From: Julian Seward Date: Thu, 19 Sep 2013 23:04:03 +0000 (+0000) Subject: More bug triage. X-Git-Tag: svn/VALGRIND_3_9_0~119 X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=f306d747e9f3b5acc8ab44dce5bfc6fba0bd4d18;p=thirdparty%2Fvalgrind.git More bug triage. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@13572 --- diff --git a/NEWS b/NEWS index d40770dbd4..a7ec222828 100644 --- a/NEWS +++ b/NEWS @@ -56,6 +56,15 @@ Release 3.9.0 (?? ?????? 201?) 'v.do expensive_sanity_check_general' that checks the sanity of various Valgrind aspects, including the Valgrind heap. + - remote debuginfo server + overhaul of debuginfo reading + + - some fixes for OSX 10.8 + + - partial-loads-ok now works for 16 and 32 byte loads + and in general improved handling of vectorised code + + + * ==================== FIXED BUGS ==================== The following bugs have been fixed or resolved. Note that "n-i-bz" @@ -230,6 +239,8 @@ m = merged into 3_8_BRANCH FIXED 13294 315545 [390] (find_TTEntry_from_hcode): Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed +321814 == 315545 + 316535 [390] Use of |signed int| instead of (unsigned) |size_t| in valgrind messages... 315959 [390] valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section 316144 [390] valgrind.1 manpage contains unknown ??? strings for some core option references @@ -512,6 +523,43 @@ FIXED r?? 322294 Add initial support for IBM Power ISA 2.07 FIXED 2740 13494, but needs close +324816 Incorrect VEX implementation for xscvspdp/xvcvspdp regarding SNaN inputs + FIXED 2760 13544 + +324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc + FIXED 13562 + +323893 SSE3 not available on amd cpus in valgrind + No action so far; maybe MJW can look at it + FIXED 2761 + +320895 add fanotify support (patch included) + FIXED 13460 + +320057 We have problems when we try to mmap more than 12 memory pages + on MIPS32 when we are using Valgrind + FIXED 13450 + +319505 [MIPSEL] Crash: unhandled UNRAY operator. + FIXED 2729 + +319494 VEX Makefile-gcc standalone build update after r2702 + FIXED 2702 + +318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16) + FIXED 2721 13396 + +318773 unhandled instruction bytes: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F + tzcnt? == 295808, but not closed + FIXED 2478 + +318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER + FIXED 13461 + +316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled + FIXED 13459 + + Release 3.8.1 (19 September 2012) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ diff --git a/docs/internals/3_8_BUGSTATUS.txt b/docs/internals/3_8_BUGSTATUS.txt index 2d3c6f1c55..9368e88016 100644 --- a/docs/internals/3_8_BUGSTATUS.txt +++ b/docs/internals/3_8_BUGSTATUS.txt @@ -327,98 +327,83 @@ gcp libnss3.so debuginfo read hang 316034 Building valgrind with gcc (4.7, trunk) fails on MacOS X 10.8 -- Mon Mar 4 11:36:17 CET 2013 +XXX 20 Sept 2013: triaged back to here -!! 294285 LoPresti 16-byte helper returns +294285 --partial-loads-ok does not work for 16-byte SSE loads + FIXED (multiple commits) + Requires regtest hookup on ppc 316382 Support for AMDs FMA4 instructions seem to be missing No action so far + Severity: high Audience: low 316528 Dump for cachegrind Wishlist 316612 A -Wunused-value warning from Clang when including memcheck.h Probably needs doc update - -316613 valgrind seg-faults in get_StackTrace_wrk under 32-bit - x86 app using makecontext - WONTFIX (is that the correct resolution?) + Severity: low Audience: med 316696 fluidanimate program of parsec 2.1 stuck (do not finish for more than 5 hours) No action so far (DRD specific) - -316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled - Has patch; looks plausible; no action so far + NEEDINFO + Severity: high Audience: low 316874 amd64->IR: 0xC4 0xE2 0xF8 0xF2 0x85 0x30 0xFF 0xFF Probably some AMD FMA insn? - QUERIED + Severity: high Audience: low 316903 memcheck/tests/origin5-bz2 FAIL with gcc-4.5+ Possibly a real bug + Severity: med Audience: med 317078 _mm_madd_epi16, undefined * 0 