1 2Bugs reported after the release of 3.8.0 3~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 4 5This file lists bugs that were reported after 3.8.0 was released, both 6in 3.8.0 and trunk, and that have not been closed. Bugs that have 7been closed as "fixed" are listed in the top level NEWS file instead. 8In other words they get listed here first and moved to NEWS if they 9get fixed. 10 11This file also tracks commits & fixes that should be pushed to the 3.8 12branch. 13 14-- PUSH TO 3_8_BRANCH 15 16(B38=2500) 2469 Fix a mixup .. 17(B38=12930) 12873 n-i-bz: Use -s USR1 instead of -10 to send SIGUSR1 signal 18(B38=12931) 12874 n-i-bz: shmat of a segment > 4Gb does not work 19(B38=12933) 12878 n-i-bz: On Darwin, use __NR_poll_nocancel rather than __NR_poll 20(B38=12934) 12879 + 12882 + 12883 (301281, OSX hang on "system") 21(B38=12936) 12881 ptrace mips stuff 22(B38=12938) 12885 (Iop_DPBtoBCD and Iop_BCDtoDPB fix) 23(B38=2502) 2475 (289584: Handle a reg-reg encoding of MOVAPS) 24(B38=2503) 2476 (305042: Implement 0F 7F encoding of movq reg-reg) 25(B38=12940) 12887 (n-i-bz: libmpiwrap.c fixes) 26(B38=2504) 2477 (305321: Re-enable 'prefetch m8' and 'prefetchw m8') 27(B38=2505) 2478 (295808: Fix LZCNT and TZCNT properly) 28(B38=2506) 2479 (305199: Implement QDADD and QDSUB) 29(B38=2507) 2480 (304867: Implement MOVBE in 32 bit mode) 30(B38=12941) 12891 (304980: Fix up range checking in ML_(addLineInfo)) 31(B38=12942) 12892 (305513: Fix skipping of block forms) 32(B38=12943) 12893 (298281: Implement a wrapper for wcslen on Linux) 33(B38=12944) 12894 (n-i-bz, exp-sgcheck memcpy) 34(B38=12945) 12897/8 (305690: Handle non-zero sem_*wait() return values correctly) 35(B38=2508) 2499 (305926: Invalid alignment checks for some AVX instructions) 36(B38=12946) 12923 (304980: Guard against negative symbol sizes) 37(B38=12948) 12947 (n-i-bz: volatile in stack check, re clang) 38(B38=12984) 12964 (306310 3.8.0 release tarball missing some files) 39(B38=2534) 2517/8 (n-i-bz: fix a couple of union tag-vs-field mixups) 40(B38=2535) 2531 (306297: ARM: STM<c>.W <Rn>{!}, <registers> too strict) 41(B38=2536) 2532 (306664: AMD64: Fix PCMPxSTRx variant $0x46) 42(B38=12985) 12982 (306612: Intercept __GI_memmove) 43(B38=2537,12986) 44 2533,12980 45 (304035: ARM: uqsub16 shadd16 uhsub8 uhsub16) 46 47-------- Bugs brought forward from 3_7_BUGSTATUS.txt -------- 48 49291310 FXSAVE instruction marks memory as undefined on amd64 50 x86 vs amd64 inconsistency? 51 **is fixed in 3.8.0, but was not marked as such 52 ** no it's not; v38BRANCH shows error, vTRUNK doesn't -- investigate 53 REOP 54 55303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped (w/ patch) 56 SYSCALL, HAS_PATCH 57 58-------- Unfixed in 3.8.0 59 60303741 sockaddr* parameter of recvfrom not handled correctly 61 (corruption not detected later on) 62 SYSCALL, HAS_TESTCASE 63 64303877 valgrind doesn't support compressed debuginfo sections. 65 QUERIED 66 67304259 support Darwin 12 (Mac OS 10.8, Mountain Lion) 68 OSX108 (Canonical) 69 70304321 m_signals.c:2314 (sync_signalhandler_from_kernel): 71 Assertion 'tid != 0' failed 72 CLOSED as UNMAINTAINED. 73 74304450 time logging tools ? 75 WISHLIST 76 77304675 make install-strip fails when cross-compiling 78 w/ one-liner patch; but needs checking (should be more general? 79 assumes auto* version?) 80 81304676 regtest hangs on mac OS X (mchelp, coolo_sigaction, system) 82 probably related to 301281 et al. Still valid. 