XML output 374 199468 Suppressions: stack size limited to 25 375 while --num-callers allows more frames 376 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM) 377 278744 cvtps2pd with redundant RexW 378 303877 valgrind doesn't support compressed debuginfo sections. 379 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5 380 348345 Assertion fails for negative lineno 381 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI 382 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1 383 351692 Dumps created by valgrind are not readable by gdb (mips32 specific) 384 351804 Crash on generating suppressions for "printf" call on OS X 10.10 385 352197 mips: mmap2() not wrapped correctly for page size > 4096 386 353083 arm64 doesn't implement various xattr system calls 387 353084 arm64 doesn't support sigpending system call 388 353137 www: update info for Supported Platforms 389 353138 www: update "The Valgrind Developers" page 390 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine 391 == 365325 392 == 357873 393 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62) 394 353398 WARNING: unhandled amd64-solaris syscall: 207 395 353660 XML in auxwhat tag not escaping reserved symbols properly 396 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN 397 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72) 398 353802 ELF debug info reader confused with multiple .rodata sections 399 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed 400 353917 unhandled amd64-solaris syscall fchdir(120) 401 353920 unhandled amd64-solaris syscall: 170 402 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3) 403 354392 unhandled amd64-solaris syscall: 171 404 354797 Vbit test does not include Iops for Power 8 instruction support 405 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11 406 == 361351 407 == 362920 408 == 366222 409 354933 Fix documentation of --kernel-variant=android-no-hw-tls option 410 355188 valgrind should intercept all malloc related global functions 411 355454 do not intercept malloc related symbols from the runtime linker 412 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained 413 356044 Dwarf line info reader misinterprets is_stmt register 414 356112 mips: replace addi with addiu 415 356393 valgrind (vex) crashes because isZeroU happened 416 == 363497 417 == 364497 418 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min) 419 356678 arm64-linux: unhandled syscall 232 (mincore) 420 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND 421 356823 Unsupported ARM instruction: stlex 422 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state 423 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL 424 357673 crash if I try to run valgrind with a binary link with libcurl 425 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+ 426 357871 pthread_spin_destroy not properly wrapped 427 357887 Calls to VG_(fclose) do not close the file descriptor 428 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm. 429 358030 support direct socket calls on x86 32bit (new in linux 4.3) 430 358478 drd/tests/std_thread.cpp doesn't build with GCC6 431 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed 432 359181 Buffer Overflow during Demangling 433 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET 434 359289 s390x: popcnt (B9E1) not implemented 435 359472 The Power PC vsubuqm instruction doesn't always give the correct result 436 359503 Add missing syscalls for aarch64 (arm64) 437 359645 "You need libc6-dbg" help message could be more helpful 438 359703 s390: wire up separate socketcalls system calls 439 359724 getsockname might crash - deref_UInt should call safe_to_deref 440 359733 amd64 implement ld.so strchr/index override like x86 441 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5 442 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses 443 uninitialized data 444 359838 arm64: Unhandled instruction 0xD5033F5F (clrex) 445 359871 Incorrect mask handling in ppoll 446 359952 Unrecognised PCMPESTRM variants (0x70, 0x19) 447 360008 Contents of Power vr registers contents is not printed correctly when 448 the --vgdb-shadow-registers=yes option is used 449 360035 POWER PC instruction bcdadd and bcdsubtract generate result with 450 non-zero shadow bits 451 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2) 452 360425 arm64 unsupported instruction ldpsw 453 == 364435 454 360519 none/tests/arm64/memory.vgtest might fail with newer gcc 455 360571 Error about the Android Runtime reading below the stack pointer on ARM 456 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64 457 360749 kludge for multiple .rodata sections on Solaris no longer needed 458 360752 raise the number of reserved fds in m_main.c from 10 to 12 459 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5 460 361226 s390x: risbgn (EC59) not implemented 461 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create' 462 361354 ppc64[le]: wire up separate socketcalls system calls 463 361615 Inconsistent termination for multithreaded process terminated by signal 464 361926 Unhandled Solaris syscall: sysfs(84) 465 362009 V dumps core on unimplemented functionality before threads are created 466 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5 467 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64) 468 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed 469 362953 Request for an update to the Valgrind Developers page 470 363680 add renameat2() support 471 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at 472 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at 473 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5 474 364058 clarify in manual limitations of array overruns detections 475 364413 pselect sycallwrapper mishandles NULL sigmask 476 364728 Power PC, missing support for several HW registers in 477 get_otrack_shadow_offset_wrk() 478 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5 479 365273 Invalid write to stack location reported after signal handler runs 480 365912 ppc64BE segfault during jm-insns test (RELRO) 481 366079 FPXX Support for MIPS32 Valgrind 482 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0) 483 366344 Multiple unhandled instruction for Aarch64 484 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060) 485 367995 Integration of memcheck with custom memory allocator 486 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer 487 368412 False positive result for altivec capability check 488 368416 Add tc06_two_races_xml.exp output for ppc64 489 368419 Perf Events ioctls not implemented 490 368461 mmapunmap test fails on ppc64 491 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target 492 369000 AMD64 fma4 instructions unsupported. 493 369169 ppc64 fails jm_int_isa_2_07 test 494 369175 jm_vec_isa_2_07 test crashes on ppc64 495 369209 valgrind loops and eats up all memory if cwd doesn't exist. 496 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr 497 369359 msghdr_foreachfield can crash when handling bad iovec 498 369360 Bad sigprocmask old or new sets can crash valgrind 499 369361 vmsplice syscall wrapper crashes on bad iovec 500 369362 Bad sigaction arguments crash valgrind 501 369383 x86 sys_modify_ldt wrapper crashes on bad ptr 502 369402 Bad set/get_thread_area pointer crashes valgrind 503 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers 504 369446 valgrind crashes on unknown fcntl command 505 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER 506 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter) 507 370265 ISA 3.0 HW cap stuff needs updating 508 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work 509 372195 Power PC, xxsel instruction is not always recognized 510 511 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64 512 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap 513 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments 514 n-i-bz Fix ppoll_alarm exclusion on OS X 515 n-i-bz Document brk segment limitation, reference manual in limit reached msg. 516 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737] 517 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206] 518 n-i-bz amd64: memcheck false positive with shr %edx 519 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]" 520 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion 521 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D} 522 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd" 523 524 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094) 525 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096) 526 (3.12.0: 21 October 2016, vex r3282, valgrind r16098) 527 528 529 530 Release 3.11.0 (22 September 2015) 531 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 532 533 3.11.0 is a feature release with many improvements and the usual 534 collection of bug fixes. 535 536 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, 537 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, 538 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android, 539 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX 540 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for 541 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux. 542 543 * ================== PLATFORM CHANGES ================= 544 545 * Support for Solaris/x86 and Solaris/amd64 has been added. 546 547 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added. 548 549 * Preliminary support for the Tilera TileGX architecture has been added. 550 551 * s390x: It is now required for the host to have the "long displacement" 552 facility. The oldest supported machine model is z990. 553 554 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a 555 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3. 556 557 * The JIT's register allocator is significantly faster, making the JIT 558 as a whole somewhat faster, so JIT-intensive activities, for example 559 program startup, are modestly faster, around 5%. 560 561 * There have been changes to the default settings of several command 562 line flags, as detailed below. 563 564 * Intel AVX2 support is more complete (64 bit targets only). On AVX2 565 capable hosts, the simulated CPUID will now indicate AVX2 support. 566 567 * ==================== TOOL CHANGES ==================== 568 569 * Memcheck: 570 571 - The default value for --leak-check-heuristics has been changed from 572 "none" to "all". This helps to reduce the number of possibly 573 lost blocks, in particular for C++ applications. 574 575 - The default value for --keep-stacktraces has been changed from 576 "malloc-then-free" to "malloc-and-free". This has a small cost in 577 memory (one word per malloc-ed block) but allows Memcheck to show the 578 3 stacktraces of a dangling reference: where the block was allocated, 579 where it was freed, and where it is acccessed after being freed. 580 581 - The default value for --partial-loads-ok has been changed from "no" to 582 "yes", so as to avoid false positive errors resulting from some kinds 583 of vectorised loops. 584 585 - A new monitor command 'xb <addr> <len>' shows the validity bits of 586 <len> bytes at <addr>. The monitor command 'xb' is easier to use 587 than get_vbits when you need to associate byte data value with 588 their corresponding validity bits. 589 590 - The 'block_list' monitor command has been enhanced: 591 o it can print a range of loss records 592 o it now accepts an optional argument 'limited <max_blocks>' 593 to control the number of blocks printed. 594 o if a block has been found using a heuristic, then 595 'block_list' now shows the heuristic after the block size. 596 o the loss records/blocks to print can be limited to the blocks 597 found via specified heuristics. 598 599 - The C helper functions used to instrument loads on 600 x86-{linux,solaris} and arm-linux (both 32-bit only) have been 601 replaced by handwritten assembly sequences. This gives speedups 602 in the region of 0% to 7% for those targets only. 603 604 - A new command line option, --expensive-definedness-checks=yes|no, 605 has been added. This is useful for avoiding occasional invalid 606 uninitialised-value errors in optimised code. Watch out for 607 runtime degradation, as this can be up to 25%. As always, though, 608 the slowdown is highly application specific. The default setting 609 is "no". 610 611 * Massif: 612 613 - A new monitor command 'all_snapshots <filename>' dumps all 614 snapshots taken so far. 615 616 * Helgrind: 617 618 - Significant memory reduction and moderate speedups for 619 --history-level=full for applications accessing a lot of memory 620 with many different stacktraces. 621 622 - The default value for --conflict-cache-size=N has been doubled to 623 2000000. Users that were not using the default value should 624 preferably also double the value they give. 625 626 The default was changed due to the changes in the "full history" 627 implementation. Doubling the value gives on average a slightly more 628 complete history and uses similar memory (or significantly less memory 629 in the worst case) than the previous implementation. 630 631 - The Helgrind monitor command 'info locks' now accepts an optional 632 argument 'lock_addr', which shows information about the lock at the 633 given address only. 634 635 - When using --history-level=full, the new Helgrind monitor command 636 'accesshistory <addr> [<len>]' will show the recorded accesses for 637 <len> (or 1) bytes at <addr>. 638 639 * ==================== OTHER CHANGES ==================== 640 641 * The default value for the --smc-check option has been changed from 642 "stack" to "all-non-file" on targets that provide automatic D-I 643 cache coherence (x86, amd64 and s390x). The result is to provide, 644 by default, transparent support for JIT generated and self-modifying 645 code on all targets. 646 647 * Mac OS X only: the default value for the --dsymutil option has been 648 changed from "no" to "yes", since any serious usage on Mac OS X 649 always required it to be "yes". 650 651 * The command line options --db-attach and --db-command have been removed. 652 They were deprecated in 3.10.0. 653 654 * When a process dies due to a signal, Valgrind now shows the signal 655 and the stacktrace at default verbosity (i.e. verbosity 1). 656 657 * The address description logic used by Memcheck and Helgrind now 658 describes addresses in anonymous segments, file mmap-ed segments, 659 shared memory segments and the brk data segment. 660 661 * The new option --error-markers=<begin>,<end> can be used to mark the 662 begin/end of errors in textual output mode, to facilitate 663 searching/extracting errors in output files that mix valgrind errors 664 with program output. 665 666 * The new option --max-threads=<number> can be used to change the number 667 of threads valgrind can handle. The default is 500 threads which 668 should be more than enough for most applications. 669 670 * The new option --valgrind-stacksize=<number> can be used to change the 671 size of the private thread stacks used by Valgrind. This is useful 672 for reducing memory use or increasing the stack size if Valgrind 673 segfaults due to stack overflow. 674 675 * The new option --avg-transtab-entry-size=<number> can be used to specify 676 the expected instrumented block size, either to reduce memory use or 677 to avoid excessive retranslation. 678 679 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later. 680 681 * New and modified GDB server monitor features: 682 683 - When a signal is reported in GDB, you can now use the GDB convenience 684 variable $_siginfo to examine detailed signal information. 685 686 - Valgrind's gdbserver now allows the user to change the signal 687 to deliver to the process. So, use 'signal SIGNAL' to continue execution 688 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to 689 continue without passing the signal to the process. 690 691 - With GDB >= 7.10, the command 'target remote' 692 will automatically load the executable file of the process running 693 under Valgrind. This means you do not need to specify the executable 694 file yourself, GDB will discover it itself. See GDB documentation about 695 'qXfer:exec-file:read' packet for more info. 696 697 * ==================== FIXED BUGS ==================== 698 699 The following bugs have been fixed or resolved. Note that "n-i-bz" 700 stands for "not in bugzilla" -- that is, a bug that was reported to us 701 but never got a bugzilla entry. We encourage you to file bugs in 702 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 703 than mailing the developers (or mailing lists) directly -- bugs that 704 are not entered into bugzilla tend to get forgotten about or ignored. 705 706 To see details of a given bug, visit 707 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 708 where XXXXXX is the bug number as listed below. 709 710 116002 VG_(printf): Problems with justification of strings and integers 711 155125 avoid cutting away file:lineno after long function name 712 197259 Unsupported arch_prtctl PR_SET_GS option 713 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269 714 201216 Fix Valgrind does not support pthread_sigmask() on OS X 715 201435 Fix Darwin: -v does not show kernel version 716 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X 717 211256 Fixed an outdated comment regarding the default platform. 718 211529 Incomplete call stacks for code compiled by newer versions of MSVC 719 211926 Avoid compilation warnings in valgrind.h with -pedantic 720 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X 721 == 263119 722 226609 Crediting upstream authors in man page 723 231257 Valgrind omits path when executing script from shebang line 724 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..] 725 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames 726 269360 s390x: Fix addressing mode selection for compare-and-swap 727 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)' 728 == 326797 729 312989 ioctl handling needs to do POST handling on generic ioctls and [..] 730 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X 731 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it) 732 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8 733 330147 libmpiwrap PMPI_Get_count returns undefined value 734 333051 mmap of huge pages fails due to incorrect alignment 735 == 339163 736 334802 valgrind does not always explain why a given option is bad 737 335618 mov.w rN, pc/sp (ARM32) 738 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd) 739 == 307399 740 == 343175 741 == 342740 742 == 346912 743 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind 744 338602 AVX2 bit in CPUID missing 745 338606 Strange message for scripts with invalid interpreter 746 338731 ppc: Fix testuite build for toolchains not supporting -maltivec 747 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL 748 339045 Getting valgrind to compile and run on OS X Yosemite (10.10) 749 == 340252 750 339156 gdbsrv not called for fatal signal 751 339215 Valgrind 3.10.0 contain 2013 in copyrights notice 752 339288 support Cavium Octeon MIPS specific BBIT*32 instructions 753 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix 754 339442 Fix testsuite build failure on OS X 10.9 755 339542 Enable compilation with Intel's ICC compiler 756 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper 757 339688 Mac-specific ASM does not support .version directive (cpuid, 758 tronical and pushfpopf tests) 759 339745 Valgrind crash when check Marmalade app (partial fix) 760 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9 761 339778 Linux/TileGx platform support for Valgrind 762 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9 763 339789 Fix none/tests/execve test on Mac OS X 10.9 764 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9 765 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42) 766 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR 767 340392 Allow user to select more accurate definedness checking in memcheck 768 to avoid invalid complaints on optimised code 769 340430 Fix some grammatical weirdness in the manual. 