should not give uninitialised result True, but probably difficult to fix without perf hit + Prio: low 317127 Fedora18/x86_64 --sanity-level=3 : aspacem segment mismatch Diagnosed (BTRFS specific); unclear what to do, if anything 317208 vex x86->IR: unhandled instruction bytes: 0xC8 0x0 0x0 0x0 "enter $0,$0" created by D language compiler - -317311 Helgrind shouldn't report errors for memories allocated - inside pthread/mutex methods - FreeBSD+Helgrind weirdness + Severity: high Audience: low 317381 helgrind warns about xchg vs suppressed store No action so far. Not sure there's an easy fix for this. + Severity: med Audience: low 317390 Patch for FBLD/FBSTP/FTST Needs cleanup+test cases. Queried. + Severity: high Audience: low 317441 Add support for AMD's LWP (LightWeight Profiling) architecture extension Contains plausible infrastructure patch; no insns so far tho + This is going to need some working through. Probably OK. + Not sure if can make 3.9. + Severity: high Audience: low 317698 parse_var_DIE: confused by: DW_TAG_compile_unit using Intel 13.0 update 3 compiler Reporter has queried Intel since this might be an ICC bug + NEEDINFO + Severity: med Audience: low 317893 massif terminates without any message - Probably just a memory limit thing. Close as a dup, but - of what? + Probably just a memory limit thing. Close as a dup, but of what? + NEEDINFO 318030 addHRegUse takes a lot of CPU time; band-aid speedup patch within No action so far + Needs review; would be nice to redo this with a more robust API 318050 libmpiwrap fails to compile with out-of-source build - Has simple-sounding fix; should commit. + FIX 3.9 (commit) Thu Apr 11 18:16:04 CEST 2013 -318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER - Has patch, looks plausible - -318505 Valgrind (callgrind) cannot handle recvmmsg system call - INVALID (old version) - -318575 Does not running. CRC mismatch wanted 0fcc7297 - Probably invalid -- inability to find ld.so symbols at startup - -318656 (dpps) vex x86->IR: 0x66 0xf 0x3a 0x40 0xc1 0xf1 - WONTFIX (SSE4 on 32-bit x86) - -318773 unhandled instruction bytes: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F - tzcnt? == 295808, but not closed - -318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16) - FIXED 2721 13396 - -319225 QA Notice: Package triggers severe warnings which indicate - that it may exhibit random runtime failures. - We should close as WONTFIX - 319237 [PATCH] Added support for ELF injection FreeBSD-specific @@ -430,59 +415,54 @@ Thu Apr 11 18:16:04 CEST 2013 319307 Patches to let test suite run in parallel Looks plausible + Wishlist, effectively (has patch) 319393 bad rounding in cvtsi2ss instruction Not started. Looks valid. (double rounding) + Severity: med Audience: low 319419 unhandled syscall: unix:369 - OSX specific - -319494 VEX Makefile-gcc standalone build update after r2702 - Has patch; should land it. - -319505 [MIPSEL] Crash: unhandled UNRAY operator. - --> petarj + OSX108-specific 319547 helgrind gives false positives on rwlocks - Needs investigation + Needs investigation. Quite possibly a real bug. + Severity: med Audience: med 319896 vex amd64->IR: 0x8C 0xE0 Probably requires proper segment support, but need to check. - -319968 disInstr(arm): 0x69746E65 (valgrind_v3.81, cortex-A9) - INVALID (needs closing) - -320058 We have problems when we try to mmap more than 12 memory pages - on MIPS32 when we are using Valgrind - --> petarj + Severity: high Audience: low 320082 disInstr(thumb): unhandled instruction on instruction LDRB Has patch LDRB; surely already supported? Queried. + MMhmh, untestable? + Severity: high Audience: low 320100 Add nanosecond timer support A bigger patch depends on it, but that has not appeared yet + Wishlist, + as yet incomplete 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64) - Has patch + FIX 3.9 (the simple patch) 320546 valgrind exits with "Assertion 'in_rx' failed" for /system/lib/libva-android.so on Android 4.0.4 Need more info (the .so); queried. + MAYBE FIX 3.9 (potentially serious) 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname' failed" on weird files - Has patch; needs resync (Queried) - -320895 add fanotify support (patch included) - Has patches but need more work; queried + Has patch; + FIX 3.9 (commit); looks very simple 320965 Unrecognised instruction __ieee754_pow_sse2 Not sure what this is. Queried. + Severity: high Audience: low (??) 