83 OSX 84 85305170 linux: add FIOCLEX and FIONCLEX ioctls 86 SYSCALL. w/ patch, looks pretty harmless 87 88n-i-bz exp-sgcheck asserts on gcc-4.6.2 generated Dwarf3 89 90305957 m_debuginfo/d3basics.c:965 (vgModuleLocal_evaluate_GX): 91 Assertion 'aMin == (Addr)0' failed. 92 QUERIED -- no info so far -- close? 93 94306004 Wishlist: port valgrind to sparc 95 WISHLIST 96 97-- Sat 1 Sept 2012 10pm 98 99306171 Boost.Context appears to cause Valgrind to crash 100 Stack switching; should fix. (Conservative unwind?) 101 102306235 unhandled syscall: 429 (freebsd) 103 Need to forward to FreeBSD port maintainers. 104 105306299 Internal error when running a winelib application 106 Conservative unwind? 107 108306340 parse_var_DIE confused by DWARF from clang-3.1 109 Might be quite simple to fix; need to investigate. 110 No action so far 111 112306535 massif: force dump result every X minutes 113 Queried 114 115306587 data cache line size is 128 and instruction cache line size is 32 116 PPC; obscure, needs investigatoin 117 118306590 not support e500v2 spe 0x216 command 119 PPC SPE insns; with plausible looking patch? Need test cases, and 120 cc Mehafy. 121 122286864 strlen function redirection error 123 Gentoo glibc "miscompilation"; is this important? 124 125-- Mon 17 Sept 2012, 10 am 126 127valgrind.spec diff at https://bugzilla.mozilla.org/show_bug.cgi?id=750856#c20 128r13008 to 382 ? check 129change of leak-check-counting? 307465 130gkw extra suppression? 131gcp libnss3.so debuginfo read hang 132 133306947 s390x: BFP rounding mode improvements 134 Proposal, no action 135 136306949 s390x: BFP condition code improvements 137 Proposal, no action 138 139211352 helgrind races in helgrind's own mythread_wrapper 140 NEEDINFO 141 142307367 amd64->IR: 0xF2 0x48 0xF 0x5F 0x3E 0xF2 0x48 0xF 143 redundant-REX, should be easy to fix 144 145307399 amd64->IR: 0xC4 0x62 0x55 0x2E 0x18 0x44 0x39 0x87 146 (vmaskmovps %ymm11,%ymm5,(%rax)) 147 CONDITIONAL STORES -- probably fixed? 148 149307415 3.8.1 build error - OSX Lion 10.7.4 duplicate symbol 150 _mach_make_memory_entry 151 OSX or XCode 4.5 152 153307500 Valgrind docs have two <title> tags 154 QUERIED (unclear how to fix) 155 Severity: low 156 157308027 Warning: negative range in .debug_loc section 158 QUERIED -- not sure this needs fixing or not 159 16079362 Debug info is lost for .so files when they are dlclose'd 161 Still alive! 162 163308158 running with --leack-check=full makes valgrind exit 164 FreeBSD -- although FreeBSD maintainers may not be aware 165 166------ Thu Oct 11 18:51:26 CEST 2012 167 168197915 80-bit floats are not supported on x86 and x86-64 169 indeed, yes. 170 171299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' 172 failed. 173 OSX Helgrind 174 175308582 massif: wrong stack usage at exit of threaded program 176 No action so far 177 178308668 pthread rwlock_rdlock doesn't understand modifications in 179 some cases (Helgrind) 180 Real; no action so far 181 182308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 183 HAS_PATCH, but NEEDINFO 184 185243232 *Inconsistent Lock Orderings report with trylock (PBOS?) 186 Almost certainly real (+ probably a dup). No action so far. 187 188237920 valgrind segfault on fork failure 189 Uninvestigated; filed by JJ 190 191308816 valgrind GPF in signal handler context 3.8.1 (allignment 192 Needs investigation 193 Maybe fixed? 194 195309100 s390x: Testcases for extended BFP 196 No action so far 197 198309430 Self hosting ppc64 encounters a vassert error on operand type 199 No action so far. Sounds a bit strange. 