Re: arcmsr + archttp64 calls dma_free_coherent() with irqs disabled - dmesg filled with warnings

[Daniel Drake <dsd@gentoo.org>]

James Bottomley wrote:
> On Sat, 2008-02-16 at 11:49 +0000, Daniel Drake wrote:
>> I assume you're aware that this patch is just a subset of commit 
>> 76d78300a6eb8 which you've already pushed up to Linus. Adding Nick Cheng 
>> (commit author) to CC so that he can go over the feedback.
> 
> Well, in case it's not obvious by now:  The way to get bad code upstream
> is to send a patch that combines many changes (the more the better) so
> that any potential reviewer has no idea which change is meant by which
> hunk and then to make sure Andrew picks it up so he'll hound the
> subsystem Maintainer until it's applied.  Best of all, mention that it
> fixes a bug and you're made.

Sorry, I didn't mean to sound that as a criticism. I'm sure you have a 
lot of patches flowing by you at any one time.

Here is a patch to address your comments.
Joshua, would you mind testing this before I submit it properly? It will 
apply cleanly to 2.6.24 on top of the previous patch you tested. I have 
compile-tested it.

> The odd thing is, it should have triggered a might_sleep() warning under
> testing ... do you know why it didn't?

No, and I can see that scsi_dispatch_command does invoke ->queuecommand 
under a spinlock so it must be atomic context...
I'm not sure which might_sleep() codepath you are looking at though. At 
a guess it depends on SLUB vs SLAB?

Daniel