On 3/30/09, Andrew Thompson <thompsa_at_freebsd.org> wrote: > On Mon, Mar 30, 2009 at 04:04:17PM +0100, Paul B. Mahol wrote: >> On 3/30/09, Andrew Thompson <thompsa_at_freebsd.org> wrote: >> > On Tue, Mar 24, 2009 at 03:49:32AM -0500, Robert Noland wrote: >> >> On Mon, 2009-03-23 at 19:40 -0300, Nenhum_de_Nos wrote: >> >> > >> >> > I had problem a while ago with via mini itx hardware, that was quite >> >> > close. If I try boot from usb (installed in usb hdd), I get to the >> >> > point >> >> > of loader not finding my disk. >> >> > >> >> > I then used a small flash disk attached to the ata (44 pin ide) >> >> > channel >> >> > and formatted /boot in there. this way I get to the point of mount >> >> > root >> >> > you said, and da0 not being alive soon enough to mount root. list >> >> > disks >> >> > also couldn't find da0 though. >> >> > >> >> > I tried current from that time, and no good. >> >> > >> >> > if this is solved, I'll be happy to try whatever patch to current. >> >> > (as >> >> > long as I can install it from another box/or its ata channel, as it >> >> > can't >> >> > boot vanilla 7.1R) >> >> >> >> So, my solution was to set kern.cam.scsi_delay=10000 >> >> in /boot/loader.conf >> > >> > The following patch should work. It creates interleaving root hold >> > tokens from the CAM probe to disk_create and geom providor tasting. >> > I had to add a malloc type flag as sleeping isnt allowed at the point I >> > added the token alloc in CAM. >> > >> > http://people.freebsd.org/~thompsa/root_wait.diff >> >> Hmm, this is supposed to fix issue when trying to boot from usb disk >> with UP kernel? > > This is to address the issue where the usb disk hasnt been attached by > the time the root filesystem is mounted. ie, you are booting from usb. > > If your problem is different then please say. On SMP booting from usb works (kern.cam.scsi_delay=2000), da0 will appear after user is asked to enter root mount point, and pressing ? will show ufs:da0s1a. On UP it doesnt work, ufs:da0s1a is not available. I thought it may be related to scsi_delay but increasing it was not solution. -- PaulReceived on Mon Mar 30 2009 - 15:03:36 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:45 UTC