[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Greetings from Sleepycat

From: Greg Hudson <ghudson_at_MIT.EDU>
Date: 2004-09-21 16:11:44 CEST

On Tue, 2004-09-21 at 08:41, Gregory Burd wrote:
> The locks are implemented in shared memory, not as signals.

That depends on whether you build with --enable-posixmutexes, doesn't
it? (In neither case are the locks implemented as signals, but I think
they might be implemented with fcntl/flock locks if you're not using
mutexes, right?)

If BDB is using mutexes, then we need to know how the underlying OS
implements inter-process mutexes in order to know whether they are
multi-uid safe. On Linux, with NPTL, they are implemented using
futex(), which is a completely separate sleep/wakeup system than the
signalling and which uses memory addresses as keys. futex doesn't care
about uids; the security model is that if you can address the same
memory as another process, you can futex with it.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Sep 21 16:12:14 2004

This is an archived mail posted to the Subversion Dev mailing list.