njn | 718d3b1 | 2006-12-16 00:54:12 +0000 | [diff] [blame] | 1 | |
njn | 9b919b6 | 2010-02-28 23:41:07 +0000 | [diff] [blame] | 2 | Release 3.6.0 (???) |
| 3 | ~~~~~~~~~~~~~~~~~~~ |
| 4 | Improvements: |
| 5 | - XXX: ARM support |
njn | 69d495d | 2010-06-30 05:23:34 +0000 | [diff] [blame] | 6 | - XXX: Mac OS 10.6 support (32 and 64 bit) |
| 7 | - XXX: Much faster startup on Mac OS 10.5 for 64-bit programs. |
njn | 9b919b6 | 2010-02-28 23:41:07 +0000 | [diff] [blame] | 8 | |
njn | e323a6b | 2010-07-01 02:35:03 +0000 | [diff] [blame^] | 9 | - Valgrind runs much faster when the --smc-check=all option is given. |
njn | 9b919b6 | 2010-02-28 23:41:07 +0000 | [diff] [blame] | 10 | |
njn | 69d495d | 2010-06-30 05:23:34 +0000 | [diff] [blame] | 11 | - Cachegrind has a new processing script, cg_diff, which finds the |
| 12 | difference between two profiles. It's very useful for evaluating the |
| 13 | performance effects of a change in a program. |
| 14 | |
| 15 | Related to this change, the meaning of cg_annotate's (rarely-used) |
| 16 | --threshold option has changed; this is unlikely to affect many people, if |
| 17 | you do use it please see the user manual for details. |
njn | 9b919b6 | 2010-02-28 23:41:07 +0000 | [diff] [blame] | 18 | |
njn | e323a6b | 2010-07-01 02:35:03 +0000 | [diff] [blame^] | 19 | - Massif has a new option, --pages-as-heap, which is disabled by default. |
| 20 | When enabled, instead of tracking allocations at the level of heap blocks |
| 21 | (as allocated with malloc/new/new[]), it instead tracks memory allocations |
| 22 | at the level of memory pages (as mapped by mmap, brk, etc). Each mapped |
| 23 | page is treated as its own block. Interpreting the page-level output is |
| 24 | harder than the heap-level output, but this option is useful if you want |
| 25 | to account for every byte of memory used by a program. |
| 26 | |
| 27 | |
njn | 9b919b6 | 2010-02-28 23:41:07 +0000 | [diff] [blame] | 28 | |
sewardj | 6bbe024 | 2009-08-19 11:38:54 +0000 | [diff] [blame] | 29 | Release 3.5.0 (19 August 2009) |
| 30 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 31 | 3.5.0 is a feature release with many significant improvements and the |
| 32 | usual collection of bug fixes. The main improvement is that Valgrind |
| 33 | now works on Mac OS X. |
| 34 | |
| 35 | This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux |
| 36 | and X86/Darwin. Support for recent distros and toolchain components |
| 37 | (glibc 2.10, gcc 4.5) has been added. |
| 38 | |
| 39 | ------------------------- |
| 40 | |
| 41 | Here is a short summary of the changes. Details are shown further |
| 42 | down: |
| 43 | |
njn | f95c30b | 2009-08-16 00:20:58 +0000 | [diff] [blame] | 44 | * Support for Mac OS X (10.5.x). |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 45 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 46 | * Improvements and simplifications to Memcheck's leak checker. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 47 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 48 | * Clarification and simplifications in various aspects of Valgrind's |
| 49 | text output. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 50 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 51 | * XML output for Helgrind and Ptrcheck. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 52 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 53 | * Performance and stability improvements for Helgrind and DRD. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 54 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 55 | * Genuinely atomic support for x86/amd64/ppc atomic instructions. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 56 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 57 | * A new experimental tool, BBV, useful for computer architecture |
| 58 | research. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 59 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 60 | * Improved Wine support, including ability to read Windows PDB |
| 61 | debuginfo. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 62 | |
| 63 | ------------------------- |
| 64 | |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 65 | Here are details of the above changes, followed by descriptions of |
| 66 | many other minor changes, and a list of fixed bugs. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 67 | |
| 68 | |
| 69 | * 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] | 70 | called "Darwin" because that is the name of the OS core, which is the |
| 71 | level that Valgrind works at.) |
njn | f76d27a | 2009-05-28 01:53:07 +0000 | [diff] [blame] | 72 | |
njn | f95c30b | 2009-08-16 00:20:58 +0000 | [diff] [blame] | 73 | Supported systems: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 74 | |
njn | f95c30b | 2009-08-16 00:20:58 +0000 | [diff] [blame] | 75 | - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned |
| 76 | because it would require work and 10.4 is only becoming less common. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 77 | |
njn | f95c30b | 2009-08-16 00:20:58 +0000 | [diff] [blame] | 78 | - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported |
| 79 | fairly well. For 10.5.x, 32-bit programs are the default even on |
| 80 | 64-bit machines, so it handles most current programs. |
| 81 | |
| 82 | - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not |
| 83 | officially supported, but simple programs at least will probably work. |
| 84 | However, start-up is slow. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 85 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 86 | - PowerPC machines are not supported. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 87 | |
njn | c6ab7b6 | 2009-06-04 22:47:21 +0000 | [diff] [blame] | 88 | Things that don't work: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 89 | |
| 90 | - The Ptrcheck tool. |
| 91 | |
| 92 | - Objective-C garbage collection. |
| 93 | |
| 94 | - --db-attach=yes. |
| 95 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 96 | - If you have Rogue Amoeba's "Instant Hijack" program installed, |
| 97 | Valgrind will fail with a SIGTRAP at start-up. See |
| 98 | https://bugs.kde.org/show_bug.cgi?id=193917 for details and a |
| 99 | simple work-around. |
njn | c6ab7b6 | 2009-06-04 22:47:21 +0000 | [diff] [blame] | 100 | |
njn | 97db761 | 2009-08-04 02:32:55 +0000 | [diff] [blame] | 101 | Usage notes: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 102 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 103 | - You will likely find --dsymutil=yes a useful option, as error |
| 104 | messages may be imprecise without it. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 105 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 106 | - 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] | 107 | Linux support. Please report any bugs you find. |
njn | 97db761 | 2009-08-04 02:32:55 +0000 | [diff] [blame] | 108 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 109 | - Threaded programs may run more slowly than on Linux. |
| 110 | |
njn | c6ab7b6 | 2009-06-04 22:47:21 +0000 | [diff] [blame] | 111 | Many thanks to Greg Parker for developing this port over several years. |
| 112 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 113 | |
| 114 | * Memcheck's leak checker has been improved. |
| 115 | |
| 116 | - The results for --leak-check=summary now match the summary results |
| 117 | for --leak-check=full. Previously they could differ because |
| 118 | --leak-check=summary counted "indirectly lost" blocks and |
| 119 | "suppressed" blocks as "definitely lost". |
| 120 | |
| 121 | - Blocks that are only reachable via at least one interior-pointer, |
| 122 | but are directly pointed to by a start-pointer, were previously |
| 123 | marked as "still reachable". They are now correctly marked as |
| 124 | "possibly lost". |
| 125 | |
| 126 | - The default value for the --leak-resolution option has been |
| 127 | changed from "low" to "high". In general, this means that more |
| 128 | leak reports will be produced, but each leak report will describe |
| 129 | fewer leaked blocks. |
| 130 | |
| 131 | - With --leak-check=full, "definitely lost" and "possibly lost" |
| 132 | leaks are now considered as proper errors, ie. they are counted |
| 133 | for the "ERROR SUMMARY" and affect the behaviour of |
| 134 | --error-exitcode. These leaks are not counted as errors if |
| 135 | --leak-check=summary is specified, however. |
| 136 | |
| 137 | - Documentation for the leak checker has been improved. |
| 138 | |
njn | f76d27a | 2009-05-28 01:53:07 +0000 | [diff] [blame] | 139 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 140 | * Various aspects of Valgrind's text output have changed. |
njn | dbebecc | 2009-07-14 01:39:54 +0000 | [diff] [blame] | 141 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 142 | - Valgrind's start-up message has changed. It is shorter but also |
| 143 | includes the command being run, which makes it easier to use |
| 144 | --trace-children=yes. An example: |
njn | 53162bf | 2009-07-29 23:34:49 +0000 | [diff] [blame] | 145 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 146 | - Valgrind's shut-down messages have also changed. This is most |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 147 | noticeable with Memcheck, where the leak summary now occurs before |
| 148 | the error summary. This change was necessary to allow leaks to be |
| 149 | counted as proper errors (see the description of the leak checker |
| 150 | changes above for more details). This was also necessary to fix a |
| 151 | longstanding bug in which uses of suppressions against leaks were |
| 152 | not "counted", leading to difficulties in maintaining suppression |
sewardj | 783e9d6 | 2009-08-19 22:18:35 +0000 | [diff] [blame] | 153 | files (see https://bugs.kde.org/show_bug.cgi?id=186790). |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 154 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 155 | - Behavior of -v has changed. In previous versions, -v printed out |
| 156 | a mixture of marginally-user-useful information, and tool/core |
| 157 | statistics. The statistics printing has now been moved to its own |
| 158 | flag, --stats=yes. This means -v is less verbose and more likely |
| 159 | to convey useful end-user information. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 160 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 161 | - The format of some (non-XML) stack trace entries has changed a |
| 162 | little. Previously there were six possible forms: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 163 | |
| 164 | 0x80483BF: really (a.c:20) |
| 165 | 0x80483BF: really (in /foo/a.out) |
| 166 | 0x80483BF: really |
| 167 | 0x80483BF: (within /foo/a.out) |
| 168 | 0x80483BF: ??? (a.c:20) |
| 169 | 0x80483BF: ??? |
| 170 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 171 | The third and fourth of these forms have been made more consistent |
| 172 | with the others. The six possible forms are now: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 173 | |
| 174 | 0x80483BF: really (a.c:20) |
| 175 | 0x80483BF: really (in /foo/a.out) |
| 176 | 0x80483BF: really (in ???) |
| 177 | 0x80483BF: ??? (in /foo/a.out) |
| 178 | 0x80483BF: ??? (a.c:20) |
| 179 | 0x80483BF: ??? |
| 180 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 181 | Stack traces produced when --xml=yes is specified are different |
| 182 | and unchanged. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 183 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 184 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 185 | * Helgrind and Ptrcheck now support XML output, so they can be used |
| 186 | from GUI tools. Also, the XML output mechanism has been |
| 187 | overhauled. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 188 | |
njn | 78b7bb3 | 2009-08-14 22:34:30 +0000 | [diff] [blame] | 189 | - The XML format has been overhauled and generalised, so it is more |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 190 | suitable for error reporting tools in general. The Memcheck |
| 191 | specific aspects of it have been removed. The new format, which |
| 192 | is an evolution of the old format, is described in |
| 193 | docs/internals/xml-output-protocol4.txt. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 194 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 195 | - Memcheck has been updated to use the new format. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 196 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 197 | - Helgrind and Ptrcheck are now able to emit output in this format. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 198 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 199 | - The XML output mechanism has been overhauled. XML is now output |
| 200 | to its own file descriptor, which means that: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 201 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 202 | * Valgrind can output text and XML independently. |
njn | b4e0687 | 2009-08-10 05:11:43 +0000 | [diff] [blame] | 203 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 204 | * The longstanding problem of XML output being corrupted by |
| 205 | unexpected un-tagged text messages is solved. |
njn | 8df80b2 | 2009-03-02 05:11:06 +0000 | [diff] [blame] | 206 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 207 | As before, the destination for text output is specified using |
| 208 | --log-file=, --log-fd= or --log-socket=. |
njn | d55f0d9 | 2009-08-03 01:38:56 +0000 | [diff] [blame] | 209 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 210 | As before, XML output for a tool is enabled using --xml=yes. |
njn | c6ab7b6 | 2009-06-04 22:47:21 +0000 | [diff] [blame] | 211 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 212 | Because there's a new XML output channel, the XML output |
| 213 | destination is now specified by --xml-file=, --xml-fd= or |
| 214 | --xml-socket=. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 215 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 216 | Initial feedback has shown this causes some confusion. To |
| 217 | clarify, the two envisaged usage scenarios are: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 218 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 219 | (1) Normal text output. In this case, do not specify --xml=yes |
| 220 | nor any of --xml-file=, --xml-fd= or --xml-socket=. |
| 221 | |
| 222 | (2) XML output. In this case, specify --xml=yes, and one of |
| 223 | --xml-file=, --xml-fd= or --xml-socket= to select the XML |
| 224 | destination, one of --log-file=, --log-fd= or --log-socket= |
| 225 | to select the destination for any remaining text messages, |
| 226 | and, importantly, -q. |
| 227 | |
| 228 | -q makes Valgrind completely silent on the text channel, |
| 229 | except in the case of critical failures, such as Valgrind |
| 230 | itself segfaulting, or failing to read debugging information. |
| 231 | Hence, in this scenario, it suffices to check whether or not |
| 232 | any output appeared on the text channel. If yes, then it is |
| 233 | likely to be a critical error which should be brought to the |
| 234 | attention of the user. If no (the text channel produced no |
| 235 | output) then it can be assumed that the run was successful. |
| 236 | |
| 237 | This allows GUIs to make the critical distinction they need to |
| 238 | make (did the run fail or not?) without having to search or |
| 239 | filter the text output channel in any way. |
| 240 | |
| 241 | It is also recommended to use --child-silent-after-fork=yes in |
| 242 | scenario (2). |
| 243 | |
| 244 | |
| 245 | * Improvements and changes in Helgrind: |
| 246 | |
| 247 | - XML output, as described above |
| 248 | |
| 249 | - Checks for consistent association between pthread condition |
| 250 | variables and their associated mutexes are now performed. |
| 251 | |
| 252 | - pthread_spinlock functions are supported. |
| 253 | |
| 254 | - Modest performance improvements. |
| 255 | |
| 256 | - Initial (skeletal) support for describing the behaviour of |
| 257 | non-POSIX synchronisation objects through ThreadSanitizer |
| 258 | compatible ANNOTATE_* macros. |
| 259 | |
| 260 | - More controllable tradeoffs between performance and the level of |
| 261 | detail of "previous" accesses in a race. There are now three |
| 262 | settings: |
| 263 | |
| 264 | * --history-level=full. This is the default, and was also the |
| 265 | default in 3.4.x. It shows both stacks involved in a race, but |
| 266 | requires a lot of memory and can be very slow in programs that |
| 267 | do many inter-thread synchronisation events. |
| 268 | |
| 269 | * --history-level=none. This only shows the later stack involved |
| 270 | in a race. This can be much faster than --history-level=full, |
| 271 | but makes it much more difficult to find the other access |
| 272 | involved in the race. |
| 273 | |
| 274 | The new intermediate setting is |
| 275 | |
| 276 | * --history-level=approx |
| 277 | |
| 278 | For the earlier (other) access, two stacks are presented. The |
| 279 | earlier access is guaranteed to be somewhere in between the two |
| 280 | program points denoted by those stacks. This is not as useful |
| 281 | as showing the exact stack for the previous access (as per |
| 282 | --history-level=full), but it is better than nothing, and it's |
| 283 | almost as fast as --history-level=none. |
| 284 | |
njn | ace90d8 | 2009-04-27 01:27:58 +0000 | [diff] [blame] | 285 | |
bart | 6a146fc | 2009-07-21 07:32:19 +0000 | [diff] [blame] | 286 | * New features and improvements in DRD: |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 287 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 288 | - The error messages printed by DRD are now easier to interpret. |
| 289 | Instead of using two different numbers to identify each thread |
| 290 | (Valgrind thread ID and DRD thread ID), DRD does now identify |
| 291 | threads via a single number (the DRD thread ID). Furthermore |
| 292 | "first observed at" information is now printed for all error |
| 293 | messages related to synchronization objects. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 294 | |
bart | fe212db | 2009-07-25 11:16:51 +0000 | [diff] [blame] | 295 | - Added support for named semaphores (sem_open() and sem_close()). |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 296 | |
bart | 6a146fc | 2009-07-21 07:32:19 +0000 | [diff] [blame] | 297 | - Race conditions between pthread_barrier_wait() and |
| 298 | pthread_barrier_destroy() calls are now reported. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 299 | |
| 300 | - Added support for custom allocators through the macros |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 301 | VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in |
| 302 | in <valgrind/valgrind.h>). An alternative for these two macros is |
| 303 | the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in |
bart | 6a146fc | 2009-07-21 07:32:19 +0000 | [diff] [blame] | 304 | <valgrind/drd.h>). |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 305 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 306 | - Added support for annotating non-POSIX synchronization objects |
| 307 | through several new ANNOTATE_*() macros. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 308 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 309 | - OpenMP: added support for the OpenMP runtime (libgomp) included |
| 310 | with gcc versions 4.3.0 and 4.4.0. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 311 | |
bart | 6a146fc | 2009-07-21 07:32:19 +0000 | [diff] [blame] | 312 | - Faster operation. |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 313 | |
| 314 | - Added two new command-line options (--first-race-only and |
bart | 6a146fc | 2009-07-21 07:32:19 +0000 | [diff] [blame] | 315 | --segment-merging-interval). |
| 316 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 317 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 318 | * Genuinely atomic support for x86/amd64/ppc atomic instructions |
| 319 | |
| 320 | Valgrind will now preserve (memory-access) atomicity of LOCK- |
| 321 | prefixed x86/amd64 instructions, and any others implying a global |
| 322 | bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions. |
| 323 | |
| 324 | This means that Valgrinded processes will "play nicely" in |
| 325 | situations where communication with other processes, or the kernel, |
| 326 | is done through shared memory and coordinated with such atomic |
| 327 | instructions. Prior to this change, such arrangements usually |
| 328 | resulted in hangs, races or other synchronisation failures, because |
| 329 | Valgrind did not honour atomicity of such instructions. |
| 330 | |
| 331 | |
| 332 | * A new experimental tool, BBV, has been added. BBV generates basic |
| 333 | block vectors for use with the SimPoint analysis tool, which allows |
| 334 | a program's overall behaviour to be approximated by running only a |
| 335 | fraction of it. This is useful for computer architecture |
| 336 | researchers. You can run BBV by specifying --tool=exp-bbv (the |
| 337 | "exp-" prefix is short for "experimental"). BBV was written by |
| 338 | Vince Weaver. |
| 339 | |
| 340 | |
| 341 | * Modestly improved support for running Windows applications under |
| 342 | Wine. In particular, initial support for reading Windows .PDB debug |
| 343 | information has been added. |
| 344 | |
| 345 | |
| 346 | * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been |
| 347 | added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks |
| 348 | instead of bytes. |
| 349 | |
| 350 | |
| 351 | * The Valgrind client requests VALGRIND_PRINTF and |
| 352 | VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously, |
| 353 | the string was always printed immediately on its own line. Now, the |
| 354 | string will be added to a buffer but not printed until a newline is |
| 355 | encountered, or other Valgrind output is printed (note that for |
| 356 | VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered |
| 357 | "other Valgrind output"). This allows you to use multiple |
| 358 | VALGRIND_PRINTF calls to build up a single output line, and also to |
| 359 | print multiple output lines with a single request (by embedding |
| 360 | multiple newlines in the string). |
| 361 | |
| 362 | |
| 363 | * The graphs drawn by Massif's ms_print program have changed slightly: |
| 364 | |
| 365 | - The half-height chars '.' and ',' are no longer drawn, because |
| 366 | they are confusing. The --y option can be used if the default |
| 367 | y-resolution is not high enough. |
| 368 | |
| 369 | - Horizontal lines are now drawn after the top of a snapshot if |
| 370 | there is a gap until the next snapshot. This makes it clear that |
| 371 | the memory usage has not dropped to zero between snapshots. |
| 372 | |
| 373 | |
| 374 | * Something that happened in 3.4.0, but wasn't clearly announced: the |
| 375 | option --read-var-info=yes can be used by some tools (Memcheck, |
| 376 | Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3 |
| 377 | variable type and location information. This makes those tools |
| 378 | start up more slowly and increases memory consumption, but |
| 379 | descriptions of data addresses in error messages become more |
| 380 | detailed. |
| 381 | |
| 382 | |
| 383 | * exp-Omega, an experimental instantaneous leak-detecting tool, was |
| 384 | disabled in 3.4.0 due to a lack of interest and maintenance, |
| 385 | although the source code was still in the distribution. The source |
| 386 | code has now been removed from the distribution. For anyone |
| 387 | interested, the removal occurred in SVN revision r10247. |
| 388 | |
njn | acda178 | 2009-06-04 23:11:50 +0000 | [diff] [blame] | 389 | |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 390 | * Some changes have been made to the build system. |
njn | 6bf365c | 2009-02-11 00:35:45 +0000 | [diff] [blame] | 391 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 392 | - VEX/ is now integrated properly into the build system. This means |
| 393 | that dependency tracking within VEX/ now works properly, "make |
| 394 | install" will work without requiring "make" before it, and |
| 395 | parallel builds (ie. 'make -j') now work (previously a |
| 396 | .NOTPARALLEL directive was used to serialize builds, ie. 'make -j' |
| 397 | was effectively ignored). |
njn | 6bf365c | 2009-02-11 00:35:45 +0000 | [diff] [blame] | 398 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 399 | - The --with-vex configure option has been removed. It was of |
| 400 | little use and removing it simplified the build system. |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 401 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 402 | - The location of some install files has changed. This should not |
| 403 | affect most users. Those who might be affected: |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 404 | |
| 405 | * For people who use Valgrind with MPI programs, the installed |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 406 | libmpiwrap.so library has moved from |
| 407 | $(INSTALL)/<platform>/libmpiwrap.so to |
| 408 | $(INSTALL)/libmpiwrap-<platform>.so. |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 409 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 410 | * For people who distribute standalone Valgrind tools, the |
| 411 | installed libraries such as $(INSTALL)/<platform>/libcoregrind.a |
| 412 | have moved to $(INSTALL)/libcoregrind-<platform>.a. |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 413 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 414 | These changes simplify the build system. |
njn | 8b68b64 | 2009-06-24 00:37:09 +0000 | [diff] [blame] | 415 | |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 416 | - Previously, all the distributed suppression (*.supp) files were |
| 417 | installed. Now, only default.supp is installed. This should not |
| 418 | affect users as the other installed suppression files were not |
| 419 | read; the fact that they were installed was a mistake. |
| 420 | |
| 421 | |
| 422 | * KNOWN LIMITATIONS: |
| 423 | |
| 424 | - Memcheck is unusable with the Intel compiler suite version 11.1, |
| 425 | when it generates code for SSE2-and-above capable targets. This |
| 426 | is because of icc's use of highly optimised inlined strlen |
| 427 | implementations. It causes Memcheck to report huge numbers of |
| 428 | false errors even in simple programs. Helgrind and DRD may also |
| 429 | have problems. |
| 430 | |
| 431 | Versions 11.0 and earlier may be OK, but this has not been |
| 432 | properly tested. |
| 433 | |
| 434 | |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 435 | The following bugs have been fixed or resolved. Note that "n-i-bz" |
| 436 | stands for "not in bugzilla" -- that is, a bug that was reported to us |
| 437 | but never got a bugzilla entry. We encourage you to file bugs in |
| 438 | bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than |
| 439 | mailing the developers (or mailing lists) directly -- bugs that are |
| 440 | not entered into bugzilla tend to get forgotten about or ignored. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 441 | |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 442 | To see details of a given bug, visit |
| 443 | https://bugs.kde.org/show_bug.cgi?id=XXXXXX |
| 444 | where XXXXXX is the bug number as listed below. |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 445 | |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 446 | 84303 How about a LockCheck tool? |
| 447 | 91633 dereference of null ptr in vgPlain_st_basetype |
| 448 | 97452 Valgrind doesn't report any pthreads problems |
| 449 | 100628 leak-check gets assertion failure when using |
| 450 | VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory |
| 451 | 108528 NPTL pthread cleanup handlers not called |
| 452 | 110126 Valgrind 2.4.1 configure.in tramples CFLAGS |
| 453 | 110128 mallinfo is not implemented... |
| 454 | 110770 VEX: Generated files not always updated when making valgrind |
| 455 | 111102 Memcheck: problems with large (memory footprint) applications |
| 456 | 115673 Vex's decoder should never assert |
| 457 | 117564 False positive: Syscall param clone(child_tidptr) contains |
| 458 | uninitialised byte(s) |
| 459 | 119404 executing ssh from inside valgrind fails |
| 460 | 133679 Callgrind does not write path names to sources with dwarf debug |
| 461 | info |
| 462 | 135847 configure.in problem with non gnu compilers (and possible fix) |
| 463 | 136154 threads.c:273 (vgCallgrind_post_signal): Assertion |
| 464 | '*(vgCallgrind_current_fn_stack.top) == 0' failed. |
| 465 | 136230 memcheck reports "possibly lost", should be "still reachable" |
| 466 | 137073 NULL arg to MALLOCLIKE_BLOCK causes crash |
| 467 | 137904 Valgrind reports a memory leak when using POSIX threads, |
| 468 | while it shouldn't |
| 469 | 139076 valgrind VT_GETSTATE error |
| 470 | 142228 complaint of elf_dynamic_do_rela in trivial usage |
| 471 | 145347 spurious warning with USBDEVFS_REAPURB |
| 472 | 148441 (wine) can't find memory leak in Wine, win32 binary |
| 473 | executable file. |
| 474 | 148742 Leak-check fails assert on exit |
| 475 | 149878 add (proper) check for calloc integer overflow |
| 476 | 150606 Call graph is broken when using callgrind control |
| 477 | 152393 leak errors produce an exit code of 0. I need some way to |
| 478 | cause leak errors to result in a nonzero exit code. |
| 479 | 157154 documentation (leak-resolution doc speaks about num-callers |
| 480 | def=4) + what is a loss record |
| 481 | 159501 incorrect handling of ALSA ioctls |
| 482 | 162020 Valgrinding an empty/zero-byte file crashes valgrind |
| 483 | 162482 ppc: Valgrind crashes while reading stabs information |
| 484 | 162718 x86: avoid segment selector 0 in sys_set_thread_area() |
| 485 | 163253 (wine) canonicaliseSymtab forgot some fields in DiSym |
| 486 | 163560 VEX/test_main.c is missing from valgrind-3.3.1 |
| 487 | 164353 malloc_usable_size() doesn't return a usable size |
| 488 | 165468 Inconsistent formatting in memcheck manual -- please fix |
| 489 | 169505 main.c:286 (endOfInstr): |
| 490 | Assertion 'ii->cost_offset == *cost_offset' failed |
| 491 | 177206 Generate default.supp during compile instead of configure |
| 492 | 177209 Configure valt_load_address based on arch+os |
| 493 | 177305 eventfd / syscall 323 patch lost |
| 494 | 179731 Tests fail to build because of inlining of non-local asm labels |
| 495 | 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion |
| 496 | 'ordxx == POrd_EQ || ordxx == POrd_LT' failed. |
| 497 | 181594 Bogus warning for empty text segment |
| 498 | 181707 dwarf doesn't require enumerations to have name |
| 499 | 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64 |
| 500 | 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree): |
| 501 | Assertion '!already_present' failed. |
| 502 | 185359 exp-ptrcheck: unhandled syscall getresuid() |
| 503 | 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64 |
| 504 | 185816 Valgrind is unable to handle debug info for files with split |
| 505 | debug info that are prelinked afterwards |
| 506 | 185980 [darwin] unhandled syscall: sem_open |
| 507 | 186238 bbToIR_AMD64: disInstr miscalculated next %rip |
| 508 | 186507 exp-ptrcheck unhandled syscalls prctl, etc. |
| 509 | 186790 Suppression pattern used for leaks are not reported |
| 510 | 186796 Symbols with length>200 in suppression files are ignored |
| 511 | 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation |
| 512 | 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid} |
| 513 | 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened |
| 514 | 188046 bashisms in the configure script |
| 515 | 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA |
| 516 | 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672 |
| 517 | (get_otrack_shadow_offset_wrk): the 'impossible' happened." |
| 518 | 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock, |
| 519 | assertion fail "!lock->heldBy" |
| 520 | 188427 Add support for epoll_create1 (with patch) |
| 521 | 188530 Support for SIOCGSTAMPNS |
| 522 | 188560 Include valgrind.spec in the tarball |
| 523 | 188572 Valgrind on Mac should suppress setenv() mem leak |
| 524 | 189054 Valgrind fails to build because of duplicate non-local asm labels |
| 525 | 189737 vex amd64->IR: unhandled instruction bytes: 0xAC |
| 526 | 189762 epoll_create syscall not handled (--tool=exp-ptrcheck) |
| 527 | 189763 drd assertion failure: s_threadinfo[tid].is_recording |
| 528 | 190219 unhandled syscall: 328 (x86-linux) |
| 529 | 190391 dup of 181394; see above |
| 530 | 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc |
| 531 | 190820 No debug information on powerpc-linux |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 532 | 191095 PATCH: Improve usbdevfs ioctl handling |
| 533 | 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks |
| 534 | or big nr of errors |
| 535 | 191189 --xml=yes should obey --gen-suppressions=all |
| 536 | 191192 syslog() needs a suppression on macosx |
| 537 | 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265 |
| 538 | 191761 getrlimit on MacOSX |
| 539 | 191992 multiple --fn-skip only works sometimes; dependent on order |
| 540 | 192634 V. reports "aspacem sync_check_mapping_callback: |
| 541 | segment mismatch" on Darwin |
| 542 | 192954 __extension__ missing on 2 client requests |
| 543 | 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29 |
| 544 | 194474 "INSTALL" file has different build instructions than "README" |
| 545 | 194671 Unhandled syscall (sem_wait?) from mac valgrind |
| 546 | 195069 memcheck: reports leak (memory still reachable) for |
| 547 | printf("%d', x) |
| 548 | 195169 drd: (vgDrd_barrier_post_wait): |
| 549 | Assertion 'r->sg[p->post_iteration]' failed. |
| 550 | 195268 valgrind --log-file doesn't accept ~/... |
| 551 | 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate |
| 552 | 195860 WARNING: unhandled syscall: unix:223 |
| 553 | 196528 need a error suppression for pthread_rwlock_init under os x? |
| 554 | 197227 Support aio_* syscalls on Darwin |
| 555 | 197456 valgrind should reject --suppressions=(directory) |
| 556 | 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10 |
| 557 | 197591 unhandled syscall 27 (mincore) |
| 558 | 197793 Merge DCAS branch to the trunk == 85756, 142103 |
| 559 | 197794 Avoid duplicate filenames in Vex |
| 560 | 197898 make check fails on current SVN |
| 561 | 197901 make check fails also under exp-ptrcheck in current SVN |
| 562 | 197929 Make --leak-resolution=high the default |
| 563 | 197930 Reduce spacing between leak reports |
| 564 | 197933 Print command line of client at start-up, and shorten preamble |
| 565 | 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck) |
| 566 | 198395 add BBV to the distribution as an experimental tool |
| 567 | 198624 Missing syscalls on Darwin: 82, 167, 281, 347 |
| 568 | 198649 callgrind_annotate doesn't cumulate counters |
| 569 | 199338 callgrind_annotate sorting/thresholds are broken for all but Ir |
| 570 | 199977 Valgrind complains about an unrecognized instruction in the |
| 571 | atomic_incs test program |
| 572 | 200029 valgrind isn't able to read Fedora 12 debuginfo |
| 573 | 200760 darwin unhandled syscall: unix:284 |
| 574 | 200827 DRD doesn't work on Mac OS X |
| 575 | 200990 VG_(read_millisecond_timer)() does not work correctly |
| 576 | 201016 Valgrind does not support pthread_kill() on Mac OS |
| 577 | 201169 Document --read-var-info |
| 578 | 201323 Pre-3.5.0 performance sanity checking |
| 579 | 201384 Review user manual for the 3.5.0 release |
| 580 | 201585 mfpvr not implemented on ppc |
| 581 | 201708 tests failing because x86 direction flag is left set |
| 582 | 201757 Valgrind doesn't handle any recent sys_futex additions |
sewardj | 6bbe024 | 2009-08-19 11:38:54 +0000 | [diff] [blame] | 583 | 204377 64-bit valgrind can not start a shell script |
| 584 | (with #!/path/to/shell) if the shell is a 32-bit executable |
sewardj | 9176c86 | 2009-08-18 13:47:31 +0000 | [diff] [blame] | 585 | n-i-bz drd: fixed assertion failure triggered by mutex reinitialization. |
| 586 | n-i-bz drd: fixed a bug that caused incorrect messages to be printed |
| 587 | about memory allocation events with memory access tracing enabled |
| 588 | n-i-bz drd: fixed a memory leak triggered by vector clock deallocation |
njn | 254dafe | 2009-02-09 03:25:04 +0000 | [diff] [blame] | 589 | |
sewardj | 48867fb | 2009-08-19 22:40:53 +0000 | [diff] [blame] | 590 | (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846). |
sewardj | 570e244 | 2009-08-14 13:42:41 +0000 | [diff] [blame] | 591 | |
njn | 6eccab9 | 2009-08-14 02:41:37 +0000 | [diff] [blame] | 592 | |
njn | 254dafe | 2009-02-09 03:25:04 +0000 | [diff] [blame] | 593 | |
njn | ad91a75 | 2009-03-13 19:55:36 +0000 | [diff] [blame] | 594 | Release 3.4.1 (28 February 2009) |
| 595 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 596 | 3.4.1 is a bug-fix release that fixes some regressions and assertion |
| 597 | failures in debug info reading in 3.4.0, most notably incorrect stack |
| 598 | traces on amd64-linux on older (glibc-2.3 based) systems. Various |
| 599 | other debug info problems are also fixed. A number of bugs in the |
| 600 | exp-ptrcheck tool introduced in 3.4.0 have been fixed. |
| 601 | |
| 602 | In view of the fact that 3.4.0 contains user-visible regressions |
| 603 | relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are |
| 604 | encouraged to ship 3.4.1 in preference to 3.4.0. |
| 605 | |
| 606 | The fixed bugs are as follows. Note that "n-i-bz" stands for "not in |
| 607 | bugzilla" -- that is, a bug that was reported to us but never got a |
| 608 | bugzilla entry. We encourage you to file bugs in bugzilla |
| 609 | (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the |
| 610 | developers (or mailing lists) directly -- bugs that are not entered |
| 611 | into bugzilla tend to get forgotten about or ignored. |
| 612 | |
| 613 | n-i-bz Fix various bugs reading icc-11 generated debug info |
| 614 | n-i-bz Fix various bugs reading gcc-4.4 generated debug info |
| 615 | n-i-bz Preliminary support for glibc-2.10 / Fedora 11 |
| 616 | n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes, |
| 617 | so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2. |
| 618 | 179618 exp-ptrcheck crashed / exit prematurely |
| 619 | 179624 helgrind: false positive races with pthread_create and |
| 620 | recv/open/close/read |
| 621 | 134207 pkg-config output contains @VG_PLATFORM@ |
| 622 | 176926 floating point exception at valgrind startup with PPC 440EPX |
| 623 | 181594 Bogus warning for empty text segment |
| 624 | 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes) |
| 625 | 181707 Dwarf3 doesn't require enumerations to have name |
| 626 | 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64 |
| 627 | 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree): |
| 628 | Assertion '!already_present' failed. |
| 629 | 185359 exp-ptrcheck unhandled syscall getresuid() |
| 630 | |
| 631 | (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253). |
| 632 | (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293). |
| 633 | |
| 634 | |
| 635 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 636 | Release 3.4.0 (2 January 2009) |
| 637 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 638 | 3.4.0 is a feature release with many significant improvements and the |
| 639 | usual collection of bug fixes. This release supports X86/Linux, |
| 640 | AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros |
| 641 | (using gcc 4.4, glibc 2.8 and 2.9) has been added. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 642 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 643 | 3.4.0 brings some significant tool improvements. Memcheck can now |
| 644 | report the origin of uninitialised values, the thread checkers |
| 645 | Helgrind and DRD are much improved, and we have a new experimental |
| 646 | tool, exp-Ptrcheck, which is able to detect overruns of stack and |
| 647 | global arrays. In detail: |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 648 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 649 | * Memcheck is now able to track the origin of uninitialised values. |
| 650 | When it reports an uninitialised value error, it will try to show |
| 651 | the origin of the value, as either a heap or stack allocation. |
| 652 | Origin tracking is expensive and so is not enabled by default. To |
| 653 | use it, specify --track-origins=yes. Memcheck's speed will be |
| 654 | essentially halved, and memory usage will be significantly |
| 655 | increased. Nevertheless it can drastically reduce the effort |
| 656 | required to identify the root cause of uninitialised value errors, |
| 657 | and so is often a programmer productivity win, despite running more |
| 658 | slowly. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 659 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 660 | * 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] | 661 | 3.4.0, will be released shortly. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 662 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 663 | * Helgrind's race detection algorithm has been completely redesigned |
| 664 | and reimplemented, to address usability and scalability concerns: |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 665 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 666 | - The new algorithm has a lower false-error rate: it is much less |
| 667 | likely to report races that do not really exist. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 668 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 669 | - Helgrind will display full call stacks for both accesses involved |
sewardj | f1fa5ed | 2008-12-23 17:04:52 +0000 | [diff] [blame] | 670 | in a race. This makes it easier to identify the root causes of |
| 671 | races. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 672 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 673 | - Limitations on the size of program that can run have been removed. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 674 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 675 | - Performance has been modestly improved, although that is very |
| 676 | workload-dependent. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 677 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 678 | - Direct support for Qt4 threading has been added. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 679 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 680 | - pthread_barriers are now directly supported. |
njn | 75010ca | 2008-02-09 05:10:45 +0000 | [diff] [blame] | 681 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 682 | - Helgrind works well on all supported Linux targets. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 683 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 684 | * The DRD thread debugging tool has seen major improvements: |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 685 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 686 | - Greatly improved performance and significantly reduced memory |
