philippe | 2d75ea2 | 2012-08-16 19:40:52 +0000 | [diff] [blame] | 1 | Release 3.9.0 (?? ?????? 201?) |
| 2 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 3 | |
| 4 | * ================== PLATFORM CHANGES ================= |
| 5 | |
| 6 | * ==================== TOOL CHANGES ==================== |
| 7 | |
| 8 | * ==================== OTHER CHANGES ==================== |
| 9 | |
| 10 | * ==================== FIXED BUGS ==================== |
| 11 | |
| 12 | The following bugs have been fixed or resolved. Note that "n-i-bz" |
| 13 | stands for "not in bugzilla" -- that is, a bug that was reported to us |
| 14 | but never got a bugzilla entry. We encourage you to file bugs in |
| 15 | bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather |
| 16 | than mailing the developers (or mailing lists) directly -- bugs that |
| 17 | are not entered into bugzilla tend to get forgotten about or ignored. |
| 18 | |
| 19 | To see details of a given bug, visit |
| 20 | https://bugs.kde.org/show_bug.cgi?id=XXXXXX |
| 21 | where XXXXXX is the bug number as listed below. |
| 22 | |
sewardj | 0d5251e | 2012-09-03 07:24:30 +0000 | [diff] [blame] | 23 | v = verified fixed in 3_8_BRANCH |
| 24 | m = merged into 3_8_BRANCH |
sewardj | 374c0dd | 2012-08-24 00:06:17 +0000 | [diff] [blame] | 25 | |
sewardj | 0d5251e | 2012-09-03 07:24:30 +0000 | [diff] [blame] | 26 | 219156 [380] handle statically linked malloc and/or other malloc libs |
florian | 0950120 | 2012-09-15 19:31:07 +0000 | [diff] [blame] | 27 | 252955 [390] Impossible to compile with ccache |
sewardj | 0d5251e | 2012-09-03 07:24:30 +0000 | [diff] [blame] | 28 | 254088 [380] Valgrind should know about UD2 instruction |
| 29 | 274695 [390] s390x: Support "compare to/from logical" instructions (z196) |
| 30 | 275800 [390] s390x: Add support for the ecag instruction (part 1) |
| 31 | 284004 mv [381] == 301281 |
| 32 | 289584 m [381] Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS) |
| 33 | 295808 m [381] amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT) |
| 34 | 298281 mv [381] wcslen causes false(?) uninitialised value warnings |
| 35 | 301280 [380] == 254088 |
| 36 | 301281 mv [381] valgrind hangs on OS X when the process calls system() |
| 37 | 301902 [380] == 254088 |
| 38 | 304346 [380] == 219156 |
| 39 | 304754 [380] NEWS blows TeX's little mind |
| 40 | 304867 m [381] implement MOVBE instruction in x86 mode |
| 41 | 304980 m [381] Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping |
| 42 | 305042 m [381] amd64: implement 0F 7F encoding of movq between two registers |
| 43 | 305199 m [381] ARM: implement QDADD and QDSUB |
| 44 | 305321 m [381] amd64->IR: 0xF 0xD 0xC (prefetchw) |
| 45 | 305513 m [381] killed by fatal signal: SIGSEGV |
| 46 | 305690 m [381] DRD reporting invalid semaphore when sem_trywait returns |
| 47 | EAGAIN or sem_timedwait returns ETIMEDOUT |
| 48 | 305926 m [381] Invalid alignment checks for some AVX instructions |
florian | f74f542 | 2012-09-13 19:41:12 +0000 | [diff] [blame] | 49 | 305948 [390] ppc64: code generation for ShlD64 / ShrD64 asserts |
florian | f9abc02 | 2012-09-06 03:26:50 +0000 | [diff] [blame] | 50 | 306054 [390] s390x: Condition code computation for convert-to-int/logical |
philippe | 7d9b41b | 2012-09-11 19:53:01 +0000 | [diff] [blame] | 51 | 306310 [390] 3.8.0 release tarball missing some files |
sewardj | 0d5251e | 2012-09-03 07:24:30 +0000 | [diff] [blame] | 52 | n-i-bz m [381] shmat of a segment > 4Gb does not work |
| 53 | n-i-bz m [381] simulate_control_c script wrong USR1 sig# on mips |
| 54 | n-i-bz m [381] vgdb ptrace calls wrong on mips [...] |
| 55 | n-i-bz m [381] Fixes for more MPI false positives |
| 56 | n-i-bz m [381] exp-sgcheck's memcpy causes programs to segfault |
| 57 | n-i-bz m [381] volatile in stack check, re clang |
| 58 | n-i-bz m [381] Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB |
sewardj | 5036a7a | 2012-08-24 16:53:02 +0000 | [diff] [blame] | 59 | |
florian | ff7efa6 | 2011-11-16 03:59:49 +0000 | [diff] [blame] | 60 | |
sewardj | 1c41a1a | 2012-08-09 15:06:36 +0000 | [diff] [blame] | 61 | Release 3.8.0 (10 August 2012) |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 62 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
sewardj | 8ab8a8e | 2012-08-07 14:54:17 +0000 | [diff] [blame] | 63 | 3.8.0 is a feature release with many improvements and the usual |
| 64 | collection of bug fixes. |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 65 | |
| 66 | This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, |
| 67 | PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android, |
| 68 | X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent |
| 69 | distros and toolchain components (glibc 2.16, gcc 4.7) has been added. |
| 70 | There is initial support for MacOSX 10.8, but it is not usable for |
| 71 | serious work at present. |
sewardj | 7d79e0b | 2012-05-21 21:57:05 +0000 | [diff] [blame] | 72 | |
florian | 8dc79ce | 2011-12-10 16:00:25 +0000 | [diff] [blame] | 73 | * ================== PLATFORM CHANGES ================= |
| 74 | |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 75 | * Support for MIPS32 platforms running Linux. Valgrind has been |
| 76 | tested on MIPS32 and MIPS32r2 platforms running different Debian |
| 77 | Squeeze and MeeGo distributions. Both little-endian and big-endian |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 78 | cores are supported. The tools Memcheck, Massif and Lackey have |
| 79 | been tested and are known to work. See README.mips for more details. |
| 80 | |
sewardj | 8ab8a8e | 2012-08-07 14:54:17 +0000 | [diff] [blame] | 81 | * Preliminary support for Android running on x86. |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 82 | |
| 83 | * Preliminary (as-yet largely unusable) support for MacOSX 10.8. |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 84 | |
| 85 | * Support for Intel AVX instructions and for AES instructions. This |
| 86 | support is available only for 64 bit code. |
| 87 | |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 88 | * Support for POWER Decimal Floating Point instructions. |
philippe | 5d5dd8e | 2012-08-05 00:08:25 +0000 | [diff] [blame] | 89 | |
florian | 8dc79ce | 2011-12-10 16:00:25 +0000 | [diff] [blame] | 90 | * ==================== TOOL CHANGES ==================== |
| 91 | |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 92 | * Non-libc malloc implementations are now supported. This is useful |
| 93 | for tools that replace malloc (Memcheck, Massif, DRD, Helgrind). |
| 94 | Using the new option --soname-synonyms, such tools can be informed |
| 95 | that the malloc implementation is either linked statically into the |
| 96 | executable, or is present in some other shared library different |
| 97 | from libc.so. This makes it possible to process statically linked |
| 98 | programs, and programs using other malloc libraries, for example |
| 99 | TCMalloc or JEMalloc. |
| 100 | |
| 101 | * For tools that provide their own replacement for malloc et al, the |
| 102 | option --redzone-size=<number> allows users to specify the size of |
| 103 | the padding blocks (redzones) added before and after each client |
| 104 | allocated block. Smaller redzones decrease the memory needed by |
| 105 | Valgrind. Bigger redzones increase the chance to detect blocks |
| 106 | overrun or underrun. Prior to this change, the redzone size was |
| 107 | hardwired to 16 bytes in Memcheck. |
philippe | 1e470b5 | 2012-05-11 19:33:46 +0000 | [diff] [blame] | 108 | |
philippe | 8423490 | 2012-01-14 13:53:13 +0000 | [diff] [blame] | 109 | * Memcheck: |
| 110 | |
| 111 | - The leak_check GDB server monitor command now can |
| 112 | control the maximum nr of loss records to output. |
| 113 | |
philippe | 1e470b5 | 2012-05-11 19:33:46 +0000 | [diff] [blame] | 114 | - Reduction of memory use for applications allocating |
philippe | 6643e96 | 2012-01-17 21:16:30 +0000 | [diff] [blame] | 115 | many blocks and/or having many partially defined bytes. |
| 116 | |
philippe | a22f59d | 2012-01-26 23:13:52 +0000 | [diff] [blame] | 117 | - Addition of GDB server monitor command 'block_list' that lists |
| 118 | the addresses/sizes of the blocks of a leak search loss record. |
| 119 | |
| 120 | - Addition of GDB server monitor command 'who_points_at' that lists |
| 121 | the locations pointing at a block. |
philippe | 6643e96 | 2012-01-17 21:16:30 +0000 | [diff] [blame] | 122 | |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 123 | - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will |
| 124 | detect an invalid access of these redzones, by marking them |
| 125 | noaccess. Similarly, if a redzone size is given for a memory |
| 126 | pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access. |
| 127 | This still allows to find some bugs if the user has forgotten to |
| 128 | mark the pool superblock noaccess. |
philippe | 11cc9ce | 2012-03-11 17:59:00 +0000 | [diff] [blame] | 129 | |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 130 | - Performance of memory leak check has been improved, especially in |
| 131 | cases where there are many leaked blocks and/or many suppression |
| 132 | rules used to suppress leak reports. |
philippe | 57a16a2 | 2012-07-18 22:26:51 +0000 | [diff] [blame] | 133 | |
sewardj | e060f6a | 2012-08-05 16:14:02 +0000 | [diff] [blame] | 134 | - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to |
philippe | 4621729 | 2012-08-05 21:35:02 +0000 | [diff] [blame] | 135 | more precise analysis, which is important for LLVM/Clang |
sewardj | e060f6a | 2012-08-05 16:14:02 +0000 | [diff] [blame] | 136 | generated code. This is at the cost of somewhat reduced |
| 137 | performance. Note there is no change to analysis precision or |
| 138 | costs on Linux targets. |
| 139 | |
bart | c3009f8 | 2012-06-16 19:02:25 +0000 | [diff] [blame] | 140 | * DRD: |
| 141 | |
bart | c3009f8 | 2012-06-16 19:02:25 +0000 | [diff] [blame] | 142 | - Added even more facilities that can help finding the cause of a data |
| 143 | race, namely the command-line option --ptrace-addr and the macro |
| 144 | DRD_STOP_TRACING_VAR(x). More information can be found in the manual. |
| 145 | |
bart | a7a5788 | 2012-08-04 09:19:11 +0000 | [diff] [blame] | 146 | - Fixed a subtle bug that could cause false positive data race reports. |
| 147 | |
florian | 8dc79ce | 2011-12-10 16:00:25 +0000 | [diff] [blame] | 148 | * ==================== OTHER CHANGES ==================== |
| 149 | |
| 150 | * The C++ demangler has been updated so as to work well with C++ |
philippe | 172a3bb | 2012-03-16 15:03:08 +0000 | [diff] [blame] | 151 | compiled by up to at least g++ 4.6. |
florian | ff7efa6 | 2011-11-16 03:59:49 +0000 | [diff] [blame] | 152 | |
bart | 1e45164 | 2012-07-28 13:06:45 +0000 | [diff] [blame] | 153 | * Tool developers can make replacement/wrapping more flexible thanks |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 154 | to the new option --soname-synonyms. This was reported above, but |
| 155 | in fact is very general and applies to all function |
| 156 | replacement/wrapping, not just to malloc-family functions. |
philippe | 1e470b5 | 2012-05-11 19:33:46 +0000 | [diff] [blame] | 157 | |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 158 | * Round-robin scheduling of threads can be selected, using the new |
| 159 | option --fair-sched= yes. Prior to this change, the pipe-based |
| 160 | thread serialisation mechanism (which is still the default) could |
| 161 | give very unfair scheduling. --fair-sched=yes improves |
| 162 | responsiveness of interactive multithreaded applications, and |
| 163 | improves repeatability of results from the thread checkers Helgrind |
| 164 | and DRD. |
philippe | 236a71a | 2012-02-22 20:23:29 +0000 | [diff] [blame] | 165 | |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 166 | * For tool developers: support to run Valgrind on Valgrind has been |
| 167 | improved. We can now routinely Valgrind on Helgrind or Memcheck. |
philippe | 72faf10 | 2012-03-11 22:24:03 +0000 | [diff] [blame] | 168 | |
philippe | ede2d6c | 2012-03-14 21:27:35 +0000 | [diff] [blame] | 169 | * gdbserver now shows the float shadow registers as integer |
| 170 | rather than float values, as the shadow values are mostly |
| 171 | used as bit patterns. |
| 172 | |
florian | 7711f9e | 2012-06-29 21:20:52 +0000 | [diff] [blame] | 173 | * Increased limit for the --num-callers command line flag to 500. |
| 174 | |
philippe | 02ccd56 | 2012-08-05 17:35:17 +0000 | [diff] [blame] | 175 | * Performance improvements for error matching when there are many |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 176 | suppression records in use. |
philippe | cbfe62c | 2012-08-04 09:08:50 +0000 | [diff] [blame] | 177 | |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 178 | * Improved support for DWARF4 debugging information (bug 284184). |
sewardj | d09354b | 2012-06-30 16:44:45 +0000 | [diff] [blame] | 179 | |
sewardj | a930ef2 | 2012-08-08 20:21:29 +0000 | [diff] [blame] | 180 | * Initial support for DWZ compressed Dwarf debug info. |
| 181 | |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 182 | * Improved control over the IR optimiser's handling of the tradeoff |
| 183 | between performance and precision of exceptions. Specifically, |
| 184 | --vex-iropt-precise-memory-exns has been removed and replaced by |
| 185 | --vex-iropt-register-updates, with extended functionality. This |
philippe | e30542d | 2012-08-05 17:37:39 +0000 | [diff] [blame] | 186 | allows the Valgrind gdbserver to always show up to date register |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 187 | values to GDB. |
| 188 | |
| 189 | * Modest performance gains through the use of translation chaining for |
| 190 | JIT-generated code. |
philippe | 0c0291a | 2012-08-01 22:03:12 +0000 | [diff] [blame] | 191 | |
florian | ff7efa6 | 2011-11-16 03:59:49 +0000 | [diff] [blame] | 192 | * ==================== FIXED BUGS ==================== |
| 193 | |
| 194 | The following bugs have been fixed or resolved. Note that "n-i-bz" |
| 195 | stands for "not in bugzilla" -- that is, a bug that was reported to us |
| 196 | but never got a bugzilla entry. We encourage you to file bugs in |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 197 | bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather |
| 198 | than mailing the developers (or mailing lists) directly -- bugs that |
| 199 | are not entered into bugzilla tend to get forgotten about or ignored. |
florian | ff7efa6 | 2011-11-16 03:59:49 +0000 | [diff] [blame] | 200 | |
| 201 | To see details of a given bug, visit |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 202 | https://bugs.kde.org/show_bug.cgi?id=XXXXXX |
florian | ff7efa6 | 2011-11-16 03:59:49 +0000 | [diff] [blame] | 203 | where XXXXXX is the bug number as listed below. |
| 204 | |
florian | ef0bca6 | 2012-05-06 03:37:25 +0000 | [diff] [blame] | 205 | 197914 Building valgrind from svn now requires automake-1.10 |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 206 | 203877 increase to 16Mb maximum allowed alignment for memalign et al |
| 207 | 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc) |
philippe | 8f927a5 | 2011-12-23 11:42:20 +0000 | [diff] [blame] | 208 | 247386 make perf does not run all performance tests |
philippe | 8d67425 | 2012-03-20 19:04:39 +0000 | [diff] [blame] | 209 | 270006 Valgrind scheduler unfair |
philippe | b79add8 | 2012-07-01 20:35:26 +0000 | [diff] [blame] | 210 | 270777 Adding MIPS/Linux port to Valgrind |
florian | 722b48e | 2012-01-16 17:26:48 +0000 | [diff] [blame] | 211 | 270796 s390x: Removed broken support for the TS insn |