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instructions are unsupported (by version 3.8.1) PCMPxSTRx $0x30, $0x40 + Severity: high Audience: low 321396 Add Support for Freescale e500v2 SPE Architecture. Is metabug? -- no patch, no details @@ -493,14 +473,8 @@ Thu Apr 11 18:16:04 CEST 2013 321401 Add Support for Freescale e6500 Architecture. No patch, no details -321444 Multi-byte fill options - Wishlist - 321681 printf false positive - Probably WONTFIX - -321814 stepping through breakpoints in attached gdb process crashes valgrind - OSX specific; queried + Severity: med Audience: low 321836 valgrind cross compilation failure on Ubuntu machine MIPS specific? @@ -509,185 +483,176 @@ Tue Jul 2 12:15:07 CEST 2013 321888 Unhandled instruction: LDRH (Thumb) Has patch; plausible (needs checking) + FIX 3.9 (commit) 321891 Unhandled instruction: LDRHT (Thumb) Has patch; plausible (needs checking) + FIX 3.9 (commit) 321892 Unhandled instruction: LDRSH (Thumb) Has patch; plausible (needs checking) + FIX 3.9 (commit) 321902 disInstr(ARM): 0xECECA102 321903 == 321902 - -322064 Instruction not recognized on i7 with gcc -march=native - INVALID (AVX) - -322131 unhandled instruction bytes: 0xC5 0xFB 0x10 0x8D 0xC8 0xFB 0xFF 0xFF - INVALID (AVX) + MAYBE 3.9 (query vgolubev) 322150 Add Support for Cavium Octeon Instructions MIPS64; has patch; assigned to Petar + MAYBE 3.9 (query Petar) 322254 Show threadname together with tid if set by application Contains plausible looking patch (what's with XML output??) + FIX 3.9 (commit) 322256 Provide more callstacks on memory pool related errors Contains plausible patch; needs looking at + Severity: low Audience: low 322258 No thread notification is printed if new thread reused tid of old thread that triggered last error Contains plausible patch. Is bug fix. (should take) - -322279 vgdb fails on OSX Mountain Lion - WONTFIX - -322499 False positive 'uninitialized bytes' error with fwrite - INVALID + FIX 3.9 (commit) 322563 vex mips->IR: unhandled instruction bytes: 0x70 0x83 0xF0 0x3A MIPS; assigned to Petar - -322566 Compilation failed on MAC OSX Darwin 10.8.0: option "-arch x86_64" - not recognized by gcc 4.6.0 - QUERIED. Maybe use of FSF gcc instead of XCode variant? + MAYBE 3.9 (query Petar) 322586 Unknown instruction (bextr) in ld-2.15.so with --march=native on AMD FX-8350 AMD specific. Uninvestigated. + Severity: high Audience: low (??) 322593 Power 8 code to implement Transactional support Implements pass-to-host scheme; not intended to land. + Severity: low (right now) Audience: low 322621 False positives with pthread_barrier_wait / helgrind - Uninvestigated. - -322691 chainXDirect_ARM: Assertion is_imm32_to_iregNo_EXACTLY2 fails - QUERIED - -322780 false positive due to use of red zone in openssl - QUERIED, but is essentially an openssl problem. Probable wontfix. - -322818 disInstr(thumb): unhandled instruction: 0xEEBB 0x0BC8 - INVALID (is already fixed?) - -322935 disInstr(arm): unhandled instruction: 0xF1010200 (SETEND BE) - Probably WONTFIX + Severity: med Audience: low 323020 more mysteries in valgrind thread stack handling QUERIED + Severity: med Audience: low 323035 Unhandled instruction: LDRSHT(Thumb) Has patch; plausible (needs checking) + FIX 3.9 (commit) 323036 Unhandled instruction: SMMLS (ARM and Thumb) Has patch; plausible (needs checking) (duplication?) + FIX 3.9 (commit) 323046 Implement ANNOTATE_THREAD_NAME() Hellgrind annotation Request for implementation QUERIED .. is 322254 good enough? + Severity: low Audience: med 323175 Unhandled instruction: SMLALD (ARM + Thumb) Has patch; plausible (needs checking) (duplication?) + FIX 3.9 (commit) 323177 Unhandled instruction: SMLSLD (ARM + Thumb) Has patch; plausible (needs checking) (duplication?) + FIX 3.9 (commit) 323178 Unhandled instruction: PLDW register (ARM) Has patch; plausible (needs checking) + FIX 3.9 (commit) 323179 Unhandled instruction: PLDW immediate (ARM) Has patch; plausible (needs checking) + FIX 3.9 (commit) 323268 New calls for custom memory allocators Obscure; needs investigation + Severity: low Audience: low -323431 vex amd64->IR: 0x8F 0xE8 0x78 0xA2 0xC1 0x40 0xC5 0xFB +323431 vex amd64->IR: 0x8F 0xE8 0x78 0xA2 0xC1 0x40 0xC5 0xFB AMD-specific? + Severity: high Audience: low (??) + FIX 3.9 if can do it at no risk and no CPUID hassle + -- ask mjw 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() with PTHREAD_COND_INITIALIZER/PTHREAD_MUTEX_INITIALIZER triggers false positive Has patch from pbos + FIX 3.9 323448 Check for memory overlap in sprintf()-like functions - Wishlist + Severity: low Wishlist 323464 VALGRIND_MEMPOOL_ALLOC problems with size >= 2^32 64-bit uncleanness; no action so far; QUERIED - -323495 Detect when declared buffer size is too big - WONTFIX + Severity: med Audience: low 323777 Documentation does not warn of sgcheck+darwin incompatibility - True; should fix. + FIX 3.9 323803 Transactional memory instructions are not supported for Power This is TM scheme 1 (immediate fail). Needs landing. - -323893 SSE3 not available on amd cpus in valgrind - No action so far; maybe MJW can look at it + FIX 3.9 (?? queried Carl) 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions Has patch; plausible (needs checking) (aren't these done already?) + FIX 3.9 (commit) 324050 Helgrind: SEGV because of unaligned stack when using movdqa Needs to be looked at. Is this real, or some Gentoo-caused breakage? + Severity: high (hg crash) Audience: low 324078 Track and show in logs which thread allocated memory Probably sync-event-loss in Helgrind + Severity: high (hg correctness bug?) Audience: low 324081 3.8.1 Valgrind periodically hangs in long-running application - NEEDINFO + NEEDINFO (requested) 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT, helgrind thinks the condvar is still waited on - Has patches; needs looking at - -324156 memcheck probably reports false memory leak in pciutils library - INVALID - -324179 massif crashes on exit, valgrind --tool=memcheck is fine - PROBABLE INVALID; QUERIED + Has patches; needs looking at (LOOKS OK) + FIX 3.9 (commit) 324181 mmap does not handle MAP_32BIT Should at least detect and fail this case + FIX 3.9 (to fail MAP_32BIT; should be easy) 324227 memcheck false positive leak when a thread calls exit+block only reachable via other thread live register Inconclusive, after discussions with Philippe + Severity: low Audience: med 324261 errorcount in xml output remains empty (no error pairs mentioned) Hmm, looks like a real bug. + Severity: med Audience: low 324421 Support for fanotify API on ARM architecture Has simple patch; should apply. - -324472 Documentation incorrect: - valgrind: --track-origins no longer accepted in 3.8.1 - INVALID + FIX 3.9 (trivial patch) 324486 WARNING: unhandled syscall: 310 + == 166355 Hmm, tricky, sys_unshare. + Severity: high Audience: low 324525 valgrind can not detect any leak PROBABLE INVALID; needs closing. - -324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc + Severity: med Audience: low 324809 show nightly regression tests results (and later performance) in a nice html page on valgrind.org In progress - -324816 Incorrect VEX implementation for xscvspdp/xvcvspdp regarding SNaN inputs - Has patch; seems plausible. Carl to review? + Severity: low Audience: low 324834 Unhandled instructions in Microsoft C run-time (msvcrt.dll) for x86_64 - Should be easy to fix + Should be easy to fix; queried + Severity: high Audience: low 324882 CPUID should depend on VEX capabilities, not on underlying CPU Middle-term cleanup; not urgent 324894 Phase 3 support for IBM Power ISA 2.07 Waiting for review (but by who?) + MAYBE 3.9 Thu Sep 19 10:34:49 CEST 2013