| 687 | usage. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 688 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 689 | - Support for several major threading libraries (Boost.Thread, Qt4, |
| 690 | glib, OpenMP) has been added. |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 691 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 692 | - Support for atomic instructions, POSIX semaphores, barriers and |
| 693 | reader-writer locks has been added. |
| 694 | |
| 695 | - Works now on PowerPC CPUs too. |
| 696 | |
| 697 | - Added support for printing thread stack usage at thread exit time. |
| 698 | |
| 699 | - Added support for debugging lock contention. |
| 700 | |
| 701 | - Added a manual for Drd. |
| 702 | |
| 703 | * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck |
| 704 | checks for misuses of pointers. In that sense it is a bit like |
| 705 | Memcheck. However, Ptrcheck can do things Memcheck can't: it can |
| 706 | detect overruns of stack and global arrays, it can detect |
| 707 | arbitrarily far out-of-bounds accesses to heap blocks, and it can |
| 708 | detect accesses heap blocks that have been freed a very long time |
| 709 | ago (millions of blocks in the past). |
| 710 | |
| 711 | Ptrcheck currently works only on x86-linux and amd64-linux. To use |
| 712 | it, use --tool=exp-ptrcheck. A simple manual is provided, as part |
| 713 | of the main Valgrind documentation. As this is an experimental |
| 714 | tool, we would be particularly interested in hearing about your |
| 715 | experiences with it. |
| 716 | |
sewardj | 974aace | 2008-12-24 18:40:57 +0000 | [diff] [blame] | 717 | * exp-Omega, an experimental instantaneous leak-detecting tool, is no |
| 718 | longer built by default, although the code remains in the repository |
| 719 | and the tarball. This is due to three factors: a perceived lack of |
| 720 | users, a lack of maintenance, and concerns that it may not be |
| 721 | possible to achieve reliable operation using the existing design. |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 722 | |
| 723 | * As usual, support for the latest Linux distros and toolchain |
| 724 | components has been added. It should work well on Fedora Core 10, |
| 725 | OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release |
| 726 | state) is supported, as is glibc-2.9. The C++ demangler has been |
| 727 | updated so as to work well with C++ compiled by even the most recent |
| 728 | g++'s. |
| 729 | |
| 730 | * You can now use frame-level wildcards in suppressions. This was a |
| 731 | frequently-requested enhancement. A line "..." in a suppression now |
| 732 | matches zero or more frames. This makes it easier to write |
| 733 | suppressions which are precise yet insensitive to changes in |
| 734 | inlining behaviour. |
| 735 | |
| 736 | * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set. |
| 737 | |
| 738 | * Very basic support for IBM Power6 has been added (64-bit processes only). |
| 739 | |
| 740 | * Valgrind is now cross-compilable. For example, it is possible to |
| 741 | cross compile Valgrind on an x86/amd64-linux host, so that it runs |
| 742 | on a ppc32/64-linux target. |
| 743 | |
sewardj | f1fa5ed | 2008-12-23 17:04:52 +0000 | [diff] [blame] | 744 | * You can set the main thread's stack size at startup using the |
| 745 | new --main-stacksize= flag (subject of course to ulimit settings). |
| 746 | This is useful for running apps that need a lot of stack space. |
| 747 | |
| 748 | * The limitation that you can't use --trace-children=yes together |
| 749 | with --db-attach=yes has been removed. |
| 750 | |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 751 | * The following bugs have been fixed. Note that "n-i-bz" stands for |
| 752 | "not in bugzilla" -- that is, a bug that was reported to us but |
| 753 | never got a bugzilla entry. We encourage you to file bugs in |
| 754 | bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than |
| 755 | mailing the developers (or mailing lists) directly. |
| 756 | |
sewardj | f1fa5ed | 2008-12-23 17:04:52 +0000 | [diff] [blame] | 757 | n-i-bz Make return types for some client requests 64-bit clean |
| 758 | n-i-bz glibc 2.9 support |
| 759 | n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865) |
| 760 | n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults |
| 761 | n-i-bz Building in an env without gdb gives bogus gdb attach |
| 762 | 92456 Tracing the origin of uninitialised memory |
| 763 | 106497 Valgrind does not demangle some C++ template symbols |
| 764 | 162222 ==106497 |
sewardj | df1fc9a | 2008-12-23 23:09:57 +0000 | [diff] [blame] | 765 | 151612 Suppression with "..." (frame-level wildcards in .supp files) |
sewardj | f1fa5ed | 2008-12-23 17:04:52 +0000 | [diff] [blame] | 766 | 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit) |
| 767 | 159285 unhandled syscall:25 (stime, on x86-linux) |
| 768 | 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig" |
| 769 | 160954 ppc build of valgrind crashes with illegal instruction (isel) |
| 770 | 160956 mallinfo implementation, w/ patch |
| 771 | 162092 Valgrind fails to start gnome-system-monitor |
| 772 | 162819 malloc_free_fill test doesn't pass on glibc2.8 x86 |
| 773 | 163794 assertion failure with "--track-origins=yes" |
| 774 | 163933 sigcontext.err and .trapno must be set together |
| 775 | 163955 remove constraint !(--db-attach=yes && --trace-children=yes) |
| 776 | 164476 Missing kernel module loading system calls |
| 777 | 164669 SVN regression: mmap() drops posix file locks |
| 778 | 166581 Callgrind output corruption when program forks |
| 779 | 167288 Patch file for missing system calls on Cell BE |
| 780 | 168943 unsupported scas instruction pentium |
| 781 | 171645 Unrecognised instruction (MOVSD, non-binutils encoding) |
| 782 | 172417 x86->IR: 0x82 ... |
| 783 | 172563 amd64->IR: 0xD9 0xF5 - fprem1 |
| 784 | 173099 .lds linker script generation error |
| 785 | 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl) |
| 786 | 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes) |
| 787 | 174532 == 173751 |
| 788 | 174908 --log-file value not expanded correctly for core file |
| 789 | 175044 Add lookup_dcookie for amd64 |
| 790 | 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding) |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 791 | |
| 792 | Developer-visible changes: |
| 793 | |
| 794 | * Valgrind's debug-info reading machinery has been majorly overhauled. |
| 795 | It can now correctly establish the addresses for ELF data symbols, |
| 796 | which is something that has never worked properly before now. |
| 797 | |
| 798 | Also, Valgrind can now read DWARF3 type and location information for |
| 799 | stack and global variables. This makes it possible to use the |
| 800 | framework to build tools that rely on knowing the type and locations |
| 801 | of stack and global variables, for example exp-Ptrcheck. |
| 802 | |
| 803 | Reading of such information is disabled by default, because most |
| 804 | tools don't need it, and because it is expensive in space and time. |
| 805 | However, you can force Valgrind to read it, using the |
| 806 | --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to |
| 807 | make use of such information, if present, to provide source-level |
| 808 | descriptions of data addresses in the error messages they create. |
| 809 | |
sewardj | 974aace | 2008-12-24 18:40:57 +0000 | [diff] [blame] | 810 | (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882). |
sewardj | b8b1664 | 2009-01-02 23:21:54 +0000 | [diff] [blame] | 811 | (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899). |
sewardj | 565fcf1 | 2008-12-23 13:14:06 +0000 | [diff] [blame] | 812 | |
sewardj | 8639f11 | 2008-12-17 12:26:24 +0000 | [diff] [blame] | 813 | |
sewardj | 9f42ba3 | 2008-06-04 11:47:51 +0000 | [diff] [blame] | 814 | |
| 815 | Release 3.3.1 (4 June 2008) |
| 816 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 817 | 3.3.1 fixes a bunch of bugs in 3.3.0, adds support for glibc-2.8 based |
| 818 | systems (openSUSE 11, Fedora Core 9), improves the existing glibc-2.7 |
| 819 | support, and adds support for the SSSE3 (Core 2) instruction set. |
| 820 | |
| 821 | 3.3.1 will likely be the last release that supports some very old |
| 822 | systems. In particular, the next major release, 3.4.0, will drop |
| 823 | support for the old LinuxThreads threading library, and for gcc |
| 824 | versions prior to 3.0. |
| 825 | |
| 826 | The fixed bugs are as follows. Note that "n-i-bz" stands for "not in |
| 827 | bugzilla" -- that is, a bug that was reported to us but never got a |
| 828 | bugzilla entry. We encourage you to file bugs in bugzilla |
| 829 | (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the |
| 830 | developers (or mailing lists) directly -- bugs that are not entered |
| 831 | into bugzilla tend to get forgotten about or ignored. |
| 832 | |
| 833 | n-i-bz Massif segfaults at exit |
| 834 | n-i-bz Memcheck asserts on Altivec code |
| 835 | n-i-bz fix sizeof bug in Helgrind |
| 836 | n-i-bz check fd on sys_llseek |
| 837 | n-i-bz update syscall lists to kernel 2.6.23.1 |
| 838 | n-i-bz support sys_sync_file_range |
| 839 | n-i-bz handle sys_sysinfo, sys_getresuid, sys_getresgid on ppc64-linux |
| 840 | n-i-bz intercept memcpy in 64-bit ld.so's |
| 841 | n-i-bz Fix wrappers for sys_{futimesat,utimensat} |
| 842 | n-i-bz Minor false-error avoidance fixes for Memcheck |
| 843 | n-i-bz libmpiwrap.c: add a wrapper for MPI_Waitany |
| 844 | n-i-bz helgrind support for glibc-2.8 |
| 845 | n-i-bz partial fix for mc_leakcheck.c:698 assert: |
| 846 | 'lc_shadows[i]->data + lc_shadows[i] ... |
| 847 | n-i-bz Massif/Cachegrind output corruption when programs fork |
| 848 | n-i-bz register allocator fix: handle spill stores correctly |
| 849 | n-i-bz add support for PA6T PowerPC CPUs |
| 850 | 126389 vex x86->IR: 0xF 0xAE (FXRSTOR) |
| 851 | 158525 ==126389 |
| 852 | 152818 vex x86->IR: 0xF3 0xAC (repz lodsb) |
| 853 | 153196 vex x86->IR: 0xF2 0xA6 (repnz cmpsb) |
| 854 | 155011 vex x86->IR: 0xCF (iret) |
| 855 | 155091 Warning [...] unhandled DW_OP_ opcode 0x23 |
| 856 | 156960 ==155901 |
| 857 | 155528 support Core2/SSSE3 insns on x86/amd64 |
| 858 | 155929 ms_print fails on massif outputs containing long lines |
| 859 | 157665 valgrind fails on shmdt(0) after shmat to 0 |
| 860 | 157748 support x86 PUSHFW/POPFW |
| 861 | 158212 helgrind: handle pthread_rwlock_try{rd,wr}lock. |
| 862 | 158425 sys_poll incorrectly emulated when RES==0 |
| 863 | 158744 vex amd64->IR: 0xF0 0x41 0xF 0xC0 (xaddb) |
| 864 | 160907 Support for a couple of recent Linux syscalls |
| 865 | 161285 Patch -- support for eventfd() syscall |
| 866 | 161378 illegal opcode in debug libm (FUCOMPP) |
| 867 | 160136 ==161378 |
| 868 | 161487 number of suppressions files is limited to 10 |
| 869 | 162386 ms_print typo in milliseconds time unit for massif |
| 870 | 161036 exp-drd: client allocated memory was never freed |
| 871 | 162663 signalfd_wrapper fails on 64bit linux |
| 872 | |
| 873 | (3.3.1.RC1: 2 June 2008, vex r1854, valgrind r8169). |
| 874 | (3.3.1: 4 June 2008, vex r1854, valgrind r8180). |
| 875 | |
| 876 | |
| 877 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 878 | Release 3.3.0 (7 December 2007) |
| 879 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 880 | 3.3.0 is a feature release with many significant improvements and the |
| 881 | usual collection of bug fixes. This release supports X86/Linux, |
| 882 | AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros |
| 883 | (using gcc 4.3, glibc 2.6 and 2.7) has been added. |
njn | f4f9cd4 | 2007-11-23 01:42:10 +0000 | [diff] [blame] | 884 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 885 | The main excitement in 3.3.0 is new and improved tools. Helgrind |
| 886 | works again, Massif has been completely overhauled and much improved, |
| 887 | Cachegrind now does branch-misprediction profiling, and a new category |
| 888 | of experimental tools has been created, containing two new tools: |
| 889 | Omega and DRD. There are many other smaller improvements. In detail: |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 890 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 891 | - Helgrind has been completely overhauled and works for the first time |
| 892 | since Valgrind 2.2.0. Supported functionality is: detection of |
| 893 | misuses of the POSIX PThreads API, detection of potential deadlocks |
| 894 | resulting from cyclic lock dependencies, and detection of data |
| 895 | races. Compared to the 2.2.0 Helgrind, the race detection algorithm |
| 896 | has some significant improvements aimed at reducing the false error |
| 897 | rate. Handling of various kinds of corner cases has been improved. |
| 898 | Efforts have been made to make the error messages easier to |
| 899 | understand. Extensive documentation is provided. |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 900 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 901 | - Massif has been completely overhauled. Instead of measuring |
| 902 | space-time usage -- which wasn't always useful and many people found |
| 903 | confusing -- it now measures space usage at various points in the |
| 904 | execution, including the point of peak memory allocation. Its |
| 905 | output format has also changed: instead of producing PostScript |
| 906 | graphs and HTML text, it produces a single text output (via the new |
| 907 | 'ms_print' script) that contains both a graph and the old textual |
| 908 | information, but in a more compact and readable form. Finally, the |
| 909 | new version should be more reliable than the old one, as it has been |
| 910 | tested more thoroughly. |
njn | 718d3b1 | 2006-12-16 00:54:12 +0000 | [diff] [blame] | 911 | |
sewardj | 8badbaa | 2007-05-08 09:20:25 +0000 | [diff] [blame] | 912 | - Cachegrind has been extended to do branch-misprediction profiling. |
| 913 | Both conditional and indirect branches are profiled. The default |
| 914 | behaviour of Cachegrind is unchanged. To use the new functionality, |
| 915 | give the option --branch-sim=yes. |
| 916 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 917 | - A new category of "experimental tools" has been created. Such tools |
| 918 | may not work as well as the standard tools, but are included because |
| 919 | some people will find them useful, and because exposure to a wider |
| 920 | user group provides tool authors with more end-user feedback. These |
| 921 | tools have a "exp-" prefix attached to their names to indicate their |
| 922 | experimental nature. Currently there are two experimental tools: |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 923 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 924 | * exp-Omega: an instantaneous leak detector. See |
| 925 | exp-omega/docs/omega_introduction.txt. |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 926 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 927 | * exp-DRD: a data race detector based on the happens-before |
sewardj | 8b09d4f | 2007-12-04 21:27:18 +0000 | [diff] [blame] | 928 | relation. See exp-drd/docs/README.txt. |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 929 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 930 | - Scalability improvements for very large programs, particularly those |
| 931 | which have a million or more malloc'd blocks in use at once. These |
| 932 | improvements mostly affect Memcheck. Memcheck is also up to 10% |
| 933 | faster for all programs, with x86-linux seeing the largest |
| 934 | improvement. |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 935 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 936 | - Works well on the latest Linux distros. Has been tested on Fedora |
| 937 | Core 8 (x86, amd64, ppc32, ppc64) and openSUSE 10.3. glibc 2.6 and |
| 938 | 2.7 are supported. gcc-4.3 (in its current pre-release state) is |
| 939 | supported. At the same time, 3.3.0 retains support for older |
| 940 | distros. |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 941 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 942 | - The documentation has been modestly reorganised with the aim of |
| 943 | making it easier to find information on common-usage scenarios. |
| 944 | Some advanced material has been moved into a new chapter in the main |
| 945 | manual, so as to unclutter the main flow, and other tidying up has |
| 946 | been done. |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 947 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 948 | - There is experimental support for AIX 5.3, both 32-bit and 64-bit |
| 949 | processes. You need to be running a 64-bit kernel to use Valgrind |
| 950 | on a 64-bit executable. |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 951 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 952 | - There have been some changes to command line options, which may |
| 953 | affect you: |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 954 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 955 | * --log-file-exactly and |
| 956 | --log-file-qualifier options have been removed. |
njn | 374a36d | 2007-11-23 01:41:32 +0000 | [diff] [blame] | 957 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 958 | To make up for this --log-file option has been made more powerful. |
| 959 | It now accepts a %p format specifier, which is replaced with the |
| 960 | process ID, and a %q{FOO} format specifier, which is replaced with |
| 961 | the contents of the environment variable FOO. |
| 962 | |
| 963 | * --child-silent-after-fork=yes|no [no] |
| 964 | |
| 965 | Causes Valgrind to not show any debugging or logging output for |
| 966 | the child process resulting from a fork() call. This can make the |
| 967 | output less confusing (although more misleading) when dealing with |
| 968 | processes that create children. |
| 969 | |
| 970 | * --cachegrind-out-file, --callgrind-out-file and --massif-out-file |
| 971 | |
| 972 | These control the names of the output files produced by |
| 973 | Cachegrind, Callgrind and Massif. They accept the same %p and %q |
| 974 | format specifiers that --log-file accepts. --callgrind-out-file |
| 975 | replaces Callgrind's old --base option. |
| 976 | |
| 977 | * Cachegrind's 'cg_annotate' script no longer uses the --<pid> |
| 978 | option to specify the output file. Instead, the first non-option |
| 979 | argument is taken to be the name of the output file, and any |
| 980 | subsequent non-option arguments are taken to be the names of |
| 981 | source files to be annotated. |
| 982 | |
| 983 | * Cachegrind and Callgrind now use directory names where possible in |
| 984 | their output files. This means that the -I option to |
| 985 | 'cg_annotate' and 'callgrind_annotate' should not be needed in |
| 986 | most cases. It also means they can correctly handle the case |
| 987 | where two source files in different directories have the same |
| 988 | name. |
| 989 | |
| 990 | - Memcheck offers a new suppression kind: "Jump". This is for |
| 991 | suppressing jump-to-invalid-address errors. Previously you had to |
| 992 | use an "Addr1" suppression, which didn't make much sense. |
| 993 | |
| 994 | - Memcheck has new flags --malloc-fill=<hexnum> and |
| 995 | --free-fill=<hexnum> which free malloc'd / free'd areas with the |
| 996 | specified byte. This can help shake out obscure memory corruption |
bart | e7b86c3 | 2009-07-21 07:35:23 +0000 | [diff] [blame] | 997 | problems. The definedness and addressability of these areas is |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 998 | unchanged -- only the contents are affected. |
njn | 718d3b1 | 2006-12-16 00:54:12 +0000 | [diff] [blame] | 999 | |
| 1000 | - The behaviour of Memcheck's client requests VALGRIND_GET_VBITS and |
| 1001 | VALGRIND_SET_VBITS have changed slightly. They no longer issue |
| 1002 | addressability errors -- if either array is partially unaddressable, |
| 1003 | they just return 3 (as before). Also, SET_VBITS doesn't report |
| 1004 | definedness errors if any of the V bits are undefined. |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1005 | |
njn | 92676d7 | 2007-12-04 21:18:06 +0000 | [diff] [blame] | 1006 | - The following Memcheck client requests have been removed: |
| 1007 | VALGRIND_MAKE_NOACCESS |
| 1008 | VALGRIND_MAKE_WRITABLE |
| 1009 | VALGRIND_MAKE_READABLE |
| 1010 | VALGRIND_CHECK_WRITABLE |
| 1011 | VALGRIND_CHECK_READABLE |
| 1012 | VALGRIND_CHECK_DEFINED |
| 1013 | They were deprecated in 3.2.0, when equivalent but better-named client |
| 1014 | requests were added. See the 3.2.0 release notes for more details. |
| 1015 | |
njn | 5a0a773 | 2007-02-12 10:37:37 +0000 | [diff] [blame] | 1016 | - The behaviour of the tool Lackey has changed slightly. First, the output |
| 1017 | from --trace-mem has been made more compact, to reduce the size of the |
| 1018 | traces. Second, a new option --trace-superblocks has been added, which |
| 1019 | shows the addresses of superblocks (code blocks) as they are executed. |
| 1020 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 1021 | - The following bugs have been fixed. Note that "n-i-bz" stands for |
| 1022 | "not in bugzilla" -- that is, a bug that was reported to us but |
| 1023 | never got a bugzilla entry. We encourage you to file bugs in |
| 1024 | bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than |
| 1025 | mailing the developers (or mailing lists) directly. |
| 1026 | |
