jseward | 4323065 | 2003-12-14 00:14:54 +0000 | [diff] [blame] | 1 | |
jseward | 9cd978c | 2004-07-17 13:38:12 +0000 | [diff] [blame] | 2 | Developer (cvs head) release 2.1.2 (18 July 2004) |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 3 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 4 | 2.1.2 contains four months worth of bug fixes and refinements. |
jseward | 9cd978c | 2004-07-17 13:38:12 +0000 | [diff] [blame] | 5 | Although officially a developer release, we believe it to be stable |
| 6 | enough for widespread day-to-day use. 2.1.2 is pretty good, so try it |
| 7 | first, although there is a chance it won't work. If so then try 2.0.0 |
| 8 | and tell us what went wrong." 2.1.2 fixes a lot of problems present |
| 9 | in 2.0.0 and is generally a much better product. |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 10 | |
jseward | 9cd978c | 2004-07-17 13:38:12 +0000 | [diff] [blame] | 11 | Relative to 2.1.1, a large number of minor problems with 2.1.1 have |
| 12 | been fixed, and so if you use 2.1.1 you should try 2.1.2. Users of |
| 13 | the last stable release, 2.0.0, might also want to try this release. |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 14 | |
| 15 | The following bugs, and probably many more, have been fixed. These |
| 16 | are listed at http://bugs.kde.org. Reporting a bug for valgrind in |
| 17 | the http://bugs.kde.org is much more likely to get you a fix than |
| 18 | mailing developers directly, so please continue to keep sending bugs |
| 19 | there. |
| 20 | |
| 21 | 76869 Crashes when running any tool under Fedora Core 2 test1 |
| 22 | This fixes the problem with returning from a signal handler |
jseward | 0400d4d | 2004-07-17 23:26:46 +0000 | [diff] [blame] | 23 | when VDSOs are turned off in FC2. |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 24 | |
jseward | 0400d4d | 2004-07-17 23:26:46 +0000 | [diff] [blame] | 25 | 69508 java 1.4.2 client fails with erroneous "stack size too small". |
| 26 | This fix makes more of the pthread stack attribute related |
| 27 | functions work properly. Java still doesn't work though. |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 28 | |
| 29 | 71906 malloc alignment should be 8, not 4 |
| 30 | All memory returned by malloc/new etc is now at least |
| 31 | 8-byte aligned. |
| 32 | |
jseward | a6017a4 | 2004-07-17 23:44:35 +0000 | [diff] [blame] | 33 | 81970 vg_alloc_ThreadState: no free slots available |
| 34 | (closed because the workaround is simple: increase |
| 35 | VG_N_THREADS, rebuild and try again.) |
| 36 | |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 37 | 78514 Conditional jump or move depends on uninitialized value(s) |
| 38 | (a slight mishanding of FP code in memcheck) |
| 39 | |
jseward | a6017a4 | 2004-07-17 23:44:35 +0000 | [diff] [blame] | 40 | 77952 pThread Support (crash) (due to initialisation-ordering probs) |
| 41 | (also 85118) |
| 42 | |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 43 | 80942 Addrcheck wasn't doing overlap checking as it should. |
| 44 | 78048 return NULL on malloc/new etc failure, instead of asserting |
| 45 | 73655 operator new() override in user .so files often doesn't get picked up |
| 46 | 83060 Valgrind does not handle native kernel AIO |
| 47 | 69872 Create proper coredumps after fatal signals |
| 48 | 82026 failure with new glibc versions: __libc_* functions are not exported |
| 49 | 70344 UNIMPLEMENTED FUNCTION: tcdrain |
| 50 | 81297 Cancellation of pthread_cond_wait does not require mutex |
| 51 | 82872 Using debug info from additional packages (wishlist) |
| 52 | 83025 Support for ioctls FIGETBSZ and FIBMAP |
| 53 | 83340 Support for ioctl HDIO_GET_IDENTITY |
| 54 | 79714 Support for the semtimedop system call. |
| 55 | 77022 Support for ioctls FBIOGET_VSCREENINFO and FBIOGET_FSCREENINFO |
| 56 | 82098 hp2ps ansification (wishlist) |
| 57 | 83573 Valgrind SIGSEGV on execve |
| 58 | 82999 show which cmdline option was erroneous (wishlist) |
| 59 | 83040 make valgrind VPATH and distcheck-clean (wishlist) |
jseward | a6017a4 | 2004-07-17 23:44:35 +0000 | [diff] [blame] | 60 | 83998 Assertion `newfd > vgPlain_max_fd' failed (see below) |