770 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14) 771 341419 Signal handler ucontext_t not filled out correctly on OS X 772 341539 VG_(describe_addr) should not describe address as belonging to client 773 segment if it is past the heap end 774 341613 Enable building of manythreads and thread-exits tests on Mac OS X 775 341615 Fix none/tests/darwin/access_extended test on Mac OS X 776 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..] 777 341789 aarch64: shmat fails with valgrind on ARMv8 778 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly 779 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode 780 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy) 781 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests 782 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine 783 342221 socket connect false positive uninit memory for unknown af family 784 342353 Allow dumping full massif output while valgrind is still running 785 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS 786 == 346476 787 == 348387 788 == 350593 789 342603 Add I2C_SMBUS ioctl support 790 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code 791 342683 Mark memory past the initial brk limit as unaddressable 792 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12" 793 342795 Internal glibc __GI_mempcpy call should be intercepted 794 342841 s390x: Support instructions fiebr(a) and fidbr(a) 795 343012 Unhandled syscall 319 (memfd_create) 796 343069 Patch updating v4l2 API support 797 343173 helgrind crash during stack unwind 798 343219 fix GET_STARTREGS for arm 799 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10 800 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option 801 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64 802 343335 unhandled instruction 0x1E638400 (fccmp) aarch64 803 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..] 804 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..] 805 343597 ppc64le: incorrect use of offseof macro 806 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..] 807 343663 OS X 10.10 Memchecj always reports a leak regardless of [..] 808 343732 Unhandled syscall 144 (setgid) on aarch64 809 343733 Unhandled syscall 187 (msgctl and related) on aarch64 810 343802 s390x: False positive "conditional jump or move depends on [..] 811 343902 --vgdb=yes doesn't break when --xml=yes is used 812 343967 Don't warn about setuid/setgid/setcap executable for directories 813 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants 814 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344) 815 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait 816 344054 www - update info for Solaris/illumos 817 344416 'make regtest' does not work cleanly on OS X 818 344235 Remove duplicate include of pub_core_aspacemgr.h 819 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled 820 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled 821 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40) 822 344314 callgrind_annotate ... warnings about commands containing newlines 823 344318 socketcall should wrap recvmmsg and sendmmsg 824 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap) 825 344416 Fix 'make regtest' does not work cleanly on OS X 826 344499 Fix compilation for Linux kernel >= 4.0.0 827 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir), 828 unix:349 (__pthread_fchdir) 829 344559 Garbage collection of unused segment names in address space manager 830 344560 Fix stack traces missing penultimate frame on OS X 831 344621 Fix memcheck/tests/err_disable4 test on OS X 832 344686 Fix suppression for pthread_rwlock_init on OS X 10.10 833 344702 Fix missing libobjc suppressions on OS X 10.10 834 == 344543 835 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10 836 344939 Fix memcheck/tests/xml1 on OS X 10.10 837 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes 838 345079 Fix build problems in VEX/useful/test_main.c 839 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT 840 345177 arm64: prfm (reg) not implemented 841 345215 Performance improvements for the register allocator 842 345248 add support for Solaris OS in valgrind 843 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux 844 345394 Fix memcheck/tests/strchr on OS X 845 345637 Fix memcheck/tests/sendmsg on OS X 846 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2 847 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode 848 345887 Fix an assertion in the address space manager 849 345928 amd64: callstack only contains current function for small stacks 850 345984 disInstr(arm): unhandled instruction: 0xEE193F1E 851 345987 MIPS64: Implement cavium LHX instruction 852 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31) 853 346185 Fix typo saving altivec register v24 854 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER() 855 and LibVEX_GuestPPC64_get_CR() 856 346270 Regression tests none/tests/jm_vec/isa_2_07 and 857 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian 858 346307 fuse filesystem syscall deadlocks 859 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions 860 346411 MIPS: SysRes::_valEx handling is incorrect 861 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls 862 346474 PPC64 Power 8, spr TEXASRU register not supported 863 346487 Compiler generates "note" about a future ABI change for PPC64 864 346562 MIPS64: lwl/lwr instructions are performing 64bit loads 865 and causing spurious "invalid read of size 8" warnings 866 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack 867 347151 Fix suppression for pthread_rwlock_init on OS X 10.8 868 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell) 869 347322 Power PC regression test cleanup 870 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8 871 == 217236 872 347389 unhandled syscall: 373 (Linux ARM syncfs) 873 347686 Patch set to cleanup PPC64 regtests 874 347978 Remove bash dependencies where not needed 875 347982 OS X: undefined symbols for architecture x86_64: "_global" [..] 876 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine) 877 == 345929 878 348102 Patch updating v4l2 API support 879 348247 amd64 front end: jno jumps wrongly when overflow is not set 880 348269 Improve mmap MAP_HUGETLB support. 881 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates 882 348345 Assertion fails for negative lineno 883 348377 Unsupported ARM instruction: yield 884 348565 Fix detection of command line option availability for clang 885 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18) 886 348728 Fix broken check for VIDIOC_G_ENC_INDEX 887 348748 Fix redundant condition 888 348890 Fix clang warning about unsupported --param inline-unit-growth=900 889 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range" 890 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK 891 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..] 892 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..] 893 349626 Implemented additional Xen hypercalls 894 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64 895 349790 Clean up of the hardware capability checking utilities. 896 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so) 897 349874 Fix typos in source code 898 349879 memcheck: add handwritten assembly for helperc_LOADV* 899 349941 di_notify_mmap might create wrong start/size DebugInfoMapping 900 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X 901 350202 Add limited param to 'monitor block_list' 902 350290 s390x: Support instructions fixbr(a) 903 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X 904 350809 Fix none/tests/async-sigs for Solaris 905 350811 Remove reference to --db-attach which has been removed. 906 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too 907 350854 hard-to-understand code in VG_(load_ELF)() 908 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented 909 351386 Solaris: Cannot run ld.so.1 under Valgrind 910 351474 Fix VG_(iseqsigset) as obvious 911 351531 Typo in /include/vki/vki-xen-physdev.h header guard 912 351756 Intercept platform_memchr$VARIANT$Haswell on OS X 913 351858 ldsoexec support on Solaris 914 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode 915 352130 helgrind reports false races for printfs using mempcpy on FILE* state 916 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf 917 352320 arm64 crash on none/tests/nestedfs 918 352765 Vbit test fails on Power 6 919 352768 The mbar instruction is missing from the Power PC support 920 352769 Power PC program priority register (PPR) is not supported 921 n-i-bz Provide implementations of certain compiler builtins to support 922 compilers that may not provide those 923 n-i-bz Old STABS code is still being compiled, but never used. Remove it. 924 n-i-bz Fix compilation on distros with glibc < 2.5 925 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts 926 n-i-bz Enable rt_sigpending syscall on ppc64 linux. 927 n-i-bz mremap did not work properly on shared memory 928 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity 929 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml 930 931 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646) 932 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667) 933 (3.11.0: 22 September 2015, vex r3195, valgrind r15674) 934 935 936 937 Release 3.10.1 (25 November 2014) 938 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 939 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0 940 and backports fixes for all reported missing AArch64 ARMv8 instructions 941 and syscalls from the trunk. If you package or deliver 3.10.0 for others 942 to use, you might want to consider upgrading to 3.10.1 instead. 943 944 The following bugs have been fixed or resolved. Note that "n-i-bz" 945 stands for "not in bugzilla" -- that is, a bug that was reported to us 946 but never got a bugzilla entry. We encourage you to file bugs in 947 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 948 than mailing the developers (or mailing lists) directly -- bugs that 949 are not entered into bugzilla tend to get forgotten about or ignored. 950 951 To see details of a given bug, visit 952 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 953 where XXXXXX is the bug number as listed below. 954 955 335440 arm64: ld1 (single structure) is not implemented 956 335713 arm64: unhanded instruction: prfm (immediate) 957 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build 958 339182 ppc64: AvSplat ought to load destination vector register with [..] 959 339336 PPC64 store quad instruction (stq) is not supposed to change [..] 960 339433 ppc64 lxvw4x instruction uses four 32-byte loads 961 339645 Use correct tag names in sys_getdents/64 wrappers 962 339706 Fix false positive for ioctl(TIOCSIG) on linux 963 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ... 964 339853 arm64 times syscall unknown 965 339855 arm64 unhandled getsid/setsid syscalls 966 339858 arm64 dmb sy not implemented 967 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64 968 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64 969 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla) 970 == 339950 971 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch 972 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants 973 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid) 974 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset) 975 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat 976 340509 arm64: unhandled instruction fcvtas 977 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized 978 340632 arm64: unhandled instruction fcvtas 979 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000" 980 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms 981 340788 warning: unhandled syscall: 318 (getrandom) 982 340807 disInstr(arm): unhandled instruction: 0xEE989B20 983 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel) 984 340922 arm64: unhandled getgroups/setgroups syscalls 985 350251 Fix typo in VEX utility program (test_main.c). 986 350407 arm64: unhandled instruction ucvtf (vector, integer) 987 350809 none/tests/async-sigs breaks when run under cron on Solaris 988 350811 update README.solaris after r15445 989 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..] 990 350854 strange code in VG_(load_ELF)() 991 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented 992 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM) 993 n-i-bz Add missing ]] to terminate CDATA. 994 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO 995 n-i-bz Enable sys_fadvise64_64 on arm32. 996 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns. 997 n-i-bz Add test cases for all known arm64 load/store instructions. 998 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..] 999 n-i-bz Add detection of old ppc32 magic instructions from bug 278808. 1000 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size". 1001 n-i-bz arm64: Implement "fcvtpu w, s". 1002 n-i-bz arm64: implement ADDP and various others 1003 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt). 1004 n-i-bz arm64: enable FCVT{A,N}S X,S. 1005 1006 (3.10.1: 25 November 2014, vex r3026, valgrind r14785) 1007 1008 1009 1010 Release 3.10.0 (10 September 2014) 1011 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1012 1013 3.10.0 is a feature release with many improvements and the usual 1014 collection of bug fixes. 1015 1016 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux, 1017 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux, 1018 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9 1019 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is 1020 significantly improved relative to the 3.9.0 release. 1021 1022 * ================== PLATFORM CHANGES ================= 1023 1024 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port 1025 is mostly complete, and is usable, but some SIMD instructions are as 1026 yet unsupported. 1027 1028 * Support for little-endian variant of the 64-bit POWER architecture. 1029 1030 * Support for Android on MIPS32. 1031 1032 * Support for 64bit FPU on MIPS32 platforms. 1033 1034 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9. 1035 1036 * Configuration for and running on Android targets has changed. 1037 See README.android in the source tree for details. 1038 1039 * ================== DEPRECATED FEATURES ================= 1040 1041 * --db-attach is now deprecated and will be removed in the next 1042 valgrind feature release. The built-in GDB server capabilities are 1043 superior and should be used instead. Learn more here: 1044 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver 1045 1046 * ==================== TOOL CHANGES ==================== 1047 1048 * Memcheck: 1049 1050 - Client code can now selectively disable and re-enable reporting of 1051 invalid address errors in specific ranges using the new client 1052 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and 1053 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE. 1054 1055 - Leak checker: there is a new leak check heuristic called 1056 "length64". This is used to detect interior pointers pointing 8 1057 bytes inside a block, on the assumption that the first 8 bytes 1058 holds the value "block size - 8". This is used by 1059 sqlite3MemMalloc, for example. 1060 1061 - Checking of system call parameters: if a syscall parameter 1062 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has 1063 several fields not initialised, an error is now reported for each 1064 field. Previously, an error was reported only for the first 1065 uninitialised field. 1066 1067 - Mismatched alloc/free checking: a new flag 1068 --show-mismatched-frees=no|yes [yes] makes it possible to turn off 1069 such checks if necessary. 1070 1071 * Helgrind: 1072 1073 - Improvements to error messages: 1074 1075 o Race condition error message involving heap allocated blocks also 1076 show the thread number that allocated the raced-on block. 1077 1078 o All locks referenced by an error message are now announced. 1079 Previously, some error messages only showed the lock addresses. 1080 1081 o The message indicating where a lock was first observed now also 1082 describes the address/location of the lock. 1083 1084 - Helgrind now understands the Ada task termination rules and 1085 creates a happens-before relationship between a terminated task 1086 and its master. This avoids some false positives and avoids a big 1087 memory leak when a lot of Ada tasks are created and terminated. 1088 The interceptions are only activated with forthcoming releases of 1089 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0. 1090 1091 - A new GDB server monitor command "info locks" giving the list of 1092 locks, their location, and their status. 1093 1094 * Callgrind: 1095 1096 - callgrind_control now supports the --vgdb-prefix argument, 1097 which is needed if valgrind was started with this same argument. 1098 1099 * ==================== OTHER CHANGES ==================== 1100 1101 * Unwinding through inlined function calls. Stack unwinding can now 1102 make use of Dwarf3 inlined-unwind information if it is available. 1103 The practical effect is that inlined calls become visible in stack 1104 traces. The suppression matching machinery has been adjusted 1105 accordingly. This is controlled by the new option 1106 --read-inline-info=yes|no. Currently this is enabled by default 1107 only on Linux and Android targets and only for the tools Memcheck, 1108 Helgrind and DRD. 1109 1110 * Valgrind can now read EXIDX unwind information on 32-bit ARM 1111 targets. If an object contains both CFI and EXIDX unwind 1112 information, Valgrind will prefer the CFI over the EXIDX. This 1113 facilitates unwinding through system libraries on arm-android 1114 targets. 1115 1116 * Address description logic has been improved and is now common 1117 between Memcheck and Helgrind, resulting in better address 1118 descriptions for some kinds of error messages. 1119 1120 * Error messages about dubious arguments (eg, to malloc or calloc) are 1121 output like other errors. This means that they can be suppressed 1122 and they have a stack trace. 1123 1124 * The C++ demangler has been updated for better C++11 support. 1125 1126 * New and modified GDB server monitor features: 1127 1128 - Thread local variables/storage (__thread) can now be displayed. 1129 1130 - The GDB server monitor command "v.info location <address>" 1131 displays information about an address. The information produced 1132 depends on the tool and on the options given to valgrind. 1133 Possibly, the following are described: global variables, local 1134 (stack) variables, allocated or freed blocks, ... 1135 1136 - The option "--vgdb-stop-at=event1,event2,..." allows the user to 1137 ask the GDB server to stop at the start of program execution, at 1138 the end of the program execution and on Valgrind internal errors. 1139 1140 - A new monitor command "v.info stats" shows various Valgrind core 1141 and tool statistics. 1142 1143 - A new monitor command "v.set hostvisibility" allows the GDB server 1144 to provide access to Valgrind internal host status/memory. 1145 1146 * A new option "--aspace-minaddr=<address>" can in some situations 1147 allow the use of more memory by decreasing the address above which 1148 Valgrind maps memory. It can also be used to solve address 1149 conflicts with system libraries by increasing the default value. 1150 See user manual for details. 1151 1152 * The amount of memory used by Valgrind to store debug info (unwind 1153 info, line number information and symbol data) has been 1154 significantly reduced, even though Valgrind now reads more 1155 information in order to support unwinding of inlined function calls. 