> > Long term I think per-request options here are essential.
> > Near term, flagging this at the device level may well be
> > the least invasive solution.
> 
> Why? I think that either whole device needs GFP_NOIO or whole device can live
> with GFP_KERNEL, but I can't imagine a device where aprt needs GFP_NOIO and
> part does not...

I can imagine a device that includes both a usb-storage
interface, used for swapping/paging/etc, and other
interfaces (say, for audio/video I/O streams).  Would
that suit as one example, or would you insist on being
able to buy one first?  :)

Or with a single usb-storage interface, and vendor-specific
control messages used to tweak something from a usermode
tool; there's no reason to have the usermode calls use NOIO,
but control messages to clear error status during paging will
need it.

More generally, my point is that the "I dare not trigger
I/O during this allocation" constraint is sensitive to the
context in which the request is made, not the device.

- Dave





_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to