florian | fdfca22 | 2012-01-17 13:16:50 +0000 | [diff] [blame] | 212 | 271438 Fix configure for proper SSE4.2 detection |
florian | da27543 | 2012-02-04 17:16:40 +0000 | [diff] [blame] | 213 | 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 214 | 273475 Add support for AVX instructions |
philippe | f7c5433 | 2012-05-17 15:32:54 +0000 | [diff] [blame] | 215 | 274078 improved configure logic for mpicc |
philippe | 15e301e | 2011-12-22 13:25:58 +0000 | [diff] [blame] | 216 | 276993 fix mremap 'no thrash checks' |
sewardj | 5d1d88f | 2012-07-16 22:39:24 +0000 | [diff] [blame] | 217 | 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 218 | 281482 memcheck incorrect byte allocation count in realloc() for silly argument |
philippe | 6643e96 | 2012-01-17 21:16:30 +0000 | [diff] [blame] | 219 | 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit |
florian | 8dc79ce | 2011-12-10 16:00:25 +0000 | [diff] [blame] | 220 | 283413 Fix wrong sanity check |
florian | 535bc7d | 2012-02-26 17:01:22 +0000 | [diff] [blame] | 221 | 283671 Robustize alignment computation in LibVEX_Alloc |
sewardj | d09354b | 2012-06-30 16:44:45 +0000 | [diff] [blame] | 222 | 283961 Adding support for some HCI IOCTLs |
| 223 | 284124 parse_type_DIE: confused by: DWARF 4 |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 224 | 284864 == 273475 (Add support for AVX instructions) |
sewardj | d09354b | 2012-06-30 16:44:45 +0000 | [diff] [blame] | 225 | 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register" |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 226 | 285662 (MacOSX): Memcheck needs to replace memcpy/memmove |
| 227 | 285725 == 273475 (Add support for AVX instructions) |
| 228 | 286261 add wrapper for linux I2C_RDWR ioctl |
philippe | 5eb378e | 2011-12-19 11:03:07 +0000 | [diff] [blame] | 229 | 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings |
| 230 | 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 231 | 286384 configure fails "checking for a supported version of gcc" |
| 232 | 286497 == 273475 (Add support for AVX instructions) |
| 233 | 286596 == 273475 (Add support for AVX instructions) |
sewardj | 7ceb70c | 2012-07-11 14:28:13 +0000 | [diff] [blame] | 234 | 286917 disInstr(arm): unhandled instruction: QADD (also QSUB) |
sewardj | 69f799d | 2012-08-04 08:41:44 +0000 | [diff] [blame] | 235 | 287175 ARM: scalar VFP fixed-point VCVT instructions not handled |
florian | e3c0a2a | 2012-01-29 02:23:07 +0000 | [diff] [blame] | 236 | 287260 Incorrect conditional jump or move depends on uninitialised value(s) |
sewardj | d09354b | 2012-06-30 16:44:45 +0000 | [diff] [blame] | 237 | 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW) |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 238 | 287307 == 273475 (Add support for AVX instructions) |
florian | 5945164 | 2011-12-24 21:50:53 +0000 | [diff] [blame] | 239 | 287858 VG_(strerror): unknown error |
sewardj | 69f799d | 2012-08-04 08:41:44 +0000 | [diff] [blame] | 240 | 288298 (MacOSX) unhandled syscall shm_unlink |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 241 | 288995 == 273475 (Add support for AVX instructions) |
sewardj | a1c8b4c | 2012-07-13 14:04:03 +0000 | [diff] [blame] | 242 | 289470 Loading of large Mach-O thin binaries fails. |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 243 | 289656 == 273475 (Add support for AVX instructions) |
philippe | 1d76a80 | 2011-12-26 21:21:37 +0000 | [diff] [blame] | 244 | 289699 vgdb connection in relay mode erroneously closed due to buffer overrun |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 245 | 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters) |
florian | 6746c4c | 2012-08-06 00:10:53 +0000 | [diff] [blame] | 246 | 289839 s390x: Provide support for unicode conversion instructions |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 247 | 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks |
sewardj | 617f1b2 | 2012-07-11 00:06:35 +0000 | [diff] [blame] | 248 | 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm" |
philippe | 6156b77 | 2012-02-14 21:35:46 +0000 | [diff] [blame] | 249 | 290655 Add support for AESKEYGENASSIST instruction |
philippe | faa4f05 | 2012-07-01 20:06:04 +0000 | [diff] [blame] | 250 | 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage |
philippe | 53b0d9a | 2012-02-02 21:33:55 +0000 | [diff] [blame] | 251 | 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM |
philippe | faa4f05 | 2012-07-01 20:06:04 +0000 | [diff] [blame] | 252 | 291253 ES register not initialised in valgrind simulation |
| 253 | 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch) |
florian | cc58cef | 2012-06-06 02:27:51 +0000 | [diff] [blame] | 254 | 291865 s390x: Support the "Compare Double and Swap" family of instructions |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 255 | 292300 == 273475 (Add support for AVX instructions) |
| 256 | 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid |
| 257 | 292493 == 273475 (Add support for AVX instructions) |
philippe | faa4f05 | 2012-07-01 20:06:04 +0000 | [diff] [blame] | 258 | 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support |
| 259 | 292627 Missing support for some SCSI ioctls |
| 260 | 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 261 | 292841 == 273475 (Add support for AVX instructions) |
philippe | faa4f05 | 2012-07-01 20:06:04 +0000 | [diff] [blame] | 262 | 292993 implement the getcpu syscall on amd64-linux |
| 263 | 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2 |
florian | 4458779 | 2012-02-15 04:05:52 +0000 | [diff] [blame] | 264 | 293088 Add some VEX sanity checks for ppc64 unhandled instructions |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 265 | 293751 == 290655 (Add support for AESKEYGENASSIST instruction) |
philippe | faa4f05 | 2012-07-01 20:06:04 +0000 | [diff] [blame] | 266 | 293754 PCMPxSTRx not implemented for 16-bit characters |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 267 | 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters) |
philippe | faa4f05 | 2012-07-01 20:06:04 +0000 | [diff] [blame] | 268 | 293808 CLFLUSH not supported by latest VEX for amd64 |
| 269 | 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...) |
| 270 | 294048 MPSADBW instruction not implemented |
florian | c6fccd6 | 2012-02-15 03:57:27 +0000 | [diff] [blame] | 271 | 294055 regtest none/tests/shell fails when locale is not set to C |
philippe | faa4f05 | 2012-07-01 20:06:04 +0000 | [diff] [blame] | 272 | 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM |
philippe | baf6964 | 2012-02-15 22:29:30 +0000 | [diff] [blame] | 273 | 294190 --vgdb-error=xxx can be out of sync with errors shown to the user |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 274 | 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions |
| 275 | 294260 disInstr_AMD64: disInstr miscalculated next %rip |
sewardj | 8a91f38 | 2012-07-05 13:26:07 +0000 | [diff] [blame] | 276 | 294523 --partial-loads-ok=yes causes false negatives |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 277 | 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A |
| 278 | 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83 |
philippe | faa4f05 | 2012-07-01 20:06:04 +0000 | [diff] [blame] | 279 | 294812 patch allowing to run (on x86 at least) helgrind/drd on tool. |
| 280 | 295089 can not annotate source for both helgrind and drd |
florian | 2011f75 | 2012-07-01 21:55:13 +0000 | [diff] [blame] | 281 | 295221 POWER Processor decimal floating point instruction support missing |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 282 | 295427 building for i386 with clang on darwin11 requires "-new_linker linker" |
florian | 2011f75 | 2012-07-01 21:55:13 +0000 | [diff] [blame] | 283 | 295428 coregrind/m_main.c has incorrect x86 assembly for darwin |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 284 | 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed |
tom | d0b8b37 | 2012-07-11 15:42:25 +0000 | [diff] [blame] | 285 | 295617 ARM - Add some missing syscalls |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 286 | 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...] |
sewardj | a1c8b4c | 2012-07-13 14:04:03 +0000 | [diff] [blame] | 287 | 296229 Linux user input device ioctls missing wrappers |
| 288 | 296318 ELF Debug info improvements (more than one rx/rw mapping) |
philippe | faa4f05 | 2012-07-01 20:06:04 +0000 | [diff] [blame] | 289 | 296422 Add translation chaining support |
sewardj | 4a2faf1 | 2012-07-04 20:30:47 +0000 | [diff] [blame] | 290 | 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES) |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 291 | 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper |
florian | 5bdb9ec | 2012-05-06 03:51:00 +0000 | [diff] [blame] | 292 | 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 293 | 297078 gdbserver signal handling problems [..] |
bart | c3009f8 | 2012-06-16 19:02:25 +0000 | [diff] [blame] | 294 | 297147 drd false positives on newly allocated memory |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 295 | 297329 disallow decoding of IBM Power DFP insns on some machines |
florian | f5a3937 | 2012-07-02 12:14:00 +0000 | [diff] [blame] | 296 | 297497 POWER Processor decimal floating point instruction support missing |
sewardj | a1c8b4c | 2012-07-13 14:04:03 +0000 | [diff] [blame] | 297 | 297701 Another alias for strncasecmp_l in libc-2.13.so |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 298 | 297911 'invalid write' not reported when using APIs for custom mem allocators. |
florian | 295bb7f | 2012-07-01 21:24:53 +0000 | [diff] [blame] | 299 | 297976 s390x: revisit EX implementation |
philippe | 8050bb7 | 2012-04-13 23:07:29 +0000 | [diff] [blame] | 300 | 297991 Valgrind interferes with mmap()+ftell() |
philippe | 8fb1b13 | 2012-04-13 17:27:40 +0000 | [diff] [blame] | 301 | 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux) |
florian | f71b35a | 2012-04-22 19:27:44 +0000 | [diff] [blame] | 302 | 297993 Fix compilation of valgrind with gcc -g3. |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 303 | 298080 POWER Processor DFP support missing, part 3 |
| 304 | 298227 == 273475 (Add support for AVX instructions) |
| 305 | 298335 == 273475 (Add support for AVX instructions) |
florian | f5a3937 | 2012-07-02 12:14:00 +0000 | [diff] [blame] | 306 | 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq) |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 307 | 298394 s390x: Don't bail out on an unknown machine model. [..] |
florian | f5a3937 | 2012-07-02 12:14:00 +0000 | [diff] [blame] | 308 | 298421 accept4() syscall (366) support is missing for ARM |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 309 | 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45 |
florian | f5a3937 | 2012-07-02 12:14:00 +0000 | [diff] [blame] | 310 | 298732 valgrind installation problem in ubuntu with kernel version 3.x |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 311 | 298862 POWER Processor DFP instruction support missing, part 4 |
florian | f5a3937 | 2012-07-02 12:14:00 +0000 | [diff] [blame] | 312 | 298864 DWARF reader mis-parses DW_FORM_ref_addr |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 313 | 298943 massif asserts with --pages-as-heap=yes when brk is changing [..] |
florian | f5a3937 | 2012-07-02 12:14:00 +0000 | [diff] [blame] | 314 | 299053 Support DWARF4 DW_AT_high_pc constant form |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 315 | 299104 == 273475 (Add support for AVX instructions) |
florian | d7ebe55 | 2012-07-02 15:03:29 +0000 | [diff] [blame] | 316 | 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed. |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 317 | 299629 dup3() syscall (358) support is missing for ARM |
| 318 | 299694 POWER Processor DFP instruction support missing, part 5 |
| 319 | 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests |
| 320 | 299803 == 273475 (Add support for AVX instructions) |
| 321 | 299804 == 273475 (Add support for AVX instructions) |
| 322 | 299805 == 273475 (Add support for AVX instructions) |
sewardj | a1c8b4c | 2012-07-13 14:04:03 +0000 | [diff] [blame] | 323 | 300140 ARM - Missing (T1) SMMUL |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 324 | 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping)) |
| 325 | 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed. |
florian | d7ebe55 | 2012-07-02 15:03:29 +0000 | [diff] [blame] | 326 | 300414 FCOM and FCOMP unimplemented for amd64 guest |
sewardj | 8a91f38 | 2012-07-05 13:26:07 +0000 | [diff] [blame] | 327 | 301204 infinite loop in canonicaliseSymtab with ifunc symbol |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 328 | 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc) |
philippe | 5d5dd8e | 2012-08-05 00:08:25 +0000 | [diff] [blame] | 329 | 301265 add x86 support to Android build |
philippe | faa4f05 | 2012-07-01 20:06:04 +0000 | [diff] [blame] | 330 | 301984 configure script doesn't detect certain versions of clang |
| 331 | 302205 Fix compiler warnings for POWER VEX code and POWER test cases |
sewardj | 5d1d88f | 2012-07-16 22:39:24 +0000 | [diff] [blame] | 332 | 302287 Unhandled movbe instruction on Atom processors |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 333 | 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result |
florian | d7ebe55 | 2012-07-02 15:03:29 +0000 | [diff] [blame] | 334 | 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0. |
sewardj | 5d1d88f | 2012-07-16 22:39:24 +0000 | [diff] [blame] | 335 | 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 336 | 302656 == 273475 (Add support for AVX instructions) |
| 337 | 302709 valgrind for ARM needs extra tls support for android emulator [..] |
tom | d0b8b37 | 2012-07-11 15:42:25 +0000 | [diff] [blame] | 338 | 302827 add wrapper for CDROM_GET_CAPABILITY |
sewardj | f745041 | 2012-07-14 10:08:40 +0000 | [diff] [blame] | 339 | 302901 Valgrind crashes with dwz optimized debuginfo |
philippe | b0e5b69 | 2012-07-06 21:42:14 +0000 | [diff] [blame] | 340 | 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite |
philippe | a6be8b1 | 2012-07-06 21:59:03 +0000 | [diff] [blame] | 341 | 303116 Add support for the POWER instruction popcntb |
philippe | 32b4a42 | 2012-07-06 22:52:09 +0000 | [diff] [blame] | 342 | 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions. |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 343 | 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code |
| 344 | 303466 == 273475 (Add support for AVX instructions) |
| 345 | 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP) |
weidendo | 0ffe10e | 2012-07-25 12:13:29 +0000 | [diff] [blame] | 346 | 303963 strstr() function produces wrong results under valgrind callgrind |
tom | 0ec79a1 | 2012-08-02 09:26:38 +0000 | [diff] [blame] | 347 | 304054 CALL_FN_xx macros need to enforce stack alignment |
sewardj | a930ef2 | 2012-08-08 20:21:29 +0000 | [diff] [blame] | 348 | 304561 tee system call not supported |
sewardj | 8ab8a8e | 2012-08-07 14:54:17 +0000 | [diff] [blame] | 349 | 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#) |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 350 | n-i-bz Add missing gdbserver xml files for shadow registers for ppc32 |
| 351 | n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts |
| 352 | n-i-bz Fix assert in gdbserver for watchpoints watching the same address |
| 353 | n-i-bz Fix false positive in sys_clone on amd64 when optional args [..] |
| 354 | n-i-bz s390x: Shadow registers can now be examined using vgdb |
