Matthew Dharm wrote:
I found the problem. It's not in usb-storage, but in the usb core.
The patch from David Brownell "USB: usbcore misc cleanup (notably for
non-dma hcds)" seems to be the problem. If I revert it out of my tree,
everything works again.
This even makes sense -- that patch changes some things in the
scatter-gather primitives, which are exactly the types of transfers which
aren't working properly.
Doesn't make sense to me ... the only change to the sglist stuff was
to stop assuming DMA was always used. The change was basically to
not map scatterlists unless the HC supports DMA mapping. Typically
it does, in which case the code behaves exactly as it did before.
What specific part of the patch are you thinking has a problem?
No problem reproduces here for me.
I have a sneaking suspicion that usb-storage may be the only code that uses
the scatter-gather stuff, which is why nothing else showed any problems
from David's patch.
The "usbtest" code uses it also, and hasn't reported problems.
But it does now work with a driver that wouldn't work before.
I just ran basic sanity tests with both "usbtest" and "mkfs",
both had no issues doing scatterlist DMA.
- Dave
-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