| 61 | 82722 Unchecked mmap in as_pad leads to mysterious failures later |
| 62 | 78958 memcheck seg faults while running Mozilla |
jseward | a2c0800 | 2004-07-18 09:24:57 +0000 | [diff] [blame] | 63 | 85416 Arguments with colon (e.g. --logsocket) ignored |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 64 | |
| 65 | |
| 66 | Additionally there are the following changes, which are not |
| 67 | connected to any bug report numbers, AFAICS: |
| 68 | |
| 69 | * Rearranged address space layout relative to 2.1.1, so that |
| 70 | Valgrind/tools will run out of memory later than currently in many |
| 71 | circumstances. This is good news esp. for Calltree. It should |
| 72 | be possible for client programs to allocate over 800MB of |
| 73 | memory when using memcheck now. |
| 74 | |
jseward | 9cd978c | 2004-07-17 13:38:12 +0000 | [diff] [blame] | 75 | * Improved checking when laying out memory. Should hopefully avoid |
| 76 | the random segmentation faults that 2.1.1 sometimes caused. |
| 77 | |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 78 | * Support for Fedora Core 2 and SuSE 9.1. Improvements to NPTL |
| 79 | support to the extent that V now works properly on NPTL-only setups. |
| 80 | |
| 81 | * Renamed the following options: |
| 82 | --logfile-fd --> --log-fd |
| 83 | --logfile --> --log-file |
| 84 | --logsocket --> --log-socket |
| 85 | to be consistent with each other and other options (esp. --input-fd). |
| 86 | |
| 87 | * Add support for SIOCGMIIPHY, SIOCGMIIREG and SIOCSMIIREG ioctls and |
| 88 | improve the checking of other interface related ioctls. |
| 89 | |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 90 | * Fix building with gcc-3.4.1. |
| 91 | |
| 92 | * Remove limit on number of semaphores supported. |
| 93 | |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 94 | * Add support for syscalls: set_tid_address (258), acct (51). |
| 95 | |
| 96 | * Support instruction "repne movs" -- not official but seems to occur. |
| 97 | |
| 98 | * Implement an emulated soft limit for file descriptors in addition to |
| 99 | the current reserved area, which effectively acts as a hard limit. The |
| 100 | setrlimit system call now simply updates the emulated limits as best |
| 101 | as possible - the hard limit is not allowed to move at all and just |
| 102 | returns EPERM if you try and change it. This should stop reductions |
| 103 | in the soft limit causing assertions when valgrind tries to allocate |
| 104 | descriptors from the reserved area. |
jseward | a6017a4 | 2004-07-17 23:44:35 +0000 | [diff] [blame] | 105 | (This actually came from bug #83998). |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 106 | |
nethercote | 5e63bfb | 2004-07-17 12:49:53 +0000 | [diff] [blame] | 107 | * Major overhaul of Cachegrind implementation. Only user visible change |
| 108 | is that cachegrind.out files are now typically 90% smaller than they |
| 109 | used to be. Code annotation times are correspondingly much smaller. |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 110 | |
| 111 | * Client requests for telling valgrind about memory pools. |
| 112 | |
| 113 | |
| 114 | |
jseward | 9cd978c | 2004-07-17 13:38:12 +0000 | [diff] [blame] | 115 | Developer (cvs head) release 2.1.1 (12 March 2004) |
| 116 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 117 | 2.1.1 contains some internal structural changes needed for V's |
jseward | e3e7704 | 2004-03-12 21:07:05 +0000 | [diff] [blame] | 118 | long-term future. These don't affect end-users. Most notable |
| 119 | user-visible changes are: |
| 120 | |
| 121 | * Greater isolation between Valgrind and the program being run, so |
| 122 | the program is less likely to inadvertently kill Valgrind by |
| 123 | doing wild writes. |
| 124 | |
| 125 | * Massif: a new space profiling tool. Try it! It's cool, and it'll |
| 126 | tell you in detail where and when your C/C++ code is allocating heap. |
| 127 | Draws pretty .ps pictures of memory use against time. A potentially |
| 128 | powerful tool for making sense of your program's space use. |
