Date: Wed, 10 Apr 2002 20:19:56 +0100 (BST) From: Doug Rabson <dfr@nlsystems.com> To: Peter Wemm <peter@wemm.org> Cc: Jake Burkholder <jake@locore.ca>, John Baldwin <jhb@freebsd.org>, Perforce Change Reviews <perforce@freebsd.org> Subject: Re: PERFORCE change 9504 for review Message-ID: <20020410201934.D16912-100000@salmon.nlsystems.com> In-Reply-To: <20020410191220.92F753810@overcee.wemm.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 10 Apr 2002, Peter Wemm wrote: > Doug Rabson wrote: > > On Wed, 10 Apr 2002, Jake Burkholder wrote: > > > > > Apparently, On Wed, Apr 10, 2002 at 09:06:18AM -0400, > > > John Baldwin said words to the effect of; > > > > > > > > > > > On 10-Apr-2002 Peter Wemm wrote: > > > > > http://people.freebsd.org/~peter/p4db/chv.cgi?CH=9504 > > > > > > > > > > Change 9504 by peter@peter_thunder on 2002/04/10 04:51:36 > > > > > > > > > > Use dfr's fix instead of my hack. I expect he'll commit this to > > > > > freefall soon. :-) > > > > > > > > Yep, much better. :) > > > > > > Well, now the flag checking code is duplicated in both the trap and syscall > > > return paths, alpha is the same. One wonders if the FRAME_SYSCALL optimiza > tion > > > is actually worth all this complication. > > > > I think its still worth it - it still does a lot less work in the common > > case. Bear in mind that calling ast() is quite rare and being forced to do > > a full exception restore (e.g. for a signal) is even rarer. A harmless > > extra call to ast() in that case is unlikely to be noticable. Having said > > that, for ia64 at least, it should be possible for the syscall to bypass > > the ast() bits at the beginning of exception_restore(). As long as syscall > > sets (p1,p2) to (1,0), it can jump to label 2 in exception_restore. > > John asked an awkward question a while ago.. What happens if the ast() > call in the back of the fast syscall return path changes the frame type? > eg: it posted a signal and turned off the FRAME_SYSCALL bit. We'll continue > with the fast return, right? Is this good or bad? Nonono. We check the frame type *after* the call to ast()... -- Doug Rabson Mail: dfr@nlsystems.com Phone: +44 20 8348 6160 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe p4-projects" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020410201934.D16912-100000>