On Mon, Feb 23, 2004 at 10:47:21PM +0100, Poul-Henning Kamp wrote: > In message <16442.29712.148987.927650_at_grasshopper.cs.duke.edu>, Andrew Gallatin > writes: > > >I'm guessing it followed a null function pointer. Maybe the > >recent device changes had some special affect on alpha. > > The fact that you can compile the kernel in the first place rules out > the most common makedev() bogosities, but it is certainly not impossible > that the console code is being creative with dev_t's. I havn't noticed > it on my way through the tree though. > Drew reminded me CVS takes more intricate date specifications (thanks Drew :-). If I do this: cvs -q -R -d /home/ncvs get -D "2004/02/21 13:55 PST" src/sys I get a working kernel. If I then follow up with this: cvs -q -R -d /home/ncvs update -AdPD "2004/02/21 13:58 PST" > update.log I get a broken kernel. I believe that means it's the last of the device megapatches that causes the problem (I'm not blaming the device megapatches, as phk says the issue might be with the Alpha code doing things it wasn't supposed to do...). I'll let you guys figure out which of those it is. :-) Given absolutely nothing comes out the serial console other than the complaint about the kernel stack my vote goes for the console code being where it dies but ... This should land you at the archived commit message: http://docs.freebsd.org/cgi/getmsg.cgi?fetch=3328755+0+archive/2004/cvs-all/20040222.cvs-all -- Ken Smith - From there to here, from here to | kensmith_at_cse.buffalo.edu there, funny things are everywhere. | - Theodore Geisel |Received on Mon Feb 23 2004 - 19:59:32 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:44 UTC