1156 1157 * Dwarf3 handling with --read-var-info=yes has been improved: 1158 1159 - Ada and C struct containing VLAs no longer cause a "bad DIE" error 1160 1161 - Code compiled with 1162 -ffunction-sections -fdata-sections -Wl,--gc-sections 1163 no longer causes assertion failures. 1164 1165 * Improved checking for the --sim-hints= and --kernel-variant= 1166 options. Unknown strings are now detected and reported to the user 1167 as a usage error. 1168 1169 * The semantics of stack start/end boundaries in the valgrind.h 1170 VALGRIND_STACK_REGISTER client request has been clarified and 1171 documented. The convention is that start and end are respectively 1172 the lowest and highest addressable bytes of the stack. 1173 1174 * ==================== FIXED BUGS ==================== 1175 1176 The following bugs have been fixed or resolved. Note that "n-i-bz" 1177 stands for "not in bugzilla" -- that is, a bug that was reported to us 1178 but never got a bugzilla entry. We encourage you to file bugs in 1179 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 1180 than mailing the developers (or mailing lists) directly -- bugs that 1181 are not entered into bugzilla tend to get forgotten about or ignored. 1182 1183 To see details of a given bug, visit 1184 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 1185 where XXXXXX is the bug number as listed below. 1186 1187 175819 Support for ipv6 socket reporting with --track-fds 1188 232510 make distcheck fails 1189 249435 Analyzing wine programs with callgrind triggers a crash 1190 278972 support for inlined function calls in stacktraces and suppression 1191 == 199144 1192 291310 FXSAVE instruction marks memory as undefined on amd64 1193 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped 1194 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall) 1195 315199 vgcore file for threaded app does not show which thread crashed 1196 315952 tun/tap ioctls are not supported 1197 323178 Unhandled instruction: PLDW register (ARM) 1198 323179 Unhandled instruction: PLDW immediate (ARM) 1199 324050 Helgrind: SEGV because of unaligned stack when using movdqa 1200 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo. 1201 325124 [MIPSEL] Compilation error 1202 325477 Phase 4 support for IBM Power ISA 2.07 1203 325538 cavium octeon mips64, valgrind reported "dumping core" [...] 1204 325628 Phase 5 support for IBM Power ISA 2.07 1205 325714 Empty vgcore but RLIMIT_CORE is big enough (too big) 1206 325751 Missing the two privileged Power PC Transactional Memory Instructions 1207 325816 Phase 6 support for IBM Power ISA 2.07 1208 325856 Make SGCheck fail gracefully on unsupported platforms 1209 326026 Iop names for count leading zeros/sign bits incorrectly imply [..] 1210 326436 DRD: False positive in libstdc++ std::list::push_back 1211 326444 Cavium MIPS Octeon Specific Load Indexed Instructions 1212 326462 Refactor vgdb to isolate invoker stuff into separate module 1213 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E) 1214 326623 DRD: false positive conflict report in a field assignment 1215 326724 Valgrind does not compile on OSX 1.9 Mavericks 1216 326816 Intercept for __strncpy_sse2_unaligned missing? 1217 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V 1218 326983 Clear direction flag after tests on amd64. 1219 327212 Do not prepend the current directory to absolute path names. 1220 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions 1221 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed 1222 327284 s390x: Fix translation of the risbg instruction 1223 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34 1224 327837 dwz compressed alternate .debug_info and .debug_str not read correctly 1225 327916 DW_TAG_typedef may have no name 1226 327943 s390x: add a redirection for the 'index' function 1227 328100 XABORT not implemented 1228 328205 Implement additional Xen hypercalls 1229 328454 add support Backtraces with ARM unwind tables (EXIDX) 1230 328455 s390x: SIGILL after emitting wrong register pair for ldxbr 1231 328711 valgrind.1 manpage "memcheck options" section is badly generated 1232 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14 1233 329612 Incorrect handling of AT_BASE for image execution 1234 329694 clang warns about using uninitialized variable 1235 329956 valgrind crashes when lmw/stmw instructions are used on ppc64 1236 330228 mmap must align to VKI_SHMLBA on mips32 1237 330257 LLVM does not support `-mno-dynamic-no-pic` option 1238 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend) 1239 330459 --track-fds=yes doesn't track eventfds 1240 330469 Add clock_adjtime syscall support 1241 330594 Missing sysalls on PowerPC / uClibc 1242 330622 Add test to regression suite for POWER instruction: dcbzl 1243 330939 Support for AMD's syscall instruction on x86 1244 == 308729 1245 330941 Typo in PRE(poll) syscall wrapper 1246 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch) 1247 331254 Fix expected output for memcheck/tests/dw4 1248 331255 Fix race condition in test none/tests/coolo_sigaction 1249 331257 Fix type of jump buffer in test none/tests/faultstatus 1250 331305 configure uses bash specific syntax 1251 331337 s390x WARNING: unhandled syscall: 326 (dup3) 1252 331380 Syscall param timer_create(evp) points to uninitialised byte(s) 1253 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64) 1254 331829 Unexpected ioctl opcode sign extension 1255 331830 ppc64: WARNING: unhandled syscall: 96/97 1256 331839 drd/tests/sem_open specifies invalid semaphore name 1257 331847 outcome of drd/tests/thread_name is nondeterministic 1258 332037 Valgrind cannot handle Thumb "add pc, reg" 1259 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and 1260 consistency checks enabled 1261 332263 intercepts for pthread_rwlock_timedrdlock and 1262 pthread_rwlock_timedwrlock are incorrect 1263 332265 drd could do with post-rwlock_init and pre-rwlock_destroy 1264 client requests 1265 332276 Implement additional Xen hypercalls 1266 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment 1267 332765 Fix ms_print to create temporary files in a proper directory 1268 333072 drd: Add semaphore annotations 1269 333145 Tests for missaligned PC+#imm access for arm 1270 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0 1271 333230 AAarch64 missing instruction encodings: dc, ic, dsb. 1272 333248 WARNING: unhandled syscall: unix:443 1273 333428 ldr.w pc [rD, #imm] instruction leads to assertion 1274 333501 cachegrind: assertion: Cache set count is not a power of two. 1275 == 336577 1276 == 292281 1277 333666 Recognize MPX instructions and bnd prefix. 1278 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch) 1279 333817 Valgrind reports the memory areas written to by the SG_IO 1280 ioctl as untouched 1281 334049 lzcnt fails silently (x86_32) 1282 334384 Valgrind does not have support Little Endian support for 1283 IBM POWER PPC 64 1284 334585 recvmmsg unhandled (+patch) (arm) 1285 334705 sendmsg and recvmsg should guard against bogus msghdr fields. 1286 334727 Build fails with -Werror=format-security 1287 334788 clarify doc about --log-file initial program directory 1288 334834 PPC64 Little Endian support, patch 2 1289 334836 PPC64 Little Endian support, patch 3 testcase fixes 1290 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls 1291 335034 Unhandled ioctl: HCIGETDEVLIST 1292 335155 vgdb, fix error print statement. 1293 335262 arm64: movi 8bit version is not supported 1294 335263 arm64: dmb instruction is not implemented 1295 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind 1296 335496 arm64: sbc/abc instructions are not implemented 1297 335554 arm64: unhandled instruction: abs 1298 335564 arm64: unhandled instruction: fcvtpu Xn, Sn 1299 335735 arm64: unhandled instruction: cnt 1300 335736 arm64: unhandled instruction: uaddlv 1301 335848 arm64: unhandled instruction: {s,u}cvtf 1302 335902 arm64: unhandled instruction: sli 1303 335903 arm64: unhandled instruction: umull (vector) 1304 336055 arm64: unhandled instruction: mov (element) 1305 336062 arm64: unhandled instruction: shrn{,2} 1306 336139 mip64: [...] valgrind hangs and spins on a single core [...] 1307 336189 arm64: unhandled Instruction: mvn 1308 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU 1309 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type 1310 336772 Make moans about unknown ioctls more informative 1311 336957 Add a section about the Solaris/illumos port on the webpage 1312 337094 ifunc wrapper is broken on ppc64 1313 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported 1314 337528 leak check heuristic for block prefixed by length as 64bit number 1315 337740 Implement additional Xen hypercalls 1316 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed. 1317 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230) 1318 337871 deprecate --db-attach 1319 338023 Add support for all V4L2/media ioctls 1320 338024 inlined functions are not shown if DW_AT_ranges is used 1321 338106 Add support for 'kcmp' syscall 1322 338115 DRD: computed conflict set differs from actual after fork 1323 338160 implement display of thread local storage in gdbsrv 1324 338205 configure.ac and check for -Wno-tautological-compare 1325 338300 coredumps are missing one byte of every segment 1326 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX 1327 338499 --sim-hints parsing broken due to wrong order in tokens 1328 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7 1329 338681 Unable to unwind through clone thread created on i386-linux 1330 338698 race condition between gdbsrv and vgdb on startup 1331 338703 helgrind on arm-linux gets false positives in dynamic loader 1332 338791 alt dwz files can be relative of debug/main file 1333 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed 1334 338932 build V-trunk with gcc-trunk 1335 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390 1336 345079 Fix build problems in VEX/useful/test_main.c 1337 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling 1338 n-i-bz s390x: Fix memory corruption for multithreaded applications 1339 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases 1340 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked 1341 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014) 1342 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread" 1343 n-i-bz Add command-line option --default-suppressions 1344 n-i-bz Add support for BLKDISCARDZEROES ioctl 1345 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions 1346 n-i-bz Add support for sys_pivot_root and sys_unshare 1347 1348 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428) 1349 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503) 1350 (3.10.0: 10 September 2014, vex r2950, valgrind r14514) 1351 1352 1353 1354 Release 3.9.0 (31 October 2013) 1355 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1356 3.9.0 is a feature release with many improvements and the usual 1357 collection of bug fixes. 1358 1359 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, 1360 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android, 1361 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for 1362 MacOSX 10.8 is significantly improved relative to the 3.8.0 release. 1363 1364 * ================== PLATFORM CHANGES ================= 1365 1366 * Support for MIPS64 LE and BE running Linux. Valgrind has been 1367 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions. 1368 1369 * Support for MIPS DSP ASE on MIPS32 platforms. 1370 1371 * Support for s390x Decimal Floating Point instructions on hosts that 1372 have the DFP facility installed. 1373 1374 * Support for POWER8 (Power ISA 2.07) instructions 1375 1376 * Support for Intel AVX2 instructions. This is available only on 64 1377 bit code. 1378 1379 * Initial support for Intel Transactional Synchronization Extensions, 1380 both RTM and HLE. 1381 1382 * Initial support for Hardware Transactional Memory on POWER. 1383 1384 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now 1385 run large GUI apps tolerably well. 1386 1387 * ==================== TOOL CHANGES ==================== 1388 1389 * Memcheck: 1390 1391 - Improvements in handling of vectorised code, leading to 1392 significantly fewer false error reports. You need to use the flag 1393 --partial-loads-ok=yes to get the benefits of these changes. 1394 1395 - Better control over the leak checker. It is now possible to 1396 specify which leak kinds (definite/indirect/possible/reachable) 1397 should be displayed, which should be regarded as errors, and which 1398 should be suppressed by a given leak suppression. This is done 1399 using the options --show-leak-kinds=kind1,kind2,.., 1400 --errors-for-leak-kinds=kind1,kind2,.. and an optional 1401 "match-leak-kinds:" line in suppression entries, respectively. 1402 1403 Note that generated leak suppressions contain this new line and 1404 are therefore more specific than in previous releases. To get the 1405 same behaviour as previous releases, remove the "match-leak-kinds:" 1406 line from generated suppressions before using them. 1407 1408 - Reduced "possible leak" reports from the leak checker by the use 1409 of better heuristics. The available heuristics provide detection 1410 of valid interior pointers to std::stdstring, to new[] allocated 1411 arrays with elements having destructors and to interior pointers 1412 pointing to an inner part of a C++ object using multiple 1413 inheritance. They can be selected individually using the 1414 option --leak-check-heuristics=heur1,heur2,... 1415 1416 - Better control of stacktrace acquisition for heap-allocated 1417 blocks. Using the --keep-stacktraces option, it is possible to 1418 control independently whether a stack trace is acquired for each 1419 allocation and deallocation. This can be used to create better 1420 "use after free" errors or to decrease Valgrind's resource 1421 consumption by recording less information. 1422 1423 - Better reporting of leak suppression usage. The list of used 1424 suppressions (shown when the -v option is given) now shows, for 1425 each leak suppressions, how many blocks and bytes it suppressed 1426 during the last leak search. 1427 1428 * Helgrind: 1429 1430 - False errors resulting from the use of statically initialised 1431 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc) 1432 have been removed. 1433 1434 - False errors resulting from the use of pthread_cond_waits that 1435 timeout, have been removed. 1436 1437 * ==================== OTHER CHANGES ==================== 1438 1439 * Some attempt to tune Valgrind's space requirements to the expected 1440 capabilities of the target: 1441 1442 - The default size of the translation cache has been reduced from 8 1443 sectors to 6 on Android platforms, since each sector occupies 1444 about 40MB when using Memcheck. 1445 1446 - The default size of the translation cache has been increased to 16 1447 sectors on all other platforms, reflecting the fact that large 1448 applications require instrumentation and storage of huge amounts 1449 of code. For similar reasons, the number of memory mapped 1450 segments that can be tracked has been increased by a factor of 6. 1451 1452 - In all cases, the maximum number of sectors in the translation 1453 cache can be controlled by the new flag --num-transtab-sectors. 1454 1455 * Changes in how debug info (line numbers, etc) is read: 1456 1457 - Valgrind no longer temporarily mmaps the entire object to read 1458 from it. Instead, reading is done through a small fixed sized 1459 buffer. This avoids virtual memory usage spikes when Valgrind 1460 reads debuginfo from large shared objects. 1461 1462 - A new experimental remote debug info server. Valgrind can read 1463 debug info from a different machine (typically, a build host) 1464 where debuginfo objects are stored. This can save a lot of time 1465 and hassle when running Valgrind on resource-constrained targets 1466 (phones, tablets) when the full debuginfo objects are stored 1467 somewhere else. This is enabled by the --debuginfo-server= 1468 option. 1469 1470 - Consistency checking between main and debug objects can be 1471 disabled using the --allow-mismatched-debuginfo option. 1472 1473 * Stack unwinding by stack scanning, on ARM. Unwinding by stack 1474 scanning can recover stack traces in some cases when the normal 1475 unwind mechanisms fail. Stack scanning is best described as "a 1476 nasty, dangerous and misleading hack" and so is disabled by default. 1477 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable 1478 and control it. 1479 1480 * Detection and merging of recursive stack frame cycles. When your 1481 program has recursive algorithms, this limits the memory used by 1482 Valgrind for recorded stack traces and avoids recording 1483 uninteresting repeated calls. This is controlled by the command 1484 line option --merge-recursive-frame and by the monitor command 1485 "v.set merge-recursive-frames". 1486 1487 * File name and line numbers for used suppressions. The list of used 1488 suppressions (shown when the -v option is given) now shows, for each 1489 used suppression, the file name and line number where the suppression 1490 is defined. 1491 1492 * New and modified GDB server monitor features: 1493 1494 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND, 1495 that can be used to execute gdbserver monitor commands from the 1496 client program. 1497 1498 - A new monitor command, "v.info open_fds", that gives the list of 1499 open file descriptors and additional details. 1500 1501 - An optional message in the "v.info n_errs_found" monitor command, 1502 for example "v.info n_errs_found test 1234 finished", allowing a 1503 comment string to be added to the process output, perhaps for the 1504 purpose of separating errors of different tests or test phases. 1505 1506 - A new monitor command "v.info execontext" that shows information 1507 about the stack traces recorded by Valgrind. 1508 1509 - A new monitor command "v.do expensive_sanity_check_general" to run 1510 some internal consistency checks. 1511 1512 * New flag --sigill-diagnostics to control whether a diagnostic 1513 message is printed when the JIT encounters an instruction it can't 1514 translate. The actual behavior -- delivery of SIGILL to the 1515 application -- is unchanged. 1516 1517 * The maximum amount of memory that Valgrind can use on 64 bit targets 1518 has been increased from 32GB to 64GB. This should make it possible 1519 to run applications on Memcheck that natively require up to about 35GB. 1520 1521 * ==================== FIXED BUGS ==================== 1522 1523 The following bugs have been fixed or resolved. Note that "n-i-bz" 1524 stands for "not in bugzilla" -- that is, a bug that was reported to us 1525 but never got a bugzilla entry. We encourage you to file bugs in 1526 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 1527 than mailing the developers (or mailing lists) directly -- bugs that 1528 are not entered into bugzilla tend to get forgotten about or ignored. 1529 1530 To see details of a given bug, visit 1531 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 1532 where XXXXXX is the bug number as listed below. 1533 1534 123837 system call: 4th argument is optional, depending on cmd 1535 135425 memcheck should tell you where Freed blocks were Mallocd 1536 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small 1537 207815 Adds some of the drm ioctls to syswrap-linux.c 1538 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP) 1539 252955 Impossible to compile with ccache 1540 253519 Memcheck reports auxv pointer accesses as invalid reads. 1541 263034 Crash when loading some PPC64 binaries 1542 269599 Increase deepest backtrace 1543 274695 s390x: Support "compare to/from logical" instructions (z196) 1544 275800 s390x: Autodetect cache info (part 2) 1545 280271 Valgrind reports possible memory leaks on still-reachable std::string 1546 284540 Memcheck shouldn't count suppressions matching still-reachable [..] 