sewardj | 9023879 | 2003-05-05 00:23:42 +0000 | [diff] [blame^] | 1 | |
| 2 | Version 1.9.5 (7 April 2003) |
| 3 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 4 | |
| 5 | It occurs to me that it would be helpful for valgrind users to record |
| 6 | in the source distribution the changes in each release. So I now |
| 7 | attempt to mend my errant ways :-) Changes in this and future releases |
| 8 | will be documented in the NEWS file in the source distribution. |
| 9 | |
| 10 | Major changes in 1.9.5: |
| 11 | |
| 12 | - (Critical bug fix): Fix a bug in the FPU simulation. This was |
| 13 | causing some floating point conditional tests not to work right. |
| 14 | Several people reported this. If you had floating point code which |
| 15 | didn't work right on 1.9.1 to 1.9.4, it's worth trying 1.9.5. |
| 16 | |
| 17 | - Partial support for Red Hat 9. RH9 uses the new Native Posix |
| 18 | Threads Library (NPTL), instead of the older LinuxThreads. |
| 19 | This potentially causes problems with V which will take some |
| 20 | time to correct. In the meantime we have partially worked around |
| 21 | this, and so 1.9.5 works on RH9. Threaded programs still work, |
| 22 | but they may deadlock, because some system calls (accept, read, |
| 23 | write, etc) which should be nonblocking, in fact do block. This |
| 24 | is a known bug which we are looking into. |
| 25 | |
| 26 | If you can, your best bet (unfortunately) is to avoid using |
| 27 | 1.9.5 on a Red Hat 9 system, or on any NPTL-based distribution. |
| 28 | If your glibc is 2.3.1 or earlier, you're almost certainly OK. |
| 29 | |
| 30 | Minor changes in 1.9.5: |
| 31 | |
| 32 | - Added some #errors to valgrind.h to ensure people don't include |
| 33 | it accidentally in their sources. This is a change from 1.0.X |
| 34 | which was never properly documented. The right thing to include |
| 35 | is now memcheck.h. Some people reported problems and strange |
| 36 | behaviour when (incorrectly) including valgrind.h in code with |
| 37 | 1.9.1 -- 1.9.4. This is no longer possible. |
| 38 | |
| 39 | - Add some __extension__ bits and pieces so that gcc configured |
| 40 | for valgrind-checking compiles even with -Werror. If you |
| 41 | don't understand this, ignore it. Of interest to gcc developers |
| 42 | only. |
| 43 | |
| 44 | - Removed a pointless check which caused problems interworking |
| 45 | with Clearcase. V would complain about shared objects whose |
| 46 | names did not end ".so", and refuse to run. This is now fixed. |
| 47 | In fact it was fixed in 1.9.4 but not documented. |
| 48 | |
| 49 | - Fixed a bug causing an assertion failure of "waiters == 1" |
| 50 | somewhere in vg_scheduler.c, when running large threaded apps, |
| 51 | notably MySQL. |
| 52 | |
| 53 | - Add support for the munlock system call (124). |
| 54 | |
| 55 | Some comments about future releases: |
| 56 | |
| 57 | 1.9.5 is, we hope, the most stable Valgrind so far. It pretty much |
| 58 | supersedes the 1.0.X branch. If you are a valgrind packager, please |
| 59 | consider making 1.9.5 available to your users. You can regard the |
| 60 | 1.0.X branch as obsolete: 1.9.5 is stable and vastly superior. There |
| 61 | are no plans at all for further releases of the 1.0.X branch. |
| 62 | |
| 63 | If you want a leading-edge valgrind, consider building the cvs head |
| 64 | (from SourceForge), or getting a snapshot of it. Current cool stuff |
| 65 | going in includes MMX support (done); SSE/SSE2 support (in progress), |
| 66 | a significant (10-20%) performance improvement (done), and the usual |
| 67 | large collection of minor changes. Hopefully we will be able to |
| 68 | improve our NPTL support, but no promises. |
| 69 | |