| 1027 | n-i-bz x86_linux_REDIR_FOR_index() broken |
| 1028 | n-i-bz guest-amd64/toIR.c:2512 (dis_op2_E_G): Assertion `0' failed. |
| 1029 | n-i-bz Support x86 INT insn (INT (0xCD) 0x40 - 0x43) |
| 1030 | n-i-bz Add sys_utimensat system call for Linux x86 platform |
| 1031 | 79844 Helgrind complains about race condition which does not exist |
| 1032 | 82871 Massif output function names too short |
| 1033 | 89061 Massif: ms_main.c:485 (get_XCon): Assertion `xpt->max_chi...' |
| 1034 | 92615 Write output from Massif at crash |
| 1035 | 95483 massif feature request: include peak allocation in report |
| 1036 | 112163 MASSIF crashed with signal 7 (SIGBUS) after running 2 days |
| 1037 | 119404 problems running setuid executables (partial fix) |
| 1038 | 121629 add instruction-counting mode for timing |
| 1039 | 127371 java vm giving unhandled instruction bytes: 0x26 0x2E 0x64 0x65 |
| 1040 | 129937 ==150380 |
| 1041 | 129576 Massif loses track of memory, incorrect graphs |
| 1042 | 132132 massif --format=html output does not do html entity escaping |
| 1043 | 132950 Heap alloc/usage summary |
| 1044 | 133962 unhandled instruction bytes: 0xF2 0x4C 0xF 0x10 |
| 1045 | 134990 use -fno-stack-protector if possible |
| 1046 | 136382 ==134990 |
| 1047 | 137396 I would really like helgrind to work again... |
| 1048 | 137714 x86/amd64->IR: 0x66 0xF 0xF7 0xC6 (maskmovq, maskmovdq) |
| 1049 | 141631 Massif: percentages don't add up correctly |
| 1050 | 142706 massif numbers don't seem to add up |
| 1051 | 143062 massif crashes on app exit with signal 8 SIGFPE |
| 1052 | 144453 (get_XCon): Assertion 'xpt->max_children != 0' failed. |
| 1053 | 145559 valgrind aborts when malloc_stats is called |
| 1054 | 145609 valgrind aborts all runs with 'repeated section!' |
| 1055 | 145622 --db-attach broken again on x86-64 |
| 1056 | 145837 ==149519 |
| 1057 | 145887 PPC32: getitimer() system call is not supported |
| 1058 | 146252 ==150678 |
| 1059 | 146456 (update_XCon): Assertion 'xpt->curr_space >= -space_delta'... |
| 1060 | 146701 ==134990 |
| 1061 | 146781 Adding support for private futexes |
| 1062 | 147325 valgrind internal error on syscall (SYS_io_destroy, 0) |
| 1063 | 147498 amd64->IR: 0xF0 0xF 0xB0 0xF (lock cmpxchg %cl,(%rdi)) |
sewardj | bc39ad2 | 2007-12-09 02:24:02 +0000 | [diff] [blame] | 1064 | 147545 Memcheck: mc_main.c:817 (get_sec_vbits8): Assertion 'n' failed. |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 1065 | 147628 SALC opcode 0xd6 unimplemented |
| 1066 | 147825 crash on amd64-linux with gcc 4.2 and glibc 2.6 (CFI) |
| 1067 | 148174 Incorrect type of freed_list_volume causes assertion [...] |
| 1068 | 148447 x86_64 : new NOP codes: 66 66 66 66 2e 0f 1f |
| 1069 | 149182 PPC Trap instructions not implemented in valgrind |
| 1070 | 149504 Assertion hit on alloc_xpt->curr_space >= -space_delta |
| 1071 | 149519 ppc32: V aborts with SIGSEGV on execution of a signal handler |
| 1072 | 149892 ==137714 |
| 1073 | 150044 SEGV during stack deregister |
| 1074 | 150380 dwarf/gcc interoperation (dwarf3 read problems) |
| 1075 | 150408 ==148447 |
| 1076 | 150678 guest-amd64/toIR.c:3741 (dis_Grp5): Assertion `sz == 4' failed |
| 1077 | 151209 V unable to execute programs for users with UID > 2^16 |
| 1078 | 151938 help on --db-command= misleading |
| 1079 | 152022 subw $0x28, %%sp causes assertion failure in memcheck |
| 1080 | 152357 inb and outb not recognized in 64-bit mode |
| 1081 | 152501 vex x86->IR: 0x27 0x66 0x89 0x45 (daa) |
| 1082 | 152818 vex x86->IR: 0xF3 0xAC 0xFC 0x9C (rep lodsb) |
| 1083 | |
njn | d5f4bef | 2006-12-24 07:54:19 +0000 | [diff] [blame] | 1084 | Developer-visible changes: |
| 1085 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 1086 | - The names of some functions and types within the Vex IR have |
| 1087 | changed. Run 'svn log -r1689 VEX/pub/libvex_ir.h' for full details. |
| 1088 | Any existing standalone tools will have to be updated to reflect |
| 1089 | these changes. The new names should be clearer. The file |
| 1090 | VEX/pub/libvex_ir.h is also much better commented. |
njn | d8366ee | 2007-02-12 22:38:50 +0000 | [diff] [blame] | 1091 | |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 1092 | - A number of new debugging command line options have been added. |
| 1093 | These are mostly of use for debugging the symbol table and line |
| 1094 | number readers: |
| 1095 | |
| 1096 | --trace-symtab-patt=<patt> limit debuginfo tracing to obj name <patt> |
| 1097 | --trace-cfi=no|yes show call-frame-info details? [no] |
| 1098 | --debug-dump=syms mimic /usr/bin/readelf --syms |
| 1099 | --debug-dump=line mimic /usr/bin/readelf --debug-dump=line |
| 1100 | --debug-dump=frames mimic /usr/bin/readelf --debug-dump=frames |
| 1101 | --sym-offsets=yes|no show syms in form 'name+offset' ? [no] |
| 1102 | |
| 1103 | - Internally, the code base has been further factorised and |
| 1104 | abstractified, particularly with respect to support for non-Linux |
| 1105 | OSs. |
| 1106 | |
sewardj | 135a7ff | 2007-12-10 22:17:42 +0000 | [diff] [blame] | 1107 | (3.3.0.RC1: 2 Dec 2007, vex r1803, valgrind r7268). |
| 1108 | (3.3.0.RC2: 5 Dec 2007, vex r1804, valgrind r7282). |
| 1109 | (3.3.0.RC3: 9 Dec 2007, vex r1804, valgrind r7288). |
sewardj | 686b5be | 2007-12-10 22:21:14 +0000 | [diff] [blame] | 1110 | (3.3.0: 10 Dec 2007, vex r1804, valgrind r7290). |
sewardj | cb2288b | 2007-12-02 02:08:17 +0000 | [diff] [blame] | 1111 | |
njn | d5f4bef | 2006-12-24 07:54:19 +0000 | [diff] [blame] | 1112 | |
| 1113 | |
njn | 9923775 | 2007-01-30 05:18:38 +0000 | [diff] [blame] | 1114 | Release 3.2.3 (29 Jan 2007) |
| 1115 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 1116 | Unfortunately 3.2.2 introduced a regression which can cause an |
| 1117 | assertion failure ("vex: the `impossible' happened: eqIRConst") when |
| 1118 | running obscure pieces of SSE code. 3.2.3 fixes this and adds one |
| 1119 | more glibc-2.5 intercept. In all other respects it is identical to |
| 1120 | 3.2.2. Please do not use (or package) 3.2.2; instead use 3.2.3. |
| 1121 | |
| 1122 | n-i-bz vex: the `impossible' happened: eqIRConst |
| 1123 | n-i-bz Add an intercept for glibc-2.5 __stpcpy_chk |
| 1124 | |
| 1125 | (3.2.3: 29 Jan 2007, vex r1732, valgrind r6560). |
| 1126 | |
| 1127 | |
njn | a7e4880 | 2007-01-23 05:58:52 +0000 | [diff] [blame] | 1128 | Release 3.2.2 (22 Jan 2007) |
| 1129 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 1130 | 3.2.2 fixes a bunch of bugs in 3.2.1, adds support for glibc-2.5 based |
| 1131 | systems (openSUSE 10.2, Fedora Core 6), improves support for icc-9.X |
| 1132 | compiled code, and brings modest performance improvements in some |
| 1133 | areas, including amd64 floating point, powerpc support, and startup |
| 1134 | responsiveness on all targets. |
| 1135 | |
| 1136 | The fixed bugs are as follows. Note that "n-i-bz" stands for "not in |
| 1137 | bugzilla" -- that is, a bug that was reported to us but never got a |
| 1138 | bugzilla entry. We encourage you to file bugs in bugzilla |
| 1139 | (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the |
| 1140 | developers (or mailing lists) directly. |
| 1141 | |
| 1142 | 129390 ppc?->IR: some kind of VMX prefetch (dstt) |
| 1143 | 129968 amd64->IR: 0xF 0xAE 0x0 (fxsave) |
| 1144 | 134319 ==129968 |
| 1145 | 133054 'make install' fails with syntax errors |
| 1146 | 118903 ==133054 |
| 1147 | 132998 startup fails in when running on UML |
| 1148 | 134207 pkg-config output contains @VG_PLATFORM@ |
| 1149 | 134727 valgrind exits with "Value too large for defined data type" |
| 1150 | n-i-bz ppc32/64: support mcrfs |
| 1151 | n-i-bz Cachegrind/Callgrind: Update cache parameter detection |
| 1152 | 135012 x86->IR: 0xD7 0x8A 0xE0 0xD0 (xlat) |
| 1153 | 125959 ==135012 |
| 1154 | 126147 x86->IR: 0xF2 0xA5 0xF 0x77 (repne movsw) |
| 1155 | 136650 amd64->IR: 0xC2 0x8 0x0 |
| 1156 | 135421 x86->IR: unhandled Grp5(R) case 6 |
| 1157 | n-i-bz Improved documentation of the IR intermediate representation |
| 1158 | n-i-bz jcxz (x86) (users list, 8 Nov) |
| 1159 | n-i-bz ExeContext hashing fix |
| 1160 | n-i-bz fix CFI reading failures ("Dwarf CFI 0:24 0:32 0:48 0:7") |
| 1161 | n-i-bz fix Cachegrind/Callgrind simulation bug |
| 1162 | n-i-bz libmpiwrap.c: fix handling of MPI_LONG_DOUBLE |
| 1163 | n-i-bz make User errors suppressible |
| 1164 | 136844 corrupted malloc line when using --gen-suppressions=yes |
| 1165 | 138507 ==136844 |
| 1166 | n-i-bz Speed up the JIT's register allocator |
| 1167 | n-i-bz Fix confusing leak-checker flag hints |
| 1168 | n-i-bz Support recent autoswamp versions |
| 1169 | n-i-bz ppc32/64 dispatcher speedups |
| 1170 | n-i-bz ppc64 front end rld/rlw improvements |
| 1171 | n-i-bz ppc64 back end imm64 improvements |
| 1172 | 136300 support 64K pages on ppc64-linux |
| 1173 | 139124 == 136300 |
| 1174 | n-i-bz fix ppc insn set tests for gcc >= 4.1 |
| 1175 | 137493 x86->IR: recent binutils no-ops |
| 1176 | 137714 x86->IR: 0x66 0xF 0xF7 0xC6 (maskmovdqu) |
| 1177 | 138424 "failed in UME with error 22" (produce a better error msg) |
| 1178 | 138856 ==138424 |
| 1179 | 138627 Enhancement support for prctl ioctls |
| 1180 | 138896 Add support for usb ioctls |
| 1181 | 136059 ==138896 |
| 1182 | 139050 ppc32->IR: mfspr 268/269 instructions not handled |
| 1183 | n-i-bz ppc32->IR: lvxl/stvxl |
| 1184 | n-i-bz glibc-2.5 support |
| 1185 | n-i-bz memcheck: provide replacement for mempcpy |
| 1186 | n-i-bz memcheck: replace bcmp in ld.so |
| 1187 | n-i-bz Use 'ifndef' in VEX's Makefile correctly |
| 1188 | n-i-bz Suppressions for MVL 4.0.1 on ppc32-linux |
| 1189 | n-i-bz libmpiwrap.c: Fixes for MPICH |
| 1190 | n-i-bz More robust handling of hinted client mmaps |
| 1191 | 139776 Invalid read in unaligned memcpy with Intel compiler v9 |
| 1192 | n-i-bz Generate valid XML even for very long fn names |
| 1193 | n-i-bz Don't prompt about suppressions for unshown reachable leaks |
| 1194 | 139910 amd64 rcl is not supported |
| 1195 | n-i-bz DWARF CFI reader: handle DW_CFA_undefined |
| 1196 | n-i-bz DWARF CFI reader: handle icc9 generated CFI info better |
| 1197 | n-i-bz fix false uninit-value errs in icc9 generated FP code |
| 1198 | n-i-bz reduce extraneous frames in libmpiwrap.c |
| 1199 | n-i-bz support pselect6 on amd64-linux |
| 1200 | |
| 1201 | (3.2.2: 22 Jan 2007, vex r1729, valgrind r6545). |
| 1202 | |
| 1203 | |
sewardj | 172bce8 | 2006-09-16 01:00:35 +0000 | [diff] [blame] | 1204 | Release 3.2.1 (16 Sept 2006) |
| 1205 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 1206 | 3.2.1 adds x86/amd64 support for all SSE3 instructions except monitor |
| 1207 | and mwait, further reduces memcheck's false error rate on all |
| 1208 | platforms, adds support for recent binutils (in OpenSUSE 10.2 and |
| 1209 | Fedora Rawhide) and fixes a bunch of bugs in 3.2.0. Some of the fixed |
| 1210 | bugs were causing large programs to segfault with --tool=callgrind and |
| 1211 | --tool=cachegrind, so an upgrade is recommended. |
| 1212 | |
| 1213 | In view of the fact that any 3.3.0 release is unlikely to happen until |
| 1214 | well into 1Q07, we intend to keep the 3.2.X line alive for a while |
| 1215 | yet, and so we tentatively plan a 3.2.2 release sometime in December |
| 1216 | 06. |
| 1217 | |
| 1218 | The fixed bugs are as follows. Note that "n-i-bz" stands for "not in |
| 1219 | bugzilla" -- that is, a bug that was reported to us but never got a |
| 1220 | bugzilla entry. |
| 1221 | |
| 1222 | n-i-bz Expanding brk() into last available page asserts |
| 1223 | n-i-bz ppc64-linux stack RZ fast-case snafu |
| 1224 | n-i-bz 'c' in --gen-supps=yes doesn't work |
| 1225 | n-i-bz VG_N_SEGMENTS too low (users, 28 June) |
| 1226 | n-i-bz VG_N_SEGNAMES too low (Stu Robinson) |
| 1227 | 106852 x86->IR: fisttp (SSE3) |
| 1228 | 117172 FUTEX_WAKE does not use uaddr2 |
| 1229 | 124039 Lacks support for VKI_[GP]IO_UNIMAP* |
| 1230 | 127521 amd64->IR: 0xF0 0x48 0xF 0xC7 (cmpxchg8b) |
| 1231 | 128917 amd64->IR: 0x66 0xF 0xF6 0xC4 (psadbw,SSE2) |
| 1232 | 129246 JJ: ppc32/ppc64 syscalls, w/ patch |
| 1233 | 129358 x86->IR: fisttpl (SSE3) |
| 1234 | 129866 cachegrind/callgrind causes executable to die |
| 1235 | 130020 Can't stat .so/.exe error while reading symbols |
| 1236 | 130388 Valgrind aborts when process calls malloc_trim() |
| 1237 | 130638 PATCH: ppc32 missing system calls |
| 1238 | 130785 amd64->IR: unhandled instruction "pushfq" |
| 1239 | 131481: (HINT_NOP) vex x86->IR: 0xF 0x1F 0x0 0xF |
| 1240 | 131298 ==131481 |
| 1241 | 132146 Programs with long sequences of bswap[l,q]s |
| 1242 | 132918 vex amd64->IR: 0xD9 0xF8 (fprem) |
| 1243 | 132813 Assertion at priv/guest-x86/toIR.c:652 fails |
| 1244 | 133051 'cfsi->len > 0 && cfsi->len < 2000000' failed |
| 1245 | 132722 valgrind header files are not standard C |
| 1246 | n-i-bz Livelocks entire machine (users list, Timothy Terriberry) |
| 1247 | n-i-bz Alex Bennee mmap problem (9 Aug) |
| 1248 | n-i-bz BartV: Don't print more lines of a stack-trace than were obtained. |
| 1249 | n-i-bz ppc32 SuSE 10.1 redir |
| 1250 | n-i-bz amd64 padding suppressions |
| 1251 | n-i-bz amd64 insn printing fix. |
| 1252 | n-i-bz ppc cmp reg,reg fix |
| 1253 | n-i-bz x86/amd64 iropt e/rflag reduction rules |
| 1254 | n-i-bz SuSE 10.1 (ppc32) minor fixes |
| 1255 | 133678 amd64->IR: 0x48 0xF 0xC5 0xC0 (pextrw?) |
| 1256 | 133694 aspacem assertion: aspacem_minAddr <= holeStart |
| 1257 | n-i-bz callgrind: fix warning about malformed creator line |
| 1258 | n-i-bz callgrind: fix annotate script for data produced with |
| 1259 | --dump-instr=yes |
| 1260 | n-i-bz callgrind: fix failed assertion when toggling |
| 1261 | instrumentation mode |
| 1262 | n-i-bz callgrind: fix annotate script fix warnings with |
| 1263 | --collect-jumps=yes |
| 1264 | n-i-bz docs path hardwired (Dennis Lubert) |
| 1265 | |
| 1266 | The following bugs were not fixed, due primarily to lack of developer |
| 1267 | time, and also because bug reporters did not answer requests for |
| 1268 | feedback in time for the release: |
| 1269 | |
| 1270 | 129390 ppc?->IR: some kind of VMX prefetch (dstt) |
| 1271 | 129968 amd64->IR: 0xF 0xAE 0x0 (fxsave) |
| 1272 | 133054 'make install' fails with syntax errors |
| 1273 | n-i-bz Signal race condition (users list, 13 June, Johannes Berg) |
| 1274 | n-i-bz Unrecognised instruction at address 0x70198EC2 (users list, |
| 1275 | 19 July, Bennee) |
| 1276 | 132998 startup fails in when running on UML |
| 1277 | |
| 1278 | The following bug was tentatively fixed on the mainline but the fix |
| 1279 | was considered too risky to push into 3.2.X: |
| 1280 | |
| 1281 | 133154 crash when using client requests to register/deregister stack |
| 1282 | |
| 1283 | (3.2.1: 16 Sept 2006, vex r1658, valgrind r6070). |
| 1284 | |
| 1285 | |
sewardj | 3ecf8f2 | 2006-06-06 21:41:22 +0000 | [diff] [blame] | 1286 | Release 3.2.0 (7 June 2006) |
sewardj | 535a235 | 2006-04-18 02:04:52 +0000 | [diff] [blame] | 1287 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1288 | 3.2.0 is a feature release with many significant improvements and the |
| 1289 | usual collection of bug fixes. This release supports X86/Linux, |
| 1290 | AMD64/Linux, PPC32/Linux and PPC64/Linux. |
njn | a3e9676 | 2006-03-27 12:23:44 +0000 | [diff] [blame] | 1291 | |
sewardj | af92499 | 2006-05-26 13:51:37 +0000 | [diff] [blame] | 1292 | Performance, especially of Memcheck, is improved, Addrcheck has been |
| 1293 | removed, Callgrind has been added, PPC64/Linux support has been added, |
| 1294 | Lackey has been improved, and MPI support has been added. In detail: |
njn | a3e9676 | 2006-03-27 12:23:44 +0000 | [diff] [blame] | 1295 | |
sewardj | f41db3c | 2006-05-26 16:06:10 +0000 | [diff] [blame] | 1296 | - Memcheck has improved speed and reduced memory use. Run times are |
| 1297 | typically reduced by 15-30%, averaging about 24% for SPEC CPU2000. |
sewardj | af92499 | 2006-05-26 13:51:37 +0000 | [diff] [blame] | 1298 | The other tools have smaller but noticeable speed improvments. We |
| 1299 | are interested to hear what improvements users get. |
njn | a3e9676 | 2006-03-27 12:23:44 +0000 | [diff] [blame] | 1300 | |
sewardj | 3ecf8f2 | 2006-06-06 21:41:22 +0000 | [diff] [blame] | 1301 | Memcheck uses less memory due to the introduction of a compressed |
sewardj | af92499 | 2006-05-26 13:51:37 +0000 | [diff] [blame] | 1302 | representation for shadow memory. The space overhead has been |
| 1303 | reduced by a factor of up to four, depending on program behaviour. |
| 1304 | This means you should be able to run programs that use more memory |
| 1305 | than before without hitting problems. |
njn | a3e9676 | 2006-03-27 12:23:44 +0000 | [diff] [blame] | 1306 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1307 | - Addrcheck has been removed. It has not worked since version 2.4.0, |
| 1308 | and the speed and memory improvements to Memcheck make it redundant. |
| 1309 | If you liked using Addrcheck because it didn't give undefined value |
| 1310 | errors, you can use the new Memcheck option --undef-value-errors=no |
| 1311 | to get the same behaviour. |
| 1312 | |
sewardj | af92499 | 2006-05-26 13:51:37 +0000 | [diff] [blame] | 1313 | - The number of undefined-value errors incorrectly reported by |
| 1314 | Memcheck has been reduced (such false reports were already very |
| 1315 | rare). In particular, efforts have been made to ensure Memcheck |
| 1316 | works really well with gcc 4.0/4.1-generated code on X86/Linux and |
| 1317 | AMD64/Linux. |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1318 | |
| 1319 | - Josef Weidendorfer's popular Callgrind tool has been added. Folding |
sewardj | af92499 | 2006-05-26 13:51:37 +0000 | [diff] [blame] | 1320 | it in was a logical step given its popularity and usefulness, and |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1321 | makes it easier for us to ensure it works "out of the box" on all |
| 1322 | supported targets. The associated KDE KCachegrind GUI remains a |
| 1323 | separate project. |
njn | eaf0ca9 | 2006-04-09 01:23:29 +0000 | [diff] [blame] | 1324 | |
sewardj | 3ecf8f2 | 2006-06-06 21:41:22 +0000 | [diff] [blame] | 1325 | - A new release of the Valkyrie GUI for Memcheck, version 1.2.0, |
| 1326 | accompanies this release. Improvements over previous releases |
| 1327 | include improved robustness, many refinements to the user interface, |
| 1328 | and use of a standard autoconf/automake build system. You can get |
| 1329 | it from http://www.valgrind.org/downloads/guis.html. |
| 1330 | |
sewardj | 535a235 | 2006-04-18 02:04:52 +0000 | [diff] [blame] | 1331 | - Valgrind now works on PPC64/Linux. As with the AMD64/Linux port, |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1332 | this supports programs using to 32G of address space. On 64-bit |
| 1333 | capable PPC64/Linux setups, you get a dual architecture build so |
| 1334 | that both 32-bit and 64-bit executables can be run. Linux on POWER5 |
| 1335 | is supported, and POWER4 is also believed to work. Both 32-bit and |
| 1336 | 64-bit DWARF2 is supported. This port is known to work well with |
| 1337 | both gcc-compiled and xlc/xlf-compiled code. |
njn | a3e9676 | 2006-03-27 12:23:44 +0000 | [diff] [blame] | 1338 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1339 | - Floating point accuracy has been improved for PPC32/Linux. |
| 1340 | Specifically, the floating point rounding mode is observed on all FP |
| 1341 | arithmetic operations, and multiply-accumulate instructions are |
| 1342 | preserved by the compilation pipeline. This means you should get FP |
| 1343 | results which are bit-for-bit identical to a native run. These |
| 1344 | improvements are also present in the PPC64/Linux port. |
njn | b293916 | 2006-04-18 00:52:37 +0000 | [diff] [blame] | 1345 | |
njn | eaf0ca9 | 2006-04-09 01:23:29 +0000 | [diff] [blame] | 1346 | - Lackey, the example tool, has been improved: |
njn | a3e9676 | 2006-03-27 12:23:44 +0000 | [diff] [blame] | 1347 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1348 | * It has a new option --detailed-counts (off by default) which |
| 1349 | causes it to print out a count of loads, stores and ALU operations |
| 1350 | done, and their sizes. |
njn | dbf7ca7 | 2006-03-31 11:57:59 +0000 | [diff] [blame] | 1351 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1352 | * It has a new option --trace-mem (off by default) which causes it |
| 1353 | to print out a trace of all memory accesses performed by a |
| 1354 | program. It's a good starting point for building Valgrind tools |
| 1355 | that need to track memory accesses. Read the comments at the top |
| 1356 | of the file lackey/lk_main.c for details. |
sewardj | 42fcf5d | 2006-04-08 13:16:36 +0000 | [diff] [blame] | 1357 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1358 | * The original instrumentation (counting numbers of instructions, |
| 1359 | jumps, etc) is now controlled by a new option --basic-counts. It |
| 1360 | is on by default. |
sewardj | 42fcf5d | 2006-04-08 13:16:36 +0000 | [diff] [blame] | 1361 | |
| 1362 | - MPI support: partial support for debugging distributed applications |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1363 | using the MPI library specification has been added. Valgrind is |
sewardj | 42fcf5d | 2006-04-08 13:16:36 +0000 | [diff] [blame] | 1364 | aware of the memory state changes caused by a subset of the MPI |
| 1365 | functions, and will carefully check data passed to the (P)MPI_ |
| 1366 | interface. |
| 1367 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1368 | - A new flag, --error-exitcode=, has been added. This allows changing |
| 1369 | the exit code in runs where Valgrind reported errors, which is |