| 129 | |
| 130 | * Fixes for many bugs, including support for more SSE2/SSE3 instructions, |
| 131 | various signal/syscall things, and various problems with debug |
| 132 | info readers. |
| 133 | |
| 134 | * Support for glibc-2.3.3 based systems. |
| 135 | |
| 136 | We are now doing automatic overnight build-and-test runs on a variety |
| 137 | of distros. As a result, we believe 2.1.1 builds and runs on: |
| 138 | Red Hat 7.2, 7.3, 8.0, 9, Fedora Core 1, SuSE 8.2, SuSE 9. |
| 139 | |
| 140 | |
| 141 | The following bugs, and probably many more, have been fixed. These |
| 142 | are listed at http://bugs.kde.org. Reporting a bug for valgrind in |
| 143 | the http://bugs.kde.org is much more likely to get you a fix than |
| 144 | mailing developers directly, so please continue to keep sending bugs |
| 145 | there. |
| 146 | |
| 147 | 69616 glibc 2.3.2 w/NPTL is massively different than what valgrind expects |
| 148 | 69856 I don't know how to instrument MMXish stuff (Helgrind) |
| 149 | 73892 valgrind segfaults starting with Objective-C debug info |
| 150 | (fix for S-type stabs) |
| 151 | 73145 Valgrind complains too much about close(<reserved fd>) |
| 152 | 73902 Shadow memory allocation seems to fail on RedHat 8.0 |
| 153 | 68633 VG_N_SEMAPHORES too low (V itself was leaking semaphores) |
| 154 | 75099 impossible to trace multiprocess programs |
| 155 | 76839 the `impossible' happened: disInstr: INT but not 0x80 ! |
| 156 | 76762 vg_to_ucode.c:3748 (dis_push_segreg): Assertion `sz == 4' failed. |
| 157 | 76747 cannot include valgrind.h in c++ program |
| 158 | 76223 parsing B(3,10) gave NULL type => impossible happens |
| 159 | 75604 shmdt handling problem |
| 160 | 76416 Problems with gcc 3.4 snap 20040225 |
| 161 | 75614 using -gstabs when building your programs the `impossible' happened |
| 162 | 75787 Patch for some CDROM ioctls CDORM_GET_MCN, CDROM_SEND_PACKET, |
| 163 | 75294 gcc 3.4 snapshot's libstdc++ have unsupported instructions. |
| 164 | (REP RET) |
| 165 | 73326 vg_symtab2.c:272 (addScopeRange): Assertion `range->size > 0' failed. |
| 166 | 72596 not recognizing __libc_malloc |
| 167 | 69489 Would like to attach ddd to running program |
| 168 | 72781 Cachegrind crashes with kde programs |
| 169 | 73055 Illegal operand at DXTCV11CompressBlockSSE2 (more SSE opcodes) |
| 170 | 73026 Descriptor leak check reports port numbers wrongly |
| 171 | 71705 README_MISSING_SYSCALL_OR_IOCTL out of date |
| 172 | 72643 Improve support for SSE/SSE2 instructions |
| 173 | 72484 valgrind leaves it's own signal mask in place when execing |
| 174 | 72650 Signal Handling always seems to restart system calls |
| 175 | 72006 The mmap system call turns all errors in ENOMEM |
| 176 | 71781 gdb attach is pretty useless |
| 177 | 71180 unhandled instruction bytes: 0xF 0xAE 0x85 0xE8 |
| 178 | 69886 writes to zero page cause valgrind to assert on exit |
| 179 | 71791 crash when valgrinding gimp 1.3 (stabs reader problem) |
| 180 | 69783 unhandled syscall: 218 |
| 181 | 69782 unhandled instruction bytes: 0x66 0xF 0x2B 0x80 |
| 182 | 70385 valgrind fails if the soft file descriptor limit is less |
| 183 | than about 828 |
| 184 | 69529 "rep; nop" should do a yield |
| 185 | 70827 programs with lots of shared libraries report "mmap failed" |
| 186 | for some of them when reading symbols |
| 187 | 71028 glibc's strnlen is optimised enough to confuse valgrind |
| 188 | |
| 189 | |
| 190 | |
| 191 | |
jseward | 4323065 | 2003-12-14 00:14:54 +0000 | [diff] [blame] | 192 | Unstable (cvs head) release 2.1.0 (15 December 2003) |
| 193 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 194 | For whatever it's worth, 2.1.0 actually seems pretty darn stable to me |
| 195 | (Julian). It looks eminently usable, and given that it fixes some |
| 196 | significant bugs, may well be worth using on a day-to-day basis. |
| 197 | 2.1.0 is known to build and pass regression tests on: SuSE 9, SuSE |
| 198 | 8.2, RedHat 8. |
| 199 | |
| 200 | 2.1.0 most notably includes Jeremy Fitzhardinge's complete overhaul of |
| 201 | handling of system calls and signals, and their interaction with |