1547 289578 Backtraces with ARM unwind tables (stack scan flags) 1548 296311 Wrong stack traces due to -fomit-frame-pointer (x86) 1549 304832 ppc32: build failure 1550 305431 Use find_buildid shdr fallback for separate .debug files 1551 305728 Add support for AVX2 instructions 1552 305948 ppc64: code generation for ShlD64 / ShrD64 asserts 1553 306035 s390x: Fix IR generation for LAAG and friends 1554 306054 s390x: Condition code computation for convert-to-int/logical 1555 306098 s390x: alternate opcode form for convert to/from fixed 1556 306587 Fix cache line detection from auxiliary vector for PPC. 1557 306783 Mips unhandled syscall : 4025 / 4079 / 4182 1558 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al) 1559 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV 1560 307101 sys_capget second argument can be NULL 1561 307103 sys_openat: If pathname is absolute, then dirfd is ignored. 1562 307106 amd64->IR: f0 0f c0 02 (lock xadd byte) 1563 307113 s390x: DFP support 1564 307141 valgrind does't work in mips-linux system 1565 307155 filter_gdb should filter out syscall-template.S T_PSEUDO 1566 307285 x86_amd64 feature test for avx in test suite is wrong 1567 307290 memcheck overlap testcase needs memcpy version filter 1568 307463 Please add "&limit=0" to the "all open bugs" link 1569 307465 --show-possibly-lost=no should reduce the error count / exit code 1570 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..] 1571 307729 pkgconfig support broken valgrind.pc 1572 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr 1573 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang 1574 308089 Unhandled syscall on ppc64: prctl 1575 308135 PPC32 MPC8xx has 16 bytes cache size 1576 308321 testsuite memcheck filter interferes with gdb_filter 1577 308333 == 307106 1578 308341 vgdb should report process exit (or fatal signal) 1579 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit 1580 308495 Remove build dependency on installed Xen headers 1581 308573 Internal error on 64-bit instruction executed in 32-bit mode 1582 308626 == 308627 1583 308627 pmovmskb validity bit propagation is imprecise 1584 308644 vgdb command for having the info for the track-fds option 1585 308711 give more info about aspacemgr and arenas in out_of_memory 1586 308717 ARM: implement fixed-point VCVT.F64.[SU]32 1587 308718 ARM implement SMLALBB family of instructions 1588 308886 Missing support for PTRACE_SET/GETREGSET 1589 308930 syscall name_to_handle_at (303 on amd64) not handled 1590 309229 V-bit tester does not report number of tests generated 1591 309323 print unrecognized instuction on MIPS 1592 309425 Provide a --sigill-diagnostics flag to suppress illegal [..] 1593 309427 SSE optimized stpncpy trigger uninitialised value [..] errors 1594 309430 Self hosting ppc64 encounters a vassert error on operand type 1595 309600 valgrind is a bit confused about 0-sized sections 1596 309823 Generate errors for still reachable blocks 1597 309921 PCMPISTRI validity bit propagation is imprecise 1598 309922 none/tests/ppc64/test_dfp5 sometimes fails 1599 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker. 1600 310424 --read-var-info does not properly describe static variables 1601 310792 search additional path for debug symbols 1602 310931 s390x: Message-security assist (MSA) instruction extension [..] 1603 311100 PPC DFP implementation of the integer operands is inconsistent [..] 1604 311318 ARM: "128-bit constant is not implemented" error message 1605 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..] 1606 311690 V crashes because it redirects branches inside of a redirected function 1607 311880 x86_64: make regtest hangs at shell_valid1 1608 311922 WARNING: unhandled syscall: 170 1609 311933 == 251569 1610 312171 ppc: insn selection for DFP 1611 312571 Rounding mode call wrong for the DFP Iops [..] 1612 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..] 1613 312913 Dangling pointers error should also report the alloc stack trace 1614 312980 Building on Mountain Lion generates some compiler warnings 1615 313267 Adding MIPS64/Linux port to Valgrind 1616 313348 == 251569 1617 313354 == 251569 1618 313811 Buffer overflow in assert_fail 1619 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix 1620 314269 ppc: dead code in insn selection 1621 314718 ARM: implement integer divide instruction (sdiv and udiv) 1622 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi= 1623 315441 sendmsg syscall should ignore unset msghdr msg_flags 1624 315534 msgrcv inside a thread causes valgrind to hang (block) 1625 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed 1626 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT) 1627 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64) 1628 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section 1629 316144 valgrind.1 manpage contains unknown ??? strings [..] 1630 316145 callgrind command line options in manpage reference (unknown) [..] 1631 316145 callgrind command line options in manpage reference [..] 1632 316181 drd: Fixed a 4x slowdown for certain applications 1633 316503 Valgrind does not support SSE4 "movntdqa" instruction 1634 316535 Use of |signed int| instead of |size_t| in valgrind messages 1635 316696 fluidanimate program of parsec 2.1 stuck 1636 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled 1637 317091 Use -Wl,-Ttext-segment when static linking if possible [..] 1638 317186 "Impossible happens" when occurs VCVT instruction on ARM 1639 317318 Support for Threading Building Blocks "scalable_malloc" 1640 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps) 1641 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs 1642 317463 bmi testcase IR SANITY CHECK FAILURE 1643 317506 memcheck/tests/vbit-test fails with unknown opcode after [..] 1644 318050 libmpiwrap fails to compile with out-of-source build 1645 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER 1646 318643 annotate_trace_memory tests infinite loop on arm and ppc [..] 1647 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F 1648 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16) 1649 318932 Add missing PPC64 and PPC32 system call support 1650 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled 1651 319395 Crash with unhandled instruction on STRT (Thumb) instructions 1652 319494 VEX Makefile-gcc standalone build update after r2702 1653 319505 [MIPSEL] Crash: unhandled UNRAY operator. 1654 319858 disInstr(thumb): unhandled instruction on instruction STRBT 1655 319932 disInstr(thumb): unhandled instruction on instruction STRHT 1656 320057 Problems when we try to mmap more than 12 memory pages on MIPS32 1657 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised 1658 320083 disInstr(thumb): unhandled instruction on instruction LDRBT 1659 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding 1660 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64) 1661 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR 1662 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname' 1663 320895 add fanotify support (patch included) 1664 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction 1665 321065 Valgrind updates for Xen 4.3 1666 321148 Unhandled instruction: PLI (Thumb 1, 2, 3) 1667 321363 Unhandled instruction: SSAX (ARM + Thumb) 1668 321364 Unhandled instruction: SXTAB16 (ARM + Thumb) 1669 321466 Unhandled instruction: SHASX (ARM + Thumb) 1670 321467 Unhandled instruction: SHSAX (ARM + Thumb) 1671 321468 Unhandled instruction: SHSUB16 (ARM + Thumb) 1672 321619 Unhandled instruction: SHSUB8 (ARM + Thumb) 1673 321620 Unhandled instruction: UASX (ARM + Thumb) 1674 321621 Unhandled instruction: USAX (ARM + Thumb) 1675 321692 Unhandled instruction: UQADD16 (ARM + Thumb) 1676 321693 Unhandled instruction: LDRSBT (Thumb) 1677 321694 Unhandled instruction: UQASX (ARM + Thumb) 1678 321696 Unhandled instruction: UQSAX (Thumb + ARM) 1679 321697 Unhandled instruction: UHASX (ARM + Thumb) 1680 321703 Unhandled instruction: UHSAX (ARM + Thumb) 1681 321704 Unhandled instruction: REVSH (ARM + Thumb) 1682 321730 Add cg_diff and cg_merge man pages 1683 321738 Add vgdb and valgrind-listener man pages 1684 321814 == 315545 1685 321891 Unhandled instruction: LDRHT (Thumb) 1686 321960 pthread_create() then alloca() causing invalid stack write errors 1687 321969 ppc32 and ppc64 don't support [lf]setxattr 1688 322254 Show threadname together with tid if set by application 1689 322294 Add initial support for IBM Power ISA 2.07 1690 322368 Assertion failure in wqthread_hijack under OS X 10.8 1691 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A 1692 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr 1693 322851 0bXXX binary literal syntax is not standard 1694 323035 Unhandled instruction: LDRSHT(Thumb) 1695 323036 Unhandled instruction: SMMLS (ARM and Thumb) 1696 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..] 1697 323175 Unhandled instruction: SMLALD (ARM + Thumb) 1698 323177 Unhandled instruction: SMLSLD (ARM + Thumb) 1699 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..] 1700 323437 Phase 2 support for IBM Power ISA 2.07 1701 323713 Support mmxext (integer sse) subset on i386 (athlon) 1702 323803 Transactional memory instructions are not supported for Power 1703 323893 SSE3 not available on amd cpus in valgrind 1704 323905 Probable false positive from Valgrind/drd on close() 1705 323912 valgrind.h header isn't compatible for mingw64 1706 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions 1707 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..] 1708 324181 mmap does not handle MAP_32BIT 1709 324227 memcheck false positive leak when a thread calls exit+block [..] 1710 324421 Support for fanotify API on ARM architecture 1711 324514 gdbserver monitor cmd output behaviour consistency [..] 1712 324518 ppc64: Emulation of dcbt instructions does not handle [..] 1713 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64 1714 324582 When access is made to freed memory, report both allocation [..] 1715 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo. 1716 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc 1717 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs 1718 324834 Unhandled instructions in Microsoft C run-time for x86_64 1719 324894 Phase 3 support for IBM Power ISA 2.07 1720 326091 drd: Avoid false race reports from optimized strlen() impls 1721 326113 valgrind libvex hwcaps error on AMD64 1722 n-i-bz Some wrong command line options could be ignored 1723 n-i-bz patch to allow fair-sched on android 1724 n-i-bz report error for vgdb snapshot requested before execution 1725 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android 1726 1727 (3.9.0: 31 October 2013, vex r2796, valgrind r13708) 1728 1729 1730 1731 Release 3.8.1 (19 September 2012) 1732 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1733 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0 1734 that occur moderately frequently in real use cases, adds support for 1735 some missing instructions on ARM, and fixes a deadlock condition on 1736 MacOSX. If you package or deliver 3.8.0 for others to use, you might 1737 want to consider upgrading to 3.8.1 instead. 1738 1739 The following bugs have been fixed or resolved. Note that "n-i-bz" 1740 stands for "not in bugzilla" -- that is, a bug that was reported to us 1741 but never got a bugzilla entry. We encourage you to file bugs in 1742 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 1743 than mailing the developers (or mailing lists) directly -- bugs that 1744 are not entered into bugzilla tend to get forgotten about or ignored. 1745 1746 To see details of a given bug, visit 1747 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 1748 where XXXXXX is the bug number as listed below. 1749 1750 284004 == 301281 1751 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS) 1752 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT) 1753 298281 wcslen causes false(?) uninitialised value warnings 1754 301281 valgrind hangs on OS X when the process calls system() 1755 304035 disInstr(arm): unhandled instruction 0xE1023053 1756 304867 implement MOVBE instruction in x86 mode 1757 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping 1758 305042 amd64: implement 0F 7F encoding of movq between two registers 1759 305199 ARM: implement QDADD and QDSUB 1760 305321 amd64->IR: 0xF 0xD 0xC (prefetchw) 1761 305513 killed by fatal signal: SIGSEGV 1762 305690 DRD reporting invalid semaphore when sem_trywait fails 1763 305926 Invalid alignment checks for some AVX instructions 1764 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C 1765 306310 3.8.0 release tarball missing some files 1766 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns 1767 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF 1768 n-i-bz shmat of a segment > 4Gb does not work 1769 n-i-bz simulate_control_c script wrong USR1 signal number on mips 1770 n-i-bz vgdb ptrace calls wrong on mips [...] 1771 n-i-bz Fixes for more MPI false positives 1772 n-i-bz exp-sgcheck's memcpy causes programs to segfault 1773 n-i-bz OSX build w/ clang: asserts at startup 1774 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB 1775 n-i-bz fix a couple of union tag-vs-field mixups 1776 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll 1777 1778 The following bugs were fixed in 3.8.0 but not listed in this NEWS 1779 file at the time: 1780 1781 254088 Valgrind should know about UD2 instruction 1782 301280 == 254088 1783 301902 == 254088 1784 304754 NEWS blows TeX's little mind 1785 1786 (3.8.1: 19 September 2012, vex r2537, valgrind r12996) 1787 1788 1789 1790 Release 3.8.0 (10 August 2012) 1791 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1792 3.8.0 is a feature release with many improvements and the usual 1793 collection of bug fixes. 1794 1795 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, 1796 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android, 1797 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent 1798 distros and toolchain components (glibc 2.16, gcc 4.7) has been added. 1799 There is initial support for MacOSX 10.8, but it is not usable for 1800 serious work at present. 1801 1802 * ================== PLATFORM CHANGES ================= 1803 1804 * Support for MIPS32 platforms running Linux. Valgrind has been 1805 tested on MIPS32 and MIPS32r2 platforms running different Debian 1806 Squeeze and MeeGo distributions. Both little-endian and big-endian 1807 cores are supported. The tools Memcheck, Massif and Lackey have 1808 been tested and are known to work. See README.mips for more details. 1809 1810 * Preliminary support for Android running on x86. 1811 1812 * Preliminary (as-yet largely unusable) support for MacOSX 10.8. 1813 1814 * Support for Intel AVX instructions and for AES instructions. This 1815 support is available only for 64 bit code. 1816 1817 * Support for POWER Decimal Floating Point instructions. 1818 1819 * ==================== TOOL CHANGES ==================== 1820 1821 * Non-libc malloc implementations are now supported. This is useful 1822 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind). 1823 Using the new option --soname-synonyms, such tools can be informed 1824 that the malloc implementation is either linked statically into the 1825 executable, or is present in some other shared library different 1826 from libc.so. This makes it possible to process statically linked 1827 programs, and programs using other malloc libraries, for example 1828 TCMalloc or JEMalloc. 1829 1830 * For tools that provide their own replacement for malloc et al, the 1831 option --redzone-size=<number> allows users to specify the size of 1832 the padding blocks (redzones) added before and after each client 1833 allocated block. Smaller redzones decrease the memory needed by 1834 Valgrind. Bigger redzones increase the chance to detect blocks 1835 overrun or underrun. Prior to this change, the redzone size was 1836 hardwired to 16 bytes in Memcheck. 1837 1838 * Memcheck: 1839 1840 - The leak_check GDB server monitor command now can 1841 control the maximum nr of loss records to output. 1842 1843 - Reduction of memory use for applications allocating 1844 many blocks and/or having many partially defined bytes. 1845 1846 - Addition of GDB server monitor command 'block_list' that lists 1847 the addresses/sizes of the blocks of a leak search loss record. 1848 1849 - Addition of GDB server monitor command 'who_points_at' that lists 1850 the locations pointing at a block. 1851 1852 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will 1853 detect an invalid access of these redzones, by marking them 1854 noaccess. Similarly, if a redzone size is given for a memory 1855 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access. 1856 This still allows to find some bugs if the user has forgotten to 1857 mark the pool superblock noaccess. 1858 1859 - Performance of memory leak check has been improved, especially in 1860 cases where there are many leaked blocks and/or many suppression 1861 rules used to suppress leak reports. 1862 1863 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to 1864 more precise analysis, which is important for LLVM/Clang 1865 generated code. This is at the cost of somewhat reduced 1866 performance. Note there is no change to analysis precision or 1867 costs on Linux targets. 1868 1869 * DRD: 1870 1871 - Added even more facilities that can help finding the cause of a data 1872 race, namely the command-line option --ptrace-addr and the macro 1873 DRD_STOP_TRACING_VAR(x). More information can be found in the manual. 1874 1875 - Fixed a subtle bug that could cause false positive data race reports. 1876 1877 * ==================== OTHER CHANGES ==================== 1878 1879 * The C++ demangler has been updated so as to work well with C++ 1880 compiled by up to at least g++ 4.6. 1881 1882 * Tool developers can make replacement/wrapping more flexible thanks 1883 to the new option --soname-synonyms. This was reported above, but 1884 in fact is very general and applies to all function 1885 replacement/wrapping, not just to malloc-family functions. 1886 1887 * Round-robin scheduling of threads can be selected, using the new 1888 option --fair-sched= yes. Prior to this change, the pipe-based 1889 thread serialisation mechanism (which is still the default) could 1890 give very unfair scheduling. --fair-sched=yes improves 1891 responsiveness of interactive multithreaded applications, and 1892 improves repeatability of results from the thread checkers Helgrind 1893 and DRD. 1894 1895 * For tool developers: support to run Valgrind on Valgrind has been 1896 improved. We can now routinely Valgrind on Helgrind or Memcheck. 1897 1898 * gdbserver now shows the float shadow registers as integer 1899 rather than float values, as the shadow values are mostly 1900 used as bit patterns. 1901 1902 * Increased limit for the --num-callers command line flag to 500. 1903 1904 * Performance improvements for error matching when there are many 1905 suppression records in use. 1906 1907 * Improved support for DWARF4 debugging information (bug 284184). 1908 1909 * Initial support for DWZ compressed Dwarf debug info. 1910 1911 * Improved control over the IR optimiser's handling of the tradeoff 1912 between performance and precision of exceptions. Specifically, 1913 --vex-iropt-precise-memory-exns has been removed and replaced by 1914 --vex-iropt-register-updates, with extended functionality. This 1915 allows the Valgrind gdbserver to always show up to date register 1916 values to GDB. 1917 1918 * Modest performance gains through the use of translation chaining for 1919 JIT-generated code. 1920 1921 * ==================== FIXED BUGS ==================== 1922 1923 The following bugs have been fixed or resolved. Note that "n-i-bz" 1924 stands for "not in bugzilla" -- that is, a bug that was reported to us 1925 but never got a bugzilla entry. We encourage you to file bugs in 1926 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather 1927 than mailing the developers (or mailing lists) directly -- bugs that 1928 are not entered into bugzilla tend to get forgotten about or ignored. 