| 1370 | useful when using Valgrind as part of an automated test suite. |
sewardj | b977908 | 2006-05-12 23:50:15 +0000 | [diff] [blame] | 1371 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1372 | - Various segfaults when reading old-style "stabs" debug information |
| 1373 | have been fixed. |
njn | eaf0ca9 | 2006-04-09 01:23:29 +0000 | [diff] [blame] | 1374 | |
sewardj | c9d09cc | 2006-05-27 03:22:04 +0000 | [diff] [blame] | 1375 | - A simple performance evaluation suite has been added. See |
njn | 5359b6f | 2006-06-02 23:57:22 +0000 | [diff] [blame] | 1376 | perf/README and README_DEVELOPERS for details. There are |
sewardj | c9d09cc | 2006-05-27 03:22:04 +0000 | [diff] [blame] | 1377 | various bells and whistles. |
| 1378 | |
sewardj | b70a613 | 2006-05-27 21:14:09 +0000 | [diff] [blame] | 1379 | - New configuration flags: |
| 1380 | --enable-only32bit |
| 1381 | --enable-only64bit |
| 1382 | By default, on 64 bit platforms (ppc64-linux, amd64-linux) the build |
| 1383 | system will attempt to build a Valgrind which supports both 32-bit |
| 1384 | and 64-bit executables. This may not be what you want, and you can |
| 1385 | override the default behaviour using these flags. |
| 1386 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1387 | Please note that Helgrind is still not working. We have made an |
| 1388 | important step towards making it work again, however, with the |
| 1389 | addition of function wrapping (see below). |
njn | eaf0ca9 | 2006-04-09 01:23:29 +0000 | [diff] [blame] | 1390 | |
| 1391 | Other user-visible changes: |
| 1392 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1393 | - Valgrind now has the ability to intercept and wrap arbitrary |
| 1394 | functions. This is a preliminary step towards making Helgrind work |
| 1395 | again, and was required for MPI support. |
njn | eaf0ca9 | 2006-04-09 01:23:29 +0000 | [diff] [blame] | 1396 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1397 | - There are some changes to Memcheck's client requests. Some of them |
| 1398 | have changed names: |
njn | dbf7ca7 | 2006-03-31 11:57:59 +0000 | [diff] [blame] | 1399 | |
| 1400 | MAKE_NOACCESS --> MAKE_MEM_NOACCESS |
| 1401 | MAKE_WRITABLE --> MAKE_MEM_UNDEFINED |
| 1402 | MAKE_READABLE --> MAKE_MEM_DEFINED |
| 1403 | |
| 1404 | CHECK_WRITABLE --> CHECK_MEM_IS_ADDRESSABLE |
| 1405 | CHECK_READABLE --> CHECK_MEM_IS_DEFINED |
| 1406 | CHECK_DEFINED --> CHECK_VALUE_IS_DEFINED |
| 1407 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1408 | The reason for the change is that the old names are subtly |
| 1409 | misleading. The old names will still work, but they are deprecated |
| 1410 | and may be removed in a future release. |
njn | dbf7ca7 | 2006-03-31 11:57:59 +0000 | [diff] [blame] | 1411 | |
| 1412 | We also added a new client request: |
| 1413 | |
| 1414 | MAKE_MEM_DEFINED_IF_ADDRESSABLE(a, len) |
| 1415 | |
| 1416 | which is like MAKE_MEM_DEFINED but only affects a byte if the byte is |
| 1417 | already addressable. |
| 1418 | |
sewardj | 3ecf8f2 | 2006-06-06 21:41:22 +0000 | [diff] [blame] | 1419 | - The way client requests are encoded in the instruction stream has |
| 1420 | changed. Unfortunately, this means 3.2.0 will not honour client |
| 1421 | requests compiled into binaries using headers from earlier versions |
| 1422 | of Valgrind. We will try to keep the client request encodings more |
| 1423 | stable in future. |
njn | eaf0ca9 | 2006-04-09 01:23:29 +0000 | [diff] [blame] | 1424 | |
njn | a3e9676 | 2006-03-27 12:23:44 +0000 | [diff] [blame] | 1425 | BUGS FIXED: |
| 1426 | |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1427 | 108258 NPTL pthread cleanup handlers not called |
| 1428 | 117290 valgrind is sigKILL'd on startup |
| 1429 | 117295 == 117290 |
| 1430 | 118703 m_signals.c:1427 Assertion 'tst->status == VgTs_WaitSys' |
| 1431 | 118466 add %reg, %reg generates incorrect validity for bit 0 |
| 1432 | 123210 New: strlen from ld-linux on amd64 |
| 1433 | 123244 DWARF2 CFI reader: unhandled CFI instruction 0:18 |
| 1434 | 123248 syscalls in glibc-2.4: openat, fstatat, symlinkat |
| 1435 | 123258 socketcall.recvmsg(msg.msg_iov[i] points to uninit |
| 1436 | 123535 mremap(new_addr) requires MREMAP_FIXED in 4th arg |
| 1437 | 123836 small typo in the doc |
| 1438 | 124029 ppc compile failed: `vor' gcc 3.3.5 |
| 1439 | 124222 Segfault: @@don't know what type ':' is |
| 1440 | 124475 ppc32: crash (syscall?) timer_settime() |
| 1441 | 124499 amd64->IR: 0xF 0xE 0x48 0x85 (femms) |
| 1442 | 124528 FATAL: aspacem assertion failed: segment_is_sane |
| 1443 | 124697 vex x86->IR: 0xF 0x70 0xC9 0x0 (pshufw) |
| 1444 | 124892 vex x86->IR: 0xF3 0xAE (REPx SCASB) |
| 1445 | 126216 == 124892 |
| 1446 | 124808 ppc32: sys_sched_getaffinity() not handled |
| 1447 | n-i-bz Very long stabs strings crash m_debuginfo |
| 1448 | n-i-bz amd64->IR: 0x66 0xF 0xF5 (pmaddwd) |
| 1449 | 125492 ppc32: support a bunch more syscalls |
| 1450 | 121617 ppc32/64: coredumping gives assertion failure |
| 1451 | 121814 Coregrind return error as exitcode patch |
| 1452 | 126517 == 121814 |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1453 | 125607 amd64->IR: 0x66 0xF 0xA3 0x2 (btw etc) |
| 1454 | 125651 amd64->IR: 0xF8 0x49 0xFF 0xE3 (clc?) |
| 1455 | 126253 x86 movx is wrong |
| 1456 | 126451 3.2 SVN doesn't work on ppc32 CPU's without FPU |
| 1457 | 126217 increase # threads |
| 1458 | 126243 vex x86->IR: popw mem |
| 1459 | 126583 amd64->IR: 0x48 0xF 0xA4 0xC2 (shld $1,%rax,%rdx) |
sewardj | b70a613 | 2006-05-27 21:14:09 +0000 | [diff] [blame] | 1460 | 126668 amd64->IR: 0x1C 0xFF (sbb $0xff,%al) |
| 1461 | 126696 support for CDROMREADRAW ioctl and CDROMREADTOCENTRY fix |
| 1462 | 126722 assertion: segment_is_sane at m_aspacemgr/aspacemgr.c:1624 |
| 1463 | 126938 bad checking for syscalls linkat, renameat, symlinkat |
sewardj | 1ede80f | 2006-05-26 00:47:13 +0000 | [diff] [blame] | 1464 | |
sewardj | 3ecf8f2 | 2006-06-06 21:41:22 +0000 | [diff] [blame] | 1465 | (3.2.0RC1: 27 May 2006, vex r1626, valgrind r5947). |
| 1466 | (3.2.0: 7 June 2006, vex r1628, valgrind r5957). |
njn | a3e9676 | 2006-03-27 12:23:44 +0000 | [diff] [blame] | 1467 | |
| 1468 | |
sewardj | 5065467 | 2006-03-16 11:33:14 +0000 | [diff] [blame] | 1469 | Release 3.1.1 (15 March 2006) |
| 1470 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 1471 | 3.1.1 fixes a bunch of bugs reported in 3.1.0. There is no new |
| 1472 | functionality. The fixed bugs are: |
| 1473 | |
| 1474 | (note: "n-i-bz" means "not in bugzilla" -- this bug does not have |
| 1475 | a bugzilla entry). |
| 1476 | |
| 1477 | n-i-bz ppc32: fsub 3,3,3 in dispatcher doesn't clear NaNs |
| 1478 | n-i-bz ppc32: __NR_{set,get}priority |
| 1479 | 117332 x86: missing line info with icc 8.1 |
| 1480 | 117366 amd64: 0xDD 0x7C fnstsw |
| 1481 | 118274 == 117366 |
| 1482 | 117367 amd64: 0xD9 0xF4 fxtract |
| 1483 | 117369 amd64: __NR_getpriority (140) |
| 1484 | 117419 ppc32: lfsu f5, -4(r11) |
| 1485 | 117419 ppc32: fsqrt |
| 1486 | 117936 more stabs problems (segfaults while reading debug info) |
| 1487 | 119914 == 117936 |
| 1488 | 120345 == 117936 |
| 1489 | 118239 amd64: 0xF 0xAE 0x3F (clflush) |
| 1490 | 118939 vm86old system call |
| 1491 | n-i-bz memcheck/tests/mempool reads freed memory |
| 1492 | n-i-bz AshleyP's custom-allocator assertion |
| 1493 | n-i-bz Dirk strict-aliasing stuff |
| 1494 | n-i-bz More space for debugger cmd line (Dan Thaler) |
| 1495 | n-i-bz Clarified leak checker output message |
| 1496 | n-i-bz AshleyP's --gen-suppressions output fix |
| 1497 | n-i-bz cg_annotate's --sort option broken |
| 1498 | n-i-bz OSet 64-bit fastcmp bug |
| 1499 | n-i-bz VG_(getgroups) fix (Shinichi Noda) |
| 1500 | n-i-bz ppc32: allocate from callee-saved FP/VMX regs |
| 1501 | n-i-bz misaligned path word-size bug in mc_main.c |
| 1502 | 119297 Incorrect error message for sse code |
| 1503 | 120410 x86: prefetchw (0xF 0xD 0x48 0x4) |
| 1504 | 120728 TIOCSERGETLSR, TIOCGICOUNT, HDIO_GET_DMA ioctls |
| 1505 | 120658 Build fixes for gcc 2.96 |
| 1506 | 120734 x86: Support for changing EIP in signal handler |
| 1507 | n-i-bz memcheck/tests/zeropage de-looping fix |
| 1508 | n-i-bz x86: fxtract doesn't work reliably |
| 1509 | 121662 x86: lock xadd (0xF0 0xF 0xC0 0x2) |
| 1510 | 121893 calloc does not always return zeroed memory |
| 1511 | 121901 no support for syscall tkill |
| 1512 | n-i-bz Suppression update for Debian unstable |
| 1513 | 122067 amd64: fcmovnu (0xDB 0xD9) |
| 1514 | n-i-bz ppc32: broken signal handling in cpu feature detection |
| 1515 | n-i-bz ppc32: rounding mode problems (improved, partial fix only) |
| 1516 | 119482 ppc32: mtfsb1 |
| 1517 | n-i-bz ppc32: mtocrf/mfocrf |
| 1518 | |
| 1519 | (3.1.1: 15 March 2006, vex r1597, valgrind r5771). |
| 1520 | |
jseward | 4323065 | 2003-12-14 00:14:54 +0000 | [diff] [blame] | 1521 | |
sewardj | 4bbe9d7 | 2005-11-16 18:39:25 +0000 | [diff] [blame] | 1522 | Release 3.1.0 (25 November 2005) |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1523 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
sewardj | 4bbe9d7 | 2005-11-16 18:39:25 +0000 | [diff] [blame] | 1524 | 3.1.0 is a feature release with a number of significant improvements: |
| 1525 | AMD64 support is much improved, PPC32 support is good enough to be |
| 1526 | usable, and the handling of memory management and address space is |
| 1527 | much more robust. In detail: |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1528 | |
sewardj | 4bbe9d7 | 2005-11-16 18:39:25 +0000 | [diff] [blame] | 1529 | - AMD64 support is much improved. The 64-bit vs. 32-bit issues in |
| 1530 | 3.0.X have been resolved, and it should "just work" now in all |
| 1531 | cases. On AMD64 machines both 64-bit and 32-bit versions of |
| 1532 | Valgrind are built. The right version will be invoked |
| 1533 | automatically, even when using --trace-children and mixing execution |
| 1534 | between 64-bit and 32-bit executables. Also, many more instructions |
| 1535 | are supported. |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1536 | |
sewardj | 4bbe9d7 | 2005-11-16 18:39:25 +0000 | [diff] [blame] | 1537 | - PPC32 support is now good enough to be usable. It should work with |
| 1538 | all tools, but please let us know if you have problems. Three |
| 1539 | classes of CPUs are supported: integer only (no FP, no Altivec), |
| 1540 | which covers embedded PPC uses, integer and FP but no Altivec |
| 1541 | (G3-ish), and CPUs capable of Altivec too (G4, G5). |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1542 | |
sewardj | 4bbe9d7 | 2005-11-16 18:39:25 +0000 | [diff] [blame] | 1543 | - Valgrind's address space management has been overhauled. As a |
| 1544 | result, Valgrind should be much more robust with programs that use |
| 1545 | large amounts of memory. There should be many fewer "memory |
| 1546 | exhausted" messages, and debug symbols should be read correctly on |
| 1547 | large (eg. 300MB+) executables. On 32-bit machines the full address |
| 1548 | space available to user programs (usually 3GB or 4GB) can be fully |
| 1549 | utilised. On 64-bit machines up to 32GB of space is usable; when |
| 1550 | using Memcheck that means your program can use up to about 14GB. |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1551 | |
| 1552 | A side effect of this change is that Valgrind is no longer protected |
| 1553 | against wild writes by the client. This feature was nice but relied |
| 1554 | on the x86 segment registers and so wasn't portable. |
| 1555 | |
| 1556 | - Most users should not notice, but as part of the address space |
sewardj | 4bbe9d7 | 2005-11-16 18:39:25 +0000 | [diff] [blame] | 1557 | manager change, the way Valgrind is built has been changed. Each |
| 1558 | tool is now built as a statically linked stand-alone executable, |
| 1559 | rather than as a shared object that is dynamically linked with the |
| 1560 | core. The "valgrind" program invokes the appropriate tool depending |
| 1561 | on the --tool option. This slightly increases the amount of disk |
| 1562 | space used by Valgrind, but it greatly simplified many things and |
| 1563 | removed Valgrind's dependence on glibc. |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1564 | |
njn | 336c89d | 2005-11-20 18:14:04 +0000 | [diff] [blame] | 1565 | Please note that Addrcheck and Helgrind are still not working. Work |
| 1566 | is underway to reinstate them (or equivalents). We apologise for the |
| 1567 | inconvenience. |
| 1568 | |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1569 | Other user-visible changes: |
| 1570 | |
| 1571 | - The --weird-hacks option has been renamed --sim-hints. |
| 1572 | |
| 1573 | - The --time-stamp option no longer gives an absolute date and time. |
| 1574 | It now prints the time elapsed since the program began. |
| 1575 | |
sewardj | 4bbe9d7 | 2005-11-16 18:39:25 +0000 | [diff] [blame] | 1576 | - It should build with gcc-2.96. |
| 1577 | |
njn | 7cce5b8 | 2005-11-16 20:12:22 +0000 | [diff] [blame] | 1578 | - Valgrind can now run itself (see README_DEVELOPERS for how). |
sewardj | 4705861 | 2005-11-16 19:13:13 +0000 | [diff] [blame] | 1579 | This is not much use to you, but it means the developers can now |
| 1580 | profile Valgrind using Cachegrind. As a result a couple of |
| 1581 | performance bad cases have been fixed. |
| 1582 | |
sewardj | 6a5a69c | 2005-11-17 00:51:36 +0000 | [diff] [blame] | 1583 | - The XML output format has changed slightly. See |
| 1584 | docs/internals/xml-output.txt. |
| 1585 | |
njn | 14c9add | 2005-11-18 17:36:01 +0000 | [diff] [blame] | 1586 | - Core dumping has been reinstated (it was disabled in 3.0.0 and 3.0.1). |
| 1587 | If your program crashes while running under Valgrind, a core file with |
| 1588 | the name "vgcore.<pid>" will be created (if your settings allow core |
| 1589 | file creation). Note that the floating point information is not all |
| 1590 | there. If Valgrind itself crashes, the OS will create a normal core |
| 1591 | file. |
| 1592 | |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1593 | The following are some user-visible changes that occurred in earlier |
| 1594 | versions that may not have been announced, or were announced but not |
njn | 7cce5b8 | 2005-11-16 20:12:22 +0000 | [diff] [blame] | 1595 | widely noticed. So we're mentioning them now. |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1596 | |
| 1597 | - The --tool flag is optional once again; if you omit it, Memcheck |
| 1598 | is run by default. |
| 1599 | |
| 1600 | - The --num-callers flag now has a default value of 12. It was |
| 1601 | previously 4. |
| 1602 | |
njn | 7cce5b8 | 2005-11-16 20:12:22 +0000 | [diff] [blame] | 1603 | - The --xml=yes flag causes Valgrind's output to be produced in XML |
| 1604 | format. This is designed to make it easy for other programs to |
| 1605 | consume Valgrind's output. The format is described in the file |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1606 | docs/internals/xml-format.txt. |
| 1607 | |
njn | 1f50399 | 2005-11-13 16:41:15 +0000 | [diff] [blame] | 1608 | - The --gen-suppressions flag supports an "all" value that causes every |
| 1609 | suppression to be printed without asking. |
| 1610 | |
njn | ee88d28 | 2005-11-20 17:19:28 +0000 | [diff] [blame] | 1611 | - The --log-file option no longer puts "pid" in the filename, eg. the |
| 1612 | old name "foo.pid12345" is now "foo.12345". |
| 1613 | |
njn | c66077e | 2005-11-20 18:02:15 +0000 | [diff] [blame] | 1614 | - There are several graphical front-ends for Valgrind, such as Valkyrie, |
| 1615 | Alleyoop and Valgui. See http://www.valgrind.org/downloads/guis.html |
| 1616 | for a list. |
| 1617 | |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1618 | BUGS FIXED: |
| 1619 | |
sewardj | 4bbe9d7 | 2005-11-16 18:39:25 +0000 | [diff] [blame] | 1620 | 109861 amd64 hangs at startup |
| 1621 | 110301 ditto |
| 1622 | 111554 valgrind crashes with Cannot allocate memory |
| 1623 | 111809 Memcheck tool doesn't start java |
| 1624 | 111901 cross-platform run of cachegrind fails on opteron |
| 1625 | 113468 (vgPlain_mprotect_range): Assertion 'r != -1' failed. |
| 1626 | 92071 Reading debugging info uses too much memory |
| 1627 | 109744 memcheck loses track of mmap from direct ld-linux.so.2 |
| 1628 | 110183 tail of page with _end |
| 1629 | 82301 FV memory layout too rigid |
| 1630 | 98278 Infinite recursion possible when allocating memory |
| 1631 | 108994 Valgrind runs out of memory due to 133x overhead |
| 1632 | 115643 valgrind cannot allocate memory |
| 1633 | 105974 vg_hashtable.c static hash table |
| 1634 | 109323 ppc32: dispatch.S uses Altivec insn, which doesn't work on POWER. |
| 1635 | 109345 ptrace_setregs not yet implemented for ppc |
| 1636 | 110831 Would like to be able to run against both 32 and 64 bit |
| 1637 | binaries on AMD64 |
| 1638 | 110829 == 110831 |
| 1639 | 111781 compile of valgrind-3.0.0 fails on my linux (gcc 2.X prob) |
| 1640 | 112670 Cachegrind: cg_main.c:486 (handleOneStatement ... |
| 1641 | 112941 vex x86: 0xD9 0xF4 (fxtract) |
| 1642 | 110201 == 112941 |
| 1643 | 113015 vex amd64->IR: 0xE3 0x14 0x48 0x83 (jrcxz) |
| 1644 | 113126 Crash with binaries built with -gstabs+/-ggdb |
| 1645 | 104065 == 113126 |
| 1646 | 115741 == 113126 |
| 1647 | 113403 Partial SSE3 support on x86 |
| 1648 | 113541 vex: Grp5(x86) (alt encoding inc/dec) case 1 |
| 1649 | 113642 valgrind crashes when trying to read debug information |
| 1650 | 113810 vex x86->IR: 66 0F F6 (66 + PSADBW == SSE PSADBW) |
| 1651 | 113796 read() and write() do not work if buffer is in shared memory |
| 1652 | 113851 vex x86->IR: (pmaddwd): 0x66 0xF 0xF5 0xC7 |
| 1653 | 114366 vex amd64 cannnot handle __asm__( "fninit" ) |
| 1654 | 114412 vex amd64->IR: 0xF 0xAD 0xC2 0xD3 (128-bit shift, shrdq?) |
| 1655 | 114455 vex amd64->IR: 0xF 0xAC 0xD0 0x1 (also shrdq) |
| 1656 | 115590: amd64->IR: 0x67 0xE3 0x9 0xEB (address size override) |
| 1657 | 115953 valgrind svn r5042 does not build with parallel make (-j3) |
| 1658 | 116057 maximum instruction size - VG_MAX_INSTR_SZB too small? |
| 1659 | 116483 shmat failes with invalid argument |
| 1660 | 102202 valgrind crashes when realloc'ing until out of memory |
| 1661 | 109487 == 102202 |
| 1662 | 110536 == 102202 |
| 1663 | 112687 == 102202 |
| 1664 | 111724 vex amd64->IR: 0x41 0xF 0xAB (more BT{,S,R,C} fun n games) |
| 1665 | 111748 vex amd64->IR: 0xDD 0xE2 (fucom) |
| 1666 | 111785 make fails if CC contains spaces |
| 1667 | 111829 vex x86->IR: sbb AL, Ib |
| 1668 | 111851 vex x86->IR: 0x9F 0x89 (lahf/sahf) |
| 1669 | 112031 iopl on AMD64 and README_MISSING_SYSCALL_OR_IOCTL update |
| 1670 | 112152 code generation for Xin_MFence on x86 with SSE0 subarch |
| 1671 | 112167 == 112152 |
| 1672 | 112789 == 112152 |
| 1673 | 112199 naked ar tool is used in vex makefile |
| 1674 | 112501 vex x86->IR: movq (0xF 0x7F 0xC1 0xF) (mmx MOVQ) |
| 1675 | 113583 == 112501 |
| 1676 | 112538 memalign crash |
| 1677 | 113190 Broken links in docs/html/ |
| 1678 | 113230 Valgrind sys_pipe on x86-64 wrongly thinks file descriptors |
| 1679 | should be 64bit |
| 1680 | 113996 vex amd64->IR: fucomp (0xDD 0xE9) |
| 1681 | 114196 vex x86->IR: out %eax,(%dx) (0xEF 0xC9 0xC3 0x90) |
| 1682 | 114289 Memcheck fails to intercept malloc when used in an uclibc environment |
| 1683 | 114756 mbind syscall support |
| 1684 | 114757 Valgrind dies with assertion: Assertion 'noLargerThan > 0' failed |
| 1685 | 114563 stack tracking module not informed when valgrind switches threads |
| 1686 | 114564 clone() and stacks |
| 1687 | 114565 == 114564 |
| 1688 | 115496 glibc crashes trying to use sysinfo page |
| 1689 | 116200 enable fsetxattr, fgetxattr, and fremovexattr for amd64 |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1690 | |
sewardj | ae17e7e | 2005-11-20 19:45:30 +0000 | [diff] [blame] | 1691 | (3.1.0RC1: 20 November 2005, vex r1466, valgrind r5224). |
sewardj | fa46202 | 2005-11-26 13:38:58 +0000 | [diff] [blame] | 1692 | (3.1.0: 26 November 2005, vex r1471, valgrind r5235). |
sewardj | ae17e7e | 2005-11-20 19:45:30 +0000 | [diff] [blame] | 1693 | |
njn | f63170c | 2005-11-12 19:09:51 +0000 | [diff] [blame] | 1694 | |
sewardj | cd62fe7 | 2005-08-29 13:24:51 +0000 | [diff] [blame] | 1695 | Release 3.0.1 (29 August 2005) |
| 1696 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 1697 | 3.0.1 fixes a bunch of bugs reported in 3.0.0. There is no new |
| 1698 | functionality. Some of the fixed bugs are critical, so if you |
sewardj | fa23389 | 2005-08-30 12:35:36 +0000 | [diff] [blame] | 1699 | use/distribute 3.0.0, an upgrade to 3.0.1 is recommended. The fixed |
sewardj | cd62fe7 | 2005-08-29 13:24:51 +0000 | [diff] [blame] | 1700 | bugs are: |
| 1701 | |
| 1702 | (note: "n-i-bz" means "not in bugzilla" -- this bug does not have |
| 1703 | a bugzilla entry). |
| 1704 | |
| 1705 | 109313 (== 110505) x86 cmpxchg8b |
| 1706 | n-i-bz x86: track but ignore changes to %eflags.AC (alignment check) |
| 1707 | 110102 dis_op2_E_G(amd64) |
| 1708 | 110202 x86 sys_waitpid(#286) |
| 1709 | 110203 clock_getres(,0) |
| 1710 | 110208 execve fail wrong retval |
| 1711 | 110274 SSE1 now mandatory for x86 |
| 1712 | 110388 amd64 0xDD 0xD1 |
| 1713 | 110464 amd64 0xDC 0x1D FCOMP |
| 1714 | 110478 amd64 0xF 0xD PREFETCH |
| 1715 | n-i-bz XML <unique> printing wrong |
| 1716 | n-i-bz Dirk r4359 (amd64 syscalls from trunk) |
| 1717 | 110591 amd64 and x86: rdtsc not implemented properly |
| 1718 | n-i-bz Nick r4384 (stub implementations of Addrcheck and Helgrind) |
| 1719 | 110652 AMD64 valgrind crashes on cwtd instruction |
| 1720 | 110653 AMD64 valgrind crashes on sarb $0x4,foo(%rip) instruction |