| 202 | threads. In general, the accuracy of the system call, thread and |
| 203 | signal simulations is much improved. Specifically: |
| 204 | |
| 205 | - Blocking system calls behave exactly as they do when running |
| 206 | natively (not on valgrind). That is, if a syscall blocks only the |
| 207 | calling thread when running natively, than it behaves the same on |
| 208 | valgrind. No more mysterious hangs because V doesn't know that some |
| 209 | syscall or other, should block only the calling thread. |
| 210 | |
| 211 | - Interrupted syscalls should now give more faithful results. |
| 212 | |
| 213 | - Finally, signal contexts in signal handlers are supported. As a |
| 214 | result, konqueror on SuSE 9 no longer segfaults when notified of |
| 215 | file changes in directories it is watching. |
| 216 | |
| 217 | Other changes: |
| 218 | |
| 219 | - Robert Walsh's file descriptor leakage checks. When enabled, |
| 220 | Valgrind will print out a list of open file descriptors on |
| 221 | exit. Along with each file descriptor, Valgrind prints out a stack |
| 222 | backtrace of where the file was opened and any details relating to the |
| 223 | file descriptor such as the file name or socket details. |
| 224 | To use, give: --track-fds=yes |
| 225 | |
| 226 | - Implemented a few more SSE/SSE2 instructions. |
| 227 | |
| 228 | - Less crud on the stack when you do 'where' inside a GDB attach. |
| 229 | |
| 230 | - Fixed the following bugs: |
| 231 | 68360: Valgrind does not compile against 2.6.0-testX kernels |
| 232 | 68525: CVS head doesn't compile on C90 compilers |
| 233 | 68566: pkgconfig support (wishlist) |
| 234 | 68588: Assertion `sz == 4' failed in vg_to_ucode.c (disInstr) |
| 235 | 69140: valgrind not able to explicitly specify a path to a binary. |
| 236 | 69432: helgrind asserts encountering a MutexErr when there are |
| 237 | EraserErr suppressions |
| 238 | |
jseward | 0b12daf | 2003-12-14 14:35:28 +0000 | [diff] [blame] | 239 | - Increase the max size of the translation cache from 200k average bbs |
| 240 | to 300k average bbs. Programs on the size of OOo (680m17) are |
| 241 | thrashing the cache at the smaller size, creating large numbers of |
| 242 | retranslations and wasting significant time as a result. |
| 243 | |
jseward | 4323065 | 2003-12-14 00:14:54 +0000 | [diff] [blame] | 244 | |
| 245 | |
| 246 | Stable release 2.0.0 (5 Nov 2003) |
| 247 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 248 | |
| 249 | 2.0.0 improves SSE/SSE2 support, fixes some minor bugs, and |
| 250 | improves support for SuSE 9 and the Red Hat "Severn" beta. |
| 251 | |
| 252 | - Further improvements to SSE/SSE2 support. The entire test suite of |
| 253 | the GNU Scientific Library (gsl-1.4) compiled with Intel Icc 7.1 |
| 254 | 20030307Z '-g -O -xW' now works. I think this gives pretty good |
| 255 | coverage of SSE/SSE2 floating point instructions, or at least the |
| 256 | subset emitted by Icc. |
| 257 | |
| 258 | - Also added support for the following instructions: |
| 259 | MOVNTDQ UCOMISD UNPCKLPS UNPCKHPS SQRTSS |
| 260 | PUSH/POP %{FS,GS}, and PUSH %CS (Nb: there is no POP %CS). |
| 261 | |
| 262 | - CFI support for GDB version 6. Needed to enable newer GDBs |
| 263 | to figure out where they are when using --gdb-attach=yes. |
| 264 | |
| 265 | - Fix this: |
| 266 | mc_translate.c:1091 (memcheck_instrument): Assertion |
| 267 | `u_in->size == 4 || u_in->size == 16' failed. |
| 268 | |
| 269 | - Return an error rather than panicing when given a bad socketcall. |
| 270 | |
| 271 | - Fix checking of syscall rt_sigtimedwait(). |
| 272 | |
| 273 | - Implement __NR_clock_gettime (syscall 265). Needed on Red Hat Severn. |
| 274 | |
| 275 | - Fixed bug in overlap check in strncpy() -- it was assuming the src was 'n' |
| 276 | bytes long, when it could be shorter, which could cause false |
| 277 | positives. |
| 278 | |
| 279 | - Support use of select() for very large numbers of file descriptors. |
| 280 | |
| 281 | - Don't fail silently if the executable is statically linked, or is |
| 282 | setuid/setgid. Print an error message instead. |
| 283 | |
