Date: Tue, 26 Sep 2000 23:26:28 +0100 (BST) From: Doug Rabson <dfr@nlsystems.com> To: Andrew Gallatin <gallatin@cs.duke.edu> Cc: jhb@osd.bsdi.com, alpha@freebsd.org Subject: RE: ithreads on UP1000 Message-ID: <Pine.BSF.4.21.0009262324350.35016-100000@salmon.nlsystems.com> In-Reply-To: <14800.47994.288410.1680@grasshopper.cs.duke.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 26 Sep 2000, Andrew Gallatin wrote: > > John Baldwin writes: > > > > On 25-Sep-00 Andrew Gallatin wrote: > > > > > > (apol;ogies if you get 2 copies of this, mail @ Freebsd is pretty flaky > > > for me right now) > > > > > > John, > > > > > > I just tried your latest patchset on an API UP1000. > > > (MD5 (alpha.ithreads.patch) = 11d809b59741189c7261be15166e2519) > > > > Try it without static hints. Some people have been having problems with > > static hints. Also, if dynamic hints works ok for you, please do an > > nm of your kernel and see where the hints_loaded variable shows up in the > > list. Esp. in relation to sched_lock and Giant in the BSS. It seems that > > we may have a bug in the mutex code that trashes the memory just before > > a mutex. :-/ It's been reported on the x86, and the mutex code is almost > > identical in the two platforms. > > It seems we have a similar, but slightly different problem. > > If I use your ithreads patch and a stock GENERIC config file, I boot > past the copyright and the machine hangs mounting root (the scsi bus > probes come back showing da0, though). This is progress ;) > > Howver, If I use any GENERIC kernel config plus > > +options DIAGNOSTIC > +options INVARIANTS > +options INVARIANT_SUPPORT > > The box locks after the copyright is printed. Eg, it locks with or > without your ithreads patch. > > I can duplicate this on an AS500 5/266 which actually has a halt > button. However, it doesn't seem to help as the box is locked so > solidly that it ignores the halt button. > > Am I the only one seeing this? I'm using a fairly old loader (Tue Aug > 29 11:44:06 EDT 2000) on the UP1000 and a truly ancient netboot loader > on the AS500 (Sat Oct 16 11:02:54 EDT 1999). For the netboot, I've > been seting the clock and sio0 hints manually.. > > Since we show every sign of falling off the edge of the universe after > returning from the first sysinit (announce), I thought we might be > running out of space on the bootstrap stack and I tried doubling the > size of the bootstrap stack in alpha_init(), but it didn't seem to > help. We are running securely on proc0's stack before the first sysinit happens (i.e. alpha_init() has returned and we have switched to proc0's context). > > Are others seeing this, or am I going crazy? I haven't seen this but I haven't tried exactly this configuration yet. I'll try to reproduce it. -- Doug Rabson Mail: dfr@nlsystems.com Phone: +44 20 8348 6160 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0009262324350.35016-100000>