| 1721 | 110656 PATH=/usr/bin::/bin valgrind foobar stats ./fooba |
| 1722 | 110657 Small test fixes |
| 1723 | 110671 vex x86->IR: unhandled instruction bytes: 0xF3 0xC3 (rep ret) |
| 1724 | n-i-bz Nick (Cachegrind should not assert when it encounters a client |
| 1725 | request.) |
| 1726 | 110685 amd64->IR: unhandled instruction bytes: 0xE1 0x56 (loope Jb) |
| 1727 | 110830 configuring with --host fails to build 32 bit on 64 bit target |
| 1728 | 110875 Assertion when execve fails |
| 1729 | n-i-bz Updates to Memcheck manual |
| 1730 | n-i-bz Fixed broken malloc_usable_size() |
| 1731 | 110898 opteron instructions missing: btq btsq btrq bsfq |
| 1732 | 110954 x86->IR: unhandled instruction bytes: 0xE2 0xF6 (loop Jb) |
| 1733 | n-i-bz Make suppressions work for "???" lines in stacktraces. |
| 1734 | 111006 bogus warnings from linuxthreads |
| 1735 | 111092 x86: dis_Grp2(Reg): unhandled case(x86) |
| 1736 | 111231 sctp_getladdrs() and sctp_getpaddrs() returns uninitialized |
| 1737 | memory |
| 1738 | 111102 (comment #4) Fixed 64-bit unclean "silly arg" message |
| 1739 | n-i-bz vex x86->IR: unhandled instruction bytes: 0x14 0x0 |
| 1740 | n-i-bz minor umount/fcntl wrapper fixes |
| 1741 | 111090 Internal Error running Massif |
| 1742 | 101204 noisy warning |
| 1743 | 111513 Illegal opcode for SSE instruction (x86 movups) |
| 1744 | 111555 VEX/Makefile: CC is set to gcc |
sewardj | fa23389 | 2005-08-30 12:35:36 +0000 | [diff] [blame] | 1745 | n-i-bz Fix XML bugs in FAQ |
sewardj | cd62fe7 | 2005-08-29 13:24:51 +0000 | [diff] [blame] | 1746 | |
sewardj | fa23389 | 2005-08-30 12:35:36 +0000 | [diff] [blame] | 1747 | (3.0.1: 29 August 05, |
| 1748 | vex/branches/VEX_3_0_BRANCH r1367, |
| 1749 | valgrind/branches/VALGRIND_3_0_BRANCH r4574). |
sewardj | 820c772 | 2005-08-29 13:44:43 +0000 | [diff] [blame] | 1750 | |
| 1751 | |
sewardj | cd62fe7 | 2005-08-29 13:24:51 +0000 | [diff] [blame] | 1752 | |
sewardj | ec39f33 | 2005-08-02 15:28:07 +0000 | [diff] [blame] | 1753 | Release 3.0.0 (3 August 2005) |
| 1754 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
sewardj | f9929e6 | 2005-08-02 21:05:58 +0000 | [diff] [blame] | 1755 | 3.0.0 is a major overhaul of Valgrind. The most significant user |
| 1756 | visible change is that Valgrind now supports architectures other than |
| 1757 | x86. The new architectures it supports are AMD64 and PPC32, and the |
| 1758 | infrastructure is present for other architectures to be added later. |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1759 | |
sewardj | f9929e6 | 2005-08-02 21:05:58 +0000 | [diff] [blame] | 1760 | AMD64 support works well, but has some shortcomings: |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1761 | |
njn | 7b1a6a4 | 2005-07-20 04:20:44 +0000 | [diff] [blame] | 1762 | - It generally won't be as solid as the x86 version. For example, |
| 1763 | support for more obscure instructions and system calls may be missing. |
| 1764 | We will fix these as they arise. |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1765 | |
sewardj | c371bd0 | 2005-07-27 09:55:24 +0000 | [diff] [blame] | 1766 | - Address space may be limited; see the point about |
njn | 7b1a6a4 | 2005-07-20 04:20:44 +0000 | [diff] [blame] | 1767 | position-independent executables below. |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1768 | |
njn | 7b1a6a4 | 2005-07-20 04:20:44 +0000 | [diff] [blame] | 1769 | - If Valgrind is built on an AMD64 machine, it will only run 64-bit |
| 1770 | executables. If you want to run 32-bit x86 executables under Valgrind |
| 1771 | on an AMD64, you will need to build Valgrind on an x86 machine and |
| 1772 | copy it to the AMD64 machine. And it probably won't work if you do |
| 1773 | something tricky like exec'ing a 32-bit program from a 64-bit program |
| 1774 | while using --trace-children=yes. We hope to improve this situation |
| 1775 | in the future. |
| 1776 | |
| 1777 | The PPC32 support is very basic. It may not work reliably even for |
sewardj | c371bd0 | 2005-07-27 09:55:24 +0000 | [diff] [blame] | 1778 | small programs, but it's a start. Many thanks to Paul Mackerras for |
| 1779 | his great work that enabled this support. We are working to make |
| 1780 | PPC32 usable as soon as possible. |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1781 | |
| 1782 | Other user-visible changes: |
| 1783 | |
sewardj | ec39f33 | 2005-08-02 15:28:07 +0000 | [diff] [blame] | 1784 | - Valgrind is no longer built by default as a position-independent |
| 1785 | executable (PIE), as this caused too many problems. |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1786 | |
| 1787 | Without PIE enabled, AMD64 programs will only be able to access 2GB of |
| 1788 | address space. We will fix this eventually, but not for the moment. |
| 1789 | |
| 1790 | Use --enable-pie at configure-time to turn this on. |
| 1791 | |
| 1792 | - Support for programs that use stack-switching has been improved. Use |
| 1793 | the --max-stackframe flag for simple cases, and the |
| 1794 | VALGRIND_STACK_REGISTER, VALGRIND_STACK_DEREGISTER and |
| 1795 | VALGRIND_STACK_CHANGE client requests for trickier cases. |
| 1796 | |
njn | 7b1a6a4 | 2005-07-20 04:20:44 +0000 | [diff] [blame] | 1797 | - Support for programs that use self-modifying code has been improved, |
| 1798 | in particular programs that put temporary code fragments on the stack. |
| 1799 | This helps for C programs compiled with GCC that use nested functions, |
njn | e772332 | 2005-07-24 17:19:09 +0000 | [diff] [blame] | 1800 | and also Ada programs. This is controlled with the --smc-check |
njn | 7b1a6a4 | 2005-07-20 04:20:44 +0000 | [diff] [blame] | 1801 | flag, although the default setting should work in most cases. |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1802 | |
njn | 7b1a6a4 | 2005-07-20 04:20:44 +0000 | [diff] [blame] | 1803 | - Output can now be printed in XML format. This should make it easier |
sewardj | c371bd0 | 2005-07-27 09:55:24 +0000 | [diff] [blame] | 1804 | for tools such as GUI front-ends and automated error-processing |
| 1805 | schemes to use Valgrind output as input. The --xml flag controls this. |
| 1806 | As part of this change, ELF directory information is read from executables, |
| 1807 | so absolute source file paths are available if needed. |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1808 | |
| 1809 | - Programs that allocate many heap blocks may run faster, due to |
| 1810 | improvements in certain data structures. |
| 1811 | |
njn | 7b1a6a4 | 2005-07-20 04:20:44 +0000 | [diff] [blame] | 1812 | - Addrcheck is currently not working. We hope to get it working again |
| 1813 | soon. Helgrind is still not working, as was the case for the 2.4.0 |
| 1814 | release. |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1815 | |
sewardj | c371bd0 | 2005-07-27 09:55:24 +0000 | [diff] [blame] | 1816 | - The JITter has been completely rewritten, and is now in a separate |
| 1817 | library, called Vex. This enabled a lot of the user-visible changes, |
| 1818 | such as new architecture support. The new JIT unfortunately translates |
| 1819 | more slowly than the old one, so programs may take longer to start. |
| 1820 | We believe the code quality is produces is about the same, so once |
| 1821 | started, programs should run at about the same speed. Feedback about |
| 1822 | this would be useful. |
| 1823 | |
| 1824 | On the plus side, Vex and hence Memcheck tracks value flow properly |
| 1825 | through floating point and vector registers, something the 2.X line |
| 1826 | could not do. That means that Memcheck is much more likely to be |
| 1827 | usably accurate on vectorised code. |
| 1828 | |
sewardj | ec39f33 | 2005-08-02 15:28:07 +0000 | [diff] [blame] | 1829 | - There is a subtle change to the way exiting of threaded programs |
sewardj | c371bd0 | 2005-07-27 09:55:24 +0000 | [diff] [blame] | 1830 | is handled. In 3.0, Valgrind's final diagnostic output (leak check, |
| 1831 | etc) is not printed until the last thread exits. If the last thread |
| 1832 | to exit was not the original thread which started the program, any |
| 1833 | other process wait()-ing on this one to exit may conclude it has |
| 1834 | finished before the diagnostic output is printed. This may not be |
| 1835 | what you expect. 2.X had a different scheme which avoided this |
| 1836 | problem, but caused deadlocks under obscure circumstances, so we |
| 1837 | are trying something different for 3.0. |
| 1838 | |
| 1839 | - Small changes in control log file naming which make it easier to |
sewardj | ec39f33 | 2005-08-02 15:28:07 +0000 | [diff] [blame] | 1840 | use valgrind for debugging MPI-based programs. The relevant |
| 1841 | new flags are --log-file-exactly= and --log-file-qualifier=. |
sewardj | c371bd0 | 2005-07-27 09:55:24 +0000 | [diff] [blame] | 1842 | |
| 1843 | - As part of adding AMD64 support, DWARF2 CFI-based stack unwinding |
| 1844 | support was added. In principle this means Valgrind can produce |
| 1845 | meaningful backtraces on x86 code compiled with -fomit-frame-pointer |
tom | 81868f0 | 2005-07-27 10:33:08 +0000 | [diff] [blame] | 1846 | providing you also compile your code with -fasynchronous-unwind-tables. |
sewardj | c371bd0 | 2005-07-27 09:55:24 +0000 | [diff] [blame] | 1847 | |
sewardj | ec39f33 | 2005-08-02 15:28:07 +0000 | [diff] [blame] | 1848 | - The documentation build system has been completely redone. |
| 1849 | The documentation masters are now in XML format, and from that |
| 1850 | HTML, PostScript and PDF documentation is generated. As a result |
| 1851 | the manual is now available in book form. Note that the |
| 1852 | documentation in the source tarballs is pre-built, so you don't need |
| 1853 | any XML processing tools to build Valgrind from a tarball. |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1854 | |
| 1855 | Changes that are not user-visible: |
| 1856 | |
sewardj | c371bd0 | 2005-07-27 09:55:24 +0000 | [diff] [blame] | 1857 | - The code has been massively overhauled in order to modularise it. |
| 1858 | As a result we hope it is easier to navigate and understand. |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1859 | |
| 1860 | - Lots of code has been rewritten. |
| 1861 | |
sewardj | f9929e6 | 2005-08-02 21:05:58 +0000 | [diff] [blame] | 1862 | BUGS FIXED: |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1863 | |
sewardj | 79bd9b6 | 2005-08-03 16:40:10 +0000 | [diff] [blame] | 1864 | 110046 sz == 4 assertion failed |
| 1865 | 109810 vex amd64->IR: unhandled instruction bytes: 0xA3 0x4C 0x70 0xD7 |
sewardj | f9929e6 | 2005-08-02 21:05:58 +0000 | [diff] [blame] | 1866 | 109802 Add a plausible_stack_size command-line parameter ? |
| 1867 | 109783 unhandled ioctl TIOCMGET (running hw detection tool discover) |
| 1868 | 109780 unhandled ioctl BLKSSZGET (running fdisk -l /dev/hda) |
| 1869 | 109718 vex x86->IR: unhandled instruction: ffreep |
| 1870 | 109429 AMD64 unhandled syscall: 127 (sigpending) |
| 1871 | 109401 false positive uninit in strchr from ld-linux.so.2 |
| 1872 | 109385 "stabs" parse failure |
| 1873 | 109378 amd64: unhandled instruction REP NOP |
| 1874 | 109376 amd64: unhandled instruction LOOP Jb |
| 1875 | 109363 AMD64 unhandled instruction bytes |
| 1876 | 109362 AMD64 unhandled syscall: 24 (sched_yield) |
| 1877 | 109358 fork() won't work with valgrind-3.0 SVN |
| 1878 | 109332 amd64 unhandled instruction: ADC Ev, Gv |
| 1879 | 109314 Bogus memcheck report on amd64 |
| 1880 | 108883 Crash; vg_memory.c:905 (vgPlain_init_shadow_range): |
| 1881 | Assertion `vgPlain_defined_init_shadow_page()' failed. |
| 1882 | 108349 mincore syscall parameter checked incorrectly |
| 1883 | 108059 build infrastructure: small update |
| 1884 | 107524 epoll_ctl event parameter checked on EPOLL_CTL_DEL |
| 1885 | 107123 Vex dies with unhandled instructions: 0xD9 0x31 0xF 0xAE |
| 1886 | 106841 auxmap & openGL problems |
| 1887 | 106713 SDL_Init causes valgrind to exit |
| 1888 | 106352 setcontext and makecontext not handled correctly |
| 1889 | 106293 addresses beyond initial client stack allocation |
| 1890 | not checked in VALGRIND_DO_LEAK_CHECK |
| 1891 | 106283 PIE client programs are loaded at address 0 |
| 1892 | 105831 Assertion `vgPlain_defined_init_shadow_page()' failed. |
| 1893 | 105039 long run-times probably due to memory manager |
| 1894 | 104797 valgrind needs to be aware of BLKGETSIZE64 |
| 1895 | 103594 unhandled instruction: FICOM |
| 1896 | 103320 Valgrind 2.4.0 fails to compile with gcc 3.4.3 and -O0 |
| 1897 | 103168 potentially memory leak in coregrind/ume.c |
| 1898 | 102039 bad permissions for mapped region at address 0xB7C73680 |
| 1899 | 101881 weird assertion problem |
| 1900 | 101543 Support fadvise64 syscalls |
| 1901 | 75247 x86_64/amd64 support (the biggest "bug" we have ever fixed) |
njn | 4a9f6f4 | 2005-07-19 23:46:08 +0000 | [diff] [blame] | 1902 | |
sewardj | 9a91951 | 2005-08-03 17:01:51 +0000 | [diff] [blame] | 1903 | (3.0RC1: 27 July 05, vex r1303, valgrind r4283). |
sewardj | 1346e13 | 2005-08-03 20:42:06 +0000 | [diff] [blame] | 1904 | (3.0.0: 3 August 05, vex r1313, valgrind r4316). |
sewardj | bfa2b99 | 2005-07-27 17:49:17 +0000 | [diff] [blame] | 1905 | |
| 1906 | |
| 1907 | |
njn | ccbe3b1 | 2009-08-20 00:22:10 +0000 | [diff] [blame] | 1908 | Stable release 2.4.1 (1 August 2005) |
| 1909 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 1910 | (The notes for this release have been lost. Sorry! It would have |
| 1911 | contained various bug fixes but no new features.) |
| 1912 | |
| 1913 | |
| 1914 | |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1915 | Stable release 2.4.0 (March 2005) -- CHANGES RELATIVE TO 2.2.0 |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1916 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 1917 | 2.4.0 brings many significant changes and bug fixes. The most |
| 1918 | significant user-visible change is that we no longer supply our own |
| 1919 | pthread implementation. Instead, Valgrind is finally capable of |
| 1920 | running the native thread library, either LinuxThreads or NPTL. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1921 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1922 | This means our libpthread has gone, along with the bugs associated |
| 1923 | with it. Valgrind now supports the kernel's threading syscalls, and |
| 1924 | lets you use your standard system libpthread. As a result: |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1925 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1926 | * There are many fewer system dependencies and strange library-related |
| 1927 | bugs. There is a small performance improvement, and a large |
| 1928 | stability improvement. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1929 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1930 | * On the downside, Valgrind can no longer report misuses of the POSIX |
| 1931 | PThreads API. It also means that Helgrind currently does not work. |
| 1932 | We hope to fix these problems in a future release. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1933 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1934 | Note that running the native thread libraries does not mean Valgrind |
| 1935 | is able to provide genuine concurrent execution on SMPs. We still |
| 1936 | impose the restriction that only one thread is running at any given |
| 1937 | time. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1938 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1939 | There are many other significant changes too: |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1940 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1941 | * Memcheck is (once again) the default tool. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1942 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1943 | * The default stack backtrace is now 12 call frames, rather than 4. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1944 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1945 | * Suppressions can have up to 25 call frame matches, rather than 4. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1946 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1947 | * Memcheck and Addrcheck use less memory. Under some circumstances, |
| 1948 | they no longer allocate shadow memory if there are large regions of |
| 1949 | memory with the same A/V states - such as an mmaped file. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1950 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1951 | * The memory-leak detector in Memcheck and Addrcheck has been |
| 1952 | improved. It now reports more types of memory leak, including |
| 1953 | leaked cycles. When reporting leaked memory, it can distinguish |
| 1954 | between directly leaked memory (memory with no references), and |
| 1955 | indirectly leaked memory (memory only referred to by other leaked |
| 1956 | memory). |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1957 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1958 | * Memcheck's confusion over the effect of mprotect() has been fixed: |
| 1959 | previously mprotect could erroneously mark undefined data as |
| 1960 | defined. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1961 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1962 | * Signal handling is much improved and should be very close to what |
| 1963 | you get when running natively. |
| 1964 | |
| 1965 | One result of this is that Valgrind observes changes to sigcontexts |
| 1966 | passed to signal handlers. Such modifications will take effect when |
| 1967 | the signal returns. You will need to run with --single-step=yes to |
| 1968 | make this useful. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1969 | |
| 1970 | * Valgrind is built in Position Independent Executable (PIE) format if |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1971 | your toolchain supports it. This allows it to take advantage of all |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1972 | the available address space on systems with 4Gbyte user address |
| 1973 | spaces. |
| 1974 | |
| 1975 | * Valgrind can now run itself (requires PIE support). |
| 1976 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1977 | * Syscall arguments are now checked for validity. Previously all |
| 1978 | memory used by syscalls was checked, but now the actual values |
| 1979 | passed are also checked. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1980 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1981 | * Syscall wrappers are more robust against bad addresses being passed |
| 1982 | to syscalls: they will fail with EFAULT rather than killing Valgrind |
| 1983 | with SIGSEGV. |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 1984 | |
njn | 1aa5ae4 | 2005-03-26 16:04:27 +0000 | [diff] [blame] | 1985 | * Because clone() is directly supported, some non-pthread uses of it |
| 1986 | will work. Partial sharing (where some resources are shared, and |
| 1987 | some are not) is not supported. |
| 1988 | |
| 1989 | * open() and readlink() on /proc/self/exe are supported. |
| 1990 | |
| 1991 | BUGS FIXED: |
| 1992 | |
| 1993 | 88520 pipe+fork+dup2 kills the main program |
| 1994 | 88604 Valgrind Aborts when using $VALGRIND_OPTS and user progra... |
| 1995 | 88614 valgrind: vg_libpthread.c:2323 (read): Assertion `read_pt... |
| 1996 | 88703 Stabs parser fails to handle ";" |
| 1997 | 88886 ioctl wrappers for TIOCMBIS and TIOCMBIC |
| 1998 | 89032 valgrind pthread_cond_timedwait fails |
| 1999 | 89106 the 'impossible' happened |
| 2000 | 89139 Missing sched_setaffinity & sched_getaffinity |
| 2001 | 89198 valgrind lacks support for SIOCSPGRP and SIOCGPGRP |
| 2002 | 89263 Missing ioctl translations for scsi-generic and CD playing |
| 2003 | 89440 tests/deadlock.c line endings |
| 2004 | 89481 `impossible' happened: EXEC FAILED |
| 2005 | 89663 valgrind 2.2.0 crash on Redhat 7.2 |
| 2006 | 89792 Report pthread_mutex_lock() deadlocks instead of returnin... |
| 2007 | 90111 statvfs64 gives invalid error/warning |
| 2008 | 90128 crash+memory fault with stabs generated by gnat for a run... |
| 2009 | 90778 VALGRIND_CHECK_DEFINED() not as documented in memcheck.h |
| 2010 | 90834 cachegrind crashes at end of program without reporting re... |
| 2011 | 91028 valgrind: vg_memory.c:229 (vgPlain_unmap_range): Assertio... |
| 2012 | 91162 valgrind crash while debugging drivel 1.2.1 |
| 2013 | 91199 Unimplemented function |
| 2014 | 91325 Signal routing does not propagate the siginfo structure |
| 2015 | 91599 Assertion `cv == ((void *)0)' |
| 2016 | 91604 rw_lookup clears orig and sends the NULL value to rw_new |
| 2017 | 91821 Small problems building valgrind with $top_builddir ne $t... |
| 2018 | 91844 signal 11 (SIGSEGV) at get_tcb (libpthread.c:86) in corec... |
| 2019 | 92264 UNIMPLEMENTED FUNCTION: pthread_condattr_setpshared |
| 2020 | 92331 per-target flags necessitate AM_PROG_CC_C_O |
| 2021 | 92420 valgrind doesn't compile with linux 2.6.8.1/9 |
| 2022 | 92513 Valgrind 2.2.0 generates some warning messages |
| 2023 | 92528 vg_symtab2.c:170 (addLoc): Assertion `loc->size > 0' failed. |
| 2024 | 93096 unhandled ioctl 0x4B3A and 0x5601 |
| 2025 | 93117 Tool and core interface versions do not match |
| 2026 | 93128 Can't run valgrind --tool=memcheck because of unimplement... |
| 2027 | 93174 Valgrind can crash if passed bad args to certain syscalls |
| 2028 | 93309 Stack frame in new thread is badly aligned |
| 2029 | 93328 Wrong types used with sys_sigprocmask() |
| 2030 | 93763 /usr/include/asm/msr.h is missing |