| 284 | - Support for old DWARF-1 format line number info. |
| 285 | |
| 286 | |
| 287 | |
| 288 | Snapshot 20031012 (12 October 2003) |
| 289 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 290 | |
| 291 | Three months worth of bug fixes, roughly. Most significant single |
| 292 | change is improved SSE/SSE2 support, mostly thanks to Dirk Mueller. |
| 293 | |
| 294 | 20031012 builds on Red Hat Fedora ("Severn") but doesn't really work |
| 295 | (curiosly, mozilla runs OK, but a modest "ls -l" bombs). I hope to |
| 296 | get a working version out soon. It may or may not work ok on the |
| 297 | forthcoming SuSE 9; I hear positive noises about it but haven't been |
| 298 | able to verify this myself (not until I get hold of a copy of 9). |
| 299 | |
| 300 | A detailed list of changes, in no particular order: |
| 301 | |
| 302 | - Describe --gen-suppressions in the FAQ. |
| 303 | |
| 304 | - Syscall __NR_waitpid supported. |
| 305 | |
| 306 | - Minor MMX bug fix. |
| 307 | |
| 308 | - -v prints program's argv[] at startup. |
| 309 | |
| 310 | - More glibc-2.3 suppressions. |
| 311 | |
| 312 | - Suppressions for stack underrun bug(s) in the c++ support library |
| 313 | distributed with Intel Icc 7.0. |
| 314 | |
| 315 | - Fix problems reading /proc/self/maps. |
| 316 | |
| 317 | - Fix a couple of messages that should have been suppressed by -q, |
| 318 | but weren't. |
| 319 | |
| 320 | - Make Addrcheck understand "Overlap" suppressions. |
| 321 | |
| 322 | - At startup, check if program is statically linked and bail out if so. |
| 323 | |
| 324 | - Cachegrind: Auto-detect Intel Pentium-M, also VIA Nehemiah |
| 325 | |
| 326 | - Memcheck/addrcheck: minor speed optimisations |
| 327 | |
| 328 | - Handle syscall __NR_brk more correctly than before. |
| 329 | |
| 330 | - Fixed incorrect allocate/free mismatch errors when using |
| 331 | operator new(unsigned, std::nothrow_t const&) |
| 332 | operator new[](unsigned, std::nothrow_t const&) |
| 333 | |
| 334 | - Support POSIX pthread spinlocks. |
| 335 | |
| 336 | - Fixups for clean compilation with gcc-3.3.1. |
daywalker | 5d945de | 2003-09-26 00:32:53 +0000 | [diff] [blame] | 337 | |
| 338 | - Implemented more opcodes: |
| 339 | - push %es |
| 340 | - push %ds |
| 341 | - pop %es |
| 342 | - pop %ds |
| 343 | - movntq |
| 344 | - sfence |
| 345 | - pshufw |
| 346 | - pavgb |
| 347 | - ucomiss |
| 348 | - enter |
daywalker | b18d253 | 2003-09-27 20:15:01 +0000 | [diff] [blame] | 349 | - mov imm32, %esp |
| 350 | - all "in" and "out" opcodes |
daywalker | 79aad84 | 2003-09-30 22:58:12 +0000 | [diff] [blame] | 351 | - inc/dec %esp |
jseward | 4323065 | 2003-12-14 00:14:54 +0000 | [diff] [blame] | 352 | - A whole bunch of SSE/SSE2 instructions |
daywalker | 5d945de | 2003-09-26 00:32:53 +0000 | [diff] [blame] | 353 | |
jseward | 4323065 | 2003-12-14 00:14:54 +0000 | [diff] [blame] | 354 | - Memcheck: don't bomb on SSE/SSE2 code. |
daywalker | 5d945de | 2003-09-26 00:32:53 +0000 | [diff] [blame] | 355 | |
sewardj | 9023879 | 2003-05-05 00:23:42 +0000 | [diff] [blame] | 356 | |
sewardj | 945f39f | 2003-07-25 21:11:40 +0000 | [diff] [blame] | 357 | Snapshot 20030725 (25 July 2003) |
| 358 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 359 | |
| 360 | Fixes some minor problems in 20030716. |
| 361 | |
| 362 | - Fix bugs in overlap checking for strcpy/memcpy etc. |
| 363 | |
| 364 | - Do overlap checking with Addrcheck as well as Memcheck. |
| 365 | |
| 366 | - Fix this: |
| 367 | Memcheck: the `impossible' happened: |
| 368 | get_error_name: unexpected type |
| 369 | |
| 370 | - Install headers needed to compile new skins. |
| 371 | |
sewardj | e3dd2e0 | 2003-07-27 08:04:48 +0000 | [diff] [blame] | 372 | - Remove leading spaces and colon in the LD_LIBRARY_PATH / LD_PRELOAD |
sewardj | 945f39f | 2003-07-25 21:11:40 +0000 | [diff] [blame] | 373 | passed to non-traced children. |
| 374 | |
sewardj | b9eda95 | 2003-07-26 21:39:05 +0000 | [diff] [blame] | 375 | - Fix file descriptor leak in valgrind-listener. |
| 376 | |
sewardj | e3dd2e0 | 2003-07-27 08:04:48 +0000 | [diff] [blame] | 377 | - Fix longstanding bug in which the allocation point of a |