1929 1930 To see details of a given bug, visit 1931 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 1932 where XXXXXX is the bug number as listed below. 1933 1934 197914 Building valgrind from svn now requires automake-1.10 1935 203877 increase to 16Mb maximum allowed alignment for memalign et al 1936 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc) 1937 247386 make perf does not run all performance tests 1938 270006 Valgrind scheduler unfair 1939 270777 Adding MIPS/Linux port to Valgrind 1940 270796 s390x: Removed broken support for the TS insn 1941 271438 Fix configure for proper SSE4.2 detection 1942 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions 1943 273475 Add support for AVX instructions 1944 274078 improved configure logic for mpicc 1945 276993 fix mremap 'no thrash checks' 1946 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag 1947 281482 memcheck incorrect byte allocation count in realloc() for silly argument 1948 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit 1949 283413 Fix wrong sanity check 1950 283671 Robustize alignment computation in LibVEX_Alloc 1951 283961 Adding support for some HCI IOCTLs 1952 284124 parse_type_DIE: confused by: DWARF 4 1953 284864 == 273475 (Add support for AVX instructions) 1954 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register" 1955 285662 (MacOSX): Memcheck needs to replace memcpy/memmove 1956 285725 == 273475 (Add support for AVX instructions) 1957 286261 add wrapper for linux I2C_RDWR ioctl 1958 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings 1959 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails 1960 286384 configure fails "checking for a supported version of gcc" 1961 286497 == 273475 (Add support for AVX instructions) 1962 286596 == 273475 (Add support for AVX instructions) 1963 286917 disInstr(arm): unhandled instruction: QADD (also QSUB) 1964 287175 ARM: scalar VFP fixed-point VCVT instructions not handled 1965 287260 Incorrect conditional jump or move depends on uninitialised value(s) 1966 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW) 1967 287307 == 273475 (Add support for AVX instructions) 1968 287858 VG_(strerror): unknown error 1969 288298 (MacOSX) unhandled syscall shm_unlink 1970 288995 == 273475 (Add support for AVX instructions) 1971 289470 Loading of large Mach-O thin binaries fails. 1972 289656 == 273475 (Add support for AVX instructions) 1973 289699 vgdb connection in relay mode erroneously closed due to buffer overrun 1974 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters) 1975 289839 s390x: Provide support for unicode conversion instructions 1976 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks 1977 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm" 1978 290655 Add support for AESKEYGENASSIST instruction 1979 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage 1980 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM 1981 291253 ES register not initialised in valgrind simulation 1982 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch) 1983 291865 s390x: Support the "Compare Double and Swap" family of instructions 1984 292300 == 273475 (Add support for AVX instructions) 1985 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid 1986 292493 == 273475 (Add support for AVX instructions) 1987 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support 1988 292627 Missing support for some SCSI ioctls 1989 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior 1990 292841 == 273475 (Add support for AVX instructions) 1991 292993 implement the getcpu syscall on amd64-linux 1992 292995 Implement the cross memory attach syscalls introduced in Linux 3.2 1993 293088 Add some VEX sanity checks for ppc64 unhandled instructions 1994 293751 == 290655 (Add support for AESKEYGENASSIST instruction) 1995 293754 PCMPxSTRx not implemented for 16-bit characters 1996 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters) 1997 293808 CLFLUSH not supported by latest VEX for amd64 1998 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...) 1999 294048 MPSADBW instruction not implemented 2000 294055 regtest none/tests/shell fails when locale is not set to C 2001 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM 2002 294190 --vgdb-error=xxx can be out of sync with errors shown to the user 2003 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions 2004 294260 disInstr_AMD64: disInstr miscalculated next %rip 2005 294523 --partial-loads-ok=yes causes false negatives 2006 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A 2007 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83 2008 294812 patch allowing to run (on x86 at least) helgrind/drd on tool. 2009 295089 can not annotate source for both helgrind and drd 2010 295221 POWER Processor decimal floating point instruction support missing 2011 295427 building for i386 with clang on darwin11 requires "-new_linker linker" 2012 295428 coregrind/m_main.c has incorrect x86 assembly for darwin 2013 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed 2014 295617 ARM - Add some missing syscalls 2015 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...] 2016 296229 Linux user input device ioctls missing wrappers 2017 296318 ELF Debug info improvements (more than one rx/rw mapping) 2018 296422 Add translation chaining support 2019 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES) 2020 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper 2021 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains 2022 297078 gdbserver signal handling problems [..] 2023 297147 drd false positives on newly allocated memory 2024 297329 disallow decoding of IBM Power DFP insns on some machines 2025 297497 POWER Processor decimal floating point instruction support missing 2026 297701 Another alias for strncasecmp_l in libc-2.13.so 2027 297911 'invalid write' not reported when using APIs for custom mem allocators. 2028 297976 s390x: revisit EX implementation 2029 297991 Valgrind interferes with mmap()+ftell() 2030 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux) 2031 297993 Fix compilation of valgrind with gcc -g3. 2032 298080 POWER Processor DFP support missing, part 3 2033 298227 == 273475 (Add support for AVX instructions) 2034 298335 == 273475 (Add support for AVX instructions) 2035 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq) 2036 298394 s390x: Don't bail out on an unknown machine model. [..] 2037 298421 accept4() syscall (366) support is missing for ARM 2038 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45 2039 298732 valgrind installation problem in ubuntu with kernel version 3.x 2040 298862 POWER Processor DFP instruction support missing, part 4 2041 298864 DWARF reader mis-parses DW_FORM_ref_addr 2042 298943 massif asserts with --pages-as-heap=yes when brk is changing [..] 2043 299053 Support DWARF4 DW_AT_high_pc constant form 2044 299104 == 273475 (Add support for AVX instructions) 2045 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed. 2046 299629 dup3() syscall (358) support is missing for ARM 2047 299694 POWER Processor DFP instruction support missing, part 5 2048 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests 2049 299803 == 273475 (Add support for AVX instructions) 2050 299804 == 273475 (Add support for AVX instructions) 2051 299805 == 273475 (Add support for AVX instructions) 2052 300140 ARM - Missing (T1) SMMUL 2053 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping)) 2054 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed. 2055 300414 FCOM and FCOMP unimplemented for amd64 guest 2056 301204 infinite loop in canonicaliseSymtab with ifunc symbol 2057 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc) 2058 301265 add x86 support to Android build 2059 301984 configure script doesn't detect certain versions of clang 2060 302205 Fix compiler warnings for POWER VEX code and POWER test cases 2061 302287 Unhandled movbe instruction on Atom processors 2062 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result 2063 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0. 2064 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess 2065 302656 == 273475 (Add support for AVX instructions) 2066 302709 valgrind for ARM needs extra tls support for android emulator [..] 2067 302827 add wrapper for CDROM_GET_CAPABILITY 2068 302901 Valgrind crashes with dwz optimized debuginfo 2069 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite 2070 303116 Add support for the POWER instruction popcntb 2071 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions. 2072 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code 2073 303466 == 273475 (Add support for AVX instructions) 2074 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP) 2075 303963 strstr() function produces wrong results under valgrind callgrind 2076 304054 CALL_FN_xx macros need to enforce stack alignment 2077 304561 tee system call not supported 2078 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#) 2079 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32 2080 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts 2081 n-i-bz Fix assert in gdbserver for watchpoints watching the same address 2082 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..] 2083 n-i-bz s390x: Shadow registers can now be examined using vgdb 2084 2085 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865) 2086 (3.8.0: 10 August 2012, vex r2465, valgrind r12866) 2087 2088 2089 2090 Release 3.7.0 (5 November 2011) 2091 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2092 3.7.0 is a feature release with many significant improvements and the 2093 usual collection of bug fixes. 2094 2095 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, 2096 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin. 2097 Support for recent distros and toolchain components (glibc 2.14, gcc 2098 4.6, MacOSX 10.7) has been added. 2099 2100 * ================== PLATFORM CHANGES ================= 2101 2102 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can 2103 analyse 64-bit programs running on z/Architecture. Most user space 2104 instructions up to and including z10 are supported. Valgrind has 2105 been tested extensively on z9, z10, and z196 machines running SLES 2106 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are 2107 known to work well. Callgrind, Helgrind, and DRD work reasonably 2108 well on z9 and later models. See README.s390 for more details. 2109 2110 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and 2111 64-bit processes are supported. Some complex threaded applications 2112 (Firefox) are observed to hang when run as 32 bit applications, 2113 whereas 64-bit versions run OK. The cause is unknown. Memcheck 2114 will likely report some false errors. In general, expect some rough 2115 spots. This release also supports MacOSX 10.6, but drops support 2116 for 10.5. 2117 2118 * Preliminary support for Android (on ARM). Valgrind can now run 2119 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See 2120 README.android for more details, plus instructions on how to get 2121 started. 2122 2123 * Support for the IBM Power ISA 2.06 (Power7 instructions) 2124 2125 * General correctness and performance improvements for ARM/Linux, and, 2126 by extension, ARM/Android. 2127 2128 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX 2129 instruction set support is under development but is not available in 2130 this release. 2131 2132 * Support for AIX5 has been removed. 2133 2134 * ==================== TOOL CHANGES ==================== 2135 2136 * Memcheck: some incremental changes: 2137 2138 - reduction of memory use in some circumstances 2139 2140 - improved handling of freed memory, which in some circumstances 2141 can cause detection of use-after-free that would previously have 2142 been missed 2143 2144 - fix of a longstanding bug that could cause false negatives (missed 2145 errors) in programs doing vector saturated narrowing instructions. 2146 2147 * Helgrind: performance improvements and major memory use reductions, 2148 particularly for large, long running applications which perform many 2149 synchronisation (lock, unlock, etc) events. Plus many smaller 2150 changes: 2151 2152 - display of locksets for both threads involved in a race 2153 2154 - general improvements in formatting/clarity of error messages 2155 2156 - addition of facilities and documentation regarding annotation 2157 of thread safe reference counted C++ classes 2158 2159 - new flag --check-stack-refs=no|yes [yes], to disable race checking 2160 on thread stacks (a performance hack) 2161 2162 - new flag --free-is-write=no|yes [no], to enable detection of races 2163 where one thread accesses heap memory but another one frees it, 2164 without any coordinating synchronisation event 2165 2166 * DRD: enabled XML output; added support for delayed thread deletion 2167 in order to detect races that occur close to the end of a thread 2168 (--join-list-vol); fixed a memory leak triggered by repeated client 2169 memory allocatation and deallocation; improved Darwin support. 2170 2171 * exp-ptrcheck: this tool has been renamed to exp-sgcheck 2172 2173 * exp-sgcheck: this tool has been reduced in scope so as to improve 2174 performance and remove checking that Memcheck does better. 2175 Specifically, the ability to check for overruns for stack and global 2176 arrays is unchanged, but the ability to check for overruns of heap 2177 blocks has been removed. The tool has accordingly been renamed to 2178 exp-sgcheck ("Stack and Global Array Checking"). 2179 2180 * ==================== OTHER CHANGES ==================== 2181 2182 * GDB server: Valgrind now has an embedded GDB server. That means it 2183 is possible to control a Valgrind run from GDB, doing all the usual 2184 things that GDB can do (single stepping, breakpoints, examining 2185 data, etc). Tool-specific functionality is also available. For 2186 example, it is possible to query the definedness state of variables 2187 or memory from within GDB when running Memcheck; arbitrarily large 2188 memory watchpoints are supported, etc. To use the GDB server, start 2189 Valgrind with the flag --vgdb-error=0 and follow the on-screen 2190 instructions. 2191 2192 * Improved support for unfriendly self-modifying code: a new option 2193 --smc-check=all-non-file is available. This adds the relevant 2194 consistency checks only to code that originates in non-file-backed 2195 mappings. In effect this confines the consistency checking only to 2196 code that is or might be JIT generated, and avoids checks on code 2197 that must have been compiled ahead of time. This significantly 2198 improves performance on applications that generate code at run time. 2199 2200 * It is now possible to build a working Valgrind using Clang-2.9 on 2201 Linux. 2202 2203 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING. 2204 These enable and disable error reporting on a per-thread, and 2205 nestable, basis. This is useful for hiding errors in particularly 2206 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c) 2207 now uses this facility. 2208 2209 * Added the --mod-funcname option to cg_diff. 2210 2211 * ==================== FIXED BUGS ==================== 2212 2213 The following bugs have been fixed or resolved. Note that "n-i-bz" 2214 stands for "not in bugzilla" -- that is, a bug that was reported to us 2215 but never got a bugzilla entry. We encourage you to file bugs in 2216 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than 2217 mailing the developers (or mailing lists) directly -- bugs that are 2218 not entered into bugzilla tend to get forgotten about or ignored. 2219 2220 To see details of a given bug, visit 2221 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 2222 where XXXXXX is the bug number as listed below. 2223 2224 79311 malloc silly arg warning does not give stack trace 2225 210935 port valgrind.h (not valgrind) to win32 to support client requests 2226 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10 2227 243404 Port to zSeries 2228 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER() 2229 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored 2230 250101 huge "free" memory usage due to m_mallocfree.c fragmentation 2231 253206 Some fixes for the faultstatus testcase 2232 255223 capget testcase fails when running as root 2233 256703 xlc_dbl_u32.c testcase broken 2234 256726 Helgrind tests have broken inline asm 2235 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp) 2236 264800 testcase compile failure on zseries 2237 265762 make public VEX headers compilable by G++ 3.x 2238 265771 assertion in jumps.c (r11523) fails with glibc-2.3 2239 266753 configure script does not give the user the option to not use QtCore 2240 266931 gen_insn_test.pl is broken 2241 266961 ld-linux.so.2 i?86-linux strlen issues 2242 266990 setns instruction causes false positive 2243 267020 Make directory for temporary files configurable at run-time. 2244 267342 == 267997 (segmentation fault on Mac OS 10.6) 2245 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed 2246 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed. 2247 267488 regtest: darwin support for 64-bit build 2248 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools 2249 267630 Add support for IBM Power ISA 2.06 -- stage 1 2250 267769 == 267997 (Darwin: memcheck triggers segmentation fault) 2251 267819 Add client request for informing the core about reallocation 2252 267925 laog data structure quadratic for a single sequence of lock 2253 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed 2254 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1 2255 268513 missed optimizations in fold_Expr 2256 268619 s390x: fpr - gpr transfer facility 2257 268620 s390x: reconsider "long displacement" requirement 2258 268621 s390x: improve IR generation for XC 2259 268715 s390x: FLOGR is not universally available 2260 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4) 2261 268930 s390x: MHY is not universally available 2262 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register) 2263 269079 Support ptrace system call on ARM 2264 269144 missing "Bad option" error message 2265 269209 conditional load and store facility (z196) 2266 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP 2267 269641 == 267997 (valgrind segfaults immediately (segmentation fault)) 2268 269736 s390x: minor code generation tweaks 2269 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..) 2270 269863 s390x: remove unused function parameters 2271 269864 s390x: tweak s390_emit_load_cc 2272 269884 == 250101 (overhead for huge blocks exhausts space too soon) 2273 270082 s390x: Make sure to point the PSW address to the next address on SIGILL 2274 270115 s390x: rewrite some testcases 2275 270309 == 267997 (valgrind crash on startup) 2276 270320 add support for Linux FIOQSIZE ioctl() call 2277 270326 segfault while trying to sanitize the environment passed to execle 2278 270794 IBM POWER7 support patch causes regression in none/tests 2279 270851 IBM POWER7 fcfidus instruction causes memcheck to fail 2280 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app 2281 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix 2282 270959 s390x: invalid use of R0 as base register 2283 271042 VSX configure check fails when it should not 2284 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21 2285 271259 s390x: fix code confusion 2286 271337 == 267997 (Valgrind segfaults on MacOS X) 2287 271385 s390x: Implement Ist_MBE 2288 271501 s390x: misc cleanups 2289 271504 s390x: promote likely and unlikely 2290 271579 ppc: using wrong enum type 2291 271615 unhandled instruction "popcnt" (arch=amd10h) 2292 271730 Fix bug when checking ioctls: duplicate check 2293 271776 s390x: provide STFLE instruction support 2294 271779 s390x: provide clock instructions like STCK 2295 271799 Darwin: ioctls without an arg report a memory error 2296 271820 arm: fix type confusion 2297 271917 pthread_cond_timedwait failure leads to not-locked false positive 2298 272067 s390x: fix DISP20 macro 2299 272615 A typo in debug output in mc_leakcheck.c 2300 272661 callgrind_annotate chokes when run from paths containing regex chars 2301 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup) 2302 272955 Unhandled syscall error for pwrite64 on ppc64 arch 2303 272967 make documentation build-system more robust 2304 272986 Fix gcc-4.6 warnings with valgrind.h 2305 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case) 2306 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 2307 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039) 2308 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...' 