| 355 | |
sewardj | 6e374f8 | 2012-08-08 23:35:35 +0000 | [diff] [blame] | 356 | (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865) |
sewardj | 1c41a1a | 2012-08-09 15:06:36 +0000 | [diff] [blame] | 357 | (3.8.0: 10 August 2012, vex r2465, valgrind r12866) |
sewardj | 10856a0 | 2012-08-05 15:24:12 +0000 | [diff] [blame] | 358 | |
sewardj | 31e8f54 | 2012-08-05 14:57:38 +0000 | [diff] [blame] | 359 | |
florian | 295bb7f | 2012-07-01 21:24:53 +0000 | [diff] [blame] | 360 | |
sewardj | 62b0d40 | 2011-11-09 23:42:41 +0000 | [diff] [blame] | 361 | Release 3.7.0 (5 November 2011) |
| 362 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
sewardj | 469b836 | 2011-10-23 10:05:47 +0000 | [diff] [blame] | 363 | 3.7.0 is a feature release with many significant improvements and the |
| 364 | usual collection of bug fixes. |
sewardj | dc873c0 | 2011-07-24 16:02:33 +0000 | [diff] [blame] | 365 | |
sewardj | 469b836 | 2011-10-23 10:05:47 +0000 | [diff] [blame] | 366 | This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, |
| 367 | PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin. |
| 368 | Support for recent distros and toolchain components (glibc 2.14, gcc |
| 369 | 4.6, MacOSX 10.7) has been added. |
| 370 | |
| 371 | * ================== PLATFORM CHANGES ================= |
| 372 | |
| 373 | * Support for IBM z/Architecture (s390x) running Linux. Valgrind can |
| 374 | analyse 64-bit programs running on z/Architecture. Most user space |
| 375 | instructions up to and including z10 are supported. Valgrind has |
| 376 | been tested extensively on z9, z10, and z196 machines running SLES |
| 377 | 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are |
| 378 | known to work well. Callgrind, Helgrind, and DRD work reasonably |
| 379 | well on z9 and later models. See README.s390 for more details. |
| 380 | |
| 381 | * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and |
| 382 | 64-bit processes are supported. Some complex threaded applications |
| 383 | (Firefox) are observed to hang when run as 32 bit applications, |
| 384 | whereas 64-bit versions run OK. The cause is unknown. Memcheck |
| 385 | will likely report some false errors. In general, expect some rough |
| 386 | spots. This release also supports MacOSX 10.6, but drops support |
| 387 | for 10.5. |
| 388 | |
| 389 | * Preliminary support for Android (on ARM). Valgrind can now run |
| 390 | large applications (eg, Firefox) on (eg) a Samsung Nexus S. See |
| 391 | README.android for more details, plus instructions on how to get |
| 392 | started. |
| 393 | |
| 394 | * Support for the IBM Power ISA 2.06 (Power7 instructions) |
| 395 | |
| 396 | * General correctness and performance improvements for ARM/Linux, and, |
| 397 | by extension, ARM/Android. |
| 398 | |
sewardj | 3bbaf48 | 2011-10-24 08:54:56 +0000 | [diff] [blame] | 399 | * Further solidification of support for SSE 4.2 in 64-bit mode. AVX |
sewardj | 469b836 | 2011-10-23 10:05:47 +0000 | [diff] [blame] | 400 | instruction set support is under development but is not available in |
| 401 | this release. |
| 402 | |
| 403 | * Support for AIX5 has been removed. |
| 404 | |
| 405 | * ==================== TOOL CHANGES ==================== |
| 406 | |
| 407 | * Memcheck: some incremental changes: |
| 408 | |
| 409 | - reduction of memory use in some circumstances |
| 410 | |
| 411 | - improved handling of freed memory, which in some circumstances |
| 412 | can cause detection of use-after-free that would previously have |
| 413 | been missed |
| 414 | |
| 415 | - fix of a longstanding bug that could cause false negatives (missed |
| 416 | errors) in programs doing vector saturated narrowing instructions. |
| 417 | |
| 418 | * Helgrind: performance improvements and major memory use reductions, |
| 419 | particularly for large, long running applications which perform many |
| 420 | synchronisation (lock, unlock, etc) events. Plus many smaller |
| 421 | changes: |
| 422 | |
| 423 | - display of locksets for both threads involved in a race |
| 424 | |
| 425 | - general improvements in formatting/clarity of error messages |
| 426 | |
| 427 | - addition of facilities and documentation regarding annotation |
| 428 | of thread safe reference counted C++ classes |
| 429 | |
| 430 | - new flag --check-stack-refs=no|yes [yes], to disable race checking |
| 431 | on thread stacks (a performance hack) |
| 432 | |
| 433 | - new flag --free-is-write=no|yes [no], to enable detection of races |
| 434 | where one thread accesses heap memory but another one frees it, |
| 435 | without any coordinating synchronisation event |
| 436 | |
| 437 | * DRD: enabled XML output; added support for delayed thread deletion |
| 438 | in order to detect races that occur close to the end of a thread |
| 439 | (--join-list-vol); fixed a memory leak triggered by repeated client |
| 440 | memory allocatation and deallocation; improved Darwin support. |
| 441 | |
florian | 0f0ed67 | 2011-11-26 04:05:47 +0000 | [diff] [blame] | 442 | * exp-ptrcheck: this tool has been renamed to exp-sgcheck |
| 443 | |
| 444 | * exp-sgcheck: this tool has been reduced in scope so as to improve |
sewardj | 469b836 | 2011-10-23 10:05:47 +0000 | [diff] [blame] | 445 | performance and remove checking that Memcheck does better. |
| 446 | Specifically, the ability to check for overruns for stack and global |
| 447 | arrays is unchanged, but the ability to check for overruns of heap |
| 448 | blocks has been removed. The tool has accordingly been renamed to |
| 449 | exp-sgcheck ("Stack and Global Array Checking"). |
| 450 | |
| 451 | * ==================== OTHER CHANGES ==================== |
| 452 | |
| 453 | * GDB server: Valgrind now has an embedded GDB server. That means it |
| 454 | is possible to control a Valgrind run from GDB, doing all the usual |
| 455 | things that GDB can do (single stepping, breakpoints, examining |
| 456 | data, etc). Tool-specific functionality is also available. For |
| 457 | example, it is possible to query the definedness state of variables |
| 458 | or memory from within GDB when running Memcheck; arbitrarily large |
| 459 | memory watchpoints are supported, etc. To use the GDB server, start |
| 460 | Valgrind with the flag --vgdb-error=0 and follow the on-screen |
| 461 | instructions. |
| 462 | |
| 463 | * Improved support for unfriendly self-modifying code: a new option |
| 464 | --smc-check=all-non-file is available. This adds the relevant |
| 465 | consistency checks only to code that originates in non-file-backed |
| 466 | mappings. In effect this confines the consistency checking only to |
| 467 | code that is or might be JIT generated, and avoids checks on code |
| 468 | that must have been compiled ahead of time. This significantly |
| 469 | improves performance on applications that generate code at run time. |
| 470 | |
| 471 | * It is now possible to build a working Valgrind using Clang-2.9 on |
sewardj | 960eef2 | 2011-09-30 17:55:13 +0000 | [diff] [blame] | 472 | Linux. |
| 473 | |
sewardj | 469b836 | 2011-10-23 10:05:47 +0000 | [diff] [blame] | 474 | * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING. |
| 475 | These enable and disable error reporting on a per-thread, and |
| 476 | nestable, basis. This is useful for hiding errors in particularly |
| 477 | troublesome pieces of code. The MPI wrapper library (libmpiwrap.c) |
| 478 | now uses this facility. |
sewardj | 960eef2 | 2011-09-30 17:55:13 +0000 | [diff] [blame] | 479 | |
sewardj | 469b836 | 2011-10-23 10:05:47 +0000 | [diff] [blame] | 480 | * Added the --mod-funcname option to cg_diff. |
sewardj | dc873c0 | 2011-07-24 16:02:33 +0000 | [diff] [blame] | 481 | |
sewardj | 469b836 | 2011-10-23 10:05:47 +0000 | [diff] [blame] | 482 | * ==================== FIXED BUGS ==================== |
sewardj | 4629a73 | 2011-05-17 15:54:31 +0000 | [diff] [blame] | 483 | |
sewardj | 520501d | 2011-10-23 09:09:54 +0000 | [diff] [blame] | 484 | The following bugs have been fixed or resolved. Note that "n-i-bz" |
| 485 | stands for "not in bugzilla" -- that is, a bug that was reported to us |
| 486 | but never got a bugzilla entry. We encourage you to file bugs in |
| 487 | bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than |
| 488 | mailing the developers (or mailing lists) directly -- bugs that are |
| 489 | not entered into bugzilla tend to get forgotten about or ignored. |
sewardj | ffd37d6 | 2011-10-13 16:41:17 +0000 | [diff] [blame] | 490 | |
sewardj | 520501d | 2011-10-23 09:09:54 +0000 | [diff] [blame] | 491 | To see details of a given bug, visit |
| 492 | https://bugs.kde.org/show_bug.cgi?id=XXXXXX |
| 493 | where XXXXXX is the bug number as listed below. |
sewardj | ffd37d6 | 2011-10-13 16:41:17 +0000 | [diff] [blame] | 494 | |
sewardj | 520501d | 2011-10-23 09:09:54 +0000 | [diff] [blame] | 495 | 210935 port valgrind.h (not valgrind) to win32 to support client requests |
| 496 | 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10 |
sewardj | 7a9ff4c | 2011-06-13 13:36:59 +0000 | [diff] [blame] | 497 | 243404 Port to zSeries |
sewardj | 520501d | 2011-10-23 09:09:54 +0000 | [diff] [blame] | 498 | 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER() |
| 499 | 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored |
| 500 | 250101 huge "free" memory usage due to m_mallocfree.c fragmentation |
| 501 | 253206 Some fixes for the faultstatus testcase |
| 502 | 255223 capget testcase fails when running as root |
| 503 | 256703 xlc_dbl_u32.c testcase broken |
| 504 | 256726 Helgrind tests have broken inline asm |
| 505 | 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp) |
sewardj | 7a9ff4c | 2011-06-13 13:36:59 +0000 | [diff] [blame] | 506 | 264800 testcase compile failure on zseries |
sewardj | 520501d | 2011-10-23 09:09:54 +0000 | [diff] [blame] | 507 | 265762 make public VEX headers compilable by G++ 3.x |
| 508 | 265771 assertion in jumps.c (r11523) fails with glibc-2.3 |
| 509 | 266753 configure script does not give the user the option to not use QtCore |
| 510 | 266931 gen_insn_test.pl is broken |
| 511 | 266961 ld-linux.so.2 i?86-linux strlen issues |
| 512 | 266990 setns instruction causes false positive |
| 513 | 267020 Make directory for temporary files configurable at run-time. |
| 514 | 267342 == 267997 (segmentation fault on Mac OS 10.6) |
| 515 | 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed |
| 516 | 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed. |
| 517 | 267488 regtest: darwin support for 64-bit build |
| 518 | 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools |
| 519 | 267630 Add support for IBM Power ISA 2.06 -- stage 1 |
| 520 | 267769 == 267997 (Darwin: memcheck triggers segmentation fault) |
| 521 | 267819 Add client request for informing the core about reallocation |
| 522 | 267925 laog data structure quadratic for a single sequence of lock |
| 523 | 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed |
| 524 | 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1 |
| 525 | 268513 missed optimizations in fold_Expr |
| 526 | 268619 s390x: fpr - gpr transfer facility |
| 527 | 268620 s390x: reconsider "long displacement" requirement |
| 528 | 268621 s390x: improve IR generation for XC |
| 529 | 268715 s390x: FLOGR is not universally available |
| 530 | 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4) |
| 531 | 268930 s390x: MHY is not universally available |
| 532 | 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register) |
| 533 | 269079 Support ptrace system call on ARM |
| 534 | 269144 missing "Bad option" error message |
| 535 | 269209 conditional load and store facility (z196) |
| 536 | 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP |
| 537 | 269641 == 267997 (valgrind segfaults immediately (segmentation fault)) |
| 538 | 269736 s390x: minor code generation tweaks |
| 539 | 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..) |
| 540 | 269863 s390x: remove unused function parameters |
| 541 | 269864 s390x: tweak s390_emit_load_cc |
| 542 | 269884 == 250101 (overhead for huge blocks exhausts space too soon) |
| 543 | 270082 s390x: Make sure to point the PSW address to the next address on SIGILL |
| 544 | 270115 s390x: rewrite some testcases |
| 545 | 270309 == 267997 (valgrind crash on startup) |
| 546 | 270320 add support for Linux FIOQSIZE ioctl() call |
| 547 | 270326 segfault while trying to sanitize the environment passed to execle |
| 548 | 270794 IBM POWER7 support patch causes regression in none/tests |
| 549 | 270851 IBM POWER7 fcfidus instruction causes memcheck to fail |
| 550 | 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app |
| 551 | 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix |
| 552 | 270959 s390x: invalid use of R0 as base register |
| 553 | 271042 VSX configure check fails when it should not |
| 554 | 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21 |
| 555 | 271259 s390x: fix code confusion |
| 556 | 271337 == 267997 (Valgrind segfaults on MacOS X) |
| 557 | 271385 s390x: Implement Ist_MBE |
| 558 | 271501 s390x: misc cleanups |
| 559 | 271504 s390x: promote likely and unlikely |
| 560 | 271579 ppc: using wrong enum type |
| 561 | 271615 unhandled instruction "popcnt" (arch=amd10h) |
| 562 | 271730 Fix bug when checking ioctls: duplicate check |
| 563 | 271776 s390x: provide STFLE instruction support |
| 564 | 271779 s390x: provide clock instructions like STCK |
| 565 | 271799 Darwin: ioctls without an arg report a memory error |
| 566 | 271820 arm: fix type confusion |
| 567 | 271917 pthread_cond_timedwait failure leads to not-locked false positive |
| 568 | 272067 s390x: fix DISP20 macro |
| 569 | 272615 A typo in debug output in mc_leakcheck.c |
| 570 | 272661 callgrind_annotate chokes when run from paths containing regex chars |
| 571 | 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup) |
| 572 | 272955 Unhandled syscall error for pwrite64 on ppc64 arch |
| 573 | 272967 make documentation build-system more robust |
| 574 | 272986 Fix gcc-4.6 warnings with valgrind.h |
| 575 | 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case) |
| 576 | 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 |
| 577 | 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039) |
| 578 | 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...' |
| 579 | 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer' |
| 580 | 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169) |
| 581 | 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction) |
| 582 | 273778 exp-ptrcheck: unhandled sysno == 259 |
| 583 | 274089 exp-ptrcheck: unhandled sysno == 208 |
| 584 | 274378 s390x: Various dispatcher tweaks |
| 585 | 274447 WARNING: unhandled syscall: 340 |
| 586 | 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66 |
| 587 | 274784 == 267997 (valgrind ls -l results in Segmentation Fault) |
| 588 | 274926 valgrind does not build against linux-3 |
| 589 | 275148 configure FAIL with glibc-2.14 |
| 590 | 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL |
| 591 | 275168 Make Valgrind work for MacOSX 10.7 Lion |
| 592 | 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al) |
| 593 | 275278 valgrind does not build on Linux kernel 3.0.* due to silly |
| 594 | 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64 |
| 595 | 275308 Fix implementation for ppc64 fres instruc |
| 596 | 275339 s390x: fix testcase compile warnings |
| 597 | 275517 s390x: Provide support for CKSM instruction |
| 598 | 275710 s390x: get rid of redundant address mode calculation |
| 599 | 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall) |