| 378 | block resized by realloc was not correctly set. This may |
| 379 | have caused confusing error messages. |
sewardj | 945f39f | 2003-07-25 21:11:40 +0000 | [diff] [blame] | 380 | |
| 381 | |
sewardj | 626fd89 | 2003-07-16 20:10:26 +0000 | [diff] [blame] | 382 | Snapshot 20030716 (16 July 2003) |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 383 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 384 | |
sewardj | 626fd89 | 2003-07-16 20:10:26 +0000 | [diff] [blame] | 385 | 20030716 is a snapshot of our current CVS head (development) branch. |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 386 | This is the branch which will become valgrind-2.0. It contains |
| 387 | significant enhancements over the 1.9.X branch. |
| 388 | |
| 389 | Despite this being a snapshot of the CVS head, it is believed to be |
| 390 | quite stable -- at least as stable as 1.9.6 or 1.0.4, if not more so |
| 391 | -- and therefore suitable for widespread use. Please let us know asap |
| 392 | if it causes problems for you. |
| 393 | |
| 394 | Two reasons for releasing a snapshot now are: |
| 395 | |
| 396 | - It's been a while since 1.9.6, and this snapshot fixes |
| 397 | various problems that 1.9.6 has with threaded programs |
| 398 | on glibc-2.3.X based systems. |
| 399 | |
| 400 | - So as to make available improvements in the 2.0 line. |
| 401 | |
sewardj | 626fd89 | 2003-07-16 20:10:26 +0000 | [diff] [blame] | 402 | Major changes in 20030716, as compared to 1.9.6: |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 403 | |
| 404 | - More fixes to threading support on glibc-2.3.1 and 2.3.2-based |
| 405 | systems (SuSE 8.2, Red Hat 9). If you have had problems |
| 406 | with inconsistent/illogical behaviour of errno, h_errno or the DNS |
sewardj | 626fd89 | 2003-07-16 20:10:26 +0000 | [diff] [blame] | 407 | resolver functions in threaded programs, 20030716 should improve |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 408 | matters. This snapshot seems stable enough to run OpenOffice.org |
| 409 | 1.1rc on Red Hat 7.3, SuSE 8.2 and Red Hat 9, and that's a big |
| 410 | threaded app if ever I saw one. |
| 411 | |
| 412 | - Automatic generation of suppression records; you no longer |
| 413 | need to write them by hand. Use --gen-suppressions=yes. |
| 414 | |
sewardj | 2151180 | 2003-07-22 17:47:42 +0000 | [diff] [blame] | 415 | - strcpy/memcpy/etc check their arguments for overlaps, when |
| 416 | running with the Memcheck or Addrcheck skins. |
| 417 | |
| 418 | - malloc_usable_size() is now supported. |
| 419 | |
| 420 | - new client requests: |
| 421 | - VALGRIND_COUNT_ERRORS, VALGRIND_COUNT_LEAKS: |
| 422 | useful with regression testing |
| 423 | - VALGRIND_NON_SIMD_CALL[0123]: for running arbitrary functions |
| 424 | on real CPU (use with caution!) |
| 425 | |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 426 | - The GDB attach mechanism is more flexible. Allow the GDB to |
| 427 | be run to be specified by --gdb-path=/path/to/gdb, and specify |
| 428 | which file descriptor V will read its input from with |
| 429 | --input-fd=<number>. |
| 430 | |
sewardj | 2151180 | 2003-07-22 17:47:42 +0000 | [diff] [blame] | 431 | - Cachegrind gives more accurate results (wasn't tracking instructions in |
| 432 | malloc() and friends previously, is now). |
| 433 | |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 434 | - Complete support for the MMX instruction set. |
| 435 | |
| 436 | - Partial support for the SSE and SSE2 instruction sets. Work for this |
| 437 | is ongoing. About half the SSE/SSE2 instructions are done, so |
| 438 | some SSE based programs may work. Currently you need to specify |
| 439 | --skin=addrcheck. Basically not suitable for real use yet. |
| 440 | |
| 441 | - Significant speedups (10%-20%) for standard memory checking. |
| 442 | |
| 443 | - Fix assertion failure in pthread_once(). |
| 444 | |
| 445 | - Fix this: |
| 446 | valgrind: vg_intercept.c:598 (vgAllRoadsLeadToRome_select): |
| 447 | Assertion `ms_end >= ms_now' failed. |
| 448 | |
| 449 | - Implement pthread_mutexattr_setpshared. |
