pci_set_consistent_dma_mask() question

pci_set_consistent_dma_mask() question

[Marin Mitov]

Hi all,

In the file: Documentation/DMA-mapping.txt is written:

pci_set_consistent_dma_mask() will always be able to set the same or a
smaller mask as pci_set_dma_mask(). However for the rare case that a
device driver only uses consistent allocations, one would have to
check the return value from pci_set_consistent_dma_mask().

grep-ing drivers/net/* shows that in many drivers 
the return value of pci_set_consistent_dma_mask() is checked
in the path where pci_set_dma_mask() was already successfull.
Sure, this is during driver's initiallysation, so it is not time critical.

My question: Is it worth to remove the unnecessary checks?
I could prepare patches if you find it worthfull.

Best regards.

Marin Mitov

Re: pci_set_consistent_dma_mask() question

[Arjan van de Ven]

On Sun, 25 May 2008 08:43:47 +0300
Marin Mitov <mitov@issp.bas.bg> wrote:

> Hi all,
> 
> In the file: Documentation/DMA-mapping.txt is written:
> 
> pci_set_consistent_dma_mask() will always be able to set the same or a
> smaller mask as pci_set_dma_mask(). However for the rare case that a
> device driver only uses consistent allocations, one would have to
> check the return value from pci_set_consistent_dma_mask().
> 
> grep-ing drivers/net/* shows that in many drivers 
> the return value of pci_set_consistent_dma_mask() is checked
> in the path where pci_set_dma_mask() was already successfull.
> Sure, this is during driver's initiallysation, so it is not time
> critical.
> 
> My question: Is it worth to remove the unnecessary checks?
> I could prepare patches if you find it worthfull.
> 

I think it would actually be harmful. Checking for errors even if they
shouldn't happen for things like this makes drivers better! If something
weird is going on it gets detected earlier... IN general, defensive
programming makes a lot of sense.