| 600 | 275852 == 250101 (valgrind uses all swap space and is killed) |
| 601 | 276784 Add support for IBM Power ISA 2.06 -- stage 3 |
| 602 | 276987 gdbsrv: fix tests following recent commits |
| 603 | 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a |
| 604 | 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink |
| 605 | 277471 Unhandled syscall: 340 |
| 606 | 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...) |
| 607 | 277653 ARM: support Thumb2 PLD instruction |
| 608 | 277663 ARM: NEON float VMUL by scalar incorrect |
| 609 | 277689 ARM: tests for VSTn with register post-index are broken |
| 610 | 277694 ARM: BLX LR instruction broken in ARM mode |
| 611 | 277780 ARM: VMOV.F32 (immediate) instruction is broken |
| 612 | 278057 fuse filesystem syscall deadlocks |
| 613 | 278078 Unimplemented syscall 280 on ppc32 |
| 614 | 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands |
| 615 | 278454 VALGRIND_STACK_DEREGISTER has wrong output type |
| 616 | 278502 == 275284 (Valgrind confuses memcpy() and memmove()) |
sewardj | 3bbaf48 | 2011-10-24 08:54:56 +0000 | [diff] [blame] | 617 | 278892 gdbsrv: factorize gdb version handling, fix doc and typos |
sewardj | 520501d | 2011-10-23 09:09:54 +0000 | [diff] [blame] | 618 | 279027 Support for MVCL and CLCL instruction |
| 619 | 279027 s390x: Provide support for CLCL and MVCL instructions |
| 620 | 279062 Remove a redundant check in the insn selector for ppc. |
| 621 | 279071 JDK creates PTEST with redundant REX.W prefix |
| 622 | 279212 gdbsrv: add monitor cmd v.info scheduler. |
| 623 | 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call |
| 624 | 279698 memcheck discards valid-bits for packuswb |
| 625 | 279795 memcheck reports uninitialised values for mincore on amd64 |
| 626 | 279994 Add support for IBM Power ISA 2.06 -- stage 3 |
| 627 | 280083 mempolicy syscall check errors |
| 628 | 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F |
| 629 | 280710 s390x: config files for nightly builds |
| 630 | 280757 /tmp dir still used by valgrind even if TMPDIR is specified |
sewardj | 3bbaf48 | 2011-10-24 08:54:56 +0000 | [diff] [blame] | 631 | 280965 Valgrind breaks fcntl locks when program does mmap |
sewardj | 520501d | 2011-10-23 09:09:54 +0000 | [diff] [blame] | 632 | 281138 WARNING: unhandled syscall: 340 |
| 633 | 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion) |
| 634 | 281304 == 275168 (Darwin: dyld "cannot load inserted library") |
| 635 | 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1) |
| 636 | 281468 s390x: handle do_clone and gcc clones in call traces |
| 637 | 281488 ARM: VFP register corruption |
| 638 | 281828 == 275284 (false memmove warning: "Source and destination overlap") |
| 639 | 281883 s390x: Fix system call wrapper for "clone". |
| 640 | 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock |
| 641 | 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp) |
| 642 | 282238 SLES10: make check fails |
| 643 | 282979 strcasestr needs replacement with recent(>=2.12) glibc |
| 644 | 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF |
| 645 | 283243 Regression in ppc64 memcheck tests |
| 646 | 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0) |
| 647 | 283427 re-connect epoll_pwait syscall on ARM linux |
| 648 | 283600 gdbsrv: android: port vgdb.c |
| 649 | 283709 none/tests/faultstatus needs to account for page size |
| 650 | 284305 filter_gdb needs enhancement to work on ppc64 |
sewardj | c2f1d96 | 2011-10-25 09:05:53 +0000 | [diff] [blame] | 651 | 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h |
sewardj | 520501d | 2011-10-23 09:09:54 +0000 | [diff] [blame] | 652 | 284472 Thumb2 ROR.W encoding T2 not implemented |
sewardj | 3bbaf48 | 2011-10-24 08:54:56 +0000 | [diff] [blame] | 653 | 284621 XML-escape process command line in XML output |
sewardj | 520501d | 2011-10-23 09:09:54 +0000 | [diff] [blame] | 654 | n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs |
| 655 | that have non-power-of-2 sizes (also AMDs) |
| 656 | n-i-bz don't be spooked by libraries mashed by elfhack |
| 657 | n-i-bz don't be spooked by libxul.so linked with gold |
sewardj | 3bbaf48 | 2011-10-24 08:54:56 +0000 | [diff] [blame] | 658 | n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED |
sewardj | 7a9ff4c | 2011-06-13 13:36:59 +0000 | [diff] [blame] | 659 | |
sewardj | 62b0d40 | 2011-11-09 23:42:41 +0000 | [diff] [blame] | 660 | (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245) |
| 661 | (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257) |
| 662 | (3.7.0: 5 November 2011, vex r2231, valgrind r12258) |
sewardj | 469b836 | 2011-10-23 10:05:47 +0000 | [diff] [blame] | 663 | |
florian | 0c865b4 | 2011-09-30 01:27:26 +0000 | [diff] [blame] | 664 | |
njn | e5930da | 2010-12-17 00:45:19 +0000 | [diff] [blame] | 665 | |
sewardj | 4fd5729 | 2011-02-17 12:20:19 +0000 | [diff] [blame] | 666 | Release 3.6.1 (16 February 2011) |
| 667 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 668 | 3.6.1 is a bug fix release. It adds support for some SSE4 |
| 669 | instructions that were omitted in 3.6.0 due to lack of time. Initial |
| 670 | support for glibc-2.13 has been added. A number of bugs causing |
| 671 | crashing or assertion failures have been fixed. |
njn | e5930da | 2010-12-17 00:45:19 +0000 | [diff] [blame] | 672 | |
sewardj | 4fd5729 | 2011-02-17 12:20:19 +0000 | [diff] [blame] | 673 | The following bugs have been fixed or resolved. Note that "n-i-bz" |
| 674 | stands for "not in bugzilla" -- that is, a bug that was reported to us |
| 675 | but never got a bugzilla entry. We encourage you to file bugs in |
| 676 | bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than |
| 677 | mailing the developers (or mailing lists) directly -- bugs that are |
| 678 | not entered into bugzilla tend to get forgotten about or ignored. |
bart | 8e0d2c1 | 2011-02-04 19:07:11 +0000 | [diff] [blame] | 679 | |
sewardj | 4fd5729 | 2011-02-17 12:20:19 +0000 | [diff] [blame] | 680 | To see details of a given bug, visit |
| 681 | https://bugs.kde.org/show_bug.cgi?id=XXXXXX |
| 682 | where XXXXXX is the bug number as listed below. |
| 683 | |
| 684 | 188572 Valgrind on Mac should suppress setenv() mem leak |
| 685 | 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support) |
| 686 | 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ) |
| 687 | 246152 callgrind internal error after pthread_cancel on 32 Bit Linux |
| 688 | 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest |
| 689 | 254420 memory pool tracking broken |
| 690 | 254957 Test code failing to compile due to changes in memcheck.h |
| 691 | 255009 helgrind/drd: crash on chmod with invalid parameter |
| 692 | 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types |
| 693 | 255355 helgrind/drd: crash on threaded programs doing fork |
| 694 | 255358 == 255355 |
| 695 | 255418 (SSE4.x) rint call compiled with ICC |
| 696 | 255822 --gen-suppressions can create invalid files: "too many callers [...]" |
| 697 | 255888 closing valgrindoutput tag outputted to log-stream on error |
| 698 | 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD) |
| 699 | 255966 Slowness when using mempool annotations |
| 700 | 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM) |
| 701 | 256600 super-optimized strcasecmp() false positive |
| 702 | 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64 |
| 703 | 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx) |
| 704 | 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW) |
| 705 | 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS) |
| 706 | 257276 Missing case in memcheck --track-origins=yes |
| 707 | 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction |
| 708 | 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L}) |
| 709 | 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX |
| 710 | 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0) |
| 711 | 263099 callgrind_annotate counts Ir improperly [...] |
| 712 | 263877 undefined coprocessor instruction on ARMv7 |
| 713 | 265964 configure FAIL with glibc-2.13 |
| 714 | n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c |
| 715 | n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al |
| 716 | n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes |
| 717 | n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler |
| 718 | n-i-bz DRD: disable --free-is-write due to implementation difficulties |
| 719 | |
| 720 | (3.6.1: 16 February 2011, vex r2103, valgrind r11561). |
| 721 | |
bart | 8e0d2c1 | 2011-02-04 19:07:11 +0000 | [diff] [blame] | 722 | |
| 723 | |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 724 | Release 3.6.0 (21 October 2010) |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 725 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 726 | 3.6.0 is a feature release with many significant improvements and the |
| 727 | usual collection of bug fixes. |
sewardj | 00d9d1e | 2010-10-13 14:05:30 +0000 | [diff] [blame] | 728 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 729 | This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux, |
| 730 | PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros |
| 731 | and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added. |
sewardj | 00d9d1e | 2010-10-13 14:05:30 +0000 | [diff] [blame] | 732 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 733 | ------------------------- |
sewardj | 00d9d1e | 2010-10-13 14:05:30 +0000 | [diff] [blame] | 734 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 735 | Here are some highlights. Details are shown further down: |
njn | 9b919b6 | 2010-02-28 23:41:07 +0000 | [diff] [blame] | 736 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 737 | * Support for ARM/Linux. |
njn | 9b919b6 | 2010-02-28 23:41:07 +0000 | [diff] [blame] | 738 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 739 | * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14. |
| 740 | |
| 741 | * Support for Mac OS X 10.6, both 32- and 64-bit executables. |
| 742 | |
| 743 | * Support for the SSE4.2 instruction set. |
| 744 | |
| 745 | * Enhancements to the Callgrind profiler, including the ability to |
| 746 | handle CPUs with three levels of cache. |
| 747 | |
| 748 | * A new experimental heap profiler, DHAT. |
| 749 | |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 750 | * A huge number of bug fixes and small enhancements. |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 751 | |
| 752 | ------------------------- |
| 753 | |
| 754 | Here are details of the above changes, together with descriptions of |
| 755 | many other changes, and a list of fixed bugs. |
| 756 | |
| 757 | * ================== PLATFORM CHANGES ================= |
| 758 | |
| 759 | * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs |
| 760 | running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10, |
| 761 | and Maemo 5, so you can run Valgrind on your Nokia N900 if you want. |
| 762 | |
| 763 | This requires a CPU capable of running the ARMv7-A instruction set |
| 764 | (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage |
| 765 | of the user space instruction set, including ARM and Thumb integer |
| 766 | code, VFPv3, NEON and V6 media instructions. The Memcheck, |
| 767 | Cachegrind and Massif tools work properly; other tools work to |
| 768 | varying degrees. |
| 769 | |
| 770 | * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along |
| 771 | with support for recent releases of the underlying toolchain |
| 772 | components, notably gcc-4.5 and glibc-2.12. |
| 773 | |
| 774 | * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit |
| 775 | support also works much better on OS X 10.5, and is as solid as |
| 776 | 32-bit support now. |
| 777 | |
| 778 | * Support for the SSE4.2 instruction set. SSE4.2 is supported in |
| 779 | 64-bit mode. In 32-bit mode, support is only available up to and |
| 780 | including SSSE3. Some exceptions: SSE4.2 AES instructions are not |
| 781 | supported in 64-bit mode, and 32-bit mode does in fact support the |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 782 | bare minimum SSE4 instructions to needed to run programs on Mac OS X |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 783 | 10.6 on 32-bit targets. |
| 784 | |
| 785 | * Support for IBM POWER6 cpus has been improved. The Power ISA up to |
| 786 | and including version 2.05 is supported. |
| 787 | |
| 788 | * ==================== TOOL CHANGES ==================== |
| 789 | |
| 790 | * Cachegrind has a new processing script, cg_diff, which finds the |
| 791 | difference between two profiles. It's very useful for evaluating |
| 792 | the performance effects of a change in a program. |
njn | 69d495d | 2010-06-30 05:23:34 +0000 | [diff] [blame] | 793 | |
| 794 | Related to this change, the meaning of cg_annotate's (rarely-used) |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 795 | --threshold option has changed; this is unlikely to affect many |
| 796 | people, if you do use it please see the user manual for details. |
njn | 9b919b6 | 2010-02-28 23:41:07 +0000 | [diff] [blame] | 797 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 798 | * Callgrind now can do branch prediction simulation, similar to |
| 799 | Cachegrind. In addition, it optionally can count the number of |
| 800 | executed global bus events. Both can be used for a better |
| 801 | approximation of a "Cycle Estimation" as derived event (you need to |
| 802 | update the event formula in KCachegrind yourself). |
njn | 2d853a1 | 2010-10-06 22:46:31 +0000 | [diff] [blame] | 803 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 804 | * Cachegrind and Callgrind now refer to the LL (last-level) cache |
| 805 | rather than the L2 cache. This is to accommodate machines with |
| 806 | three levels of caches -- if Cachegrind/Callgrind auto-detects the |
| 807 | cache configuration of such a machine it will run the simulation as |
| 808 | if the L2 cache isn't present. This means the results are less |
| 809 | likely to match the true result for the machine, but |
| 810 | Cachegrind/Callgrind's results are already only approximate, and |
| 811 | should not be considered authoritative. The results are still |
| 812 | useful for giving a general idea about a program's locality. |
njn | 2d853a1 | 2010-10-06 22:46:31 +0000 | [diff] [blame] | 813 | |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 814 | * Massif has a new option, --pages-as-heap, which is disabled by |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 815 | default. When enabled, instead of tracking allocations at the level |
| 816 | of heap blocks (as allocated with malloc/new/new[]), it instead |
| 817 | tracks memory allocations at the level of memory pages (as mapped by |
| 818 | mmap, brk, etc). Each mapped page is treated as its own block. |
| 819 | Interpreting the page-level output is harder than the heap-level |
| 820 | output, but this option is useful if you want to account for every |
| 821 | byte of memory used by a program. |
njn | e323a6b | 2010-07-01 02:35:03 +0000 | [diff] [blame] | 822 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 823 | * DRD has two new command-line options: --free-is-write and |
| 824 | --trace-alloc. The former allows to detect reading from already freed |
| 825 | memory, and the latter allows tracing of all memory allocations and |
bart | 5cf4c39 | 2010-10-10 18:54:44 +0000 | [diff] [blame] | 826 | deallocations. |
| 827 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 828 | * DRD has several new annotations. Custom barrier implementations can |