| 450 | |
| 451 | - Understand Pentium 4 branch hints. Also implemented a couple more |
| 452 | obscure x86 instructions. |
| 453 | |
| 454 | - Lots of other minor bug fixes. |
| 455 | |
sewardj | 626fd89 | 2003-07-16 20:10:26 +0000 | [diff] [blame] | 456 | - We have a decent regression test system, for the first time. |
| 457 | This doesn't help you directly, but it does make it a lot easier |
| 458 | for us to track the quality of the system, especially across |
| 459 | multiple linux distributions. |
| 460 | |
| 461 | You can run the regression tests with 'make regtest' after 'make |
| 462 | install' completes. On SuSE 8.2 and Red Hat 9 I get this: |
| 463 | |
| 464 | == 84 tests, 0 stderr failures, 0 stdout failures == |
| 465 | |
| 466 | On Red Hat 8, I get this: |
| 467 | |
| 468 | == 84 tests, 2 stderr failures, 1 stdout failure == |
| 469 | corecheck/tests/res_search (stdout) |
| 470 | memcheck/tests/sigaltstack (stderr) |
| 471 | |
| 472 | sigaltstack is probably harmless. res_search doesn't work |
| 473 | on R H 8 even running natively, so I'm not too worried. |
| 474 | |
| 475 | On Red Hat 7.3, a glibc-2.2.5 system, I get these harmless failures: |
| 476 | |
| 477 | == 84 tests, 2 stderr failures, 1 stdout failure == |
| 478 | corecheck/tests/pth_atfork1 (stdout) |
| 479 | corecheck/tests/pth_atfork1 (stderr) |
| 480 | memcheck/tests/sigaltstack (stderr) |
| 481 | |
| 482 | You need to run on a PII system, at least, since some tests |
| 483 | contain P6-specific instructions, and the test machine needs |
| 484 | access to the internet so that corecheck/tests/res_search |
| 485 | (a test that the DNS resolver works) can function. |
| 486 | |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 487 | As ever, thanks for the vast amount of feedback :) and bug reports :( |
| 488 | We may not answer all messages, but we do at least look at all of |
| 489 | them, and tend to fix the most frequently reported bugs. |
| 490 | |
| 491 | |
| 492 | |
sewardj | 3791882 | 2003-05-05 01:05:09 +0000 | [diff] [blame] | 493 | Version 1.9.6 (7 May 2003 or thereabouts) |
| 494 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 495 | |
| 496 | Major changes in 1.9.6: |
| 497 | |
| 498 | - Improved threading support for glibc >= 2.3.2 (SuSE 8.2, |
| 499 | RedHat 9, to name but two ...) It turned out that 1.9.5 |
| 500 | had problems with threading support on glibc >= 2.3.2, |
| 501 | usually manifested by threaded programs deadlocking in system calls, |
| 502 | or running unbelievably slowly. Hopefully these are fixed now. 1.9.6 |
| 503 | is the first valgrind which gives reasonable support for |
| 504 | glibc-2.3.2. Also fixed a 2.3.2 problem with pthread_atfork(). |
| 505 | |
| 506 | - Majorly expanded FAQ.txt. We've added workarounds for all |
| 507 | common problems for which a workaround is known. |
| 508 | |
| 509 | Minor changes in 1.9.6: |
| 510 | |
| 511 | - Fix identification of the main thread's stack. Incorrect |
| 512 | identification of it was causing some on-stack addresses to not get |
| 513 | identified as such. This only affected the usefulness of some error |
| 514 | messages; the correctness of the checks made is unchanged. |
| 515 | |
| 516 | - Support for kernels >= 2.5.68. |
| 517 | |
| 518 | - Dummy implementations of __libc_current_sigrtmin, |
| 519 | __libc_current_sigrtmax and __libc_allocate_rtsig, hopefully |
| 520 | good enough to keep alive programs which previously died for lack of |
| 521 | them. |
| 522 | |
| 523 | - Fix bug in the VALGRIND_DISCARD_TRANSLATIONS client request. |
| 524 | |
sewardj | 3d47b79 | 2003-05-05 22:15:35 +0000 | [diff] [blame] | 525 | - Fix bug in the DWARF2 debug line info loader, when instructions |
| 526 | following each other have source lines far from each other |
| 527 | (e.g. with inlined functions). |
| 528 | |
sewardj | 3791882 | 2003-05-05 01:05:09 +0000 | [diff] [blame] | 529 | - Debug info reading: read symbols from both "symtab" and "dynsym" |
| 530 | sections, rather than merely from the one that comes last in the |
| 531 | file. |
| 532 | |
| 533 | - New syscall support: prctl(), creat(), lookup_dcookie(). |
