Hi,
Pandita, Vikram a écrit :
>
>> -----Original Message-----
>> From: Felipe Balbi [mailto:felipe.balbi@xxxxxxxxx]
>> Sent: Thursday, January 14, 2010 5:28 AM
>> To: ext Matthieu CASTET
>> Cc: Balbi Felipe (Nokia-D/Helsinki); Gupta, Ajay Kumar; Pandita, Vikram; matthieu castet; linux-
>> usb@xxxxxxxxxxxxxxx; Alan Stern
>> Subject: Re: [RFC] gadget ethernet dma align fix
>>
>> Hi,
>>
>> On Thu, Jan 14, 2010 at 09:40:43AM +0100, ext Matthieu CASTET wrote:
>>> Do you know if it is related to the align patch ?
>>>
>>> May be it could be interesting to enable debug in ethernet gadget driver.
>> yeah I can do that, but it seems to be related to this patch. On an
>> older version of musb (rtl 1.4) which doesn't suffer badly from this,
>> this script works fine.
>>
>> will run again and let's see what's going on.
>
> Not working:
> with this alignment patch v3 [1] in g_ether,
> ping does not go through for size >= 1471, as pointed out by Felipe and Ajay.
> This needs to be root caused and obviously a problem with introducing memmove() in g-ether.
>
> Works fine:
> With the alignment fix in the MUSB controller driver [2] rather than g-ether,
> the problem with ping size >= 1471 is not observed.
>
I found the problem. I was on rx path were skb->len were wrong.
I attach a new patch that fix the issue.
Regards,
Matthieu
diff --git a/drivers/usb/gadget/gadget_chips.h b/drivers/usb/gadget/gadget_chips.h
index f2d270b..5cd8320 100644
--- a/drivers/usb/gadget/gadget_chips.h
+++ b/drivers/usb/gadget/gadget_chips.h
@@ -273,4 +273,14 @@ static inline bool gadget_supports_altsettings(struct usb_gadget *gadget)
return true;
}
+/**
+ * gadget_dma32 - return true if we want buffer aligned on 32 bits (for dma)
+ * @gadget: the gadget in question
+ */
+static inline bool gadget_dma32(struct usb_gadget *gadget)
+{
+ if (gadget_is_musbhdrc(gadget))
+ return true;
+ return false;
+}
#endif /* __GADGET_CHIPS_H */
diff --git a/drivers/usb/gadget/u_ether.c b/drivers/usb/gadget/u_ether.c
index 2fc02bd..bd90859 100644
--- a/drivers/usb/gadget/u_ether.c
+++ b/drivers/usb/gadget/u_ether.c
@@ -249,7 +249,12 @@ rx_submit(struct eth_dev *dev, struct usb_request *req, gfp_t gfp_flags)
* but on at least one, checksumming fails otherwise. Note:
* RNDIS headers involve variable numbers of LE32 values.
*/
- skb_reserve(skb, NET_IP_ALIGN);
+ /*
+ * RX: Do not move data by IP_ALIGN:
+ * if your DMA controller cannot handle it
+ */
+ if (!gadget_dma32(dev->gadget))
+ skb_reserve(skb, NET_IP_ALIGN);
req->buf = skb->data;
req->length = size;
@@ -282,6 +287,12 @@ static void rx_complete(struct usb_ep *ep, struct usb_request *req)
/* normal completion */
case 0:
skb_put(skb, req->actual);
+ if (gadget_dma32(dev->gadget) && NET_IP_ALIGN) {
+ u8 *data = skb->data;
+ size_t len = skb_headlen(skb);
+ skb_reserve(skb, NET_IP_ALIGN);
+ memmove(skb->data, data, len);
+ }
if (dev->unwrap) {
unsigned long flags;
@@ -573,6 +584,24 @@ static netdev_tx_t eth_start_xmit(struct sk_buff *skb,
length = skb->len;
}
+
+ /*
+ * Align data to 32bit if the dma controller requires it
+ */
+ if (gadget_dma32(dev->gadget)) {
+ unsigned long align = (unsigned long)skb->data & 3;
+ if (WARN_ON(skb_headroom(skb) < align)) {
+ dev_kfree_skb_any(skb);
+ goto drop;
+ } else if (align) {
+ u8 *data = skb->data;
+ size_t len = skb_headlen(skb);
+ skb->data -= align;
+ memmove(skb->data, data, len);
+ skb_set_tail_pointer(skb, len);
+ }
+ }
+
req->buf = skb->data;
req->context = skb;
req->complete = tx_complete;
