Re: [PATCH 3/6] mv643xx_eth: fix hw checksum generation on transmit

["Dale Farnsworth" <dale@farnsworth.org>]

On Tue, Dec 14, 2004 at 11:15:55PM +0000, Christoph Hellwig wrote:
> > +			dev_kfree_skb_irq((struct sk_buff *)
> > +                                                  pkt_info.return_info);
> 
> pkt_info.return_info already is a pointer to struct sk_buff

Yep.  This line was moved, but not changed by my patch.  I'll remove the
cast.

> > +			/* CPU already calculated pseudo header checksum. */
> > +			if (skb->nh.iph->protocol == IPPROTO_UDP) {
> > +				pkt_info.cmd_sts |= ETH_UDP_FRAME;
> > +				pkt_info.l4i_chk = skb->h.uh->check;
> > +			}
> > +			else if (skb->nh.iph->protocol == IPPROTO_TCP)
> > +				pkt_info.l4i_chk = skb->h.th->check;
> > +			else {
> 
> 			} else if (skb->nh.iph->protocol == IPPROTO_TCP) {
> 				pkt_info.l4i_chk = skb->h.th->check;
> 			} else {
> 
> > +	pkt_info.buf_ptr = pci_map_single(0, skb->data, skb->len,
> > +							PCI_DMA_TODEVICE);
> 
> s/0/NULL/ to avoid sparse warnings

Another line that was moved but not changed.  In patch 4/6 I changed it
to dma_map_single(NULL, ...

> > +	/*
> > +	 * The hardware requires that each buffer that is <= 8 bytes
> > +	 * in length must be aligned on an 8 byte boundary.
> > +	 */
> > +        if (p_pkt_info->byte_cnt <= 8 && p_pkt_info->buf_ptr & 0x7) {
> 
> please use tabs, not spaces for indentation.

Again, I didn't change the indentation on this line.  The driver is
replete with whitespace problems.  I don't want to address them now,
but as soon as it's confirmed that these patches are "in the queue",
I'll submit several cosmetic patches:  white space cleanup, rename
from 64340 to 643xx, rename MV_READ/MV_WRITE to mv_read/mv_write,
and a few localized code simplifications.

> >  #ifdef MV64340_CHECKSUM_OFFLOAD_TX
> > -        int tx_first_desc;
> > +        int tx_busy_desc = mp->tx_first_desc_q;
> 
> Again.

Thanks,
-Dale