On Mon, Nov 15, 2010 at 04:24:01AM -0600, Kalliguddi, Hema wrote:
Buffer is mapped to dma when dma channel is allocated. buffer needs
to be unmapped when fallback to PIO mode if dma channel_program
fails.
g_ether use to fail on OMAp3630 as there were unaligned buffers being sent.
With this patch and [1] g_ether works fine.
[1]:http://www.spinics.net/lists/linux-usb/msg38400.html
Signed-off-by: Hema HK <hemahk@xxxxxx>
Cc: Felipe Balbi <balbi@xxxxxx>
applied, thanks.
I brushed the commit log a little bit, please verify it's ok:
commit fd542305e10f85b1f46c71daf8b6ddd12f15fa82
Author: Hema Kalliguddi <hemahk@xxxxxx>
Date: Mon Nov 15 04:24:01 2010 -0600
usb: musb: unmap dma buffer when switching to PIO
Buffer is mapped to dma when dma channel is
allocated. If, for some reason, dma channel
programming fails, musb code will fallback
to PIO mode to transfer that request. In
that case, we need to unmap the buffer
back to CPU.
MUSB RTL1.8 cannot handle buffers which are
not 32bit aligned. That happens to every
request sent by g_ether gadget driver. Since
the buffer sent was unaligned, we need to
fallback to PIO.
Because of that, g_ether was failing due
to missing buffer unmapping.
With this patch and [1] g_ether works fine
with all MUSB revisions.
Verified with OMAP3630 board, which has
MUSB RTL1.8 using g_ether and g_zero.
[1] http://www.spinics.net/lists/linux-usb/msg38400.html
Signed-off-by: Hema HK <hemahk@xxxxxx>
Signed-off-by: Felipe Balbi <balbi@xxxxxx>
--
balbi
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
