|
From: Philippe W. <phi...@sk...> - 2011-05-01 12:16:39
|
> Greetings.
>
> It's time to think a bit about a 3.7.0 release. The following are
> proposals, for contemplation/discussion.
For 3.7.0, I work or can work or provide help on:
* (of course :) 214909 gdbserver integration (and the delta leak to follow)
I have just uploaded a new version (upgraded to last SVN + retested on many platforms)
* 250101 (quadratic memory usage for some realloc patterns of big blocks)
Some days ago, uploaded a new patch version updated to a recent SVN
Once this is integrated, if I have time, I might work on
250065 Handling large allocations
(I find this one important, as doing big alloc/free decreases significantly the
probability to find dangling references).
* 267925 laog data structure quadratic for a single sequence of lock
I have a working patch doing garbage collection of the laog data structure.
Good results obtained (i.e. helgrind now works on a big appl. at my work,
doing many locks. And it found some wrong lock ordering :).
However, I believe a simpler data structure might be possible, so will investigate
further before posting the patch.
I can't really help but a fix or integration of a patch would be nice for the following:
219156 Valgrind does not handle custom allocation functions
A patch is attached to this, to support tcmalloc. At work, we will very probably switch
to tcmalloc, so ...
271438 sse4-64.c does not compile on at least 2 distrib
This is quite annoying when trying to run the tests.
|