<br><br><div class="gmail_quote">On Tue, Jun 23, 2009 at 07:33, Andrew Haley <span dir="ltr"><<a href="mailto:aph@redhat.com">aph@redhat.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Isn't there some point at which you have to say to a Linux user "Your<br>
system is simply misconfigured. Fix the overcommit parameter and this<br>
problem will go away" ?<br>
<font color="#888888"></font></blockquote><div><br>99% of users will use the default value of the overcommit parameter <br>and we have to try to make that work well.<br><br>I have thought about what we could do to fix Linux generally.<br>
Perhaps we could have a variant of fork() that promised the kernel<br>that we are about to exec. Then the COW'ed pages after fork wouldn't <br>count towards overcommit. If memory was *very* tight, one could<br>suspend all the threads in the parent process until the child exec'ed,<br>
to minimize the number of pages that were written to in the parent.<br><br>Martin<br> </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<font color="#888888"><br>
Andrew.<br>
</font></blockquote></div><br>