Menno Lageman wrote: > On 07/29/09 03:52, Shawn Walker wrote: >> I followed the directions at: >> >> http://www.opensolaris.org/os/project/indiana/building_on >> >> ...after building my own non-debug BFU archives. >> >> I'm on a b118 system, and am attempting to BFU to onnv bits from this >> past weekend. I'm hacking on a kernel driver, so yes, I really do >> need to BFU for now :P >> >> However, when selecting the new be based on these bits, I get a panic >> on start: >> >> ... >> NOTICE: Can not read the pool label from >> '/pci at 0,0/pci-ide at b/ide at 0/cmdk at 0,0:a' >> NOTICE: spa_import_rootpool: error 5 >> Cannot mount root on /pci at 0,0/pci-ide at b/ide at 0/cmdk at 0,0:a fstype >> zfs >> .... >> genunix:vfs_mount_root >> genunix:main >> unix:_locore_start >> ... >> >> I suspect I'm missing something simple, or there's something in the >> flag days that I missed between b118 and this weekend, but I don't >> know what it is. >> >> The only oddity I have to mention is that when I ran acr, it said: >> >> etc/init.d/devlinks upgrade_default_copy FAIL >> >> But when I checked the 'allresults' file in /tmp/..., there were no >> errors. >> >> Any ideas? >> > > Might be: 6865463 Unable to boot zfs after bfuing to daily.0703
Confirmed. I can't use kdb from the panic prompt (no working keyboard and no serial port; sigh) but I can tell from some cmn_err statements I added that it is failing in vdev_disk_read_rootlabel when ldi_open_by_devid and ldi_open_by_name are called. However, they appear to be failing with ENXIO instead of EIO, while bug 6865463 is about an EIO failure in vdev_disk_read_rootlabel. So, I went back to rev 10020 of the gate just before the extended partition putback, and I was able to successfully build a BFU archive and boot to it. Just to verify, I went to rev 10021 where expended partition support was putback, and sure enough, I could no longer boot my bfu'd environment. I'll update the bug with my information. Thanks! -- Shawn Walker