2309 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer' 2310 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169) 2311 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction) 2312 273778 exp-ptrcheck: unhandled sysno == 259 2313 274089 exp-ptrcheck: unhandled sysno == 208 2314 274378 s390x: Various dispatcher tweaks 2315 274447 WARNING: unhandled syscall: 340 2316 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66 2317 274784 == 267997 (valgrind ls -l results in Segmentation Fault) 2318 274926 valgrind does not build against linux-3 2319 275148 configure FAIL with glibc-2.14 2320 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL 2321 275168 Make Valgrind work for MacOSX 10.7 Lion 2322 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al) 2323 275278 valgrind does not build on Linux kernel 3.0.* due to silly 2324 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64 2325 275308 Fix implementation for ppc64 fres instruc 2326 275339 s390x: fix testcase compile warnings 2327 275517 s390x: Provide support for CKSM instruction 2328 275710 s390x: get rid of redundant address mode calculation 2329 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall) 2330 275852 == 250101 (valgrind uses all swap space and is killed) 2331 276784 Add support for IBM Power ISA 2.06 -- stage 3 2332 276987 gdbsrv: fix tests following recent commits 2333 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a 2334 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink 2335 277471 Unhandled syscall: 340 2336 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...) 2337 277653 ARM: support Thumb2 PLD instruction 2338 277663 ARM: NEON float VMUL by scalar incorrect 2339 277689 ARM: tests for VSTn with register post-index are broken 2340 277694 ARM: BLX LR instruction broken in ARM mode 2341 277780 ARM: VMOV.F32 (immediate) instruction is broken 2342 278057 fuse filesystem syscall deadlocks 2343 278078 Unimplemented syscall 280 on ppc32 2344 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands 2345 278454 VALGRIND_STACK_DEREGISTER has wrong output type 2346 278502 == 275284 (Valgrind confuses memcpy() and memmove()) 2347 278892 gdbsrv: factorize gdb version handling, fix doc and typos 2348 279027 Support for MVCL and CLCL instruction 2349 279027 s390x: Provide support for CLCL and MVCL instructions 2350 279062 Remove a redundant check in the insn selector for ppc. 2351 279071 JDK creates PTEST with redundant REX.W prefix 2352 279212 gdbsrv: add monitor cmd v.info scheduler. 2353 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call 2354 279698 memcheck discards valid-bits for packuswb 2355 279795 memcheck reports uninitialised values for mincore on amd64 2356 279994 Add support for IBM Power ISA 2.06 -- stage 3 2357 280083 mempolicy syscall check errors 2358 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F 2359 280710 s390x: config files for nightly builds 2360 280757 /tmp dir still used by valgrind even if TMPDIR is specified 2361 280965 Valgrind breaks fcntl locks when program does mmap 2362 281138 WARNING: unhandled syscall: 340 2363 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion) 2364 281304 == 275168 (Darwin: dyld "cannot load inserted library") 2365 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1) 2366 281468 s390x: handle do_clone and gcc clones in call traces 2367 281488 ARM: VFP register corruption 2368 281828 == 275284 (false memmove warning: "Source and destination overlap") 2369 281883 s390x: Fix system call wrapper for "clone". 2370 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock 2371 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp) 2372 282238 SLES10: make check fails 2373 282979 strcasestr needs replacement with recent(>=2.12) glibc 2374 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF 2375 283243 Regression in ppc64 memcheck tests 2376 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0) 2377 283427 re-connect epoll_pwait syscall on ARM linux 2378 283600 gdbsrv: android: port vgdb.c 2379 283709 none/tests/faultstatus needs to account for page size 2380 284305 filter_gdb needs enhancement to work on ppc64 2381 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h 2382 284472 Thumb2 ROR.W encoding T2 not implemented 2383 284621 XML-escape process command line in XML output 2384 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs 2385 that have non-power-of-2 sizes (also AMDs) 2386 n-i-bz don't be spooked by libraries mashed by elfhack 2387 n-i-bz don't be spooked by libxul.so linked with gold 2388 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED 2389 2390 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245) 2391 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257) 2392 (3.7.0: 5 November 2011, vex r2231, valgrind r12258) 2393 2394 2395 2396 Release 3.6.1 (16 February 2011) 2397 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2398 3.6.1 is a bug fix release. It adds support for some SSE4 2399 instructions that were omitted in 3.6.0 due to lack of time. Initial 2400 support for glibc-2.13 has been added. A number of bugs causing 2401 crashing or assertion failures have been fixed. 2402 2403 The following bugs have been fixed or resolved. Note that "n-i-bz" 2404 stands for "not in bugzilla" -- that is, a bug that was reported to us 2405 but never got a bugzilla entry. We encourage you to file bugs in 2406 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than 2407 mailing the developers (or mailing lists) directly -- bugs that are 2408 not entered into bugzilla tend to get forgotten about or ignored. 2409 2410 To see details of a given bug, visit 2411 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 2412 where XXXXXX is the bug number as listed below. 2413 2414 188572 Valgrind on Mac should suppress setenv() mem leak 2415 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support) 2416 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ) 2417 246152 callgrind internal error after pthread_cancel on 32 Bit Linux 2418 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest 2419 254420 memory pool tracking broken 2420 254957 Test code failing to compile due to changes in memcheck.h 2421 255009 helgrind/drd: crash on chmod with invalid parameter 2422 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types 2423 255355 helgrind/drd: crash on threaded programs doing fork 2424 255358 == 255355 2425 255418 (SSE4.x) rint call compiled with ICC 2426 255822 --gen-suppressions can create invalid files: "too many callers [...]" 2427 255888 closing valgrindoutput tag outputted to log-stream on error 2428 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD) 2429 255966 Slowness when using mempool annotations 2430 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM) 2431 256600 super-optimized strcasecmp() false positive 2432 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64 2433 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx) 2434 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW) 2435 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS) 2436 257276 Missing case in memcheck --track-origins=yes 2437 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction 2438 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L}) 2439 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX 2440 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0) 2441 263099 callgrind_annotate counts Ir improperly [...] 2442 263877 undefined coprocessor instruction on ARMv7 2443 265964 configure FAIL with glibc-2.13 2444 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c 2445 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al 2446 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes 2447 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler 2448 n-i-bz DRD: disable --free-is-write due to implementation difficulties 2449 2450 (3.6.1: 16 February 2011, vex r2103, valgrind r11561). 2451 2452 2453 2454 Release 3.6.0 (21 October 2010) 2455 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2456 3.6.0 is a feature release with many significant improvements and the 2457 usual collection of bug fixes. 2458 2459 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, 2460 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros 2461 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added. 2462 2463 ------------------------- 2464 2465 Here are some highlights. Details are shown further down: 2466 2467 * Support for ARM/Linux. 2468 2469 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14. 2470 2471 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 2472 2473 * Support for the SSE4.2 instruction set. 2474 2475 * Enhancements to the Callgrind profiler, including the ability to 2476 handle CPUs with three levels of cache. 2477 2478 * A new experimental heap profiler, DHAT. 2479 2480 * A huge number of bug fixes and small enhancements. 2481 2482 ------------------------- 2483 2484 Here are details of the above changes, together with descriptions of 2485 many other changes, and a list of fixed bugs. 2486 2487 * ================== PLATFORM CHANGES ================= 2488 2489 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs 2490 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10, 2491 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want. 2492 2493 This requires a CPU capable of running the ARMv7-A instruction set 2494 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage 2495 of the user space instruction set, including ARM and Thumb integer 2496 code, VFPv3, NEON and V6 media instructions. The Memcheck, 2497 Cachegrind and Massif tools work properly; other tools work to 2498 varying degrees. 2499 2500 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along 2501 with support for recent releases of the underlying toolchain 2502 components, notably gcc-4.5 and glibc-2.12. 2503 2504 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit 2505 support also works much better on OS X 10.5, and is as solid as 2506 32-bit support now. 2507 2508 * Support for the SSE4.2 instruction set. SSE4.2 is supported in 2509 64-bit mode. In 32-bit mode, support is only available up to and 2510 including SSSE3. Some exceptions: SSE4.2 AES instructions are not 2511 supported in 64-bit mode, and 32-bit mode does in fact support the 2512 bare minimum SSE4 instructions to needed to run programs on Mac OS X 2513 10.6 on 32-bit targets. 2514 2515 * Support for IBM POWER6 cpus has been improved. The Power ISA up to 2516 and including version 2.05 is supported. 2517 2518 * ==================== TOOL CHANGES ==================== 2519 2520 * Cachegrind has a new processing script, cg_diff, which finds the 2521 difference between two profiles. It's very useful for evaluating 2522 the performance effects of a change in a program. 2523 2524 Related to this change, the meaning of cg_annotate's (rarely-used) 2525 --threshold option has changed; this is unlikely to affect many 2526 people, if you do use it please see the user manual for details. 2527 2528 * Callgrind now can do branch prediction simulation, similar to 2529 Cachegrind. In addition, it optionally can count the number of 2530 executed global bus events. Both can be used for a better 2531 approximation of a "Cycle Estimation" as derived event (you need to 2532 update the event formula in KCachegrind yourself). 2533 2534 * Cachegrind and Callgrind now refer to the LL (last-level) cache 2535 rather than the L2 cache. This is to accommodate machines with 2536 three levels of caches -- if Cachegrind/Callgrind auto-detects the 2537 cache configuration of such a machine it will run the simulation as 2538 if the L2 cache isn't present. This means the results are less 2539 likely to match the true result for the machine, but 2540 Cachegrind/Callgrind's results are already only approximate, and 2541 should not be considered authoritative. The results are still 2542 useful for giving a general idea about a program's locality. 2543 2544 * Massif has a new option, --pages-as-heap, which is disabled by 2545 default. When enabled, instead of tracking allocations at the level 2546 of heap blocks (as allocated with malloc/new/new[]), it instead 2547 tracks memory allocations at the level of memory pages (as mapped by 2548 mmap, brk, etc). Each mapped page is treated as its own block. 2549 Interpreting the page-level output is harder than the heap-level 2550 output, but this option is useful if you want to account for every 2551 byte of memory used by a program. 2552 2553 * DRD has two new command-line options: --free-is-write and 2554 --trace-alloc. The former allows to detect reading from already freed 2555 memory, and the latter allows tracing of all memory allocations and 2556 deallocations. 2557 2558 * DRD has several new annotations. Custom barrier implementations can 2559 now be annotated, as well as benign races on static variables. 2560 2561 * DRD's happens before / happens after annotations have been made more 2562 powerful, so that they can now also be used to annotate e.g. a smart 2563 pointer implementation. 2564 2565 * Helgrind's annotation set has also been drastically improved, so as 2566 to provide to users a general set of annotations to describe locks, 2567 semaphores, barriers and condition variables. Annotations to 2568 describe thread-safe reference counted heap objects have also been 2569 added. 2570 2571 * Memcheck has a new command-line option, --show-possibly-lost, which 2572 is enabled by default. When disabled, the leak detector will not 2573 show possibly-lost blocks. 2574 2575 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool), 2576 has been added. DHAT keeps track of allocated heap blocks, and also 2577 inspects every memory reference to see which block (if any) is being 2578 accessed. This gives a lot of insight into block lifetimes, 2579 utilisation, turnover, liveness, and the location of hot and cold 2580 fields. You can use DHAT to do hot-field profiling. 2581 2582 * ==================== OTHER CHANGES ==================== 2583 2584 * Improved support for unfriendly self-modifying code: the extra 2585 overhead incurred by --smc-check=all has been reduced by 2586 approximately a factor of 5 as compared with 3.5.0. 2587 2588 * Ability to show directory names for source files in error messages. 2589 This is combined with a flexible mechanism for specifying which 2590 parts of the paths should be shown. This is enabled by the new flag 2591 --fullpath-after. 2592 2593 * A new flag, --require-text-symbol, which will stop the run if a 2594 specified symbol is not found it a given shared object when it is 2595 loaded into the process. This makes advanced working with function 2596 intercepting and wrapping safer and more reliable. 2597 2598 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output 2599 and control of Valgrind is now available for the tools Memcheck and 2600 Helgrind. XML output from Valgrind is available for Memcheck, 2601 Helgrind and exp-Ptrcheck. 2602 2603 * More reliable stack unwinding on amd64-linux, particularly in the 2604 presence of function wrappers, and with gcc-4.5 compiled code. 2605 2606 * Modest scalability (performance improvements) for massive 2607 long-running applications, particularly for those with huge amounts 2608 of code. 2609 2610 * Support for analyzing programs running under Wine with has been 2611 improved. The header files <valgrind/valgrind.h>, 2612 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in 2613 Windows-programs compiled with MinGW or one of the Microsoft Visual 2614 Studio compilers. 2615 2616 * A rare but serious error in the 64-bit x86 CPU simulation was fixed. 2617 The 32-bit simulator was not affected. This did not occur often, 2618 but when it did would usually crash the program under test. 2619 Bug 245925. 2620 2621 * A large number of bugs were fixed. These are shown below. 2622 2623 * A number of bugs were investigated, and were candidates for fixing, 2624 but are not fixed in 3.6.0, due to lack of developer time. They may 2625 get fixed in later releases. They are: 2626 2627 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64) 2628 212419 false positive "lock order violated" (A+B vs A) 2629 213685 Undefined value propagates past dependency breaking instruction 2630 216837 Incorrect instrumentation of NSOperationQueue on Darwin 2631 237920 valgrind segfault on fork failure 2632 242137 support for code compiled by LLVM-2.8 2633 242423 Another unknown Intel cache config value 2634 243232 Inconsistent Lock Orderings report with trylock 2635 243483 ppc: callgrind triggers VEX assertion failure 2636 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong 2637 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion 2638 'thr' failed. 2639 246152 callgrind internal error after pthread_cancel on 32 Bit Linux 2640 249435 Analyzing wine programs with callgrind triggers a crash 2641 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest 2642 250065 Handling large allocations 2643 250101 huge "free" memory usage due to m_mallocfree.c 2644 "superblocks fragmentation" 2645 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP) 2646 252091 Callgrind on ARM does not detect function returns correctly 2647 252600 [PATCH] Allow lhs to be a pointer for shl/shr 2648 254420 memory pool tracking broken 2649 n-i-bz support for adding symbols for JIT generated code 2650 2651 2652 The following bugs have been fixed or resolved. Note that "n-i-bz" 2653 stands for "not in bugzilla" -- that is, a bug that was reported to us 2654 but never got a bugzilla entry. We encourage you to file bugs in 2655 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than 2656 mailing the developers (or mailing lists) directly -- bugs that are 2657 not entered into bugzilla tend to get forgotten about or ignored. 2658 2659 To see details of a given bug, visit 2660 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 2661 where XXXXXX is the bug number as listed below. 