On 01/01/2011 06:57 PM, Colin Watson wrote: > On Sat, Jan 01, 2011 at 06:35:52PM -0500, P. J. McDermott wrote: > >> Considering the machines this is affecting (machines with the Poulsbo >> chipset) and the configuration workaround, I'd bet the issue lies in the >> video subsystem. So far, the changes in the video/ directory appear to >> be mostly code re-factoring, so I think my next step will be to add some >> debugging output to 20100702-1's kernel and see where GRUB hangs up or >> what isn't being done. Don't expect any miracles though, as this is my >> first time in the GRUB 2 code. ;) If anyone has any thoughts on what the >> problem might be or familiarity with the GRUB kernel's initialization of >> the video subsystem, that would be helpful. >> > I think (with respect) that Vladimir was mistaken when he said that the > video changes in 20100702-1 only affected the EFI port. > > One effect of these changes was to load the video_cirrus and video_bochs > modules by default (you can test whether this is the culprit by > commenting them out in grub.cfg). I've seen a handful of systems that > hang while trying to enumerate the PCI bus in GRUB; it so happens that > those are the only modules that usually trigger GRUB's PCI bus > enumeration in normal circumstances ... > > You can also verify this at a lower level by trying 'lspci' at a GRUB > prompt. If it's the same problem, this will hang. >
I noticed the new bochs and cirrus files, but I didn't think they would affect a Poulsbo system. You're right though; I commented out those lines, and I'm now looking at a graphical menu for "GNU GRUB version 1.98+20100804-11" (as installed by Squeeze beta1's debian-installer) on an AO751h. I suppose the problem then is in either grub_pci_iterate() or the hook functions passed to it by the cirrus and bochs modules? > Unfortunately, I haven't had my hands on an affected machine for more > than a day or so, and that was when I was under deadline pressure so a > workaround was the best I could manage. Steve, is your affected system > at home? Maybe I could visit at some point (for others on the bug, > we're in the same city) ... > I can do testing and debugging on my system as well, if it helps. I might poke around grub_pci_iterate() and the hook functions tonight or tomorrow sometime. P. J. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org