tag 560823 fixed-upstream thanks Productive debug session at Axel's helped to discover the problems. I've applied the patches upstream. When you package please make sure that all the .pf2 on big-endian systems are overwrtitten by new version. .pf2 should be identical on big-endian and little-endian systems, if not it's a bug. On 01/12/2011 02:14 AM, Axel Beckert wrote: > Hi, > > Adam D. Barratt wrote: > >> On Mon, 2010-07-19 at 23:42 +0200, Axel Beckert wrote: >> >>> Colin Watson wrote: >>> >>>>> Since #550160 has been fixed, the grub menu appears, I can edit the >>>>> entries on the console, etc. >>>>> >>>>> But as soon as I want to boot one entry, I get the error message >>>>> "Memory Address not Alligned" and I'm back to the boot prompt "ok" of >>>>> OpenPROM/OpenBoot. >>>>> >>>> Vladimir Serbinenko (upstream) told me that he believes he's fixed this: >>>> >> [...] >> >>>> If you still have the >>>> affected system in operation, could you please try upgrading to the >>>> current version in unstable >>>> >> Is this issue still exhibited with the current version of grub2 in >> unstable? >> > First try with version from unstable (1.98+20100804-13), no manual > interventions: > > grub-mkdevicemap generated strange device.map (attached). > update-grub used the binary carp in there and put it in grub.cfg > (attached). grub-install on the only non-floppy device in the > device.map worked without problems. > > Booting resulted in 8x the following output: > > ---snip--- > Boot device: disk File and args: > GRUB Loading kernel... > Welcome to GRUB! > > Can't read disk label. > Can't open disk label package. > Can't read disk label. > Can't open disk label package. > Can't read disk label. > Can't open disk label package. > Can't read disk label. > Can't open disk label package. > Can't read disk label. > Can't open disk label package. > Can't read disk label. > Can't open disk label package. > Can't read disk label. > Can't open disk label package. > Can't read disk label. > Can't open disk label package. > Fast Data Access MMU Miss > ok > ---snap--- > > So basically something different happens, but we're far away than we > already were, i.e. no more grub menu or commandline available. > > Warm or cold boot doesn't seem to make a difference. > > Next try with grub from experimental (1.99~20110106-1): But basically > the same happens, just with three small differences: > > a) no "Fast Data Access MMU Miss" but "error: out of memory. error: no > suitable mode found. Illegal instruction" instead. > > b) grub-installed issued a warning about embedding not possible in > this setup and that I need to use blocklist instead. It though says > also installation succeeded without error: > > 6/0/0 r...@hz:pts/0 02:01:32 [~] # grub-install '(hd0)' > /usr/sbin/grub-setup: warn: Embedding is not possible. GRUB can only be > installed in this setup by using blocklists. However, blocklists are > UNRELIABLE and their use is discouraged.. > Installation finished. No error reported. > 7/0/0 r...@hz:pts/0 02:02:41 [~] # grub-install > /dev/disk/by-id/ata-ST39120A_7BV080CK > /usr/sbin/grub-setup: warn: Embedding is not possible. GRUB can only be > installed in this setup by using blocklists. However, blocklists are > UNRELIABLE and their use is discouraged.. > Installation finished. No error reported. > 8/0/0 r...@hz:pts/0 02:03:43 [~] # > > But since grub behaves differently afterwards, I think it did write > it on disk. > > c) different binary garbage in the device.map (attached) > > I'll now try to see if fixing the device map manually (no binary > garbage but hd0 in there instead) helps. > > I really wonder where this binary garbage (see attachement) comes > from. > > Anyway, next try was to manually edit the device.map to look like I'd > expect it to look: > > (fd0) /dev/fd0 > (hd0) /dev/disk/by-id/ata-ST39120A_7BV080CK > > grub-install still throws the same error messages (in case of the > version from experimental) or not (in case of unstable). > > No big change though, just again "Fast Data Access MMU Miss" instead > of "Illegal instruction". Maybe those are interchangable in some way. > > Any advice what I could try else? I'll at least stay in contact with > Vladimir for further debugging. > > Regards, Axel >
-- Regards Vladimir 'φ-coder/phcoder' Serbinenko
signature.asc
Description: OpenPGP digital signature