| 2031 | 93776 valgrind: vg_memory.c:508 (vgPlain_find_map_space): Asser... |
| 2032 | 93810 fcntl() argument checking a bit too strict |
| 2033 | 94378 Assertion `tst->sigqueue_head != tst->sigqueue_tail' failed. |
| 2034 | 94429 valgrind 2.2.0 segfault with mmap64 in glibc 2.3.3 |
| 2035 | 94645 Impossible happened: PINSRW mem |
| 2036 | 94953 valgrind: the `impossible' happened: SIGSEGV |
| 2037 | 95667 Valgrind does not work with any KDE app |
| 2038 | 96243 Assertion 'res==0' failed |
| 2039 | 96252 stage2 loader of valgrind fails to allocate memory |
| 2040 | 96520 All programs crashing at _dl_start (in /lib/ld-2.3.3.so) ... |
| 2041 | 96660 ioctl CDROMREADTOCENTRY causes bogus warnings |
| 2042 | 96747 After looping in a segfault handler, the impossible happens |
| 2043 | 96923 Zero sized arrays crash valgrind trace back with SIGFPE |
| 2044 | 96948 valgrind stops with assertion failure regarding mmap2 |
| 2045 | 96966 valgrind fails when application opens more than 16 sockets |
| 2046 | 97398 valgrind: vg_libpthread.c:2667 Assertion failed |
| 2047 | 97407 valgrind: vg_mylibc.c:1226 (vgPlain_safe_fd): Assertion `... |
| 2048 | 97427 "Warning: invalid file descriptor -1 in syscall close()" ... |
| 2049 | 97785 missing backtrace |
| 2050 | 97792 build in obj dir fails - autoconf / makefile cleanup |
| 2051 | 97880 pthread_mutex_lock fails from shared library (special ker... |
| 2052 | 97975 program aborts without ang VG messages |
| 2053 | 98129 Failed when open and close file 230000 times using stdio |
| 2054 | 98175 Crashes when using valgrind-2.2.0 with a program using al... |
| 2055 | 98288 Massif broken |
| 2056 | 98303 UNIMPLEMENTED FUNCTION pthread_condattr_setpshared |
| 2057 | 98630 failed--compilation missing warnings.pm, fails to make he... |
| 2058 | 98756 Cannot valgrind signal-heavy kdrive X server |
| 2059 | 98966 valgrinding the JVM fails with a sanity check assertion |
| 2060 | 99035 Valgrind crashes while profiling |
| 2061 | 99142 loops with message "Signal 11 being dropped from thread 0... |
| 2062 | 99195 threaded apps crash on thread start (using QThread::start... |
| 2063 | 99348 Assertion `vgPlain_lseek(core_fd, 0, 1) == phdrs[i].p_off... |
| 2064 | 99568 False negative due to mishandling of mprotect |
| 2065 | 99738 valgrind memcheck crashes on program that uses sigitimer |
| 2066 | 99923 0-sized allocations are reported as leaks |
| 2067 | 99949 program seg faults after exit() |
| 2068 | 100036 "newSuperblock's request for 1048576 bytes failed" |
| 2069 | 100116 valgrind: (pthread_cond_init): Assertion `sizeof(* cond) ... |
| 2070 | 100486 memcheck reports "valgrind: the `impossible' happened: V... |
| 2071 | 100833 second call to "mremap" fails with EINVAL |
| 2072 | 101156 (vgPlain_find_map_space): Assertion `(addr & ((1 << 12)-1... |
| 2073 | 101173 Assertion `recDepth >= 0 && recDepth < 500' failed |
| 2074 | 101291 creating threads in a forked process fails |
| 2075 | 101313 valgrind causes different behavior when resizing a window... |
| 2076 | 101423 segfault for c++ array of floats |
| 2077 | 101562 valgrind massif dies on SIGINT even with signal handler r... |
| 2078 | |
njn | 8c92730 | 2005-03-12 16:45:01 +0000 | [diff] [blame] | 2079 | |
jseward | 34ed74f | 2004-08-30 18:04:42 +0000 | [diff] [blame] | 2080 | Stable release 2.2.0 (31 August 2004) -- CHANGES RELATIVE TO 2.0.0 |
| 2081 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
jseward | 7d180d9 | 2004-08-30 18:15:37 +0000 | [diff] [blame] | 2082 | 2.2.0 brings nine months worth of improvements and bug fixes. We |
| 2083 | believe it to be a worthy successor to 2.0.0. There are literally |
| 2084 | hundreds of bug fixes and minor improvements. There are also some |
| 2085 | fairly major user-visible changes: |
| 2086 | |
| 2087 | * A complete overhaul of handling of system calls and signals, and |
| 2088 | their interaction with threads. In general, the accuracy of the |
| 2089 | system call, thread and signal simulations is much improved: |
| 2090 | |
| 2091 | - Blocking system calls behave exactly as they do when running |
| 2092 | natively (not on valgrind). That is, if a syscall blocks only the |
| 2093 | calling thread when running natively, than it behaves the same on |
| 2094 | valgrind. No more mysterious hangs because V doesn't know that some |
| 2095 | syscall or other, should block only the calling thread. |
| 2096 | |
| 2097 | - Interrupted syscalls should now give more faithful results. |
| 2098 | |
| 2099 | - Signal contexts in signal handlers are supported. |
| 2100 | |
| 2101 | * Improvements to NPTL support to the extent that V now works |
| 2102 | properly on NPTL-only setups. |
| 2103 | |
| 2104 | * Greater isolation between Valgrind and the program being run, so |
| 2105 | the program is less likely to inadvertently kill Valgrind by |
| 2106 | doing wild writes. |
| 2107 | |
| 2108 | * Massif: a new space profiling tool. Try it! It's cool, and it'll |
| 2109 | tell you in detail where and when your C/C++ code is allocating heap. |
| 2110 | Draws pretty .ps pictures of memory use against time. A potentially |
| 2111 | powerful tool for making sense of your program's space use. |
| 2112 | |
| 2113 | * File descriptor leakage checks. When enabled, Valgrind will print out |
| 2114 | a list of open file descriptors on exit. |
| 2115 | |
| 2116 | * Improved SSE2/SSE3 support. |
| 2117 | |
jseward | 2eb002b | 2004-08-31 00:14:02 +0000 | [diff] [blame] | 2118 | * Time-stamped output; use --time-stamp=yes |
| 2119 | |
jseward | 34ed74f | 2004-08-30 18:04:42 +0000 | [diff] [blame] | 2120 | |
| 2121 | |
| 2122 | Stable release 2.2.0 (31 August 2004) -- CHANGES RELATIVE TO 2.1.2 |
| 2123 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 2124 | 2.2.0 is not much different from 2.1.2, released seven weeks ago. |
| 2125 | A number of bugs have been fixed, most notably #85658, which gave |
| 2126 | problems for quite a few people. There have been many internal |
| 2127 | cleanups, but those are not user visible. |
| 2128 | |
| 2129 | The following bugs have been fixed since 2.1.2: |
| 2130 | |
| 2131 | 85658 Assert in coregrind/vg_libpthread.c:2326 (open64) != |
| 2132 | (void*)0 failed |
| 2133 | This bug was reported multiple times, and so the following |
| 2134 | duplicates of it are also fixed: 87620, 85796, 85935, 86065, |
| 2135 | 86919, 86988, 87917, 88156 |
| 2136 | |
| 2137 | 80716 Semaphore mapping bug caused by unmap (sem_destroy) |
| 2138 | (Was fixed prior to 2.1.2) |
| 2139 | |
| 2140 | 86987 semctl and shmctl syscalls family is not handled properly |
| 2141 | |
| 2142 | 86696 valgrind 2.1.2 + RH AS2.1 + librt |
| 2143 | |
| 2144 | 86730 valgrind locks up at end of run with assertion failure |
| 2145 | in __pthread_unwind |
| 2146 | |
| 2147 | 86641 memcheck doesn't work with Mesa OpenGL/ATI on Suse 9.1 |
| 2148 | (also fixes 74298, a duplicate of this) |
| 2149 | |
| 2150 | 85947 MMX/SSE unhandled instruction 'sfence' |
| 2151 | |
| 2152 | 84978 Wrong error "Conditional jump or move depends on |
| 2153 | uninitialised value" resulting from "sbbl %reg, %reg" |
| 2154 | |
| 2155 | 86254 ssort() fails when signed int return type from comparison is |
| 2156 | too small to handle result of unsigned int subtraction |
| 2157 | |
| 2158 | 87089 memalign( 4, xxx) makes valgrind assert |
| 2159 | |
jseward | 2eb002b | 2004-08-31 00:14:02 +0000 | [diff] [blame] | 2160 | 86407 Add support for low-level parallel port driver ioctls. |
jseward | 34ed74f | 2004-08-30 18:04:42 +0000 | [diff] [blame] | 2161 | |
| 2162 | 70587 Add timestamps to Valgrind output? (wishlist) |
| 2163 | |
| 2164 | 84937 vg_libpthread.c:2505 (se_remap): Assertion `res == 0' |
| 2165 | (fixed prior to 2.1.2) |
| 2166 | |
| 2167 | 86317 cannot load libSDL-1.2.so.0 using valgrind |
| 2168 | |
| 2169 | 86989 memcpy from mac_replace_strmem.c complains about |
| 2170 | uninitialized pointers passed when length to copy is zero |
| 2171 | |
| 2172 | 85811 gnu pascal symbol causes segmentation fault; ok in 2.0.0 |
| 2173 | |
| 2174 | 79138 writing to sbrk()'d memory causes segfault |
| 2175 | |
| 2176 | 77369 sched deadlock while signal received during pthread_join |
| 2177 | and the joined thread exited |
| 2178 | |
| 2179 | 88115 In signal handler for SIGFPE, siginfo->si_addr is wrong |
| 2180 | under Valgrind |
| 2181 | |
| 2182 | 78765 Massif crashes on app exit if FP exceptions are enabled |
| 2183 | |
| 2184 | Additionally there are the following changes, which are not |
| 2185 | connected to any bug report numbers, AFAICS: |
| 2186 | |
| 2187 | * Fix scary bug causing mis-identification of SSE stores vs |
| 2188 | loads and so causing memcheck to sometimes give nonsense results |
| 2189 | on SSE code. |
| 2190 | |
| 2191 | * Add support for the POSIX message queue system calls. |
| 2192 | |
| 2193 | * Fix to allow 32-bit Valgrind to run on AMD64 boxes. Note: this does |
| 2194 | NOT allow Valgrind to work with 64-bit executables - only with 32-bit |
| 2195 | executables on an AMD64 box. |
| 2196 | |
| 2197 | * At configure time, only check whether linux/mii.h can be processed |
| 2198 | so that we don't generate ugly warnings by trying to compile it. |
| 2199 | |
jseward | 2eb002b | 2004-08-31 00:14:02 +0000 | [diff] [blame] | 2200 | * Add support for POSIX clocks and timers. |
| 2201 | |
jseward | 34ed74f | 2004-08-30 18:04:42 +0000 | [diff] [blame] | 2202 | |
| 2203 | |
jseward | 9cd978c | 2004-07-17 13:38:12 +0000 | [diff] [blame] | 2204 | Developer (cvs head) release 2.1.2 (18 July 2004) |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2205 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 2206 | 2.1.2 contains four months worth of bug fixes and refinements. |
jseward | 9cd978c | 2004-07-17 13:38:12 +0000 | [diff] [blame] | 2207 | Although officially a developer release, we believe it to be stable |
| 2208 | enough for widespread day-to-day use. 2.1.2 is pretty good, so try it |
| 2209 | first, although there is a chance it won't work. If so then try 2.0.0 |
| 2210 | and tell us what went wrong." 2.1.2 fixes a lot of problems present |
| 2211 | in 2.0.0 and is generally a much better product. |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2212 | |
jseward | 9cd978c | 2004-07-17 13:38:12 +0000 | [diff] [blame] | 2213 | Relative to 2.1.1, a large number of minor problems with 2.1.1 have |
| 2214 | been fixed, and so if you use 2.1.1 you should try 2.1.2. Users of |
| 2215 | 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] | 2216 | |
| 2217 | The following bugs, and probably many more, have been fixed. These |
| 2218 | are listed at http://bugs.kde.org. Reporting a bug for valgrind in |
| 2219 | the http://bugs.kde.org is much more likely to get you a fix than |
| 2220 | mailing developers directly, so please continue to keep sending bugs |
| 2221 | there. |
| 2222 | |
| 2223 | 76869 Crashes when running any tool under Fedora Core 2 test1 |
| 2224 | This fixes the problem with returning from a signal handler |
jseward | 0400d4d | 2004-07-17 23:26:46 +0000 | [diff] [blame] | 2225 | when VDSOs are turned off in FC2. |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2226 | |
jseward | 0400d4d | 2004-07-17 23:26:46 +0000 | [diff] [blame] | 2227 | 69508 java 1.4.2 client fails with erroneous "stack size too small". |
| 2228 | This fix makes more of the pthread stack attribute related |
| 2229 | functions work properly. Java still doesn't work though. |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2230 | |
| 2231 | 71906 malloc alignment should be 8, not 4 |
| 2232 | All memory returned by malloc/new etc is now at least |
| 2233 | 8-byte aligned. |
| 2234 | |
jseward | a6017a4 | 2004-07-17 23:44:35 +0000 | [diff] [blame] | 2235 | 81970 vg_alloc_ThreadState: no free slots available |
| 2236 | (closed because the workaround is simple: increase |
| 2237 | VG_N_THREADS, rebuild and try again.) |
| 2238 | |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2239 | 78514 Conditional jump or move depends on uninitialized value(s) |
| 2240 | (a slight mishanding of FP code in memcheck) |
| 2241 | |
jseward | a6017a4 | 2004-07-17 23:44:35 +0000 | [diff] [blame] | 2242 | 77952 pThread Support (crash) (due to initialisation-ordering probs) |
| 2243 | (also 85118) |
| 2244 | |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2245 | 80942 Addrcheck wasn't doing overlap checking as it should. |
| 2246 | 78048 return NULL on malloc/new etc failure, instead of asserting |
| 2247 | 73655 operator new() override in user .so files often doesn't get picked up |
| 2248 | 83060 Valgrind does not handle native kernel AIO |
| 2249 | 69872 Create proper coredumps after fatal signals |
| 2250 | 82026 failure with new glibc versions: __libc_* functions are not exported |
| 2251 | 70344 UNIMPLEMENTED FUNCTION: tcdrain |
| 2252 | 81297 Cancellation of pthread_cond_wait does not require mutex |
| 2253 | 82872 Using debug info from additional packages (wishlist) |
| 2254 | 83025 Support for ioctls FIGETBSZ and FIBMAP |
| 2255 | 83340 Support for ioctl HDIO_GET_IDENTITY |
| 2256 | 79714 Support for the semtimedop system call. |
| 2257 | 77022 Support for ioctls FBIOGET_VSCREENINFO and FBIOGET_FSCREENINFO |
| 2258 | 82098 hp2ps ansification (wishlist) |
| 2259 | 83573 Valgrind SIGSEGV on execve |
| 2260 | 82999 show which cmdline option was erroneous (wishlist) |
| 2261 | 83040 make valgrind VPATH and distcheck-clean (wishlist) |
jseward | a6017a4 | 2004-07-17 23:44:35 +0000 | [diff] [blame] | 2262 | 83998 Assertion `newfd > vgPlain_max_fd' failed (see below) |
| 2263 | 82722 Unchecked mmap in as_pad leads to mysterious failures later |
| 2264 | 78958 memcheck seg faults while running Mozilla |
jseward | a2c0800 | 2004-07-18 09:24:57 +0000 | [diff] [blame] | 2265 | 85416 Arguments with colon (e.g. --logsocket) ignored |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2266 | |
| 2267 | |
| 2268 | Additionally there are the following changes, which are not |
| 2269 | connected to any bug report numbers, AFAICS: |
| 2270 | |
| 2271 | * Rearranged address space layout relative to 2.1.1, so that |
| 2272 | Valgrind/tools will run out of memory later than currently in many |
| 2273 | circumstances. This is good news esp. for Calltree. It should |
| 2274 | be possible for client programs to allocate over 800MB of |
| 2275 | memory when using memcheck now. |
| 2276 | |
jseward | 9cd978c | 2004-07-17 13:38:12 +0000 | [diff] [blame] | 2277 | * Improved checking when laying out memory. Should hopefully avoid |
| 2278 | the random segmentation faults that 2.1.1 sometimes caused. |
| 2279 | |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2280 | * Support for Fedora Core 2 and SuSE 9.1. Improvements to NPTL |
| 2281 | support to the extent that V now works properly on NPTL-only setups. |
| 2282 | |
| 2283 | * Renamed the following options: |
| 2284 | --logfile-fd --> --log-fd |
| 2285 | --logfile --> --log-file |
| 2286 | --logsocket --> --log-socket |
| 2287 | to be consistent with each other and other options (esp. --input-fd). |
| 2288 | |
| 2289 | * Add support for SIOCGMIIPHY, SIOCGMIIREG and SIOCSMIIREG ioctls and |
| 2290 | improve the checking of other interface related ioctls. |
| 2291 | |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2292 | * Fix building with gcc-3.4.1. |
| 2293 | |
| 2294 | * Remove limit on number of semaphores supported. |
| 2295 | |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2296 | * Add support for syscalls: set_tid_address (258), acct (51). |
| 2297 | |
| 2298 | * Support instruction "repne movs" -- not official but seems to occur. |
| 2299 | |
| 2300 | * Implement an emulated soft limit for file descriptors in addition to |
| 2301 | the current reserved area, which effectively acts as a hard limit. The |
| 2302 | setrlimit system call now simply updates the emulated limits as best |
| 2303 | as possible - the hard limit is not allowed to move at all and just |
| 2304 | returns EPERM if you try and change it. This should stop reductions |
| 2305 | in the soft limit causing assertions when valgrind tries to allocate |
| 2306 | descriptors from the reserved area. |
jseward | a6017a4 | 2004-07-17 23:44:35 +0000 | [diff] [blame] | 2307 | (This actually came from bug #83998). |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2308 | |
nethercote | 76e8fd9 | 2004-07-21 10:37:31 +0000 | [diff] [blame] | 2309 | * Major overhaul of Cachegrind implementation. First user-visible change |
nethercote | 5e63bfb | 2004-07-17 12:49:53 +0000 | [diff] [blame] | 2310 | is that cachegrind.out files are now typically 90% smaller than they |
nethercote | 76e8fd9 | 2004-07-21 10:37:31 +0000 | [diff] [blame] | 2311 | used to be; code annotation times are correspondingly much smaller. |
| 2312 | Second user-visible change is that hit/miss counts for code that is |
| 2313 | unloaded at run-time is no longer dumped into a single "discard" pile, |
| 2314 | but accurately preserved. |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2315 | |
| 2316 | * Client requests for telling valgrind about memory pools. |
| 2317 | |
| 2318 | |
| 2319 | |
jseward | 9cd978c | 2004-07-17 13:38:12 +0000 | [diff] [blame] | 2320 | Developer (cvs head) release 2.1.1 (12 March 2004) |
| 2321 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
jseward | bd0a6e4 | 2004-07-17 12:19:43 +0000 | [diff] [blame] | 2322 | 2.1.1 contains some internal structural changes needed for V's |
jseward | e3e7704 | 2004-03-12 21:07:05 +0000 | [diff] [blame] | 2323 | long-term future. These don't affect end-users. Most notable |
| 2324 | user-visible changes are: |
| 2325 | |
| 2326 | * Greater isolation between Valgrind and the program being run, so |
| 2327 | the program is less likely to inadvertently kill Valgrind by |
| 2328 | doing wild writes. |
| 2329 | |
| 2330 | * Massif: a new space profiling tool. Try it! It's cool, and it'll |
| 2331 | tell you in detail where and when your C/C++ code is allocating heap. |
| 2332 | Draws pretty .ps pictures of memory use against time. A potentially |
| 2333 | powerful tool for making sense of your program's space use. |
| 2334 | |
| 2335 | * Fixes for many bugs, including support for more SSE2/SSE3 instructions, |
| 2336 | various signal/syscall things, and various problems with debug |
| 2337 | info readers. |
| 2338 | |
| 2339 | * Support for glibc-2.3.3 based systems. |
| 2340 | |
| 2341 | We are now doing automatic overnight build-and-test runs on a variety |
| 2342 | of distros. As a result, we believe 2.1.1 builds and runs on: |
| 2343 | Red Hat 7.2, 7.3, 8.0, 9, Fedora Core 1, SuSE 8.2, SuSE 9. |
| 2344 | |
| 2345 | |
| 2346 | The following bugs, and probably many more, have been fixed. These |
| 2347 | are listed at http://bugs.kde.org. Reporting a bug for valgrind in |
| 2348 | the http://bugs.kde.org is much more likely to get you a fix than |
| 2349 | mailing developers directly, so please continue to keep sending bugs |
| 2350 | there. |
| 2351 | |
| 2352 | 69616 glibc 2.3.2 w/NPTL is massively different than what valgrind expects |
| 2353 | 69856 I don't know how to instrument MMXish stuff (Helgrind) |
| 2354 | 73892 valgrind segfaults starting with Objective-C debug info |
| 2355 | (fix for S-type stabs) |
| 2356 | 73145 Valgrind complains too much about close(<reserved fd>) |
| 2357 | 73902 Shadow memory allocation seems to fail on RedHat 8.0 |
| 2358 | 68633 VG_N_SEMAPHORES too low (V itself was leaking semaphores) |
| 2359 | 75099 impossible to trace multiprocess programs |
| 2360 | 76839 the `impossible' happened: disInstr: INT but not 0x80 ! |
| 2361 | 76762 vg_to_ucode.c:3748 (dis_push_segreg): Assertion `sz == 4' failed. |
| 2362 | 76747 cannot include valgrind.h in c++ program |
| 2363 | 76223 parsing B(3,10) gave NULL type => impossible happens |
| 2364 | 75604 shmdt handling problem |
| 2365 | 76416 Problems with gcc 3.4 snap 20040225 |
| 2366 | 75614 using -gstabs when building your programs the `impossible' happened |
| 2367 | 75787 Patch for some CDROM ioctls CDORM_GET_MCN, CDROM_SEND_PACKET, |
| 2368 | 75294 gcc 3.4 snapshot's libstdc++ have unsupported instructions. |
| 2369 | (REP RET) |
| 2370 | 73326 vg_symtab2.c:272 (addScopeRange): Assertion `range->size > 0' failed. |
| 2371 | 72596 not recognizing __libc_malloc |
| 2372 | 69489 Would like to attach ddd to running program |
| 2373 | 72781 Cachegrind crashes with kde programs |
| 2374 | 73055 Illegal operand at DXTCV11CompressBlockSSE2 (more SSE opcodes) |
| 2375 | 73026 Descriptor leak check reports port numbers wrongly |
| 2376 | 71705 README_MISSING_SYSCALL_OR_IOCTL out of date |
| 2377 | 72643 Improve support for SSE/SSE2 instructions |
| 2378 | 72484 valgrind leaves it's own signal mask in place when execing |
| 2379 | 72650 Signal Handling always seems to restart system calls |
| 2380 | 72006 The mmap system call turns all errors in ENOMEM |
| 2381 | 71781 gdb attach is pretty useless |
| 2382 | 71180 unhandled instruction bytes: 0xF 0xAE 0x85 0xE8 |
| 2383 | 69886 writes to zero page cause valgrind to assert on exit |
| 2384 | 71791 crash when valgrinding gimp 1.3 (stabs reader problem) |
| 2385 | 69783 unhandled syscall: 218 |
| 2386 | 69782 unhandled instruction bytes: 0x66 0xF 0x2B 0x80 |
| 2387 | 70385 valgrind fails if the soft file descriptor limit is less |
| 2388 | than about 828 |
| 2389 | 69529 "rep; nop" should do a yield |
| 2390 | 70827 programs with lots of shared libraries report "mmap failed" |
| 2391 | for some of them when reading symbols |
| 2392 | 71028 glibc's strnlen is optimised enough to confuse valgrind |
| 2393 | |
| 2394 | |
| 2395 | |
| 2396 | |
jseward | 4323065 | 2003-12-14 00:14:54 +0000 | [diff] [blame] | 2397 | Unstable (cvs head) release 2.1.0 (15 December 2003) |
