Re: proposal for bus_dma(9) change

[David Young <dyoung%pobox.com@localhost>]

On Wed, Mar 04, 2009 at 03:38:57PM +0100, Christoph Egger wrote:
> On Wednesday 04 March 2009 14:56:33 Izumi Tsutsui wrote:
> > Does it require less effort to fix all bus_dmamap_create()
> > than fixing each driver?
> 
> After the implementation of bus_dmamap_create() has been found,
> find its error path and there set the dmamap pointer to NULL.
> I already did that in sys/arch/x86/x86/bus_dma.c rev. 1.47
> but had to back it out again (rev. 1.48).
> 
> I also fixed two drivers: nfe(4) and age(4). See
> sys/dev/pci/if_nfe.c rev. 1.42 and sys/dev/pci/if_age.c rev. 1.25.
> 
> Look at this those revisions. This should allow you to
> make an estimation of the required effort for a) changing bus_dma(9)
> or b) fixing all drivers

Christoph,

We have to fix all drivers, anyway.  Your proposed change to
bus_dma(9) would not have been sufficient to fix either nfe(4) or
age(4).

Dave

-- 
David Young             OJC Technologies
dyoung%ojctech.com@localhost      Urbana, IL * (217) 278-3933