200 201309554 Unhandled syscall remap_file_pages (216) 202 No action so far 203 Possibly difficult to implement 204 205309721 add client request returning current value of a statistic (e.g. Ir) 206 Wishlist. No action yet. 207 208309783 Valgrind does not support separate source and build directories 209 Has patch, but needs further checking 210 Does not appear to be much enthusiasm for this 211 212309827 *add support for nvidia (nvmem) ioctls 213 A bunch of Android hackery from VladV; needs pulling apart 214 FIX 3.9 (commit) 215 216309831 3.8.1: disInstr(arm): unhandled instruction: 0xECECA102 217 (stcl, gdb disassembles to stfp) .. IWMMXT 218 ==309604 219 220310188 need better diagnostic when mmap() fails with EINVAL 221 No patch so far 222 Severity: low 223 224310692 [coregrind] Thread scheduling regression? 225 Spinlockery 226 Severity: med Audience: low 227 228311093 massif and memcheck fail to redirect malloc calls on 229 statically linked application 230 Probably INVALID 231 232-- Wed Dec 5 23:41:20 CET 2012 233 234311544 sys_fork() calls VG_(do_atfork_pre)() AFTER forking() 235 FIX 3.9 (easy) 236 237311649 exhausting fds can leak created files 238 Probably true; NEEDINFO; possible WONTFIX; QUERIED 239 240311655 --log-file=FILE leads to apparent fd leak ("Open file 241 descriptor 3: $LOGFILE") 242 Probably true; should fix. 243 Severity: med Audience: low 244 245311827 OS X 10.8.2 (eq_SyscallStatus): the 'impossible' happened. 246 OSX10.8 Maybe fixed in the trunk already. QUERIED 247 248312989 ioctl handling needs to do POST handling on generic ioctls and 249 needs to handle BPF ioctls 250 OSX10.8 FIX 3.9 (commit) 251 252313275 False positives for shared memory semaphores/mutexes/conditions 253 No action so far. Has testcase patches. 254 Potentially important to fix, but does not look easy. 255 256313276 helgrind does not recognize shared memory initialized 257 semaphores, mutexes, conditions 258 == 313275, pretty much 259 Severity: med Audience: low 260 261313875 cr_libinit.c:183 cri_init: sigaction() failed: Invalid argument 262 No action so far 263 Severity: high Audience: low 264 265** [Bug 291310] FXSAVE instruction marks memory as undefined on amd64 266 Severity: med Audience: low 267 268314236 OS X 10.8 testsuite build failure: test million.o 269 OSX108 No action so far. 270 Severity: low Audience: med 271 272314365 enable VEX to run asm helpers that do callee register saving 273 In progress; not sure whether this is a good idea 274 275315199 vgcore file for threaded app does not show which thread crashed 276 HAS_PATCH; needs review 277 Severity: low Audience: med 278 279-- Thu Feb 28 09:54:56 CET 2013 280 281315936 UNKNOWN task message [id 3403, to mach_task_self(), reply 282 0x2603] when running fork()ing tests on Mac OS X 283 OSX10.8 284 285315952 tun/tap ioctls are not supported 286 QUERIED; leaving open for now 287 Severity: med Audience: low 288 289316034 Building valgrind with gcc (4.7, trunk) fails on MacOS X 10.8 290 Probably WONTFIX. 291 292-- Mon Mar 4 11:36:17 CET 2013 293 294294285 --partial-loads-ok does not work for 16-byte SSE loads 295 FIXED (multiple commits) 296 Requires regtest hookup on ppc 297 298316382 Support for AMDs FMA4 instructions seem to be missing 299 No action so far 300 Severity: high Audience: low 301 302316528 Dump for cachegrind 303 Wishlist 304 305316612 A -Wunused-value warning from Clang when including memcheck.h 306 Probably needs doc update 307 Severity: low Audience: med 308 309316696 fluidanimate program of parsec 2.1 stuck (do not finish 310 for more than 5 hours) 311 No action so far (DRD specific) 312 NEEDINFO 313 Severity: high Audience: low 314 315316874 amd64->IR: 0xC4 0xE2 0xF8 0xF2 0x85 0x30 0xFF 0xFF 316 Probably