| 2398 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 2399 | For whatever it's worth, 2.1.0 actually seems pretty darn stable to me |
| 2400 | (Julian). It looks eminently usable, and given that it fixes some |
| 2401 | significant bugs, may well be worth using on a day-to-day basis. |
| 2402 | 2.1.0 is known to build and pass regression tests on: SuSE 9, SuSE |
| 2403 | 8.2, RedHat 8. |
| 2404 | |
| 2405 | 2.1.0 most notably includes Jeremy Fitzhardinge's complete overhaul of |
| 2406 | handling of system calls and signals, and their interaction with |
| 2407 | threads. In general, the accuracy of the system call, thread and |
| 2408 | signal simulations is much improved. Specifically: |
| 2409 | |
| 2410 | - Blocking system calls behave exactly as they do when running |
| 2411 | natively (not on valgrind). That is, if a syscall blocks only the |
| 2412 | calling thread when running natively, than it behaves the same on |
| 2413 | valgrind. No more mysterious hangs because V doesn't know that some |
| 2414 | syscall or other, should block only the calling thread. |
| 2415 | |
| 2416 | - Interrupted syscalls should now give more faithful results. |
| 2417 | |
| 2418 | - Finally, signal contexts in signal handlers are supported. As a |
| 2419 | result, konqueror on SuSE 9 no longer segfaults when notified of |
| 2420 | file changes in directories it is watching. |
| 2421 | |
| 2422 | Other changes: |
| 2423 | |
| 2424 | - Robert Walsh's file descriptor leakage checks. When enabled, |
| 2425 | Valgrind will print out a list of open file descriptors on |
| 2426 | exit. Along with each file descriptor, Valgrind prints out a stack |
| 2427 | backtrace of where the file was opened and any details relating to the |
| 2428 | file descriptor such as the file name or socket details. |
| 2429 | To use, give: --track-fds=yes |
| 2430 | |
| 2431 | - Implemented a few more SSE/SSE2 instructions. |
| 2432 | |
| 2433 | - Less crud on the stack when you do 'where' inside a GDB attach. |
| 2434 | |
| 2435 | - Fixed the following bugs: |
| 2436 | 68360: Valgrind does not compile against 2.6.0-testX kernels |
| 2437 | 68525: CVS head doesn't compile on C90 compilers |
| 2438 | 68566: pkgconfig support (wishlist) |
| 2439 | 68588: Assertion `sz == 4' failed in vg_to_ucode.c (disInstr) |
| 2440 | 69140: valgrind not able to explicitly specify a path to a binary. |
| 2441 | 69432: helgrind asserts encountering a MutexErr when there are |
| 2442 | EraserErr suppressions |
| 2443 | |
jseward | 0b12daf | 2003-12-14 14:35:28 +0000 | [diff] [blame] | 2444 | - Increase the max size of the translation cache from 200k average bbs |
| 2445 | to 300k average bbs. Programs on the size of OOo (680m17) are |
| 2446 | thrashing the cache at the smaller size, creating large numbers of |
| 2447 | retranslations and wasting significant time as a result. |
| 2448 | |
jseward | 4323065 | 2003-12-14 00:14:54 +0000 | [diff] [blame] | 2449 | |
| 2450 | |
| 2451 | Stable release 2.0.0 (5 Nov 2003) |
| 2452 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 2453 | |
| 2454 | 2.0.0 improves SSE/SSE2 support, fixes some minor bugs, and |
| 2455 | improves support for SuSE 9 and the Red Hat "Severn" beta. |
| 2456 | |
| 2457 | - Further improvements to SSE/SSE2 support. The entire test suite of |
| 2458 | the GNU Scientific Library (gsl-1.4) compiled with Intel Icc 7.1 |
| 2459 | 20030307Z '-g -O -xW' now works. I think this gives pretty good |
| 2460 | coverage of SSE/SSE2 floating point instructions, or at least the |
| 2461 | subset emitted by Icc. |
| 2462 | |
| 2463 | - Also added support for the following instructions: |
| 2464 | MOVNTDQ UCOMISD UNPCKLPS UNPCKHPS SQRTSS |
| 2465 | PUSH/POP %{FS,GS}, and PUSH %CS (Nb: there is no POP %CS). |
| 2466 | |
| 2467 | - CFI support for GDB version 6. Needed to enable newer GDBs |
| 2468 | to figure out where they are when using --gdb-attach=yes. |
| 2469 | |
| 2470 | - Fix this: |
| 2471 | mc_translate.c:1091 (memcheck_instrument): Assertion |
| 2472 | `u_in->size == 4 || u_in->size == 16' failed. |
| 2473 | |
| 2474 | - Return an error rather than panicing when given a bad socketcall. |
| 2475 | |
| 2476 | - Fix checking of syscall rt_sigtimedwait(). |
| 2477 | |
| 2478 | - Implement __NR_clock_gettime (syscall 265). Needed on Red Hat Severn. |
| 2479 | |
| 2480 | - Fixed bug in overlap check in strncpy() -- it was assuming the src was 'n' |
| 2481 | bytes long, when it could be shorter, which could cause false |
| 2482 | positives. |
| 2483 | |
| 2484 | - Support use of select() for very large numbers of file descriptors. |
| 2485 | |
| 2486 | - Don't fail silently if the executable is statically linked, or is |
| 2487 | setuid/setgid. Print an error message instead. |
| 2488 | |
| 2489 | - Support for old DWARF-1 format line number info. |
| 2490 | |
| 2491 | |
| 2492 | |
| 2493 | Snapshot 20031012 (12 October 2003) |
| 2494 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 2495 | |
| 2496 | Three months worth of bug fixes, roughly. Most significant single |
| 2497 | change is improved SSE/SSE2 support, mostly thanks to Dirk Mueller. |
| 2498 | |
| 2499 | 20031012 builds on Red Hat Fedora ("Severn") but doesn't really work |
| 2500 | (curiosly, mozilla runs OK, but a modest "ls -l" bombs). I hope to |
| 2501 | get a working version out soon. It may or may not work ok on the |
| 2502 | forthcoming SuSE 9; I hear positive noises about it but haven't been |
| 2503 | able to verify this myself (not until I get hold of a copy of 9). |
| 2504 | |
| 2505 | A detailed list of changes, in no particular order: |
| 2506 | |
| 2507 | - Describe --gen-suppressions in the FAQ. |
| 2508 | |
| 2509 | - Syscall __NR_waitpid supported. |
| 2510 | |
| 2511 | - Minor MMX bug fix. |
| 2512 | |
| 2513 | - -v prints program's argv[] at startup. |
| 2514 | |
| 2515 | - More glibc-2.3 suppressions. |
| 2516 | |
| 2517 | - Suppressions for stack underrun bug(s) in the c++ support library |
| 2518 | distributed with Intel Icc 7.0. |
| 2519 | |
| 2520 | - Fix problems reading /proc/self/maps. |
| 2521 | |
| 2522 | - Fix a couple of messages that should have been suppressed by -q, |
| 2523 | but weren't. |
| 2524 | |
| 2525 | - Make Addrcheck understand "Overlap" suppressions. |
| 2526 | |
| 2527 | - At startup, check if program is statically linked and bail out if so. |
| 2528 | |
| 2529 | - Cachegrind: Auto-detect Intel Pentium-M, also VIA Nehemiah |
| 2530 | |
| 2531 | - Memcheck/addrcheck: minor speed optimisations |
| 2532 | |
| 2533 | - Handle syscall __NR_brk more correctly than before. |
| 2534 | |
| 2535 | - Fixed incorrect allocate/free mismatch errors when using |
| 2536 | operator new(unsigned, std::nothrow_t const&) |
| 2537 | operator new[](unsigned, std::nothrow_t const&) |
| 2538 | |
| 2539 | - Support POSIX pthread spinlocks. |
| 2540 | |
| 2541 | - Fixups for clean compilation with gcc-3.3.1. |
daywalker | 5d945de | 2003-09-26 00:32:53 +0000 | [diff] [blame] | 2542 | |
| 2543 | - Implemented more opcodes: |
| 2544 | - push %es |
| 2545 | - push %ds |
| 2546 | - pop %es |
| 2547 | - pop %ds |
| 2548 | - movntq |
| 2549 | - sfence |
| 2550 | - pshufw |
| 2551 | - pavgb |
| 2552 | - ucomiss |
| 2553 | - enter |
daywalker | b18d253 | 2003-09-27 20:15:01 +0000 | [diff] [blame] | 2554 | - mov imm32, %esp |
| 2555 | - all "in" and "out" opcodes |
daywalker | 79aad84 | 2003-09-30 22:58:12 +0000 | [diff] [blame] | 2556 | - inc/dec %esp |
jseward | 4323065 | 2003-12-14 00:14:54 +0000 | [diff] [blame] | 2557 | - A whole bunch of SSE/SSE2 instructions |
daywalker | 5d945de | 2003-09-26 00:32:53 +0000 | [diff] [blame] | 2558 | |
jseward | 4323065 | 2003-12-14 00:14:54 +0000 | [diff] [blame] | 2559 | - Memcheck: don't bomb on SSE/SSE2 code. |
daywalker | 5d945de | 2003-09-26 00:32:53 +0000 | [diff] [blame] | 2560 | |
sewardj | 9023879 | 2003-05-05 00:23:42 +0000 | [diff] [blame] | 2561 | |
sewardj | 945f39f | 2003-07-25 21:11:40 +0000 | [diff] [blame] | 2562 | Snapshot 20030725 (25 July 2003) |
| 2563 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 2564 | |
| 2565 | Fixes some minor problems in 20030716. |
| 2566 | |
| 2567 | - Fix bugs in overlap checking for strcpy/memcpy etc. |
| 2568 | |
| 2569 | - Do overlap checking with Addrcheck as well as Memcheck. |
| 2570 | |
| 2571 | - Fix this: |
| 2572 | Memcheck: the `impossible' happened: |
| 2573 | get_error_name: unexpected type |
| 2574 | |
| 2575 | - Install headers needed to compile new skins. |
| 2576 | |
sewardj | e3dd2e0 | 2003-07-27 08:04:48 +0000 | [diff] [blame] | 2577 | - Remove leading spaces and colon in the LD_LIBRARY_PATH / LD_PRELOAD |
sewardj | 945f39f | 2003-07-25 21:11:40 +0000 | [diff] [blame] | 2578 | passed to non-traced children. |
| 2579 | |
sewardj | b9eda95 | 2003-07-26 21:39:05 +0000 | [diff] [blame] | 2580 | - Fix file descriptor leak in valgrind-listener. |
| 2581 | |
sewardj | e3dd2e0 | 2003-07-27 08:04:48 +0000 | [diff] [blame] | 2582 | - Fix longstanding bug in which the allocation point of a |
| 2583 | block resized by realloc was not correctly set. This may |
| 2584 | have caused confusing error messages. |
sewardj | 945f39f | 2003-07-25 21:11:40 +0000 | [diff] [blame] | 2585 | |
| 2586 | |
sewardj | 626fd89 | 2003-07-16 20:10:26 +0000 | [diff] [blame] | 2587 | Snapshot 20030716 (16 July 2003) |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 2588 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 2589 | |
sewardj | 626fd89 | 2003-07-16 20:10:26 +0000 | [diff] [blame] | 2590 | 20030716 is a snapshot of our current CVS head (development) branch. |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 2591 | This is the branch which will become valgrind-2.0. It contains |
| 2592 | significant enhancements over the 1.9.X branch. |
| 2593 | |
| 2594 | Despite this being a snapshot of the CVS head, it is believed to be |
| 2595 | quite stable -- at least as stable as 1.9.6 or 1.0.4, if not more so |
| 2596 | -- and therefore suitable for widespread use. Please let us know asap |
| 2597 | if it causes problems for you. |
| 2598 | |
| 2599 | Two reasons for releasing a snapshot now are: |
| 2600 | |
| 2601 | - It's been a while since 1.9.6, and this snapshot fixes |
| 2602 | various problems that 1.9.6 has with threaded programs |
| 2603 | on glibc-2.3.X based systems. |
| 2604 | |
| 2605 | - So as to make available improvements in the 2.0 line. |
| 2606 | |
sewardj | 626fd89 | 2003-07-16 20:10:26 +0000 | [diff] [blame] | 2607 | Major changes in 20030716, as compared to 1.9.6: |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 2608 | |
| 2609 | - More fixes to threading support on glibc-2.3.1 and 2.3.2-based |
| 2610 | systems (SuSE 8.2, Red Hat 9). If you have had problems |
| 2611 | with inconsistent/illogical behaviour of errno, h_errno or the DNS |
sewardj | 626fd89 | 2003-07-16 20:10:26 +0000 | [diff] [blame] | 2612 | resolver functions in threaded programs, 20030716 should improve |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 2613 | matters. This snapshot seems stable enough to run OpenOffice.org |
| 2614 | 1.1rc on Red Hat 7.3, SuSE 8.2 and Red Hat 9, and that's a big |
| 2615 | threaded app if ever I saw one. |
| 2616 | |
| 2617 | - Automatic generation of suppression records; you no longer |
| 2618 | need to write them by hand. Use --gen-suppressions=yes. |
| 2619 | |
sewardj | 2151180 | 2003-07-22 17:47:42 +0000 | [diff] [blame] | 2620 | - strcpy/memcpy/etc check their arguments for overlaps, when |
| 2621 | running with the Memcheck or Addrcheck skins. |
| 2622 | |
| 2623 | - malloc_usable_size() is now supported. |
| 2624 | |
| 2625 | - new client requests: |
| 2626 | - VALGRIND_COUNT_ERRORS, VALGRIND_COUNT_LEAKS: |
| 2627 | useful with regression testing |
| 2628 | - VALGRIND_NON_SIMD_CALL[0123]: for running arbitrary functions |
| 2629 | on real CPU (use with caution!) |
| 2630 | |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 2631 | - The GDB attach mechanism is more flexible. Allow the GDB to |
| 2632 | be run to be specified by --gdb-path=/path/to/gdb, and specify |
| 2633 | which file descriptor V will read its input from with |
| 2634 | --input-fd=<number>. |
| 2635 | |
sewardj | 2151180 | 2003-07-22 17:47:42 +0000 | [diff] [blame] | 2636 | - Cachegrind gives more accurate results (wasn't tracking instructions in |
| 2637 | malloc() and friends previously, is now). |
| 2638 | |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 2639 | - Complete support for the MMX instruction set. |
| 2640 | |
| 2641 | - Partial support for the SSE and SSE2 instruction sets. Work for this |
| 2642 | is ongoing. About half the SSE/SSE2 instructions are done, so |
| 2643 | some SSE based programs may work. Currently you need to specify |
| 2644 | --skin=addrcheck. Basically not suitable for real use yet. |
| 2645 | |
| 2646 | - Significant speedups (10%-20%) for standard memory checking. |
| 2647 | |
| 2648 | - Fix assertion failure in pthread_once(). |
| 2649 | |
| 2650 | - Fix this: |
| 2651 | valgrind: vg_intercept.c:598 (vgAllRoadsLeadToRome_select): |
| 2652 | Assertion `ms_end >= ms_now' failed. |
| 2653 | |
| 2654 | - Implement pthread_mutexattr_setpshared. |
| 2655 | |
| 2656 | - Understand Pentium 4 branch hints. Also implemented a couple more |
| 2657 | obscure x86 instructions. |
| 2658 | |
| 2659 | - Lots of other minor bug fixes. |
| 2660 | |
sewardj | 626fd89 | 2003-07-16 20:10:26 +0000 | [diff] [blame] | 2661 | - We have a decent regression test system, for the first time. |
| 2662 | This doesn't help you directly, but it does make it a lot easier |
| 2663 | for us to track the quality of the system, especially across |
| 2664 | multiple linux distributions. |
| 2665 | |
| 2666 | You can run the regression tests with 'make regtest' after 'make |
| 2667 | install' completes. On SuSE 8.2 and Red Hat 9 I get this: |
| 2668 | |
| 2669 | == 84 tests, 0 stderr failures, 0 stdout failures == |
| 2670 | |
| 2671 | On Red Hat 8, I get this: |
| 2672 | |
| 2673 | == 84 tests, 2 stderr failures, 1 stdout failure == |
| 2674 | corecheck/tests/res_search (stdout) |
| 2675 | memcheck/tests/sigaltstack (stderr) |
| 2676 | |
| 2677 | sigaltstack is probably harmless. res_search doesn't work |
| 2678 | on R H 8 even running natively, so I'm not too worried. |
| 2679 | |
| 2680 | On Red Hat 7.3, a glibc-2.2.5 system, I get these harmless failures: |
| 2681 | |
| 2682 | == 84 tests, 2 stderr failures, 1 stdout failure == |
| 2683 | corecheck/tests/pth_atfork1 (stdout) |
| 2684 | corecheck/tests/pth_atfork1 (stderr) |
| 2685 | memcheck/tests/sigaltstack (stderr) |
| 2686 | |
| 2687 | You need to run on a PII system, at least, since some tests |
| 2688 | contain P6-specific instructions, and the test machine needs |
| 2689 | access to the internet so that corecheck/tests/res_search |
| 2690 | (a test that the DNS resolver works) can function. |
| 2691 | |
sewardj | 9d916ed | 2003-07-14 23:38:40 +0000 | [diff] [blame] | 2692 | As ever, thanks for the vast amount of feedback :) and bug reports :( |
| 2693 | We may not answer all messages, but we do at least look at all of |
| 2694 | them, and tend to fix the most frequently reported bugs. |
| 2695 | |
| 2696 | |
| 2697 | |
sewardj | 3791882 | 2003-05-05 01:05:09 +0000 | [diff] [blame] | 2698 | Version 1.9.6 (7 May 2003 or thereabouts) |
| 2699 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 2700 | |
| 2701 | Major changes in 1.9.6: |
| 2702 | |
| 2703 | - Improved threading support for glibc >= 2.3.2 (SuSE 8.2, |
| 2704 | RedHat 9, to name but two ...) It turned out that 1.9.5 |
| 2705 | had problems with threading support on glibc >= 2.3.2, |
| 2706 | usually manifested by threaded programs deadlocking in system calls, |
| 2707 | or running unbelievably slowly. Hopefully these are fixed now. 1.9.6 |
| 2708 | is the first valgrind which gives reasonable support for |
| 2709 | glibc-2.3.2. Also fixed a 2.3.2 problem with pthread_atfork(). |
| 2710 | |
| 2711 | - Majorly expanded FAQ.txt. We've added workarounds for all |
| 2712 | common problems for which a workaround is known. |
| 2713 | |
| 2714 | Minor changes in 1.9.6: |
| 2715 | |
| 2716 | - Fix identification of the main thread's stack. Incorrect |
| 2717 | identification of it was causing some on-stack addresses to not get |
| 2718 | identified as such. This only affected the usefulness of some error |
| 2719 | messages; the correctness of the checks made is unchanged. |
| 2720 | |
| 2721 | - Support for kernels >= 2.5.68. |
| 2722 | |
| 2723 | - Dummy implementations of __libc_current_sigrtmin, |
| 2724 | __libc_current_sigrtmax and __libc_allocate_rtsig, hopefully |
| 2725 | good enough to keep alive programs which previously died for lack of |
| 2726 | them. |
| 2727 | |
| 2728 | - Fix bug in the VALGRIND_DISCARD_TRANSLATIONS client request. |
| 2729 | |
sewardj | 3d47b79 | 2003-05-05 22:15:35 +0000 | [diff] [blame] | 2730 | - Fix bug in the DWARF2 debug line info loader, when instructions |
| 2731 | following each other have source lines far from each other |
| 2732 | (e.g. with inlined functions). |
| 2733 | |
sewardj | 3791882 | 2003-05-05 01:05:09 +0000 | [diff] [blame] | 2734 | - Debug info reading: read symbols from both "symtab" and "dynsym" |
| 2735 | sections, rather than merely from the one that comes last in the |
| 2736 | file. |
| 2737 | |
| 2738 | - New syscall support: prctl(), creat(), lookup_dcookie(). |
| 2739 | |
| 2740 | - When checking calls to accept(), recvfrom(), getsocketopt(), |
| 2741 | don't complain if buffer values are NULL. |
| 2742 | |
| 2743 | - Try and avoid assertion failures in |
| 2744 | mash_LD_PRELOAD_and_LD_LIBRARY_PATH. |
| 2745 | |
| 2746 | - Minor bug fixes in cg_annotate. |
| 2747 | |
| 2748 | |
| 2749 | |
sewardj | 9023879 | 2003-05-05 00:23:42 +0000 | [diff] [blame] | 2750 | Version 1.9.5 (7 April 2003) |
| 2751 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 2752 | |
| 2753 | It occurs to me that it would be helpful for valgrind users to record |
| 2754 | in the source distribution the changes in each release. So I now |
| 2755 | attempt to mend my errant ways :-) Changes in this and future releases |
| 2756 | will be documented in the NEWS file in the source distribution. |
| 2757 | |
| 2758 | Major changes in 1.9.5: |
| 2759 | |
| 2760 | - (Critical bug fix): Fix a bug in the FPU simulation. This was |
| 2761 | causing some floating point conditional tests not to work right. |
| 2762 | Several people reported this. If you had floating point code which |
| 2763 | didn't work right on 1.9.1 to 1.9.4, it's worth trying 1.9.5. |
| 2764 | |
| 2765 | - Partial support for Red Hat 9. RH9 uses the new Native Posix |
| 2766 | Threads Library (NPTL), instead of the older LinuxThreads. |
| 2767 | This potentially causes problems with V which will take some |
| 2768 | time to correct. In the meantime we have partially worked around |
| 2769 | this, and so 1.9.5 works on RH9. Threaded programs still work, |
| 2770 | but they may deadlock, because some system calls (accept, read, |
| 2771 | write, etc) which should be nonblocking, in fact do block. This |
| 2772 | is a known bug which we are looking into. |
| 2773 | |
| 2774 | If you can, your best bet (unfortunately) is to avoid using |
| 2775 | 1.9.5 on a Red Hat 9 system, or on any NPTL-based distribution. |
| 2776 | If your glibc is 2.3.1 or earlier, you're almost certainly OK. |
| 2777 | |
| 2778 | Minor changes in 1.9.5: |
| 2779 | |
| 2780 | - Added some #errors to valgrind.h to ensure people don't include |
| 2781 | it accidentally in their sources. This is a change from 1.0.X |
| 2782 | which was never properly documented. The right thing to include |
| 2783 | is now memcheck.h. Some people reported problems and strange |
| 2784 | behaviour when (incorrectly) including valgrind.h in code with |
| 2785 | 1.9.1 -- 1.9.4. This is no longer possible. |
| 2786 | |
| 2787 | - Add some __extension__ bits and pieces so that gcc configured |
| 2788 | for valgrind-checking compiles even with -Werror. If you |
| 2789 | don't understand this, ignore it. Of interest to gcc developers |
| 2790 | only. |
| 2791 | |
| 2792 | - Removed a pointless check which caused problems interworking |
| 2793 | with Clearcase. V would complain about shared objects whose |
| 2794 | names did not end ".so", and refuse to run. This is now fixed. |
| 2795 | In fact it was fixed in 1.9.4 but not documented. |
| 2796 | |
| 2797 | - Fixed a bug causing an assertion failure of "waiters == 1" |
| 2798 | somewhere in vg_scheduler.c, when running large threaded apps, |
| 2799 | notably MySQL. |
| 2800 | |
| 2801 | - Add support for the munlock system call (124). |
| 2802 | |
| 2803 | Some comments about future releases: |
| 2804 | |
| 2805 | 1.9.5 is, we hope, the most stable Valgrind so far. It pretty much |
| 2806 | supersedes the 1.0.X branch. If you are a valgrind packager, please |
| 2807 | consider making 1.9.5 available to your users. You can regard the |
| 2808 | 1.0.X branch as obsolete: 1.9.5 is stable and vastly superior. There |
| 2809 | are no plans at all for further releases of the 1.0.X branch. |
| 2810 | |
| 2811 | If you want a leading-edge valgrind, consider building the cvs head |
| 2812 | (from SourceForge), or getting a snapshot of it. Current cool stuff |
| 2813 | going in includes MMX support (done); SSE/SSE2 support (in progress), |
| 2814 | a significant (10-20%) performance improvement (done), and the usual |
| 2815 | large collection of minor changes. Hopefully we will be able to |
| 2816 | improve our NPTL support, but no promises. |
| 2817 | |