From: Julian Seward Date: Tue, 13 May 2008 19:21:52 +0000 (+0000) Subject: Update. X-Git-Tag: svn/VALGRIND_3_4_0~576 X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=386dc0a9d0ef937cda1df82e3f40aea6e7a28925;p=thirdparty%2Fvalgrind.git Update. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@8066 --- diff --git a/docs/internals/3_3_BUGSTATUS.txt b/docs/internals/3_3_BUGSTATUS.txt index 554ce7e66d..34f558665c 100644 --- a/docs/internals/3_3_BUGSTATUS.txt +++ b/docs/internals/3_3_BUGSTATUS.txt @@ -17,6 +17,7 @@ PRI: 33 = fix this for 3.3.1 ### merge fix to 3.3 branch ??? marginal NNN won't fix +Vfd fix has been verified on 3.3.X branch -- Last updated 9 May 08 @@ -32,9 +33,6 @@ and 7324 ? get rid of pointless constraint !(--db-attach=yes && --trace-children=yes) (users, 14 feb) -utimensat syswrap bug (Kegel, users, 24 Feb) -### FIXED r8047 - Consider potential space/time effects of r7383 (see reply to the commit message). @@ -60,35 +58,10 @@ disInstr(ppc): unhandled instruction: 0x7C1F42A6 --- start 27 March 08 --- -92456: Tracing the origin of uninitialised memory - FIXED - -126389 (fxrstr) vex x86->IR: unhandled instruction bytes: -### 0xF 0xAE 0x4D 0xF0 - == 158525 Valgrind doesn't recognise fxrstor instruction - FIXED trunk r1836, (test) r1837 - MERGE - 149878 add (proper) check for calloc integer overflow *** (possibly with a viable patch now) FIXME -152818 (repz lodsb) vex x86->IR: unhandled instruction bytes: -### 0xF3 0xAC 0xFC 0x9C - FIXED trunk r1838 (+ test case bug152818_2.c ?) - -153196 (repnz cmpsb) vex x86->IR: unhandled instruction bytes: 0xF2 -### 0xA6 0x89 0x45 - FIXED trunk r1834 - -155528 valgrind cupinfo reports a CPU, which does not reflect -### valgrind's instruction set. - Core2/ssse3 support in the trunk: - vx1808, vx1809, vx1810, - r7384, r7385, r7387, r7389. - (in essence #155528) - FIXED, MERGE - 156065 VALGRIND_MEMPOOL_ALLOC vs. zero-size allocs triggers false *** "overlapping" error FIXME @@ -97,21 +70,9 @@ disInstr(ppc): unhandled instruction: 0x7C1F42A6 ??? openoffice on suse 10.3 readlink stack problem POTENTIALLY SERIOUS (need to extend root stack before syscall) -157748 valgrind seems to not recognize what would be sse2 -### instructions - FIXED vx1835, MERGE - 158042 --db-attach at invalid free() gives broken stack trace on x86_64 -158212 Locks taken by pthread_rwlock_tryrdlock and -### pthread_rwlock_trywrlock not recognised by helgrind - r7454: Intercept pthread_rwlock_try{rd,wr}lock. - MERGE - -158425 sys_poll incorrectly emulated when RES==0 -#### FIXED r8049 - 158730 ppc: illegal opcode crash on instruction fctiwz ??? Not sure this does have anything to do with fctiwz. VEX generates code that doesn't run on e500 @@ -143,27 +104,14 @@ NNN (no info so far, probably a threading bug in client) 160872 race errors reported where libpthread.so ??? (might be a problem with pthread_setspecific) -160907 Support for a couple of recent Linux syscalls, syscall wrapper -### bugfixes - FIXED 7917,8051 160954 ppc build of valgrind crashes with illegal instruction *** 0x7D6B4F9E == isel r11,r11,r9,30 == 161251 -160956 PATCH: mallinfo implementation -### FIXED r7888 - -161285 Patch -- support for eventfd() syscall -### 7917 - 161374 SYS_readdir syscall not implemented ??? readdir is pretty prehistoric -161378 illegal opcode in debug libm - == 160136 -### FIXED vx1826 - 161487 number of suppressions files is limited to 10 *** Patch exists, or a very lazy fix @@ -173,22 +121,20 @@ NNN (no info so far, probably a threading bug in client) --- end 8 May 08 --- -### r1816 (x86 CondNS after SUBB) -### r1817 (amd64 CondNS after SUBB) -### r1833 (both, CondNZ after SUBW) - -*** !!ppc32-linux memcheck craps out on some Altivec code! - (none/tests/ppc32/jm-insns -a -v) - (I think this is fixed in trunk now; check) - -*** !!massif segfaults on ooo on amd64 - 23-deep recursive calls to pp_snapshot_SXPt +r1816 (x86 CondNS after SUBB) +r1817 (amd64 CondNS after SUBB) +r1833 (both, CondNZ after SUBW) +merged; 1847 r7795 (handle r-x mapped .bss sections - merge to trunk) TRUNK 33BRANCH PRI BUG# WHAT +r8058 r8059 Vfd n-i-bz Massif segfaults at exit + +vx1832 vx1839 Vfd n-i-bz Memcheck asserts on Altivec code + r7299 pending 33 n-i-bz stupid sizeof bug in Helgrind r7322 pending 33 n-i-bz check fd on sys_llseek r7323 pending 33 n-i-bz update syscall lists to kernel 2.6.23.1 @@ -199,17 +145,57 @@ r7334 pending 33 n-i-bz handle sys_sysinfo, sys_getresuid, r7350 pending 33 n-i-bz intercept memcpy in 64-bit ld.so's -vx1807 pending 33 155011 vex x86->IR: 0xCF (iret) +r8047 pending 33 n-i-bz Fix wrappers for sys_{futimesat,utimensat} + +many wontfix Vfd 92456 Tracing the origin of uninitialised memory + +vx1836/7 vx1845 Vfd 126389 vex x86->IR: 0xF 0xAE (FXRSTOR) + == 158525 + +vx1838 vx1846 Vfd 152818 vex x86->IR: 0xF3 0xAC (repz lodsb) + +vx1834 vx1843 Vfd 153196 vex x86->IR: 0xF2 0xA6 (repnz cmpsb) + +vx1807 vx1840 Mgd 155011 vex x86->IR: 0xCF (iret) + r7347 pending 33 155091 Warning [...] unhandled DW_OP_ opcode 0x23 (patch proposed, waiting for feedback) ==156960 -r7355 r7356 33 155929 ms_print fails on massif outputs - containing long lines +vx1808/9/10,r7384/5/7/9 + vx1841,r8056 + Vfd 155528 support Core2/SSSE3 insns on x86/amd64 -r7361 r7362 33 n-i-bz ms_print broken for --time-unit=ms +r7355 r7356 Mgd 155929 ms_print fails on massif outputs + containing long lines r7399 pending 33 157665 valgrind fails on shmdt(0) after shmat to 0 +vx1835 vx1844 Vfd 157748 support x86 PUSHFW/POPFW + +r7454 pending 33 158212 hg: handle pthread_rwlock_try{rd,wr}lock. + +r8049 pending 33 158425 sys_poll incorrectly emulated when RES==0 + +r7917,8051 + pending 33 160907 Support for a couple of recent Linux syscalls + +r7888 pending 33 160956 PATCH: mallinfo implementation + +r7917 pending 33 161285 Patch -- support for eventfd() syscall + +vx1826 vx1842 Vfd 161378 illegal opcode in debug libm (FUCOMPP) + == 160136 + + + + +r7361 r7362 Mgd n-i-bz ms_print broken for --time-unit=ms + + + +r7755 pending ?? n-i-bz exp-drd: fixed assertion failure upon delete[] + of an array of size zero. + r7895 pending 33 161036 exp-drd: client allocated memory was never freed