Re: Unifying /dev/{mem,kmem,zero,null} implementations

[Chris Gilbert <chris%dokein.co.uk@localhost>]

Joerg Sonnenberger wrote:
> On Sat, Feb 23, 2008 at 03:36:27PM +0000, Chris Gilbert wrote:
>   
> > In dev_mem_readwrite() pmap_kenter_pa and pmap_kremove calls require a 
> > pmap_update(pmap_kernel()) after them otherwise the updates may not have 
> > been sync'd into the hardware.
> >     
>
> I'll add it after the kenter. I don't think we need it after the
> pmap_kremove -- it can be thrown away lazily, can't it?
>   

The documentation says every pmap_kremove needs a pmap_update, relying 
on something later on doing a pmap_update is risky.

In theory pmap_kremove could be optimised to not change anything in 
hardware, and defer the work to be done till the pmap_update, while I 
don't believe any platform has gone to that level yet, we shouldn't 
exclude the possibility by not following the documented API.

Note I earlier today I did find a few places that didn't do a 
pmap_update for pmap_kenter_pa and pmap_kremove calls and I fixed them.

Thanks,
Chris