Re: Bug in SCSI async probing

From: Alan Stern
Date: Tue May 26 2009 - 17:53:05 EST

On Tue, 26 May 2009, James Bottomley wrote:

> > Also, I do still think that wait_scan_init() needs an extra call to
> > async_synchronize_full() at the end.
> I need to understand it before commenting ... I'm slowly working my way
> through the async patches.

There are two lists of queued asynchronous tasks: Arjan's list and the
SCSI scanning list. As things stand now, routines in each list are
capable of submitting new jobs for the other list. That is, an async
task may discover and register a new SCSI host (requiring scanning),
and the probe of a SCSI disk discovered during scanning will submit
a sd_probe_async task.

This makes it difficult to wait for both lists to become idle. Maybe
the whole idea behind scsi_wait_scan.c needs some rethinking...

Alan Stern

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at