some AMD FMA insn? 317 Severity: high Audience: low 318 319316903 memcheck/tests/origin5-bz2 FAIL with gcc-4.5+ 320 Possibly a real bug 321 Severity: med Audience: med 322 323317078 _mm_madd_epi16, undefined * 0 should not give uninitialised result 324 True, but probably difficult to fix without perf hit 325 Prio: low 326 327317127 Fedora18/x86_64 --sanity-level=3 : aspacem segment mismatch 328 Diagnosed (BTRFS specific); unclear what to do, if anything 329 330317208 vex x86->IR: unhandled instruction bytes: 0xC8 0x0 0x0 0x0 331 "enter $0,$0" created by D language compiler 332 Severity: high Audience: low 333 334317381 helgrind warns about xchg vs suppressed store 335 No action so far. Not sure there's an easy fix for this. 336 Severity: med Audience: low 337 338317390 Patch for FBLD/FBSTP/FTST 339 Needs cleanup+test cases. Queried. 340 Severity: high Audience: low 341 342317441 Add support for AMD's LWP (LightWeight Profiling) architecture 343 extension 344 Contains plausible infrastructure patch; no insns so far tho 345 This is going to need some working through. Probably OK. 346 Not sure if can make 3.9. 347 Severity: high Audience: low 348 349317698 parse_var_DIE: confused by: DW_TAG_compile_unit using 350 Intel 13.0 update 3 compiler 351 Reporter has queried Intel since this might be an ICC bug 352 NEEDINFO 353 Severity: med Audience: low 354 355317893 massif terminates without any message 356 Probably just a memory limit thing. Close as a dup, but of what? 357 NEEDINFO 358 359318030 addHRegUse takes a lot of CPU time; band-aid speedup 360 patch within 361 No action so far 362 Needs review; would be nice to redo this with a more robust API 363 364Thu Apr 11 18:16:04 CEST 2013 365 366319237 [PATCH] Added support for ELF injection 367 FreeBSD-specific 368 369319274 unhandled syscall: unix:410 370 OSX108-specific 371 372319306 gdbserver tests call vgdb without path; patch within 373 Needs further discussion? 374 375319307 Patches to let test suite run in parallel 376 Looks plausible 377 Wishlist, effectively (has patch) 378 379319393 bad rounding in cvtsi2ss instruction 380 Not started. Looks valid. (double rounding) 381 Severity: med Audience: low 382 383319419 unhandled syscall: unix:369 384 OSX108-specific 385 386319547 helgrind gives false positives on rwlocks 387 Needs investigation. Quite possibly a real bug. 388 Severity: med Audience: med 389 390319896 vex amd64->IR: 0x8C 0xE0 391 Probably requires proper segment support, but need to check. 392 Severity: high Audience: low 393 394320082 disInstr(thumb): unhandled instruction on instruction LDRB 395 Has patch 396 LDRB; surely already supported? Queried. 397 MMhmh, untestable? 398 Severity: high Audience: low 399 400320100 Add nanosecond timer support 401 A bigger patch depends on it, but that has not appeared yet 402 Wishlist, + as yet incomplete 403 404320546 valgrind exits with "Assertion 'in_rx' failed" 405 for /system/lib/libva-android.so on Android 4.0.4 406 Need more info (the .so); queried. 407 MAYBE FIX 3.9 (potentially serious) 408 409320965 Unrecognised instruction __ieee754_pow_sse2 410 Not sure what this is. Queried. 411 Severity: high Audience: low (??) 412 413321396 Add Support for Freescale e500v2 SPE Architecture. 414 Is metabug? -- no patch, no details 415 416321400 Add Support for Freescale e5500 Architecture. 417 No patch, no details 418 419321401 Add Support for Freescale e6500 Architecture. 420 No patch, no details 421 422321681 printf false positive 423 Severity: med Audience: low 424 425321836 valgrind cross compilation failure on Ubuntu machine 426 MIPS specific? 