| 534 | |
| 535 | - When checking calls to accept(), recvfrom(), getsocketopt(), |
| 536 | don't complain if buffer values are NULL. |
| 537 | |
| 538 | - Try and avoid assertion failures in |
| 539 | mash_LD_PRELOAD_and_LD_LIBRARY_PATH. |
| 540 | |
| 541 | - Minor bug fixes in cg_annotate. |
| 542 | |
| 543 | |
| 544 | |
sewardj | 9023879 | 2003-05-05 00:23:42 +0000 | [diff] [blame] | 545 | Version 1.9.5 (7 April 2003) |
| 546 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 547 | |
| 548 | It occurs to me that it would be helpful for valgrind users to record |
| 549 | in the source distribution the changes in each release. So I now |
| 550 | attempt to mend my errant ways :-) Changes in this and future releases |
| 551 | will be documented in the NEWS file in the source distribution. |
| 552 | |
| 553 | Major changes in 1.9.5: |
| 554 | |
| 555 | - (Critical bug fix): Fix a bug in the FPU simulation. This was |
| 556 | causing some floating point conditional tests not to work right. |
| 557 | Several people reported this. If you had floating point code which |
| 558 | didn't work right on 1.9.1 to 1.9.4, it's worth trying 1.9.5. |
| 559 | |
| 560 | - Partial support for Red Hat 9. RH9 uses the new Native Posix |
| 561 | Threads Library (NPTL), instead of the older LinuxThreads. |
| 562 | This potentially causes problems with V which will take some |
| 563 | time to correct. In the meantime we have partially worked around |
| 564 | this, and so 1.9.5 works on RH9. Threaded programs still work, |
| 565 | but they may deadlock, because some system calls (accept, read, |
| 566 | write, etc) which should be nonblocking, in fact do block. This |
| 567 | is a known bug which we are looking into. |
| 568 | |
| 569 | If you can, your best bet (unfortunately) is to avoid using |
| 570 | 1.9.5 on a Red Hat 9 system, or on any NPTL-based distribution. |
| 571 | If your glibc is 2.3.1 or earlier, you're almost certainly OK. |
| 572 | |
| 573 | Minor changes in 1.9.5: |
| 574 | |
| 575 | - Added some #errors to valgrind.h to ensure people don't include |
| 576 | it accidentally in their sources. This is a change from 1.0.X |
| 577 | which was never properly documented. The right thing to include |
| 578 | is now memcheck.h. Some people reported problems and strange |
| 579 | behaviour when (incorrectly) including valgrind.h in code with |
| 580 | 1.9.1 -- 1.9.4. This is no longer possible. |
| 581 | |
| 582 | - Add some __extension__ bits and pieces so that gcc configured |
| 583 | for valgrind-checking compiles even with -Werror. If you |
| 584 | don't understand this, ignore it. Of interest to gcc developers |
| 585 | only. |
| 586 | |
| 587 | - Removed a pointless check which caused problems interworking |
| 588 | with Clearcase. V would complain about shared objects whose |
| 589 | names did not end ".so", and refuse to run. This is now fixed. |
| 590 | In fact it was fixed in 1.9.4 but not documented. |
| 591 | |
| 592 | - Fixed a bug causing an assertion failure of "waiters == 1" |
| 593 | somewhere in vg_scheduler.c, when running large threaded apps, |
| 594 | notably MySQL. |
| 595 | |
| 596 | - Add support for the munlock system call (124). |
| 597 | |
| 598 | Some comments about future releases: |
| 599 | |
| 600 | 1.9.5 is, we hope, the most stable Valgrind so far. It pretty much |
| 601 | supersedes the 1.0.X branch. If you are a valgrind packager, please |
| 602 | consider making 1.9.5 available to your users. You can regard the |
| 603 | 1.0.X branch as obsolete: 1.9.5 is stable and vastly superior. There |
| 604 | are no plans at all for further releases of the 1.0.X branch. |
| 605 | |
| 606 | If you want a leading-edge valgrind, consider building the cvs head |
| 607 | (from SourceForge), or getting a snapshot of it. Current cool stuff |
| 608 | going in includes MMX support (done); SSE/SSE2 support (in progress), |
| 609 | a significant (10-20%) performance improvement (done), and the usual |
| 610 | large collection of minor changes. Hopefully we will be able to |
| 611 | improve our NPTL support, but no promises. |
| 612 | |