Re: [PATCH 10 of 20] ipath - support for userspace apps using coredriver

[Andrew Morton]

Roland Dreier <rdreier@xxxxxxxxx> wrote:
>
>     Andrew> Someone left PG_private set on this page (!)
> 
> How does PG_private get set?  Could doing a > 1 page kmalloc set it
> (because we end up with a compound page)?

I can only think that Brian went and set it.

>     Andrew> You need to decide who "owns" these pages.  Once that's
>     Andrew> decided, it tells you who should release them.
> 
>     [... really good guide to mapping pages into userspace snipped ...]
> 
> How about the case where one wants to map pages from
> dma_alloc_coherent() into userspace?  It seems one should do
> get_page() in .nopage, and then the driver can do dma_free_coherent()
> when the vma is released.

Well, where were the pages allocated?  mmap()?  In that case the pages
might be considered driver-owned and yes, I guess the right place to free
them would be in the file's ->release() method.  (Sorry to sound vague, but
I haven't really worked on this stuff for a couple of years, and Hugh keeps
on megachanging it).

Lots of sound drivers do this sort of thing for their capture buffers - you
could check them to see what (or, knowing sound drivers, what not) to do.

> Or maybe it's just simpler to use vm_insert_page() in the .mmap method
> and not try to be fancy with .nopage?
> 

One would need to work out what to do with these pages when they're shared,
after a fork - a ->nopage() handler would still be needed there, assuming
the VMA is marked VM_DONTCOPY.  Because we don't copy all the pte's on a
VM_DONTCOPY vma at fork()-time.  (I think we _could_, but we don't)

vm_insert_page() mucks around with rmap-named functions which don't
actually do rmap and sports apparently-incorrect comments wrt
PageReserved().  I don't know how well-cared-for it is...
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/