On the distinguished day of Feb 21, Steven M. Bellovin wrote:

> In message <20040221170100.GE661@goldberry.poofy.goof.com>, "Aaron J. Grier" wr
> ites:
> >On Fri, Feb 20, 2004 at 01:31:54PM -0800, Wolfgang S. Rupprecht wrote:
> >> FYI.  I'm using stock current.  Looking at res_send.c:send_dg() I
> >> don't immediately see anything that looks like a LOCK()/UNLOCK()
> >> around where it opens a new socket and where it uses it.  Hopefully
> >> someone that understands this will point out that there is implied
> >> locking because threads can't switch on certain kernel-calls or
> >> something like that.
> >
> >as another data point, I see similar behaviour in mozilla 1.6 from
> >pkgsrc under 1.6.1.  however, MozillaFirebird compiled from the same
> >day's CVS of pkgsrc works fine.
> >
>
> The problem does indeed seem to be related to locking in Mozilla 1.6.
> Kouichirou Hiratsuka's solution (see
> http://mail-index.netbsd.org/tech-pkg/2004/02/20/0001.html ) seems to
> have solved the problem for me.

This patch definitely seems to have helped me out; I haven't been able to
run mozilla for months, and with this patch, it's been running for the
longest since the problem first started happening.

-bp
--