2662 2663 135264 dcbzl instruction missing 2664 142688 == 250799 2665 153699 Valgrind should report unaligned reads with movdqa 2666 180217 == 212335 2667 190429 Valgrind reports lost of errors in ld.so 2668 with x86_64 2.9.90 glibc 2669 197266 valgrind appears to choke on the xmms instruction 2670 "roundsd" on x86_64 2671 197988 Crash when demangling very large symbol names 2672 202315 unhandled syscall: 332 (inotify_init1) 2673 203256 Add page-level profiling to Massif 2674 205093 dsymutil=yes needs quotes, locking (partial fix) 2675 205241 Snow Leopard 10.6 support (partial fix) 2676 206600 Leak checker fails to upgrade indirect blocks when their 2677 parent becomes reachable 2678 210935 port valgrind.h (not valgrind) to win32 so apps run under 2679 wine can make client requests 2680 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89 2681 within Linux ip-stack checksum functions 2682 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0 2683 (lzcnt %eax,%eax) 2684 213685 Undefined value propagates past dependency breaking instruction 2685 (partial fix) 2686 215914 Valgrind inserts bogus empty environment variable 2687 217863 == 197988 2688 219538 adjtimex syscall wrapper wrong in readonly adjtime mode 2689 222545 shmat fails under valgind on some arm targets 2690 222560 ARM NEON support 2691 230407 == 202315 2692 231076 == 202315 2693 232509 Docs build fails with formatting inside <title></title> elements 2694 232793 == 202315 2695 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls 2696 236546 vex x86->IR: 0x66 0xF 0x3A 0xA 2697 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B 2698 237371 better support for VALGRIND_MALLOCLIKE_BLOCK 2699 237485 symlink (syscall 57) is not supported on Mac OS 2700 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened: 2701 unhandled syscall 2702 238208 is_just_below_ESP doesn't take into account red-zone 2703 238345 valgrind passes wrong $0 when executing a shell script 2704 238679 mq_timedreceive syscall doesn't flag the reception buffer 2705 as "defined" 2706 238696 fcntl command F_DUPFD_CLOEXEC not supported 2707 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6 2708 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6 2709 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure 2710 says "Altivec off" 2711 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48 2712 240488 == 197988 2713 240639 == 212335 2714 241377 == 236546 2715 241903 == 202315 2716 241920 == 212335 2717 242606 unhandled syscall: setegid (in Ptrcheck) 2718 242814 Helgrind "Impossible has happened" during 2719 QApplication::initInstance(); 2720 243064 Valgrind attempting to read debug information from iso 2721 243270 Make stack unwinding in Valgrind wrappers more reliable 2722 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall 2723 sysno = 277 (mq_open) 2724 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd 2725 244493 ARM VFP d16-d31 registers support 2726 244670 add support for audit_session_self syscall on Mac OS 10.6 2727 244921 The xml report of helgrind tool is not well format 2728 244923 In the xml report file, the <preamble> not escape the 2729 xml char, eg '<','&','>' 2730 245535 print full path names in plain text reports 2731 245925 x86-64 red zone handling problem 2732 246258 Valgrind not catching integer underruns + new [] s 2733 246311 reg/reg cmpxchg doesn't work on amd64 2734 246549 unhandled syscall unix:277 while testing 32-bit Darwin app 2735 246888 Improve Makefile.vex.am 2736 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed 2737 to [f]chmod_extended 2738 247526 IBM POWER6 (ISA 2.05) support is incomplete 2739 247561 Some leak testcases fails due to reachable addresses in 2740 caller save regs 2741 247875 sizeofIRType to handle Ity_I128 2742 247894 [PATCH] unhandled syscall sys_readahead 2743 247980 Doesn't honor CFLAGS passed to configure 2744 248373 darwin10.supp is empty in the trunk 2745 248822 Linux FIBMAP ioctl has int parameter instead of long 2746 248893 [PATCH] make readdwarf.c big endianess safe to enable 2747 unwinding on big endian systems 2748 249224 Syscall 336 not supported (SYS_proc_info) 2749 249359 == 245535 2750 249775 Incorrect scheme for detecting NEON capabilities of host CPU 2751 249943 jni JVM init fails when using valgrind 2752 249991 Valgrind incorrectly declares AESKEYGENASSIST support 2753 since VEX r2011 2754 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64) 2755 250799 frexp$fenv_access_off function generates SIGILL 2756 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E 2757 251251 support pclmulqdq insn 2758 251362 valgrind: ARM: attach to debugger either fails or provokes 2759 kernel oops 2760 251674 Unhandled syscall 294 2761 251818 == 254550 2762 2763 254257 Add support for debugfiles found by build-id 2764 254550 [PATCH] Implement DW_ATE_UTF (DWARF4) 2765 254646 Wrapped functions cause stack misalignment on OS X 2766 (and possibly Linux) 2767 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0 2768 2769 (3.6.0: 21 October 2010, vex r2068, valgrind r11471). 2770 2771 2772 2773 Release 3.5.0 (19 August 2009) 2774 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2775 3.5.0 is a feature release with many significant improvements and the 2776 usual collection of bug fixes. The main improvement is that Valgrind 2777 now works on Mac OS X. 2778 2779 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux 2780 and X86/Darwin. Support for recent distros and toolchain components 2781 (glibc 2.10, gcc 4.5) has been added. 2782 2783 ------------------------- 2784 2785 Here is a short summary of the changes. Details are shown further 2786 down: 2787 2788 * Support for Mac OS X (10.5.x). 2789 2790 * Improvements and simplifications to Memcheck's leak checker. 2791 2792 * Clarification and simplifications in various aspects of Valgrind's 2793 text output. 2794 2795 * XML output for Helgrind and Ptrcheck. 2796 2797 * Performance and stability improvements for Helgrind and DRD. 2798 2799 * Genuinely atomic support for x86/amd64/ppc atomic instructions. 2800 2801 * A new experimental tool, BBV, useful for computer architecture 2802 research. 2803 2804 * Improved Wine support, including ability to read Windows PDB 2805 debuginfo. 2806 2807 ------------------------- 2808 2809 Here are details of the above changes, followed by descriptions of 2810 many other minor changes, and a list of fixed bugs. 2811 2812 2813 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes 2814 called "Darwin" because that is the name of the OS core, which is the 2815 level that Valgrind works at.) 2816 2817 Supported systems: 2818 2819 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned 2820 because it would require work and 10.4 is only becoming less common. 2821 2822 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported 2823 fairly well. For 10.5.x, 32-bit programs are the default even on 2824 64-bit machines, so it handles most current programs. 2825 2826 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not 2827 officially supported, but simple programs at least will probably work. 2828 However, start-up is slow. 2829 2830 - PowerPC machines are not supported. 2831 2832 Things that don't work: 2833 2834 - The Ptrcheck tool. 2835 2836 - Objective-C garbage collection. 2837 2838 - --db-attach=yes. 2839 2840 - If you have Rogue Amoeba's "Instant Hijack" program installed, 2841 Valgrind will fail with a SIGTRAP at start-up. See 2842 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a 2843 simple work-around. 2844 2845 Usage notes: 2846 2847 - You will likely find --dsymutil=yes a useful option, as error 2848 messages may be imprecise without it. 2849 2850 - Mac OS X support is new and therefore will be less robust than the 2851 Linux support. Please report any bugs you find. 2852 2853 - Threaded programs may run more slowly than on Linux. 2854 2855 Many thanks to Greg Parker for developing this port over several years. 2856 2857 2858 * Memcheck's leak checker has been improved. 2859 2860 - The results for --leak-check=summary now match the summary results 2861 for --leak-check=full. Previously they could differ because 2862 --leak-check=summary counted "indirectly lost" blocks and 2863 "suppressed" blocks as "definitely lost". 2864 2865 - Blocks that are only reachable via at least one interior-pointer, 2866 but are directly pointed to by a start-pointer, were previously 2867 marked as "still reachable". They are now correctly marked as 2868 "possibly lost". 2869 2870 - The default value for the --leak-resolution option has been 2871 changed from "low" to "high". In general, this means that more 2872 leak reports will be produced, but each leak report will describe 2873 fewer leaked blocks. 2874 2875 - With --leak-check=full, "definitely lost" and "possibly lost" 2876 leaks are now considered as proper errors, ie. they are counted 2877 for the "ERROR SUMMARY" and affect the behaviour of 2878 --error-exitcode. These leaks are not counted as errors if 2879 --leak-check=summary is specified, however. 2880 2881 - Documentation for the leak checker has been improved. 2882 2883 2884 * Various aspects of Valgrind's text output have changed. 2885 2886 - Valgrind's start-up message has changed. It is shorter but also 2887 includes the command being run, which makes it easier to use 2888 --trace-children=yes. An example: 2889 2890 - Valgrind's shut-down messages have also changed. This is most 2891 noticeable with Memcheck, where the leak summary now occurs before 2892 the error summary. This change was necessary to allow leaks to be 2893 counted as proper errors (see the description of the leak checker 2894 changes above for more details). This was also necessary to fix a 2895 longstanding bug in which uses of suppressions against leaks were 2896 not "counted", leading to difficulties in maintaining suppression 2897 files (see https://bugs.kde.org/show_bug.cgi?id=186790). 2898 2899 - Behavior of -v has changed. In previous versions, -v printed out 2900 a mixture of marginally-user-useful information, and tool/core 2901 statistics. The statistics printing has now been moved to its own 2902 flag, --stats=yes. This means -v is less verbose and more likely 2903 to convey useful end-user information. 2904 2905 - The format of some (non-XML) stack trace entries has changed a 2906 little. Previously there were six possible forms: 2907 2908 0x80483BF: really (a.c:20) 2909 0x80483BF: really (in /foo/a.out) 2910 0x80483BF: really 2911 0x80483BF: (within /foo/a.out) 2912 0x80483BF: ??? (a.c:20) 2913 0x80483BF: ??? 2914 2915 The third and fourth of these forms have been made more consistent 2916 with the others. The six possible forms are now: 2917 2918 0x80483BF: really (a.c:20) 2919 0x80483BF: really (in /foo/a.out) 2920 0x80483BF: really (in ???) 2921 0x80483BF: ??? (in /foo/a.out) 2922 0x80483BF: ??? (a.c:20) 2923 0x80483BF: ??? 2924 2925 Stack traces produced when --xml=yes is specified are different 2926 and unchanged. 2927 2928 2929 * Helgrind and Ptrcheck now support XML output, so they can be used 2930 from GUI tools. Also, the XML output mechanism has been 2931 overhauled. 2932 2933 - The XML format has been overhauled and generalised, so it is more 2934 suitable for error reporting tools in general. The Memcheck 2935 specific aspects of it have been removed. The new format, which 2936 is an evolution of the old format, is described in 2937 docs/internals/xml-output-protocol4.txt. 2938 2939 - Memcheck has been updated to use the new format. 2940 2941 - Helgrind and Ptrcheck are now able to emit output in this format. 2942 2943 - The XML output mechanism has been overhauled. XML is now output 2944 to its own file descriptor, which means that: 2945 2946 * Valgrind can output text and XML independently. 2947 2948 * The longstanding problem of XML output being corrupted by 2949 unexpected un-tagged text messages is solved. 2950 2951 As before, the destination for text output is specified using 2952 --log-file=, --log-fd= or --log-socket=. 2953 2954 As before, XML output for a tool is enabled using --xml=yes. 2955 2956 Because there's a new XML output channel, the XML output 2957 destination is now specified by --xml-file=, --xml-fd= or 2958 --xml-socket=. 2959 2960 Initial feedback has shown this causes some confusion. To 2961 clarify, the two envisaged usage scenarios are: 2962 2963 (1) Normal text output. In this case, do not specify --xml=yes 2964 nor any of --xml-file=, --xml-fd= or --xml-socket=. 2965 2966 (2) XML output. In this case, specify --xml=yes, and one of 2967 --xml-file=, --xml-fd= or --xml-socket= to select the XML 2968 destination, one of --log-file=, --log-fd= or --log-socket= 2969 to select the destination for any remaining text messages, 2970 and, importantly, -q. 2971 2972 -q makes Valgrind completely silent on the text channel, 2973 except in the case of critical failures, such as Valgrind 2974 itself segfaulting, or failing to read debugging information. 2975 Hence, in this scenario, it suffices to check whether or not 2976 any output appeared on the text channel. If yes, then it is 2977 likely to be a critical error which should be brought to the 2978 attention of the user. If no (the text channel produced no 2979 output) then it can be assumed that the run was successful. 2980 2981 This allows GUIs to make the critical distinction they need to 2982 make (did the run fail or not?) without having to search or 2983 filter the text output channel in any way. 2984 2985 It is also recommended to use --child-silent-after-fork=yes in 2986 scenario (2). 2987 2988 2989 * Improvements and changes in Helgrind: 2990 2991 - XML output, as described above 2992 2993 - Checks for consistent association between pthread condition 2994 variables and their associated mutexes are now performed. 2995 2996 - pthread_spinlock functions are supported. 2997 2998 - Modest performance improvements. 2999 3000 - Initial (skeletal) support for describing the behaviour of 3001 non-POSIX synchronisation objects through ThreadSanitizer 3002 compatible ANNOTATE_* macros. 3003 3004 - More controllable tradeoffs between performance and the level of 3005 detail of "previous" accesses in a race. There are now three 3006 settings: 3007 3008 * --history-level=full. This is the default, and was also the 3009 default in 3.4.x. It shows both stacks involved in a race, but 3010 requires a lot of memory and can be very slow in programs that 3011 do many inter-thread synchronisation events. 3012 3013 * --history-level=none. This only shows the later stack involved 3014 in a race. This can be much faster than --history-level=full, 3015 but makes it much more difficult to find the other access 3016 involved in the race. 3017 3018 The new intermediate setting is 3019 3020 * --history-level=approx 3021 3022 For the earlier (other) access, two stacks are presented. The 3023 earlier access is guaranteed to be somewhere in between the two 3024 program points denoted by those stacks. This is not as useful 3025 as showing the exact stack for the previous access (as per 3026 --history-level=full), but it is better than nothing, and it's 3027 almost as fast as --history-level=none. 3028 3029 3030 * New features and improvements in DRD: 3031 3032 - The error messages printed by DRD are now easier to interpret. 3033 Instead of using two different numbers to identify each thread 3034 (Valgrind thread ID and DRD thread ID), DRD does now identify 3035 threads via a single number (the DRD thread ID). Furthermore 3036 "first observed at" information is now printed for all error 3037 messages related to synchronization objects. 3038 3039 - Added support for named semaphores (sem_open() and sem_close()). 3040 3041 - Race conditions between pthread_barrier_wait() and 3042 pthread_barrier_destroy() calls are now reported. 3043 3044 - Added support for custom allocators through the macros 3045 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in 3046 in <valgrind/valgrind.h>). An alternative for these two macros is 3047 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in 3048 <valgrind/drd.h>). 3049 3050 - Added support for annotating non-POSIX synchronization objects 3051 through several new ANNOTATE_*() macros. 3052 3053 - OpenMP: added support for the OpenMP runtime (libgomp) included 3054 with gcc versions 4.3.0 and 4.4.0. 3055 3056 - Faster operation. 3057 3058 - Added two new command-line options (--first-race-only and 3059 --segment-merging-interval). 3060 3061 3062 * Genuinely atomic support for x86/amd64/ppc atomic instructions 3063 3064 Valgrind will now preserve (memory-access) atomicity of LOCK- 3065 prefixed x86/amd64 instructions, and any others implying a global 3066 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions. 3067 3068 This means that Valgrinded processes will "play nicely" in 3069 situations where communication with other processes, or the kernel, 3070 is done through shared memory and coordinated with such atomic 3071 instructions. Prior to this change, such arrangements usually 3072 resulted in hangs, races or other synchronisation failures, because 3073 Valgrind did not honour atomicity of such instructions. 3074 3075 3076 * A new experimental tool, BBV, has been added. BBV generates basic 3077 block vectors for use with the SimPoint analysis tool, which allows 3078 a program's overall behaviour to be approximated by running only a 3079 fraction of it. This is useful for computer architecture 3080 researchers. You can run BBV by specifying --tool=exp-bbv (the 3081 "exp-" prefix is short for "experimental"). BBV was written by 3082 Vince Weaver. 3083 3084 3085 * Modestly improved support for running Windows applications under 3086 Wine. In particular, initial support for reading Windows .PDB debug 3087 information has been added. 3088 3089 3090 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been 3091 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks 3092 instead of bytes. 3093 3094 3095 * The Valgrind client requests VALGRIND_PRINTF and 3096 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously, 3097 the string was always printed immediately on its own line. Now, the 3098 string will be added to a buffer but not printed until a newline is 3099 encountered, or other Valgrind output is printed (note that for 3100 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered 3101 "other Valgrind output"). This allows you to use multiple 3102 VALGRIND_PRINTF calls to build up a single output line, and also to 3103 print multiple output lines with a single request (by embedding 3104 multiple newlines in the string). 3105 3106 3107 * The graphs drawn by Massif's ms_print program have changed slightly: 3108 3109 - The half-height chars '.' and ',' are no longer drawn, because 3110 they are confusing. The --y option can be used if the default 3111 y-resolution is not high enough. 3112 3113 - Horizontal lines are now drawn after the top of a snapshot if 3114 there is a gap until the next snapshot. This makes it clear that 3115 the memory usage has not dropped to zero between snapshots. 3116 3117 3118 * Something that happened in 3.4.0, but wasn't clearly announced: the 3119 option --read-var-info=yes can be used by some tools (Memcheck, 3120 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3 3121 variable type and location information. This makes those tools 3122 start up more slowly and increases memory consumption, but 3123 descriptions of data addresses in error messages become more 3124 detailed. 3125 3126 3127 * exp-Omega, an experimental instantaneous leak-detecting tool, was 3128 disabled in 3.4.0 due to a lack of interest and maintenance, 3129 although the source code was still in the distribution. The source 3130 code has now been removed from the distribution. For anyone 3131 interested, the removal occurred in SVN revision r10247. 3132 3133 3134 * Some changes have been made to the build system. 3135 3136 - VEX/ is now integrated properly into the build system. This means 3137 that dependency tracking within VEX/ now works properly, "make 3138 install" will work without requiring "make" before it, and 3139 parallel builds (ie. 'make -j') now work (previously a 3140 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j' 3141 was effectively ignored). 3142 3143 - The --with-vex configure option has been removed. It was of 3144 little use and removing it simplified the build system. 3145 3146 - The location of some install files has changed. This should not 3147 affect most users. Those who might be affected: 3148 3149 * For people who use Valgrind with MPI programs, the installed 3150 libmpiwrap.so library has moved from 3151 $(INSTALL)/<platform>/libmpiwrap.so to 3152 $(INSTALL)/libmpiwrap-<platform>.so. 3153 3154 * For people who distribute standalone Valgrind tools, the 3155 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a 3156 have moved to $(INSTALL)/libcoregrind-<platform>.a. 3157 3158 These changes simplify the build system. 3159 3160 - Previously, all the distributed suppression (*.supp) files were 3161 installed. Now, only default.supp is installed. This should not 3162 affect users as the other installed suppression files were not 3163 read; the fact that they were installed was a mistake. 