scsi_wait_scan doesn't wait

Hi there,

I'm having a problem with 2.6.36-rc8+ on HP PARISC machine (HP 9000 RP3410) where I intend to wait for SCSI enumeration during boot before LVM2 kicks in and attempts to activate the root volume. I modprobe scsi_wait_scan in the initramdisk before lvm2 script is run. However the modprobe doesn't wait. I'm on Debian testing, fwiw.

[...]

[    2.368000] Freeing unused kernel memory: 252k freed
Loading, please wait...
[    3.028000] udev: starting version 160
[    4.104000] SCSI subsystem initialized
[    4.592000] sym0: <1010-66> rev 0x1 at pci 0000:20:01.0 irq 69
[    4.668000] sym0: No NVRAM, ID 7, Fast-80, LVD, parity checking
[    4.700000] sym0: SCSI BUS has been reset.
[    4.708000] scsi0 : sym-2.2.3
[    4.868000] sym1: <1010-66> rev 0x1 at pci 0000:20:01.1 irq 70
[    4.868000] sym1: No NVRAM, ID 7, Fast-80, LVD, parity checking
[    4.904000] sym1: SCSI BUS has been reset.
[    4.908000] scsi1 : sym-2.2.3
[    5.068000] sym2: <1010-66> rev 0x1 at pci 0000:80:01.0 irq 72
[    5.072000] sym2: No NVRAM, ID 7, Fast-80, LVD, parity checking
[    5.108000] sym2: SCSI BUS has been reset.
[    5.112000] scsi2 : sym-2.2.3
[    5.276000] sym3: <1010-66> rev 0x1 at pci 0000:80:01.1 irq 73
[    5.276000] sym3: No NVRAM, ID 7, Fast-80, LVD, parity checking
[    5.312000] sym3: SCSI BUS has been reset.
[    5.316000] scsi3 : sym-2.2.3

Begin: Loading essential drivers ... done.
Begin: Running /scripts/init-premount ... SCSI: Waiting to settle...

######### here comes modprobe scsi_wait_scan #########

SCSI: done
done.
Begin: Mounting root file system ... Begin: Running /scripts/local-top
... [ 6.496000] device-mapper: uevent: version 1.0.3

[    6.496000] device-mapper: ioctl: 4.18.0-ioctl (2010-06-29)

initialised: [email protected]
Volume group "lethal" not found
Skipping volume group lethal
Unable to find LVM volume lethal/root
done.
Begin: Waiting for root file system ... [ 7.728000] scsi 0:0:0:0:
Direct-Access HP 36.4G ST336754LC HPC2 PQ: 0 ANSI: 3

[    7.728000] scsi target0:0:0: tagged command queuing enabled, command
queue depth 16.
[    7.728000] scsi target0:0:0: Beginning Domain Validation
[    7.740000] scsi target0:0:0: FAST-80 WIDE SCSI 160.0 MB/s DT (12.5
ns, offset 31)
[    7.748000] scsi target0:0:0: Ending Domain Validation
[    7.752000] scsi 0:0:1:0: Direct-Access     HP 36.4G MAS3367NC
 HPC3 PQ: 0 ANSI: 3
[    7.752000] scsi target0:0:1: tagged command queuing enabled, command
queue depth 16.
[    7.760000] scsi target0:0:1: Beginning Domain Validation
[    7.764000] scsi target0:0:1: FAST-80 WIDE SCSI 160.0 MB/s DT (12.5
ns, offset 31)
[    7.772000] scsi target0:0:1: Ending Domain Validation
[   11.864000] sd 0:0:0:0: [sda] 71132960 512-byte logical blocks: (36.4
GB/33.9 GiB)
[   11.864000] sd 0:0:1:0: [sdb] 71132960 512-byte logical blocks: (36.4
GB/33.9 GiB)
[   11.868000] sd 0:0:0:0: [sda] Write Protect is off
[   11.892000] sd 0:0:1:0: [sdb] Write Protect is off
[   11.900000] sd 0:0:0:0: [sda] Write cache: enabled, read cache:
enabled, supports DPO and FUA
[   11.900000] sd 0:0:1:0: [sdb] Write cache: enabled, read cache:
enabled, supports DPO and FUA
[   11.924000]  sda: unknown partition table
[   11.940000]  sdb: sdb1 sdb2 sdb3
[   11.944000] sd 0:0:0:0: [sda] Attached SCSI disk
[   11.964000] sd 0:0:1:0: [sdb] Attached SCSI disk

done.
Gave up waiting for root device. Common problems:
- Boot args (cat /proc/cmdline)
- Check rootdelay= (did the system wait long enough?)
- Check root= (did the system wait for the right device?)
- Missing modules (cat /proc/modules; ls /dev)
ALERT! /dev/mapper/lethal-root does not exist. Dropping to a shell!

Apparently /dev/sda and /dev/sdb were discovered long after scsi_wait_scan was loaded.

Here are the relevant modules:
[...]

ext3                  203543  2
jbd                    63990  1 ext3
dm_mod                112954  9
scsi_wait_scan           983  0
zalon7xx               38131  0
lasi700                 2427  0
53c700                 39777  1 lasi700
hilkbd                  5901  0
sd_mod                 49383  3
crc_t10dif              1564  1 sd_mod
sym53c8xx             104303  2
scsi_transport_spi     34731  3 zalon7xx,53c700,sym53c8xx
scsi_mod              193802  7

libata,zalon7xx,lasi700,53c700,sd_mod,sym53c8xx,scsi_transport_spi

When I unset CONFIG_SCSI_SCAN_ASYNC it works just fine and all the disks are found in time for LVM.

Is that a bug in the SCSI driver? Or am I doing something wrong? It's not a recent regression, at least 2.6.35 behaves the same on that machine.

Thanks!

Michal

If you're interested in helping, the best thing you could do is back up to a kernel that works and then try to track down what commit broke the wait scan. It's a bit tedious, but fairly easy.

Cheers,
Carlos.

Yeah, I know. However a kernel rebuild takes nearly 2 hours on that machine so it may be a couple of months before I find the commit ;-) I'll see if I can come up with an automated process for that.

Michal

Hi Michal,

I'd propose that you use a cross-compiler (i686 to hppa) instead.
Then you can TFTP-boot the built kernel from your PC.
This is what I do, and it speeds up everything a lot and it's not too hard to set it up.

Helge