Date: Thu, 27 Nov 2003 11:03:37 -0800 From: Marcel Moolenaar <marcel@xcllnt.net> To: deischen@freebsd.org Cc: threads@freebsd.org Subject: Re: KSE/ia64 & thr_spinlock.c:1.18: problem identified Message-ID: <20031127190337.GB14849@dhcp01.pn.xcllnt.net> In-Reply-To: <Pine.GSO.4.10.10311271004350.16129-100000@pcnet5.pcnet.com> References: <20031127055545.GB12712@dhcp01.pn.xcllnt.net> <Pine.GSO.4.10.10311271004350.16129-100000@pcnet5.pcnet.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Nov 27, 2003 at 10:07:25AM -0500, Daniel Eischen wrote: > On Wed, 26 Nov 2003, Marcel Moolenaar wrote: > > > Ok, > > > > I found what the problem is. Syscalls that got interrupted were not > > restarted when they should be. This is the result of using the KSE > > internal context functions, when we should be going through the > > kernel. So, the end result is that we do in fact need a syscall to > > switch KSE contexts. Attached a patch to add such syscall. Please > > review (ia64 specific changes to make use of the syscall are not > > included). > > The patch looks OK. Is it possible to roll _ia64_break_setcontext() > into kse_switchin() and have ia64's set_mcontext() know the difference > between the two contexts? I'm working on it. I had this done right from the start after creating the syscall, but it reintroduced old failures related to asynchronous contexts. So, my working version now has 3 ways to restore a context (_ia64_break_setcontext(), _ia64_restore_context() and kse_switchin()) and it demonstrates that the problem was indeed that we were using _ia64_restore_context() in cases where this would be invalid. So yes, the next step is to remove the ia64_break_setcontext() hack and use kse_switchin() for asynchronous contexts as well. -- Marcel Moolenaar USPA: A-39004 marcel@xcllnt.net
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20031127190337.GB14849>