Hello, I wrote:
The DMA is supported by current git for dm6446 usb.
I've used for quite a long period.
I don't advise using it. It surely has a big issue with sending data -- DMA IRQ happens before the data is actually sent, and the next URB is likely to flush the last packet from FIFO before being actually sent. The upstream USB driver doesn't contain a workaround for that issue.
Oops, sorry for misinforming you. I should have also looked into cppi_dma.c... it has a workaround, it's just not what I thought.
I have to ascertain that it does not work, at least under the heavy load (I used bonnie++ on a flash drive). I'm preparing a fix fot this issue but it needs some more stress testing (note for TI people -- this is *not* the tasklet based fix).
David
WBR, Sergei _______________________________________________ Davinci-linux-open-source mailing list [email protected] http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