| 829 | now be annotated, as well as benign races on static variables. |
bart | 5cf4c39 | 2010-10-10 18:54:44 +0000 | [diff] [blame] | 830 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 831 | * DRD's happens before / happens after annotations have been made more |
| 832 | powerful, so that they can now also be used to annotate e.g. a smart |
bart | 5cf4c39 | 2010-10-10 18:54:44 +0000 | [diff] [blame] | 833 | pointer implementation. |
| 834 | |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 835 | * Helgrind's annotation set has also been drastically improved, so as |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 836 | to provide to users a general set of annotations to describe locks, |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 837 | semaphores, barriers and condition variables. Annotations to |
| 838 | describe thread-safe reference counted heap objects have also been |
| 839 | added. |
| 840 | |
| 841 | * Memcheck has a new command-line option, --show-possibly-lost, which |
| 842 | is enabled by default. When disabled, the leak detector will not |
| 843 | show possibly-lost blocks. |
| 844 | |
| 845 | * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool), |
| 846 | has been added. DHAT keeps track of allocated heap blocks, and also |
| 847 | inspects every memory reference to see which block (if any) is being |
| 848 | accessed. This gives a lot of insight into block lifetimes, |
| 849 | utilisation, turnover, liveness, and the location of hot and cold |
| 850 | fields. You can use DHAT to do hot-field profiling. |
| 851 | |
| 852 | * ==================== OTHER CHANGES ==================== |
| 853 | |
| 854 | * Improved support for unfriendly self-modifying code: the extra |
| 855 | overhead incurred by --smc-check=all has been reduced by |
| 856 | approximately a factor of 5 as compared with 3.5.0. |
| 857 | |
| 858 | * Ability to show directory names for source files in error messages. |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 859 | This is combined with a flexible mechanism for specifying which |
| 860 | parts of the paths should be shown. This is enabled by the new flag |
| 861 | --fullpath-after. |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 862 | |
| 863 | * A new flag, --require-text-symbol, which will stop the run if a |
| 864 | specified symbol is not found it a given shared object when it is |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 865 | loaded into the process. This makes advanced working with function |
| 866 | intercepting and wrapping safer and more reliable. |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 867 | |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 868 | * Improved support for the Valkyrie GUI, version 2.0.0. GUI output |
| 869 | and control of Valgrind is now available for the tools Memcheck and |
| 870 | Helgrind. XML output from Valgrind is available for Memcheck, |
| 871 | Helgrind and exp-Ptrcheck. |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 872 | |
| 873 | * More reliable stack unwinding on amd64-linux, particularly in the |
| 874 | presence of function wrappers, and with gcc-4.5 compiled code. |
| 875 | |
| 876 | * Modest scalability (performance improvements) for massive |
| 877 | long-running applications, particularly for those with huge amounts |
| 878 | of code. |
| 879 | |
| 880 | * Support for analyzing programs running under Wine with has been |
| 881 | improved. The header files <valgrind/valgrind.h>, |
| 882 | <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in |
| 883 | Windows-programs compiled with MinGW or one of the Microsoft Visual |
| 884 | Studio compilers. |
| 885 | |
| 886 | * A rare but serious error in the 64-bit x86 CPU simulation was fixed. |
| 887 | The 32-bit simulator was not affected. This did not occur often, |
| 888 | but when it did would usually crash the program under test. |
| 889 | Bug 245925. |
| 890 | |
| 891 | * A large number of bugs were fixed. These are shown below. |
| 892 | |
| 893 | * A number of bugs were investigated, and were candidates for fixing, |
| 894 | but are not fixed in 3.6.0, due to lack of developer time. They may |
| 895 | get fixed in later releases. They are: |
| 896 | |
| 897 | 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64) |
| 898 | 212419 false positive "lock order violated" (A+B vs A) |
| 899 | 213685 Undefined value propagates past dependency breaking instruction |
| 900 | 216837 Incorrect instrumentation of NSOperationQueue on Darwin |
| 901 | 237920 valgrind segfault on fork failure |
| 902 | 242137 support for code compiled by LLVM-2.8 |
| 903 | 242423 Another unknown Intel cache config value |
| 904 | 243232 Inconsistent Lock Orderings report with trylock |
| 905 | 243483 ppc: callgrind triggers VEX assertion failure |
| 906 | 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong |
| 907 | 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion |
| 908 | 'thr' failed. |
| 909 | 246152 callgrind internal error after pthread_cancel on 32 Bit Linux |
| 910 | 249435 Analyzing wine programs with callgrind triggers a crash |
| 911 | 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest |
| 912 | 250065 Handling large allocations |
| 913 | 250101 huge "free" memory usage due to m_mallocfree.c |
| 914 | "superblocks fragmentation" |
| 915 | 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP) |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 916 | 252091 Callgrind on ARM does not detect function returns correctly |
| 917 | 252600 [PATCH] Allow lhs to be a pointer for shl/shr |
| 918 | 254420 memory pool tracking broken |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 919 | n-i-bz support for adding symbols for JIT generated code |
| 920 | |
| 921 | |
| 922 | The following bugs have been fixed or resolved. Note that "n-i-bz" |
| 923 | stands for "not in bugzilla" -- that is, a bug that was reported to us |
| 924 | but never got a bugzilla entry. We encourage you to file bugs in |
| 925 | bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than |
| 926 | mailing the developers (or mailing lists) directly -- bugs that are |
| 927 | not entered into bugzilla tend to get forgotten about or ignored. |
| 928 | |
| 929 | To see details of a given bug, visit |
| 930 | https://bugs.kde.org/show_bug.cgi?id=XXXXXX |
| 931 | where XXXXXX is the bug number as listed below. |
| 932 | |
| 933 | 135264 dcbzl instruction missing |
| 934 | 142688 == 250799 |
| 935 | 153699 Valgrind should report unaligned reads with movdqa |
| 936 | 180217 == 212335 |
| 937 | 190429 Valgrind reports lost of errors in ld.so |
| 938 | with x86_64 2.9.90 glibc |
| 939 | 197266 valgrind appears to choke on the xmms instruction |
| 940 | "roundsd" on x86_64 |
| 941 | 197988 Crash when demangling very large symbol names |
| 942 | 202315 unhandled syscall: 332 (inotify_init1) |
| 943 | 203256 Add page-level profiling to Massif |
| 944 | 205093 dsymutil=yes needs quotes, locking (partial fix) |
| 945 | 205241 Snow Leopard 10.6 support (partial fix) |
| 946 | 206600 Leak checker fails to upgrade indirect blocks when their |
| 947 | parent becomes reachable |
| 948 | 210935 port valgrind.h (not valgrind) to win32 so apps run under |
| 949 | wine can make client requests |
| 950 | 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89 |
| 951 | within Linux ip-stack checksum functions |
| 952 | 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0 |
| 953 | (lzcnt %eax,%eax) |
| 954 | 213685 Undefined value propagates past dependency breaking instruction |
| 955 | (partial fix) |
| 956 | 215914 Valgrind inserts bogus empty environment variable |
| 957 | 217863 == 197988 |
| 958 | 219538 adjtimex syscall wrapper wrong in readonly adjtime mode |
| 959 | 222545 shmat fails under valgind on some arm targets |
| 960 | 222560 ARM NEON support |
| 961 | 230407 == 202315 |
| 962 | 231076 == 202315 |
| 963 | 232509 Docs build fails with formatting inside <title></title> elements |
| 964 | 232793 == 202315 |
| 965 | 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls |
| 966 | 236546 vex x86->IR: 0x66 0xF 0x3A 0xA |
| 967 | 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B |
| 968 | 237371 better support for VALGRIND_MALLOCLIKE_BLOCK |
| 969 | 237485 symlink (syscall 57) is not supported on Mac OS |
| 970 | 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened: |
| 971 | unhandled syscall |
| 972 | 238208 is_just_below_ESP doesn't take into account red-zone |
| 973 | 238345 valgrind passes wrong $0 when executing a shell script |
| 974 | 238679 mq_timedreceive syscall doesn't flag the reception buffer |
| 975 | as "defined" |
| 976 | 238696 fcntl command F_DUPFD_CLOEXEC not supported |
| 977 | 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6 |
| 978 | 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6 |
| 979 | 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure |
| 980 | says "Altivec off" |
| 981 | 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48 |
| 982 | 240488 == 197988 |
| 983 | 240639 == 212335 |
| 984 | 241377 == 236546 |
| 985 | 241903 == 202315 |
| 986 | 241920 == 212335 |
| 987 | 242606 unhandled syscall: setegid (in Ptrcheck) |
| 988 | 242814 Helgrind "Impossible has happened" during |
| 989 | QApplication::initInstance(); |
| 990 | 243064 Valgrind attempting to read debug information from iso |
| 991 | 243270 Make stack unwinding in Valgrind wrappers more reliable |
| 992 | 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall |
| 993 | sysno = 277 (mq_open) |
| 994 | 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd |
| 995 | 244493 ARM VFP d16-d31 registers support |
| 996 | 244670 add support for audit_session_self syscall on Mac OS 10.6 |
| 997 | 244921 The xml report of helgrind tool is not well format |
| 998 | 244923 In the xml report file, the <preamble> not escape the |
| 999 | xml char, eg '<','&','>' |
| 1000 | 245535 print full path names in plain text reports |
| 1001 | 245925 x86-64 red zone handling problem |
| 1002 | 246258 Valgrind not catching integer underruns + new [] s |
| 1003 | 246311 reg/reg cmpxchg doesn't work on amd64 |
| 1004 | 246549 unhandled syscall unix:277 while testing 32-bit Darwin app |
| 1005 | 246888 Improve Makefile.vex.am |
| 1006 | 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed |
| 1007 | to [f]chmod_extended |
| 1008 | 247526 IBM POWER6 (ISA 2.05) support is incomplete |
| 1009 | 247561 Some leak testcases fails due to reachable addresses in |
| 1010 | caller save regs |
| 1011 | 247875 sizeofIRType to handle Ity_I128 |
| 1012 | 247894 [PATCH] unhandled syscall sys_readahead |
| 1013 | 247980 Doesn't honor CFLAGS passed to configure |
| 1014 | 248373 darwin10.supp is empty in the trunk |
| 1015 | 248822 Linux FIBMAP ioctl has int parameter instead of long |
| 1016 | 248893 [PATCH] make readdwarf.c big endianess safe to enable |
| 1017 | unwinding on big endian systems |
| 1018 | 249224 Syscall 336 not supported (SYS_proc_info) |
| 1019 | 249359 == 245535 |
| 1020 | 249775 Incorrect scheme for detecting NEON capabilities of host CPU |
| 1021 | 249943 jni JVM init fails when using valgrind |
| 1022 | 249991 Valgrind incorrectly declares AESKEYGENASSIST support |
| 1023 | since VEX r2011 |
| 1024 | 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64) |
| 1025 | 250799 frexp$fenv_access_off function generates SIGILL |
| 1026 | 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E |
| 1027 | 251251 support pclmulqdq insn |
| 1028 | 251362 valgrind: ARM: attach to debugger either fails or provokes |
| 1029 | kernel oops |
| 1030 | 251674 Unhandled syscall 294 |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 1031 | 251818 == 254550 |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 1032 | |
sewardj | 695f13a | 2010-10-20 17:07:01 +0000 | [diff] [blame] | 1033 | 254257 Add support for debugfiles found by build-id |
| 1034 | 254550 [PATCH] Implement DW_ATE_UTF (DWARF4) |
| 1035 | 254646 Wrapped functions cause stack misalignment on OS X |
| 1036 | (and possibly Linux) |
| 1037 | 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0 |
| 1038 | |
sewardj | e07c2b5 | 2010-10-22 06:44:59 +0000 | [diff] [blame] | 1039 | (3.6.0: 21 October 2010, vex r2068, valgrind r11471). |
sewardj | 18a0892 | 2010-10-14 00:26:42 +0000 | [diff] [blame] | 1040 | |
| 1041 | |
njn | 9b919b6 | 2010-02-28 23:41:07 +0000 | [diff] [blame] | 1042 | |
sewardj | 6bbe024 | 2009-08-19 11:38:54 +0000 | [diff] [blame] | 1043 | Release 3.5.0 (19 August 2009) |
| 1044 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1045 | 3.5.0 is a feature release with many significant improvements and the |
| 1046 | usual collection of bug fixes. The main improvement is that Valgrind |
| 1047 | now works on Mac OS X. |
| 1048 | |
| 1049 | This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux |
| 1050 | and X86/Darwin. Support for recent distros and toolchain components |
| 1051 | (glibc 2.10, gcc 4.5) has been added. |
| 1052 | |
| 1053 | ------------------------- |
| 1054 | |
| 1055 | Here is a short summary of the changes. Details are shown further |
| 1056 | down: |
| 1057 | |
njn | f95c30b | 2009-08-16 00:20:58 +0000 | [diff] [blame] | 1058 | * Support for Mac OS X (10.5.x). |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1059 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 1060 | * Improvements and simplifications to Memcheck's leak checker. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1061 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 1062 | * Clarification and simplifications in various aspects of Valgrind's |
| 1063 | text output. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1064 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 1065 | * XML output for Helgrind and Ptrcheck. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1066 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 1067 | * Performance and stability improvements for Helgrind and DRD. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1068 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 1069 | * Genuinely atomic support for x86/amd64/ppc atomic instructions. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1070 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 1071 | * A new experimental tool, BBV, useful for computer architecture |
| 1072 | research. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1073 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 1074 | * Improved Wine support, including ability to read Windows PDB |
| 1075 | debuginfo. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1076 | |
| 1077 | ------------------------- |
| 1078 | |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 1079 | Here are details of the above changes, followed by descriptions of |
| 1080 | many other minor changes, and a list of fixed bugs. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1081 | |
| 1082 | |
| 1083 | * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes |
njn | f95c30b | 2009-08-16 00:20:58 +0000 | [diff] [blame] | 1084 | called "Darwin" because that is the name of the OS core, which is the |
| 1085 | level that Valgrind works at.) |
njn | f76d27a | 2009-05-28 01:53:07 +0000 | [diff] [blame] | 1086 | |
njn | f95c30b | 2009-08-16 00:20:58 +0000 | [diff] [blame] | 1087 | Supported systems: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1088 | |
njn | f95c30b | 2009-08-16 00:20:58 +0000 | [diff] [blame] | 1089 | - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned |
| 1090 | because it would require work and 10.4 is only becoming less common. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1091 | |
