Hi Marc, On Mon, Jun 15, 2009 at 04:14:32PM -0600, Marc Jones wrote: > Very good catch. I am suprised we didn't see this sooner. > > The bug is that the code was always updating the passed value to the > next link offset even when it was on the requested link (cap_count). I > think that your patch has a slight problem in that it skips the CapID > and CapType check on the first link, link_no == 0. > > Attached is a slightly different fix (untested). I think that this > function could be rewritten to be more clear but this is what you get > when tying to keep code similarity when going from asm to C..... > > Please review and test.
This does not appear to fix the hang for my board. Here's a boot log with lots of register dumping enabled: http://ward.vandewege.net/coreboot/h8dmr/fam10/h8dmr-ae.cap And here's one without the very lengthy dumps: http://ward.vandewege.net/coreboot/h8dmr/fam10/h8dmr-af.cap I see very little difference with the older dumps. Thanks, Ward. -- Ward Vandewege <[email protected]> -- coreboot mailing list: [email protected] http://www.coreboot.org/mailman/listinfo/coreboot
