On 27.12.2010 19:16, Blair Zajac wrote:
On 12/27/10 7:45 AM, Daniel Shahaf wrote:
Ping...
Blair Zajac wrote on Tue, Nov 09, 2010 at 20:05:13 -0800:
On Nov 9, 2010, at 1:03 PM, hwri...@apache.org wrote:
Author: hwright
Date: Tue Nov 9 21:03:16 2010
New Revision: 1033228
URL: http://svn.apache.org/viewvc?rev=1033228&view=rev
Log:
Merge r1022675 from trunk:
* r1022675 (r982355 from ^subversion/branches/performance)
Limit the amount of unused memory fragments held by the root pools.
Justification:
In combination with blame -g server side memory leaks, see
r1032808 and
http://svn.haxx.se/dev/archive-2010-11/0102.shtml, svnserve's
memory
usage can get bloated and stay that way, see
see http://svn.haxx.se/dev/archive-2010-11/0165.shtml.
Notes:
Use --accept=mine-full to avoid bizzare branch root property
conflicts
on 'svn:ignore' and 'bugtraq:logregex'. This happens for me
with both
1.6.14 and tr...@1032984.
Votes:
+1: pburba, cmpilato, hwright
--- subversion/branches/1.6.x/subversion/svnserve/main.c (original)
+++ subversion/branches/1.6.x/subversion/svnserve/main.c Tue Nov 9
21:03:16 2010
@@ -356,6 +356,7 @@ int main(int argc, const char *argv[])
apr_sockaddr_t *sa;
apr_pool_t *pool;
apr_pool_t *connection_pool;
+ apr_allocator_t *allocator;
svn_error_t *err;
apr_getopt_t *os;
int opt;
@@ -747,10 +748,22 @@ int main(int argc, const char *argv[])
return ERROR_SUCCESS;
#endif
+ /* If we are using fulltext caches etc., we will allocate
many large
+ chunks of memory of various sizes outside the cachde for
those
+ fulltexts. Make sure, we use the memory wisely: use an
allocator
+ that causes memory fragments to be given back to the OS
early. */
+
+ if (apr_allocator_create(&allocator))
+ return EXIT_FAILURE;
+
+ apr_allocator_max_free_set(allocator,
SVN_ALLOCATOR_RECOMMENDED_MAX_FREE);
+
/* Non-standard pool handling. The main thread never blocks
to join
the connection threads so it cannot clean up after each
one. So
separate pools, that can be cleared at thread exit, are
used */
- connection_pool = svn_pool_create(NULL);
+
+ connection_pool = svn_pool_create_ex(NULL, allocator);
+ apr_allocator_owner_set(allocator, connection_pool);
Shouldn't the allocator have a mutex registered with it, given that
svnserve is multithreaded?
We had a chat on IRC and it looks like the code is fine. I'd have to
dig through IRC logs to fine the conversation.
Blair
Every thread creates its own allocator and root pool object
(as indicated by the comment). Both are then only accessed
by the same thread that created them.
-- Stefan^2.