njn | f95c30b | 2009-08-16 00:20:58 +0000 | [diff] [blame] | 1092 | - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported |
| 1093 | fairly well. For 10.5.x, 32-bit programs are the default even on |
| 1094 | 64-bit machines, so it handles most current programs. |
| 1095 | |
| 1096 | - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not |
| 1097 | officially supported, but simple programs at least will probably work. |
| 1098 | However, start-up is slow. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1099 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1100 | - PowerPC machines are not supported. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1101 | |
njn | c6ab7b6 | 2009-06-04 22:47:21 +0000 | [diff] [blame] | 1102 | Things that don't work: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1103 | |
| 1104 | - The Ptrcheck tool. |
| 1105 | |
| 1106 | - Objective-C garbage collection. |
| 1107 | |
| 1108 | - --db-attach=yes. |
| 1109 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1110 | - If you have Rogue Amoeba's "Instant Hijack" program installed, |
| 1111 | Valgrind will fail with a SIGTRAP at start-up. See |
| 1112 | https://bugs.kde.org/show_bug.cgi?id=193917 for details and a |
| 1113 | simple work-around. |
njn | c6ab7b6 | 2009-06-04 22:47:21 +0000 | [diff] [blame] | 1114 | |
njn | 97db761 | 2009-08-04 02:32:55 +0000 | [diff] [blame] | 1115 | Usage notes: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1116 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1117 | - You will likely find --dsymutil=yes a useful option, as error |
| 1118 | messages may be imprecise without it. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1119 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1120 | - Mac OS X support is new and therefore will be less robust than the |
njn | 1205e02 | 2009-08-14 02:12:54 +0000 | [diff] [blame] | 1121 | Linux support. Please report any bugs you find. |
njn | 97db761 | 2009-08-04 02:32:55 +0000 | [diff] [blame] | 1122 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1123 | - Threaded programs may run more slowly than on Linux. |
| 1124 | |
njn | c6ab7b6 | 2009-06-04 22:47:21 +0000 | [diff] [blame] | 1125 | Many thanks to Greg Parker for developing this port over several years. |
| 1126 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1127 | |
| 1128 | * Memcheck's leak checker has been improved. |
| 1129 | |
| 1130 | - The results for --leak-check=summary now match the summary results |
| 1131 | for --leak-check=full. Previously they could differ because |
| 1132 | --leak-check=summary counted "indirectly lost" blocks and |
| 1133 | "suppressed" blocks as "definitely lost". |
| 1134 | |
| 1135 | - Blocks that are only reachable via at least one interior-pointer, |
| 1136 | but are directly pointed to by a start-pointer, were previously |
| 1137 | marked as "still reachable". They are now correctly marked as |
| 1138 | "possibly lost". |
| 1139 | |
| 1140 | - The default value for the --leak-resolution option has been |
| 1141 | changed from "low" to "high". In general, this means that more |
| 1142 | leak reports will be produced, but each leak report will describe |
| 1143 | fewer leaked blocks. |
| 1144 | |
| 1145 | - With --leak-check=full, "definitely lost" and "possibly lost" |
| 1146 | leaks are now considered as proper errors, ie. they are counted |
| 1147 | for the "ERROR SUMMARY" and affect the behaviour of |
| 1148 | --error-exitcode. These leaks are not counted as errors if |
| 1149 | --leak-check=summary is specified, however. |
| 1150 | |
| 1151 | - Documentation for the leak checker has been improved. |
| 1152 | |
njn | f76d27a | 2009-05-28 01:53:07 +0000 | [diff] [blame] | 1153 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1154 | * Various aspects of Valgrind's text output have changed. |
njn | dbebecc | 2009-07-14 01:39:54 +0000 | [diff] [blame] | 1155 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1156 | - Valgrind's start-up message has changed. It is shorter but also |
| 1157 | includes the command being run, which makes it easier to use |
| 1158 | --trace-children=yes. An example: |
njn | 53162bf | 2009-07-29 23:34:49 +0000 | [diff] [blame] | 1159 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1160 | - Valgrind's shut-down messages have also changed. This is most |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1161 | noticeable with Memcheck, where the leak summary now occurs before |
| 1162 | the error summary. This change was necessary to allow leaks to be |
| 1163 | counted as proper errors (see the description of the leak checker |
| 1164 | changes above for more details). This was also necessary to fix a |
| 1165 | longstanding bug in which uses of suppressions against leaks were |
| 1166 | not "counted", leading to difficulties in maintaining suppression |
sewardj | 783e9d6 | 2009-08-19 22:18:35 +0000 | [diff] [blame] | 1167 | files (see https://bugs.kde.org/show_bug.cgi?id=186790). |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1168 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1169 | - Behavior of -v has changed. In previous versions, -v printed out |
| 1170 | a mixture of marginally-user-useful information, and tool/core |
| 1171 | statistics. The statistics printing has now been moved to its own |
| 1172 | flag, --stats=yes. This means -v is less verbose and more likely |
| 1173 | to convey useful end-user information. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1174 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1175 | - The format of some (non-XML) stack trace entries has changed a |
| 1176 | little. Previously there were six possible forms: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1177 | |
| 1178 | 0x80483BF: really (a.c:20) |
| 1179 | 0x80483BF: really (in /foo/a.out) |
| 1180 | 0x80483BF: really |
| 1181 | 0x80483BF: (within /foo/a.out) |
| 1182 | 0x80483BF: ??? (a.c:20) |
| 1183 | 0x80483BF: ??? |
| 1184 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1185 | The third and fourth of these forms have been made more consistent |
| 1186 | with the others. The six possible forms are now: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1187 | |
| 1188 | 0x80483BF: really (a.c:20) |
| 1189 | 0x80483BF: really (in /foo/a.out) |
| 1190 | 0x80483BF: really (in ???) |
| 1191 | 0x80483BF: ??? (in /foo/a.out) |
| 1192 | 0x80483BF: ??? (a.c:20) |
| 1193 | 0x80483BF: ??? |
| 1194 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1195 | Stack traces produced when --xml=yes is specified are different |
| 1196 | and unchanged. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1197 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1198 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1199 | * Helgrind and Ptrcheck now support XML output, so they can be used |
| 1200 | from GUI tools. Also, the XML output mechanism has been |
| 1201 | overhauled. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1202 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 1203 | - The XML format has been overhauled and generalised, so it is more |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1204 | suitable for error reporting tools in general. The Memcheck |
| 1205 | specific aspects of it have been removed. The new format, which |
| 1206 | is an evolution of the old format, is described in |
| 1207 | docs/internals/xml-output-protocol4.txt. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1208 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1209 | - Memcheck has been updated to use the new format. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1210 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1211 | - Helgrind and Ptrcheck are now able to emit output in this format. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1212 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1213 | - The XML output mechanism has been overhauled. XML is now output |
| 1214 | to its own file descriptor, which means that: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1215 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1216 | * Valgrind can output text and XML independently. |
njn | b4e0687 | 2009-08-10 05:11:43 +0000 | [diff] [blame] | 1217 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1218 | * The longstanding problem of XML output being corrupted by |
| 1219 | unexpected un-tagged text messages is solved. |
njn | 8df80b2 | 2009-03-02 05:11:06 +0000 | [diff] [blame] | 1220 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1221 | As before, the destination for text output is specified using |
| 1222 | --log-file=, --log-fd= or --log-socket=. |
njn | d55f0d9 | 2009-08-03 01:38:56 +0000 | [diff] [blame] | 1223 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1224 | As before, XML output for a tool is enabled using --xml=yes. |
njn | c6ab7b6 | 2009-06-04 22:47:21 +0000 | [diff] [blame] | 1225 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1226 | Because there's a new XML output channel, the XML output |
| 1227 | destination is now specified by --xml-file=, --xml-fd= or |
| 1228 | --xml-socket=. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1229 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1230 | Initial feedback has shown this causes some confusion. To |
| 1231 | clarify, the two envisaged usage scenarios are: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1232 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1233 | (1) Normal text output. In this case, do not specify --xml=yes |
| 1234 | nor any of --xml-file=, --xml-fd= or --xml-socket=. |
| 1235 | |
| 1236 | (2) XML output. In this case, specify --xml=yes, and one of |
| 1237 | --xml-file=, --xml-fd= or --xml-socket= to select the XML |
| 1238 | destination, one of --log-file=, --log-fd= or --log-socket= |
| 1239 | to select the destination for any remaining text messages, |
| 1240 | and, importantly, -q. |
| 1241 | |
| 1242 | -q makes Valgrind completely silent on the text channel, |
| 1243 | except in the case of critical failures, such as Valgrind |
| 1244 | itself segfaulting, or failing to read debugging information. |
| 1245 | Hence, in this scenario, it suffices to check whether or not |
| 1246 | any output appeared on the text channel. If yes, then it is |
| 1247 | likely to be a critical error which should be brought to the |
| 1248 | attention of the user. If no (the text channel produced no |
| 1249 | output) then it can be assumed that the run was successful. |
| 1250 | |
| 1251 | This allows GUIs to make the critical distinction they need to |
| 1252 | make (did the run fail or not?) without having to search or |
| 1253 | filter the text output channel in any way. |
| 1254 | |
| 1255 | It is also recommended to use --child-silent-after-fork=yes in |
| 1256 | scenario (2). |
| 1257 | |
| 1258 | |
| 1259 | * Improvements and changes in Helgrind: |
| 1260 | |
| 1261 | - XML output, as described above |
| 1262 | |
| 1263 | - Checks for consistent association between pthread condition |
| 1264 | variables and their associated mutexes are now performed. |
| 1265 | |
| 1266 | - pthread_spinlock functions are supported. |
| 1267 | |
| 1268 | - Modest performance improvements. |
| 1269 | |
| 1270 | - Initial (skeletal) support for describing the behaviour of |
| 1271 | non-POSIX synchronisation objects through ThreadSanitizer |
| 1272 | compatible ANNOTATE_* macros. |
| 1273 | |
| 1274 | - More controllable tradeoffs between performance and the level of |
| 1275 | detail of "previous" accesses in a race. There are now three |
| 1276 | settings: |
| 1277 | |
| 1278 | * --history-level=full. This is the default, and was also the |
| 1279 | default in 3.4.x. It shows both stacks involved in a race, but |
| 1280 | requires a lot of memory and can be very slow in programs that |
| 1281 | do many inter-thread synchronisation events. |
| 1282 | |
| 1283 | * --history-level=none. This only shows the later stack involved |
| 1284 | in a race. This can be much faster than --history-level=full, |
| 1285 | but makes it much more difficult to find the other access |
| 1286 | involved in the race. |
| 1287 | |
| 1288 | The new intermediate setting is |
| 1289 | |
| 1290 | * --history-level=approx |
| 1291 | |
| 1292 | For the earlier (other) access, two stacks are presented. The |
| 1293 | earlier access is guaranteed to be somewhere in between the two |
| 1294 | program points denoted by those stacks. This is not as useful |
| 1295 | as showing the exact stack for the previous access (as per |
| 1296 | --history-level=full), but it is better than nothing, and it's |
| 1297 | almost as fast as --history-level=none. |
| 1298 | |
njn | ace90d8 | 2009-04-27 01:27:58 +0000 | [diff] [blame] | 1299 | |
bart | 6a146fc | 2009-07-21 07:32:19 +0000 | [diff] [blame] | 1300 | * New features and improvements in DRD: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1301 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1302 | - The error messages printed by DRD are now easier to interpret. |
| 1303 | Instead of using two different numbers to identify each thread |
| 1304 | (Valgrind thread ID and DRD thread ID), DRD does now identify |
| 1305 | threads via a single number (the DRD thread ID). Furthermore |
| 1306 | "first observed at" information is now printed for all error |
| 1307 | messages related to synchronization objects. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1308 | |
bart | fe212db | 2009-07-25 11:16:51 +0000 | [diff] [blame] | 1309 | - Added support for named semaphores (sem_open() and sem_close()). |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1310 | |
bart | 6a146fc | 2009-07-21 07:32:19 +0000 | [diff] [blame] | 1311 | - Race conditions between pthread_barrier_wait() and |
| 1312 | pthread_barrier_destroy() calls are now reported. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1313 | |
| 1314 | - Added support for custom allocators through the macros |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1315 | VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in |
| 1316 | in <valgrind/valgrind.h>). An alternative for these two macros is |
| 1317 | the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in |
bart | 6a146fc | 2009-07-21 07:32:19 +0000 | [diff] [blame] | 1318 | <valgrind/drd.h>). |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1319 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1320 | - Added support for annotating non-POSIX synchronization objects |
| 1321 | through several new ANNOTATE_*() macros. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1322 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1323 | - OpenMP: added support for the OpenMP runtime (libgomp) included |
| 1324 | with gcc versions 4.3.0 and 4.4.0. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1325 | |
bart | 6a146fc | 2009-07-21 07:32:19 +0000 | [diff] [blame] | 1326 | - Faster operation. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1327 | |
| 1328 | - Added two new command-line options (--first-race-only and |
bart | 6a146fc | 2009-07-21 07:32:19 +0000 | [diff] [blame] | 1329 | --segment-merging-interval). |
| 1330 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1331 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1332 | * Genuinely atomic support for x86/amd64/ppc atomic instructions |
| 1333 | |
| 1334 | Valgrind will now preserve (memory-access) atomicity of LOCK- |
| 1335 | prefixed x86/amd64 instructions, and any others implying a global |
| 1336 | bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions. |
