Date: Sat, 25 Apr 1998 19:13:00 -0500 From: Chris Csanady <ccsanady@friley585.res.iastate.edu> To: John Birrell <jb@cimlogic.com.au> Cc: dyson@FreeBSD.ORG, freebsd-hackers@FreeBSD.ORG Subject: Re: threads performance Message-ID: <199804260013.TAA09376@friley585.res.iastate.edu> In-Reply-To: Your message of "Sun, 26 Apr 1998 09:54:03 %2B1000." <199804252354.JAA09319@cimlogic.com.au>
next in thread | previous in thread | raw e-mail | index | archive | help
>John S. Dyson wrote: >> > Why this hurts so much be comparison to other platforms (which >> > supposedly also do this) is another question entirely. >> > >> We need to use a deferred mechanism, a lot like our interrupt >> code. > >The issue of blocking syscalls makes this "not worth doing". It would >only be possible for -current, anyway. > >I'd prefer that we concentrate on the kernel thread interface so that the >blocking syscall issue goes away. And with it goes the need to block >signals. I'm not sure how related this is, but has any thought been given to using an async call gate? It seems like it would be the Right Thing to do (eg. for the aio calls, etc.) Without it, wouldn't apps that do a lot of async calls get badly hurt by repeatedly yeilding to the scheduler? Perhaps I am missing something, this is not an area I am intimately familiar with.. Chris To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199804260013.TAA09376>