RE: [PATCH] bna: Implement ethtool flash_device entry point.

[Ben Hutchings <bhutchings@solarflare.com>]

On Wed, 2012-02-01 at 14:10 -0800, Krishna Gudipati wrote:
> Thanks Ben for the review.
> 
> I am making the changes as per the review comments to return proper error code.
> 
> > +	init_completion(&fcomp.comp);
> > +	spin_lock_irqsave(&bnad->bna_lock, flags);
> 
> No need to save flags, this is not called in IRQ context!
> 
> [Krishna]: bna_lock is used in Interrupt context as well, by using spin_lock_irq instead of
> 	     spin_lock_irqsave() - we will lose the state of the interrupts & will
> 	     unconditionally enable interrupts. We plan to stay with using spin_lock_irqsave()
> 	     as it's safe.

bnad_flash_device() will be called in process context, therefore it is
correct to unconditionally enable interrupts.  It is safe but odd to use
spin_lock_irqsave()/spin_lock_irqrestore() in such a function.

Ben.

-- 
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.