| 1337 | |
| 1338 | This means that Valgrinded processes will "play nicely" in |
| 1339 | situations where communication with other processes, or the kernel, |
| 1340 | is done through shared memory and coordinated with such atomic |
| 1341 | instructions. Prior to this change, such arrangements usually |
| 1342 | resulted in hangs, races or other synchronisation failures, because |
| 1343 | Valgrind did not honour atomicity of such instructions. |
| 1344 | |
| 1345 | |
| 1346 | * A new experimental tool, BBV, has been added. BBV generates basic |
| 1347 | block vectors for use with the SimPoint analysis tool, which allows |
| 1348 | a program's overall behaviour to be approximated by running only a |
| 1349 | fraction of it. This is useful for computer architecture |
| 1350 | researchers. You can run BBV by specifying --tool=exp-bbv (the |
| 1351 | "exp-" prefix is short for "experimental"). BBV was written by |
| 1352 | Vince Weaver. |
| 1353 | |
| 1354 | |
| 1355 | * Modestly improved support for running Windows applications under |
| 1356 | Wine. In particular, initial support for reading Windows .PDB debug |
| 1357 | information has been added. |
| 1358 | |
| 1359 | |
| 1360 | * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been |
| 1361 | added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks |
| 1362 | instead of bytes. |
| 1363 | |
| 1364 | |
| 1365 | * The Valgrind client requests VALGRIND_PRINTF and |
| 1366 | VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously, |
| 1367 | the string was always printed immediately on its own line. Now, the |
| 1368 | string will be added to a buffer but not printed until a newline is |
| 1369 | encountered, or other Valgrind output is printed (note that for |
| 1370 | VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered |
| 1371 | "other Valgrind output"). This allows you to use multiple |
| 1372 | VALGRIND_PRINTF calls to build up a single output line, and also to |
| 1373 | print multiple output lines with a single request (by embedding |
| 1374 | multiple newlines in the string). |
| 1375 | |
| 1376 | |
| 1377 | * The graphs drawn by Massif's ms_print program have changed slightly: |
| 1378 | |
| 1379 | - The half-height chars '.' and ',' are no longer drawn, because |
| 1380 | they are confusing. The --y option can be used if the default |
| 1381 | y-resolution is not high enough. |
| 1382 | |
| 1383 | - Horizontal lines are now drawn after the top of a snapshot if |
| 1384 | there is a gap until the next snapshot. This makes it clear that |
| 1385 | the memory usage has not dropped to zero between snapshots. |
| 1386 | |
| 1387 | |
| 1388 | * Something that happened in 3.4.0, but wasn't clearly announced: the |
| 1389 | option --read-var-info=yes can be used by some tools (Memcheck, |
| 1390 | Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3 |
| 1391 | variable type and location information. This makes those tools |
| 1392 | start up more slowly and increases memory consumption, but |
| 1393 | descriptions of data addresses in error messages become more |
| 1394 | detailed. |
| 1395 | |
| 1396 | |
| 1397 | * exp-Omega, an experimental instantaneous leak-detecting tool, was |
| 1398 | disabled in 3.4.0 due to a lack of interest and maintenance, |
| 1399 | although the source code was still in the distribution. The source |
| 1400 | code has now been removed from the distribution. For anyone |
| 1401 | interested, the removal occurred in SVN revision r10247. |
| 1402 | |
njn | acda178 | 2009-06-04 23:11:50 +0000 | [diff] [blame] | 1403 | |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 1404 | * Some changes have been made to the build system. |
njn | 6bf365c | 2009-02-11 00:35:45 +0000 | [diff] [blame] | 1405 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1406 | - VEX/ is now integrated properly into the build system. This means |
| 1407 | that dependency tracking within VEX/ now works properly, "make |
| 1408 | install" will work without requiring "make" before it, and |
| 1409 | parallel builds (ie. 'make -j') now work (previously a |
| 1410 | .NOTPARALLEL directive was used to serialize builds, ie. 'make -j' |
| 1411 | was effectively ignored). |
njn | 6bf365c | 2009-02-11 00:35:45 +0000 | [diff] [blame] | 1412 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1413 | - The --with-vex configure option has been removed. It was of |
| 1414 | little use and removing it simplified the build system. |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 1415 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1416 | - The location of some install files has changed. This should not |
| 1417 | affect most users. Those who might be affected: |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 1418 | |
| 1419 | * For people who use Valgrind with MPI programs, the installed |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1420 | libmpiwrap.so library has moved from |
| 1421 | $(INSTALL)/<platform>/libmpiwrap.so to |
| 1422 | $(INSTALL)/libmpiwrap-<platform>.so. |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 1423 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1424 | * For people who distribute standalone Valgrind tools, the |
| 1425 | installed libraries such as $(INSTALL)/<platform>/libcoregrind.a |
| 1426 | have moved to $(INSTALL)/libcoregrind-<platform>.a. |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 1427 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1428 | These changes simplify the build system. |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 1429 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1430 | - Previously, all the distributed suppression (*.supp) files were |
| 1431 | installed. Now, only default.supp is installed. This should not |
| 1432 | affect users as the other installed suppression files were not |
| 1433 | read; the fact that they were installed was a mistake. |
| 1434 | |
| 1435 | |
| 1436 | * KNOWN LIMITATIONS: |
| 1437 | |
| 1438 | - Memcheck is unusable with the Intel compiler suite version 11.1, |
| 1439 | when it generates code for SSE2-and-above capable targets. This |
| 1440 | is because of icc's use of highly optimised inlined strlen |
| 1441 | implementations. It causes Memcheck to report huge numbers of |
| 1442 | false errors even in simple programs. Helgrind and DRD may also |
| 1443 | have problems. |
| 1444 | |
| 1445 | Versions 11.0 and earlier may be OK, but this has not been |
| 1446 | properly tested. |
| 1447 | |
| 1448 | |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 1449 | The following bugs have been fixed or resolved. Note that "n-i-bz" |
| 1450 | stands for "not in bugzilla" -- that is, a bug that was reported to us |
| 1451 | but never got a bugzilla entry. We encourage you to file bugs in |
| 1452 | bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than |
| 1453 | mailing the developers (or mailing lists) directly -- bugs that are |
| 1454 | not entered into bugzilla tend to get forgotten about or ignored. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1455 | |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 1456 | To see details of a given bug, visit |
| 1457 | https://bugs.kde.org/show_bug.cgi?id=XXXXXX |
| 1458 | where XXXXXX is the bug number as listed below. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1459 | |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 1460 | 84303 How about a LockCheck tool? |
| 1461 | 91633 dereference of null ptr in vgPlain_st_basetype |
| 1462 | 97452 Valgrind doesn't report any pthreads problems |
| 1463 | 100628 leak-check gets assertion failure when using |
| 1464 | VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory |
| 1465 | 108528 NPTL pthread cleanup handlers not called |
| 1466 | 110126 Valgrind 2.4.1 configure.in tramples CFLAGS |
| 1467 | 110128 mallinfo is not implemented... |
| 1468 | 110770 VEX: Generated files not always updated when making valgrind |
| 1469 | 111102 Memcheck: problems with large (memory footprint) applications |
| 1470 | 115673 Vex's decoder should never assert |
| 1471 | 117564 False positive: Syscall param clone(child_tidptr) contains |
| 1472 | uninitialised byte(s) |
| 1473 | 119404 executing ssh from inside valgrind fails |
| 1474 | 133679 Callgrind does not write path names to sources with dwarf debug |
| 1475 | info |
| 1476 | 135847 configure.in problem with non gnu compilers (and possible fix) |
| 1477 | 136154 threads.c:273 (vgCallgrind_post_signal): Assertion |
| 1478 | '*(vgCallgrind_current_fn_stack.top) == 0' failed. |
| 1479 | 136230 memcheck reports "possibly lost", should be "still reachable" |
| 1480 | 137073 NULL arg to MALLOCLIKE_BLOCK causes crash |
| 1481 | 137904 Valgrind reports a memory leak when using POSIX threads, |
| 1482 | while it shouldn't |
| 1483 | 139076 valgrind VT_GETSTATE error |
| 1484 | 142228 complaint of elf_dynamic_do_rela in trivial usage |
| 1485 | 145347 spurious warning with USBDEVFS_REAPURB |
| 1486 | 148441 (wine) can't find memory leak in Wine, win32 binary |
| 1487 | executable file. |
| 1488 | 148742 Leak-check fails assert on exit |
| 1489 | 149878 add (proper) check for calloc integer overflow |
| 1490 | 150606 Call graph is broken when using callgrind control |
| 1491 | 152393 leak errors produce an exit code of 0. I need some way to |
| 1492 | cause leak errors to result in a nonzero exit code. |
| 1493 | 157154 documentation (leak-resolution doc speaks about num-callers |
| 1494 | def=4) + what is a loss record |
| 1495 | 159501 incorrect handling of ALSA ioctls |
| 1496 | 162020 Valgrinding an empty/zero-byte file crashes valgrind |
| 1497 | 162482 ppc: Valgrind crashes while reading stabs information |
| 1498 | 162718 x86: avoid segment selector 0 in sys_set_thread_area() |
| 1499 | 163253 (wine) canonicaliseSymtab forgot some fields in DiSym |
| 1500 | 163560 VEX/test_main.c is missing from valgrind-3.3.1 |
| 1501 | 164353 malloc_usable_size() doesn't return a usable size |
| 1502 | 165468 Inconsistent formatting in memcheck manual -- please fix |
| 1503 | 169505 main.c:286 (endOfInstr): |
| 1504 | Assertion 'ii->cost_offset == *cost_offset' failed |
| 1505 | 177206 Generate default.supp during compile instead of configure |
| 1506 | 177209 Configure valt_load_address based on arch+os |
| 1507 | 177305 eventfd / syscall 323 patch lost |
| 1508 | 179731 Tests fail to build because of inlining of non-local asm labels |
| 1509 | 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion |
| 1510 | 'ordxx == POrd_EQ || ordxx == POrd_LT' failed. |
| 1511 | 181594 Bogus warning for empty text segment |
| 1512 | 181707 dwarf doesn't require enumerations to have name |
| 1513 | 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64 |
| 1514 | 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree): |
| 1515 | Assertion '!already_present' failed. |
| 1516 | 185359 exp-ptrcheck: unhandled syscall getresuid() |
| 1517 | 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64 |
| 1518 | 185816 Valgrind is unable to handle debug info for files with split |
| 1519 | debug info that are prelinked afterwards |
| 1520 | 185980 [darwin] unhandled syscall: sem_open |
| 1521 | 186238 bbToIR_AMD64: disInstr miscalculated next %rip |
| 1522 | 186507 exp-ptrcheck unhandled syscalls prctl, etc. |
| 1523 | 186790 Suppression pattern used for leaks are not reported |
| 1524 | 186796 Symbols with length>200 in suppression files are ignored |
| 1525 | 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation |
| 1526 | 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid} |
| 1527 | 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened |
| 1528 | 188046 bashisms in the configure script |
| 1529 | 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA |
| 1530 | 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672 |
| 1531 | (get_otrack_shadow_offset_wrk): the 'impossible' happened." |
| 1532 | 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock, |
| 1533 | assertion fail "!lock->heldBy" |
| 1534 | 188427 Add support for epoll_create1 (with patch) |
| 1535 | 188530 Support for SIOCGSTAMPNS |
| 1536 | 188560 Include valgrind.spec in the tarball |
| 1537 | 188572 Valgrind on Mac should suppress setenv() mem leak |
| 1538 | 189054 Valgrind fails to build because of duplicate non-local asm labels |
| 1539 | 189737 vex amd64->IR: unhandled instruction bytes: 0xAC |
| 1540 | 189762 epoll_create syscall not handled (--tool=exp-ptrcheck) |
| 1541 | 189763 drd assertion failure: s_threadinfo[tid].is_recording |
| 1542 | 190219 unhandled syscall: 328 (x86-linux) |
| 1543 | 190391 dup of 181394; see above |
| 1544 | 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc |
| 1545 | 190820 No debug information on powerpc-linux |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 1546 | 191095 PATCH: Improve usbdevfs ioctl handling |
| 1547 | 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks |
| 1548 | or big nr of errors |
| 1549 | 191189 --xml=yes should obey --gen-suppressions=all |
| 1550 | 191192 syslog() needs a suppression on macosx |
| 1551 | 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265 |
| 1552 | 191761 getrlimit on MacOSX |
| 1553 | 191992 multiple --fn-skip only works sometimes; dependent on order |
| 1554 | 192634 V. reports "aspacem sync_check_mapping_callback: |
| 1555 | segment mismatch" on Darwin |
| 1556 | 192954 __extension__ missing on 2 client requests |
| 1557 | 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29 |
| 1558 | 194474 "INSTALL" file has different build instructions than "README" |
| 1559 | 194671 Unhandled syscall (sem_wait?) from mac valgrind |
| 1560 | 195069 memcheck: reports leak (memory still reachable) for |
| 1561 | printf("%d', x) |
| 1562 | 195169 drd: (vgDrd_barrier_post_wait): |
| 1563 | Assertion 'r->sg[p->post_iteration]' failed. |
| 1564 | 195268 valgrind --log-file doesn't accept ~/... |
| 1565 | 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate |
| 1566 | 195860 WARNING: unhandled syscall: unix:223 |
| 1567 | 196528 need a error suppression for pthread_rwlock_init under os x? |
| 1568 | 197227 Support aio_* syscalls on Darwin |
| 1569 | 197456 valgrind should reject --suppressions=(directory) |
| 1570 | 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10 |
| 1571 | 197591 unhandled syscall 27 (mincore) |
| 1572 | 197793 Merge DCAS branch to the trunk == 85756, 142103 |
| 1573 | 197794 Avoid duplicate filenames in Vex |
| 1574 | 197898 make check fails on current SVN |
| 1575 | 197901 make check fails also under exp-ptrcheck in current SVN |
| 1576 | 197929 Make --leak-resolution=high the default |
| 1577 | 197930 Reduce spacing between leak reports |
| 1578 | 197933 Print command line of client at start-up, and shorten preamble |
| 1579 | 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck) |
| 1580 | 198395 add BBV to the distribution as an experimental tool |
| 1581 | 198624 Missing syscalls on Darwin: 82, 167, 281, 347 |
| 1582 | 198649 callgrind_annotate doesn't cumulate counters |
| 1583 | 199338 callgrind_annotate sorting/thresholds are broken for all but Ir |
| 1584 | 199977 Valgrind complains about an unrecognized instruction in the |
| 1585 | atomic_incs test program |
| 1586 | 200029 valgrind isn't able to read Fedora 12 debuginfo |
| 1587 | 200760 darwin unhandled syscall: unix:284 |
| 1588 | 200827 DRD doesn't work on Mac OS X |
| 1589 | 200990 VG_(read_millisecond_timer)() does not work correctly |
| 1590 | 201016 Valgrind does not support pthread_kill() on Mac OS |
| 1591 | 201169 Document --read-var-info |
