Home » Mailing list archives » linux-kernel » 2007 » March » 27

Re: CONFIG_SCSI_WAIT_SCAN=m

Previous thread: Corrupt XFS -Filesystems on new Hardware and Kernel by Oliver Joa on Tuesday, March 27, 2007 - 9:16 am. (17 messages)

Next thread: 2.6.20.3-rt8 - DMA suffers excessive delay periodically by John on Tuesday, March 27, 2007 - 9:32 am. (8 messages)
[view original message]
From: Hugh Dickins
Subject: CONFIG_SCSI_WAIT_SCAN=m
Date: Tuesday, March 27, 2007 - 9:25 am

I'm not using an initrd, most of my kernel is builtin,
just a few modules for occasional filesystems.

CONFIG_MODULES=y
CONFIG_SCSI=y
CONFIG_SCSI_SCAN_ASYNC=y
CONFIG_SCSI_WAIT_SCAN=m

2.6.21-rc5-mm2 VFS panics unable to find my root on /dev/sda2,
but boots okay if I change drivers/scsi/Kconfig to "default y"
instead of "default m" for SCSI_WAIT_SCAN.

I'm sure that's not the right fix, but I don't know what is.

I've had no trouble with SCSI_SCAN_ASYNC=y until now: I could
turn it off, but others may have trouble with the same regression.

Hugh
-


[ message continues ]
[view original message]
From: James Bottomley
Subject: Re: CONFIG_SCSI_WAIT_SCAN=m
Date: Tuesday, March 27, 2007 - 9:34 am

Hmm ... it sounds like there's a missing scsi_complete_async_scans() in
your environment?  What's your configuration ... which SCSI drivers, and
which are modules?

scsi_wait_async_scan is supposed to be unnecessary unless you have
modular SCSI drivers that the initrd needs to wait for root on.

James


-


[ message continues ]
[view original message]
From: Hugh Dickins
Subject: Re: CONFIG_SCSI_WAIT_SCAN=m
Date: Tuesday, March 27, 2007 - 10:16 am

Fusion MPT: neither that nor the rest of SCSI modular (apart from this
> modular SCSI drivers that the initrd needs to wait for root on.
[ message continues ]
[view original message]
From: Hugh Dickins
Subject: Re: CONFIG_SCSI_WAIT_SCAN=m
Date: Tuesday, April 3, 2007 - 9:54 am

Situation unchanged in 2.6.21-rc5-mm4.  By what route do you expect
scsi_complete_async_scans() to get called usually, in a monolithic
kernel?  Before you made scsi_wait_scan a separate loadable module,
it was called via that late_initcall(wait_scan_init), so perhaps now
we need...


Make sure there's a late_initcall to scsi_complete_async_scans when
it's built in, so a monolithic SCSI_SCAN_ASYNC kernel can rely on the
scans being completed before trying to mount root, even if they're slow.

Signed-off-by: Hugh Dickins <hugh@veritas.com>

--- 2.6.21-rc5-mm4/drivers/scsi/scsi_scan.c	2007-04-03 15:29:21.000000000 +0100
+++ linux/drivers/scsi/scsi_scan.c	2007-04-03 17:24:24.000000000 +0100
@@ -183,6 +183,7 @@ int scsi_complete_async_scans(void)
 
 /* Only exported for the benefit of scsi_wait_scan */
 EXPORT_SYMBOL_GPL(scsi_complete_async_scans);
+late_initcall(scsi_complete_async_scans);
 
 /**
  * scsi_unlock_floptical - unlock device via a special MODE SENSE command
-


[ message continues ]
[view original message]
From: James Bottomley
Subject: Re: CONFIG_SCSI_WAIT_SCAN=m
Date: Friday, April 6, 2007 - 8:51 am

Sorry, was waiting for a reply on this one ... didn't realise it hadn't
cc'd linux-scsi ...

James


-


[ message continues ]
Previous thread: Corrupt XFS -Filesystems on new Hardware and Kernel by Oliver Joa on Tuesday, March 27, 2007 - 9:16 am. (17 messages)

Next thread: 2.6.20.3-rt8 - DMA suffers excessive delay periodically by John on Tuesday, March 27, 2007 - 9:32 am. (8 messages)