3164 3165 3166 * KNOWN LIMITATIONS: 3167 3168 - Memcheck is unusable with the Intel compiler suite version 11.1, 3169 when it generates code for SSE2-and-above capable targets. This 3170 is because of icc's use of highly optimised inlined strlen 3171 implementations. It causes Memcheck to report huge numbers of 3172 false errors even in simple programs. Helgrind and DRD may also 3173 have problems. 3174 3175 Versions 11.0 and earlier may be OK, but this has not been 3176 properly tested. 3177 3178 3179 The following bugs have been fixed or resolved. Note that "n-i-bz" 3180 stands for "not in bugzilla" -- that is, a bug that was reported to us 3181 but never got a bugzilla entry. We encourage you to file bugs in 3182 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than 3183 mailing the developers (or mailing lists) directly -- bugs that are 3184 not entered into bugzilla tend to get forgotten about or ignored. 3185 3186 To see details of a given bug, visit 3187 https://bugs.kde.org/show_bug.cgi?id=XXXXXX 3188 where XXXXXX is the bug number as listed below. 3189 3190 84303 How about a LockCheck tool? 3191 91633 dereference of null ptr in vgPlain_st_basetype 3192 97452 Valgrind doesn't report any pthreads problems 3193 100628 leak-check gets assertion failure when using 3194 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory 3195 108528 NPTL pthread cleanup handlers not called 3196 110126 Valgrind 2.4.1 configure.in tramples CFLAGS 3197 110128 mallinfo is not implemented... 3198 110770 VEX: Generated files not always updated when making valgrind 3199 111102 Memcheck: problems with large (memory footprint) applications 3200 115673 Vex's decoder should never assert 3201 117564 False positive: Syscall param clone(child_tidptr) contains 3202 uninitialised byte(s) 3203 119404 executing ssh from inside valgrind fails 3204 133679 Callgrind does not write path names to sources with dwarf debug 3205 info 3206 135847 configure.in problem with non gnu compilers (and possible fix) 3207 136154 threads.c:273 (vgCallgrind_post_signal): Assertion 3208 '*(vgCallgrind_current_fn_stack.top) == 0' failed. 3209 136230 memcheck reports "possibly lost", should be "still reachable" 3210 137073 NULL arg to MALLOCLIKE_BLOCK causes crash 3211 137904 Valgrind reports a memory leak when using POSIX threads, 3212 while it shouldn't 3213 139076 valgrind VT_GETSTATE error 3214 142228 complaint of elf_dynamic_do_rela in trivial usage 3215 145347 spurious warning with USBDEVFS_REAPURB 3216 148441 (wine) can't find memory leak in Wine, win32 binary 3217 executable file. 3218 148742 Leak-check fails assert on exit 3219 149878 add (proper) check for calloc integer overflow 3220 150606 Call graph is broken when using callgrind control 3221 152393 leak errors produce an exit code of 0. I need some way to 3222 cause leak errors to result in a nonzero exit code. 3223 157154 documentation (leak-resolution doc speaks about num-callers 3224 def=4) + what is a loss record 3225 159501 incorrect handling of ALSA ioctls 3226 162020 Valgrinding an empty/zero-byte file crashes valgrind 3227 162482 ppc: Valgrind crashes while reading stabs information 3228 162718 x86: avoid segment selector 0 in sys_set_thread_area() 3229 163253 (wine) canonicaliseSymtab forgot some fields in DiSym 3230 163560 VEX/test_main.c is missing from valgrind-3.3.1 3231 164353 malloc_usable_size() doesn't return a usable size 3232 165468 Inconsistent formatting in memcheck manual -- please fix 3233 169505 main.c:286 (endOfInstr): 3234 Assertion 'ii->cost_offset == *cost_offset' failed 3235 177206 Generate default.supp during compile instead of configure 3236 177209 Configure valt_load_address based on arch+os 3237 177305 eventfd / syscall 323 patch lost 3238 179731 Tests fail to build because of inlining of non-local asm labels 3239 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion 3240 'ordxx == POrd_EQ || ordxx == POrd_LT' failed. 3241 181594 Bogus warning for empty text segment 3242 181707 dwarf doesn't require enumerations to have name 3243 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64 3244 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree): 3245 Assertion '!already_present' failed. 3246 185359 exp-ptrcheck: unhandled syscall getresuid() 3247 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64 3248 185816 Valgrind is unable to handle debug info for files with split 3249 debug info that are prelinked afterwards 3250 185980 [darwin] unhandled syscall: sem_open 3251 186238 bbToIR_AMD64: disInstr miscalculated next %rip 3252 186507 exp-ptrcheck unhandled syscalls prctl, etc. 3253 186790 Suppression pattern used for leaks are not reported 3254 186796 Symbols with length>200 in suppression files are ignored 3255 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation 3256 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid} 3257 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened 3258 188046 bashisms in the configure script 3259 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA 3260 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672 3261 (get_otrack_shadow_offset_wrk): the 'impossible' happened." 3262 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock, 3263 assertion fail "!lock->heldBy" 3264 188427 Add support for epoll_create1 (with patch) 3265 188530 Support for SIOCGSTAMPNS 3266 188560 Include valgrind.spec in the tarball 3267 188572 Valgrind on Mac should suppress setenv() mem leak 3268 189054 Valgrind fails to build because of duplicate non-local asm labels 3269 189737 vex amd64->IR: unhandled instruction bytes: 0xAC 3270 189762 epoll_create syscall not handled (--tool=exp-ptrcheck) 3271 189763 drd assertion failure: s_threadinfo[tid].is_recording 3272 190219 unhandled syscall: 328 (x86-linux) 3273 190391 dup of 181394; see above 3274 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc 3275 190820 No debug information on powerpc-linux 3276 191095 PATCH: Improve usbdevfs ioctl handling 3277 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks 3278 or big nr of errors 3279 191189 --xml=yes should obey --gen-suppressions=all 3280 191192 syslog() needs a suppression on macosx 3281 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265 3282 191761 getrlimit on MacOSX 3283 191992 multiple --fn-skip only works sometimes; dependent on order 3284 192634 V. reports "aspacem sync_check_mapping_callback: 3285 segment mismatch" on Darwin 3286 192954 __extension__ missing on 2 client requests 3287 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29 3288 194474 "INSTALL" file has different build instructions than "README" 3289 194671 Unhandled syscall (sem_wait?) from mac valgrind 3290 195069 memcheck: reports leak (memory still reachable) for 3291 printf("%d', x) 3292 195169 drd: (vgDrd_barrier_post_wait): 3293 Assertion 'r->sg[p->post_iteration]' failed. 3294 195268 valgrind --log-file doesn't accept ~/... 3295 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate 3296 195860 WARNING: unhandled syscall: unix:223 3297 196528 need a error suppression for pthread_rwlock_init under os x? 3298 197227 Support aio_* syscalls on Darwin 3299 197456 valgrind should reject --suppressions=(directory) 3300 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10 3301 197591 unhandled syscall 27 (mincore) 3302 197793 Merge DCAS branch to the trunk == 85756, 142103 3303 197794 Avoid duplicate filenames in Vex 3304 197898 make check fails on current SVN 3305 197901 make check fails also under exp-ptrcheck in current SVN 3306 197929 Make --leak-resolution=high the default 3307 197930 Reduce spacing between leak reports 3308 197933 Print command line of client at start-up, and shorten preamble 3309 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck) 3310 198395 add BBV to the distribution as an experimental tool 3311 198624 Missing syscalls on Darwin: 82, 167, 281, 347 3312 198649 callgrind_annotate doesn't cumulate counters 3313 199338 callgrind_annotate sorting/thresholds are broken for all but Ir 3314 199977 Valgrind complains about an unrecognized instruction in the 3315 atomic_incs test program 3316 200029 valgrind isn't able to read Fedora 12 debuginfo 3317 200760 darwin unhandled syscall: unix:284 3318 200827 DRD doesn't work on Mac OS X 3319 200990 VG_(read_millisecond_timer)() does not work correctly 3320 201016 Valgrind does not support pthread_kill() on Mac OS 3321 201169 Document --read-var-info 3322 201323 Pre-3.5.0 performance sanity checking 3323 201384 Review user manual for the 3.5.0 release 3324 201585 mfpvr not implemented on ppc 3325 201708 tests failing because x86 direction flag is left set 3326 201757 Valgrind doesn't handle any recent sys_futex additions 3327 204377 64-bit valgrind can not start a shell script 3328 (with #!/path/to/shell) if the shell is a 32-bit executable 3329 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization. 3330 n-i-bz drd: fixed a bug that caused incorrect messages to be printed 3331 about memory allocation events with memory access tracing enabled 3332 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation 3333 3334 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846). 3335 3336 3337 3338 Release 3.4.1 (28 February 2009) 3339 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 3340 3.4.1 is a bug-fix release that fixes some regressions and assertion 3341 failures in debug info reading in 3.4.0, most notably incorrect stack 3342 traces on amd64-linux on older (glibc-2.3 based) systems. Various 3343 other debug info problems are also fixed. A number of bugs in the 3344 exp-ptrcheck tool introduced in 3.4.0 have been fixed. 3345 3346 In view of the fact that 3.4.0 contains user-visible regressions 3347 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are 3348 encouraged to ship 3.4.1 in preference to 3.4.0. 3349 3350 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in 3351 bugzilla" -- that is, a bug that was reported to us but never got a 3352 bugzilla entry. We encourage you to file bugs in bugzilla 3353 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the 3354 developers (or mailing lists) directly -- bugs that are not entered 3355 into bugzilla tend to get forgotten about or ignored. 3356 3357 n-i-bz Fix various bugs reading icc-11 generated debug info 3358 n-i-bz Fix various bugs reading gcc-4.4 generated debug info 3359 n-i-bz Preliminary support for glibc-2.10 / Fedora 11 3360 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes, 3361 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2. 3362 179618 exp-ptrcheck crashed / exit prematurely 3363 179624 helgrind: false positive races with pthread_create and 3364 recv/open/close/read 3365 134207 pkg-config output contains @VG_PLATFORM@ 3366 176926 floating point exception at valgrind startup with PPC 440EPX 3367 181594 Bogus warning for empty text segment 3368 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes) 3369 181707 Dwarf3 doesn't require enumerations to have name 3370 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64 3371 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree): 3372 Assertion '!already_present' failed. 3373 185359 exp-ptrcheck unhandled syscall getresuid() 3374 3375 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253). 3376 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293). 3377 3378 3379 3380 Release 3.4.0 (2 January 2009) 3381 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 3382 3.4.0 is a feature release with many significant improvements and the 3383 usual collection of bug fixes. This release supports X86/Linux, 3384 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros 3385 (using gcc 4.4, glibc 2.8 and 2.9) has been added. 3386 3387 3.4.0 brings some significant tool improvements. Memcheck can now 3388 report the origin of uninitialised values, the thread checkers 3389 Helgrind and DRD are much improved, and we have a new experimental 3390 tool, exp-Ptrcheck, which is able to detect overruns of stack and 3391 global arrays. In detail: 3392 3393 * Memcheck is now able to track the origin of uninitialised values. 3394 When it reports an uninitialised value error, it will try to show 3395 the origin of the value, as either a heap or stack allocation. 3396 Origin tracking is expensive and so is not enabled by default. To 3397 use it, specify --track-origins=yes. Memcheck's speed will be 3398 essentially halved, and memory usage will be significantly 3399 increased. Nevertheless it can drastically reduce the effort 3400 required to identify the root cause of uninitialised value errors, 3401 and so is often a programmer productivity win, despite running more 3402 slowly. 3403 3404 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in 3405 3.4.0, will be released shortly. 3406 3407 * Helgrind's race detection algorithm has been completely redesigned 3408 and reimplemented, to address usability and scalability concerns: 3409 3410 - The new algorithm has a lower false-error rate: it is much less 3411 likely to report races that do not really exist. 3412 3413 - Helgrind will display full call stacks for both accesses involved 3414 in a race. This makes it easier to identify the root causes of 3415 races. 3416 3417 - Limitations on the size of program that can run have been removed. 3418 3419 - Performance has been modestly improved, although that is very 3420 workload-dependent. 3421 3422 - Direct support for Qt4 threading has been added. 3423 3424 - pthread_barriers are now directly supported. 3425 3426 - Helgrind works well on all supported Linux targets. 3427 3428 * The DRD thread debugging tool has seen major improvements: 3429 3430 - Greatly improved performance and significantly reduced memory 3431 usage. 3432 3433 - Support for several major threading libraries (Boost.Thread, Qt4, 3434 glib, OpenMP) has been added. 3435 3436 - Support for atomic instructions, POSIX semaphores, barriers and 3437 reader-writer locks has been added. 3438 3439 - Works now on PowerPC CPUs too. 3440 3441 - Added support for printing thread stack usage at thread exit time. 3442 3443 - Added support for debugging lock contention. 3444 3445 - Added a manual for Drd. 3446 3447 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck 3448 checks for misuses of pointers. In that sense it is a bit like 3449 Memcheck. However, Ptrcheck can do things Memcheck can't: it can 3450 detect overruns of stack and global arrays, it can detect 3451 arbitrarily far out-of-bounds accesses to heap blocks, and it can 3452 detect accesses heap blocks that have been freed a very long time 3453 ago (millions of blocks in the past). 3454 3455 Ptrcheck currently works only on x86-linux and amd64-linux. To use 3456 it, use --tool=exp-ptrcheck. A simple manual is provided, as part 3457 of the main Valgrind documentation. As this is an experimental 3458 tool, we would be particularly interested in hearing about your 3459 experiences with it. 3460 3461 * exp-Omega, an experimental instantaneous leak-detecting tool, is no 3462 longer built by default, although the code remains in the repository 3463 and the tarball. This is due to three factors: a perceived lack of 3464 users, a lack of maintenance, and concerns that it may not be 3465 possible to achieve reliable operation using the existing design. 3466 3467 * As usual, support for the latest Linux distros and toolchain 3468 components has been added. It should work well on Fedora Core 10, 3469 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release 3470 state) is supported, as is glibc-2.9. The C++ demangler has been 3471 updated so as to work well with C++ compiled by even the most recent 3472 g++'s. 3473 3474 * You can now use frame-level wildcards in suppressions. This was a 3475 frequently-requested enhancement. A line "..." in a suppression now 3476 matches zero or more frames. This makes it easier to write 3477 suppressions which are precise yet insensitive to changes in 3478 inlining behaviour. 3479 3480 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set. 3481 3482 * Very basic support for IBM Power6 has been added (64-bit processes only). 3483 3484 * Valgrind is now cross-compilable. For example, it is possible to 3485 cross compile Valgrind on an x86/amd64-linux host, so that it runs 3486 on a ppc32/64-linux target. 3487 3488 * You can set the main thread's stack size at startup using the 3489 new --main-stacksize= flag (subject of course to ulimit settings). 3490 This is useful for running apps that need a lot of stack space. 3491 3492 * The limitation that you can't use --trace-children=yes together 3493 with --db-attach=yes has been removed. 3494 3495 * The following bugs have been fixed. Note that "n-i-bz" stands for 3496 "not in bugzilla" -- that is, a bug that was reported to us but 3497 never got a bugzilla entry. We encourage you to file bugs in 3498 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than 3499 mailing the developers (or mailing lists) directly. 3500 3501 n-i-bz Make return types for some client requests 64-bit clean 3502 n-i-bz glibc 2.9 support 3503 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865) 3504 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults 3505 n-i-bz Building in an env without gdb gives bogus gdb attach 3506 92456 Tracing the origin of uninitialised memory 3507 106497 Valgrind does not demangle some C++ template symbols 3508 162222 ==106497 3509 151612 Suppression with "..." (frame-level wildcards in .supp files) 3510 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit) 3511 159285 unhandled syscall:25 (stime, on x86-linux) 3512 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig" 3513 160954 ppc build of valgrind crashes with illegal instruction (isel) 3514 160956 mallinfo implementation, w/ patch 3515 162092 Valgrind fails to start gnome-system-monitor 3516 162819 malloc_free_fill test doesn't pass on glibc2.8 x86 3517 163794 assertion failure with "--track-origins=yes" 3518 163933 sigcontext.err and .trapno must be set together 3519 163955 remove constraint !(--db-attach=yes && --trace-children=yes) 3520 164476 Missing kernel module loading system calls 3521 164669 SVN regression: mmap() drops posix file locks 3522 166581 Callgrind output corruption when program forks 3523 167288 Patch file for missing system calls on Cell BE 3524 168943 unsupported scas instruction pentium 3525 171645 Unrecognised instruction (MOVSD, non-binutils encoding) 3526 172417 x86->IR: 0x82 ... 3527 172563 amd64->IR: 0xD9 0xF5 - fprem1 3528 173099 .lds linker script generation error 3529 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl) 3530 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes) 3531 174532 == 173751 3532 174908 --log-file value not expanded correctly for core file 3533 175044 Add lookup_dcookie for amd64 3534 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding) 3535 3536 Developer-visible changes: 3537 3538 * Valgrind's debug-info reading machinery has been majorly overhauled. 3539 It can now correctly establish the addresses for ELF data symbols, 3540 which is something that has never worked properly before now. 3541 3542 Also, Valgrind can now read DWARF3 type and location information for 3543 stack and global variables. This makes it possible to use the 3544 framework to build tools that rely on knowing the type and locations 3545 of stack and global variables, for example exp-Ptrcheck. 3546 3547 Reading of such information is disabled by default, because most 3548 tools don't need it, and because it is expensive in space and time. 3549 However, you can force Valgrind to read it, using the 3550 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to 3551 make use of such information, if present, to provide source-level 3552 descriptions of data addresses in the error messages they create. 3553 3554 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882). 3555 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899). 3556