| 1592 | 201323 Pre-3.5.0 performance sanity checking |
| 1593 | 201384 Review user manual for the 3.5.0 release |
| 1594 | 201585 mfpvr not implemented on ppc |
| 1595 | 201708 tests failing because x86 direction flag is left set |
| 1596 | 201757 Valgrind doesn't handle any recent sys_futex additions |
sewardj | 6bbe024 | 2009-08-19 11:38:54 +0000 | [diff] [blame] | 1597 | 204377 64-bit valgrind can not start a shell script |
| 1598 | (with #!/path/to/shell) if the shell is a 32-bit executable |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 1599 | n-i-bz drd: fixed assertion failure triggered by mutex reinitialization. |
| 1600 | n-i-bz drd: fixed a bug that caused incorrect messages to be printed |
| 1601 | about memory allocation events with memory access tracing enabled |
| 1602 | n-i-bz drd: fixed a memory leak triggered by vector clock deallocation |
njn | 254dafe | 2009-02-09 03:25:04 +0000 | [diff] [blame] | 1603 | |
sewardj | 48867fb | 2009-08-19 22:40:53 +0000 | [diff] [blame] | 1604 | (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846). |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 1605 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 1606 | |
njn | 254dafe | 2009-02-09 03:25:04 +0000 | [diff] [blame] | 1607 | |
njn | ad91a75 | 2009-03-13 19:55:36 +0000 | [diff] [blame] | 1608 | Release 3.4.1 (28 February 2009) |
| 1609 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 1610 | 3.4.1 is a bug-fix release that fixes some regressions and assertion |
| 1611 | failures in debug info reading in 3.4.0, most notably incorrect stack |
| 1612 | traces on amd64-linux on older (glibc-2.3 based) systems. Various |
| 1613 | other debug info problems are also fixed. A number of bugs in the |
| 1614 | exp-ptrcheck tool introduced in 3.4.0 have been fixed. |
| 1615 | |
| 1616 | In view of the fact that 3.4.0 contains user-visible regressions |
| 1617 | relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are |
| 1618 | encouraged to ship 3.4.1 in preference to 3.4.0. |
| 1619 | |
| 1620 | The fixed bugs are as follows. Note that "n-i-bz" stands for "not in |
| 1621 | bugzilla" -- that is, a bug that was reported to us but never got a |
| 1622 | bugzilla entry. We encourage you to file bugs in bugzilla |
| 1623 | (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the |
| 1624 | developers (or mailing lists) directly -- bugs that are not entered |
| 1625 | into bugzilla tend to get forgotten about or ignored. |
| 1626 | |
| 1627 | n-i-bz Fix various bugs reading icc-11 generated debug info |
| 1628 | n-i-bz Fix various bugs reading gcc-4.4 generated debug info |
| 1629 | n-i-bz Preliminary support for glibc-2.10 / Fedora 11 |
| 1630 | n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes, |
| 1631 | so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2. |
| 1632 | 179618 exp-ptrcheck crashed / exit prematurely |
| 1633 | 179624 helgrind: false positive races with pthread_create and |
| 1634 | recv/open/close/read |
| 1635 | 134207 pkg-config output contains @VG_PLATFORM@ |
| 1636 | 176926 floating point exception at valgrind startup with PPC 440EPX |
| 1637 | 181594 Bogus warning for empty text segment |
| 1638 | 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes) |
| 1639 | 181707 Dwarf3 doesn't require enumerations to have name |
| 1640 | 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64 |
| 1641 | 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree): |
| 1642 | Assertion '!already_present' failed. |
| 1643 | 185359 exp-ptrcheck unhandled syscall getresuid() |
| 1644 | |
| 1645 | (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253). |
| 1646 | (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293). |
| 1647 | |
| 1648 | |
| 1649 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1650 | Release 3.4.0 (2 January 2009) |
| 1651 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 1652 | 3.4.0 is a feature release with many significant improvements and the |
| 1653 | usual collection of bug fixes. This release supports X86/Linux, |
| 1654 | AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros |
| 1655 | (using gcc 4.4, glibc 2.8 and 2.9) has been added. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1656 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1657 | 3.4.0 brings some significant tool improvements. Memcheck can now |
| 1658 | report the origin of uninitialised values, the thread checkers |
| 1659 | Helgrind and DRD are much improved, and we have a new experimental |
| 1660 | tool, exp-Ptrcheck, which is able to detect overruns of stack and |
| 1661 | global arrays. In detail: |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1662 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1663 | * Memcheck is now able to track the origin of uninitialised values. |
| 1664 | When it reports an uninitialised value error, it will try to show |
| 1665 | the origin of the value, as either a heap or stack allocation. |
| 1666 | Origin tracking is expensive and so is not enabled by default. To |
| 1667 | use it, specify --track-origins=yes. Memcheck's speed will be |
| 1668 | essentially halved, and memory usage will be significantly |
| 1669 | increased. Nevertheless it can drastically reduce the effort |
| 1670 | required to identify the root cause of uninitialised value errors, |
| 1671 | and so is often a programmer productivity win, despite running more |
| 1672 | slowly. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1673 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1674 | * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in |
sewardj | b8b1664 | 2009-01-02 23:21:54 +0000 | [diff] [blame] | 1675 | 3.4.0, will be released shortly. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1676 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1677 | * Helgrind's race detection algorithm has been completely redesigned |
| 1678 | and reimplemented, to address usability and scalability concerns: |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1679 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1680 | - The new algorithm has a lower false-error rate: it is much less |
| 1681 | likely to report races that do not really exist. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1682 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1683 | - Helgrind will display full call stacks for both accesses involved |
sewardj | f1fa5ed | 2008-12-23 17:04:52 +0000 | [diff] [blame] | 1684 | in a race. This makes it easier to identify the root causes of |
| 1685 | races. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1686 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1687 | - Limitations on the size of program that can run have been removed. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1688 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1689 | - Performance has been modestly improved, although that is very |
| 1690 | workload-dependent. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1691 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1692 | - Direct support for Qt4 threading has been added. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1693 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1694 | - pthread_barriers are now directly supported. |
njn | 75010ca | 2008-02-09 05:10:45 +0000 | [diff] [blame] | 1695 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1696 | - Helgrind works well on all supported Linux targets. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1697 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1698 | * The DRD thread debugging tool has seen major improvements: |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1699 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1700 | - Greatly improved performance and significantly reduced memory |
| 1701 | usage. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1702 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1703 | - Support for several major threading libraries (Boost.Thread, Qt4, |
| 1704 | glib, OpenMP) has been added. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 1705 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1706 | - Support for atomic instructions, POSIX semaphores, barriers and |
| 1707 | reader-writer locks has been added. |
| 1708 | |
| 1709 | - Works now on PowerPC CPUs too. |
| 1710 | |
| 1711 | - Added support for printing thread stack usage at thread exit time. |
| 1712 | |
| 1713 | - Added support for debugging lock contention. |
| 1714 | |
| 1715 | - Added a manual for Drd. |
| 1716 | |
| 1717 | * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck |
| 1718 | checks for misuses of pointers. In that sense it is a bit like |
| 1719 | Memcheck. However, Ptrcheck can do things Memcheck can't: it can |
| 1720 | detect overruns of stack and global arrays, it can detect |
| 1721 | arbitrarily far out-of-bounds accesses to heap blocks, and it can |
| 1722 | detect accesses heap blocks that have been freed a very long time |
| 1723 | ago (millions of blocks in the past). |
| 1724 | |
| 1725 | Ptrcheck currently works only on x86-linux and amd64-linux. To use |
| 1726 | it, use --tool=exp-ptrcheck. A simple manual is provided, as part |
| 1727 | of the main Valgrind documentation. As this is an experimental |
| 1728 | tool, we would be particularly interested in hearing about your |
| 1729 | experiences with it. |
| 1730 | |
sewardj | 974aace | 2008-12-24 18:40:57 +0000 | [diff] [blame] | 1731 | * exp-Omega, an experimental instantaneous leak-detecting tool, is no |
| 1732 | longer built by default, although the code remains in the repository |
| 1733 | and the tarball. This is due to three factors: a perceived lack of |
| 1734 | users, a lack of maintenance, and concerns that it may not be |
| 1735 | possible to achieve reliable operation using the existing design. |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1736 | |
| 1737 | * As usual, support for the latest Linux distros and toolchain |
| 1738 | components has been added. It should work well on Fedora Core 10, |
| 1739 | OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release |
| 1740 | state) is supported, as is glibc-2.9. The C++ demangler has been |
| 1741 | updated so as to work well with C++ compiled by even the most recent |
| 1742 | g++'s. |
| 1743 | |
| 1744 | * You can now use frame-level wildcards in suppressions. This was a |
| 1745 | frequently-requested enhancement. A line "..." in a suppression now |
| 1746 | matches zero or more frames. This makes it easier to write |
| 1747 | suppressions which are precise yet insensitive to changes in |
| 1748 | inlining behaviour. |
| 1749 | |
| 1750 | * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set. |
| 1751 | |
| 1752 | * Very basic support for IBM Power6 has been added (64-bit processes only). |
| 1753 | |
| 1754 | * Valgrind is now cross-compilable. For example, it is possible to |
| 1755 | cross compile Valgrind on an x86/amd64-linux host, so that it runs |
| 1756 | on a ppc32/64-linux target. |
| 1757 | |
sewardj | f1fa5ed | 2008-12-23 17:04:52 +0000 | [diff] [blame] | 1758 | * You can set the main thread's stack size at startup using the |
| 1759 | new --main-stacksize= flag (subject of course to ulimit settings). |
| 1760 | This is useful for running apps that need a lot of stack space. |
| 1761 | |
| 1762 | * The limitation that you can't use --trace-children=yes together |
| 1763 | with --db-attach=yes has been removed. |
| 1764 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1765 | * The following bugs have been fixed. Note that "n-i-bz" stands for |
| 1766 | "not in bugzilla" -- that is, a bug that was reported to us but |
| 1767 | never got a bugzilla entry. We encourage you to file bugs in |
| 1768 | bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than |
| 1769 | mailing the developers (or mailing lists) directly. |
| 1770 | |
sewardj | f1fa5ed | 2008-12-23 17:04:52 +0000 | [diff] [blame] | 1771 | n-i-bz Make return types for some client requests 64-bit clean |
| 1772 | n-i-bz glibc 2.9 support |
| 1773 | n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865) |
| 1774 | n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults |
| 1775 | n-i-bz Building in an env without gdb gives bogus gdb attach |
| 1776 | 92456 Tracing the origin of uninitialised memory |
| 1777 | 106497 Valgrind does not demangle some C++ template symbols |
| 1778 | 162222 ==106497 |
sewardj | df1fc9a | 2008-12-23 23:09:57 +0000 | [diff] [blame] | 1779 | 151612 Suppression with "..." (frame-level wildcards in .supp files) |
sewardj | f1fa5ed | 2008-12-23 17:04:52 +0000 | [diff] [blame] | 1780 | 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit) |
| 1781 | 159285 unhandled syscall:25 (stime, on x86-linux) |
| 1782 | 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig" |
| 1783 | 160954 ppc build of valgrind crashes with illegal instruction (isel) |
| 1784 | 160956 mallinfo implementation, w/ patch |
| 1785 | 162092 Valgrind fails to start gnome-system-monitor |
| 1786 | 162819 malloc_free_fill test doesn't pass on glibc2.8 x86 |
| 1787 | 163794 assertion failure with "--track-origins=yes" |
| 1788 | 163933 sigcontext.err and .trapno must be set together |
| 1789 | 163955 remove constraint !(--db-attach=yes && --trace-children=yes) |
| 1790 | 164476 Missing kernel module loading system calls |
| 1791 | 164669 SVN regression: mmap() drops posix file locks |
| 1792 | 166581 Callgrind output corruption when program forks |
| 1793 | 167288 Patch file for missing system calls on Cell BE |
| 1794 | 168943 unsupported scas instruction pentium |
| 1795 | 171645 Unrecognised instruction (MOVSD, non-binutils encoding) |
| 1796 | 172417 x86->IR: 0x82 ... |
| 1797 | 172563 amd64->IR: 0xD9 0xF5 - fprem1 |
| 1798 | 173099 .lds linker script generation error |
| 1799 | 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl) |
| 1800 | 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes) |
| 1801 | 174532 == 173751 |
| 1802 | 174908 --log-file value not expanded correctly for core file |
| 1803 | 175044 Add lookup_dcookie for amd64 |
| 1804 | 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding) |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 1805 | |
| 1806 | Developer-visible changes: |
| 1807 | |
| 1808 | * Valgrind's debug-info reading machinery has been majorly overhauled. |
| 1809 | It can now correctly establish the addresses for ELF data symbols, |
| 1810 | which is something that has never worked properly before now. |
| 1811 | |
| 1812 | Also, Valgrind can now read DWARF3 type and location information for |
| 1813 | stack and global variables. This makes it possible to use the |
| 1814 | framework to build tools that rely on knowing the type and locations |
| 1815 | of stack and global variables, for example exp-Ptrcheck. |
| 1816 | |
| 1817 | Reading of such information is disabled by default, because most |
| 1818 | tools don't need it, and because it is expensive in space and time. |
| 1819 | However, you can force Valgrind to read it, using the |
| 1820 | --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to |
| 1821 | make use of such information, if present, to provide source-level |
| 1822 | descriptions of data addresses in the error messages they create. |
| 1823 | |
sewardj | 974aace | 2008-12-24 18:40:57 +0000 | [diff] [blame] | 1824 | (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882). |
sewardj | b8b1664 | 2009-01-02 23:21:54 +0000 | [diff] [blame] | 1825 | (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899). |