427 428Tue Jul 2 12:15:07 CEST 2013 429 430321888 *Unhandled instruction: LDRH (Thumb) 431 Has patch; plausible (needs checking) 432 FIX 3.9 (commit) 433 434321892 *Unhandled instruction: LDRSH (Thumb) 435 Has patch; plausible (needs checking) 436 FIX 3.9 (commit) 437 438321902 *disInstr(ARM): 0xECECA102 439321903 == 321902 440 MAYBE 3.9 (query vgolubev) 441 442322150 Add Support for Cavium Octeon Instructions 443 MIPS64; has patch; assigned to Petar 444 MAYBE 3.9 (query Petar) 445 446322256 Provide more callstacks on memory pool related errors 447 Contains plausible patch; needs looking at 448 Severity: low Audience: low 449 450322258 No thread notification is printed if new thread reused tid of 451 old thread that triggered last error 452 Contains plausible patch. Is bug fix. (should take) 453 FIX 3.9 (commit) 454 455322586 Unknown instruction (bextr) in ld-2.15.so with --march=native on AMD FX-8350 456 AMD specific. Uninvestigated. 457 Severity: high Audience: low (??) 458 459322593 Power 8 code to implement Transactional support 460 Implements pass-to-host scheme; not intended to land. 461 Severity: low (right now) Audience: low 462 463322621 False positives with pthread_barrier_wait / helgrind 464 Severity: med Audience: low 465 466323020 more mysteries in valgrind thread stack handling 467 QUERIED 468 Severity: med Audience: low 469 470323046 Implement ANNOTATE_THREAD_NAME() Hellgrind annotation 471 Request for implementation 472 QUERIED .. is 322254 good enough? 473 Severity: low Audience: med 474 475323178 *Unhandled instruction: PLDW register (ARM) 476 Has patch; plausible (needs checking) 477 FIX 3.9 (commit) 478 479323179 *Unhandled instruction: PLDW immediate (ARM) 480 Has patch; plausible (needs checking) 481 FIX 3.9 (commit) 482 483323268 New calls for custom memory allocators 484 Obscure; needs investigation 485 Severity: low Audience: low 486 487323431 vex amd64->IR: 0x8F 0xE8 0x78 0xA2 0xC1 0x40 0xC5 0xFB 488 AMD-specific? 489 Severity: high Audience: low (??) 490 FIX 3.9 if can do it at no risk and no CPUID hassle 491 -- ask mjw 492 493323448 Check for memory overlap in sprintf()-like functions 494 Severity: low Wishlist 495 496323464 VALGRIND_MEMPOOL_ALLOC problems with size >= 2^32 497 64-bit uncleanness; no action so far; QUERIED 498 Severity: med Audience: low 499 500323777 *Documentation does not warn of sgcheck+darwin incompatibility 501 FIX 3.9 502 503324050 Helgrind: SEGV because of unaligned stack when using movdqa 504 Needs to be looked at. Is this real, or some Gentoo-caused 505 breakage? 506 Severity: high (hg crash) Audience: low 507 508324078 Track and show in logs which thread allocated memory 509 Probably sync-event-loss in Helgrind 510 Severity: high (hg correctness bug?) Audience: low 511 512324081 3.8.1 Valgrind periodically hangs in long-running application 513 NEEDINFO (requested) 514 515324261 errorcount in xml output remains empty (no error pairs mentioned) 516 Hmm, looks like a real bug. 517 Severity: med Audience: low 518 519324486 WARNING: unhandled syscall: 310 520 == 166355 521 Hmm, tricky, sys_unshare. 522 Severity: high Audience: low 523 524324525 valgrind can not detect any leak 525 PROBABLE INVALID; needs closing. 526 Severity: med Audience: low 527 528324809 show nightly regression tests results (and later performance) in a 529 nice html page on valgrind.org 530 In progress 531 Severity: low Audience: low 532 533324882 CPUID should depend on VEX capabilities, not on underlying CPU 534 Middle-term cleanup; not urgent 535 536Thu Sep 19 10:34:49 CEST 2013 537