On 3/20/06, <b class="gmail_sendername">Xavier Leroy</b> &lt;<a href="mailto:Xavier.Leroy@inria.fr">Xavier.Leroy@inria.fr</a>&gt; wrote:<div><span class="gmail_quote"></span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Short explanation: atomic instructions are dog slow.</blockquote><div><br>
Thanks for the explanation.&nbsp; I'd never have guessed that atomic
instructions could be responsible for such a deterioration of
performance.<br>
</div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">So, it's time to go back to the drawing board.&nbsp;&nbsp;Fortunately, it<br>appears that reliable polling of signals is possible without atomic
<br>processor instructions.&nbsp;&nbsp;Expect a fix in 3.09.2 at the latest, and<br>probably within a couple of weeks in the CVS.<br></blockquote></div><br>
Great!&nbsp; Btw., since we are at it, you could also make us really
happy by fixing issue 3906 in the next release, too.&nbsp; Now that the
reason is clear this should be very straightforward, and would save
people who write certain kinds of threaded code a lot of headaches,
because this bug can cause all sorts of weird problems in long-running
applications (freezes, crashes, execution of random code, etc.), and
was extremely hard to trigger and track down.<br><br>
Best regards,<br>
Markus<br>
<br>-- <br>Markus
Mottl&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<a href="http://www.ocaml.info">http://www.ocaml.info</a>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<a href="mailto:markus.mottl@gmail.com">markus.mottl@gmail.com</a>