Hi Linus,

On 12 October 2012 23:32, Linus Walleij [off-list ref] wrote:

On Fri, Oct 12, 2012 at 5:33 PM, Ulf Hansson [off-list ref] wrote:

quoted

From: Ulf Hansson <redacted>

The cookie is now used to indicate if dma_unmap_sg shall be
done in post_request. At DMA errors, the DMA job is immediately
not only terminated but also unmapped. To indicate that this
has been done the cookie is reset to zero. post_request will
thus only do dma_umap_sg for requests which has a cookie not set
to zero.

Some corresponding duplicated code could then be removed and
moreover some corrections at DMA errors for terminating the same
DMA job twice has also been fixed.

Signed-off-by: Ulf Hansson <redacted>
Signed-off-by: Per Forlin <redacted>

It looks like it's both factoring out code and also adding some unmapping
in hithereto unhandled cases, correct? It looks OK to me now atleast so
Acked-by: Linus Walleij <redacted>

This code has not been tested on a "legacy" ARM PL180 but only for
ux500 boards. Even if it should affect DMA handling we should test
this properly. Would be great if you were able to help out, I guess
you still have available hardware for these tests?

Kind regards
Ulf Hansson