Re: [PATCH net-next] net: Implement fast csum_partial for x86_64

2016-01-04 Thread David Miller
From: Tom Herbert Date: Sun, 3 Jan 2016 15:22:27 -0800 > Implement assembly routine for csum_partial for 64 bit x86. This > primarily speeds up checksum calculation for smaller lengths such as > those that are present when doing skb_postpull_rcsum when getting > CHECKSUM_COMPLETE from device or a

Re: [PATCH net-next] net: Implement fast csum_partial for x86_64

2016-01-04 Thread Tom Herbert
On Mon, Jan 4, 2016 at 3:52 PM, Eric Dumazet wrote: > On Mon, 2016-01-04 at 15:34 -0800, Tom Herbert wrote: >> On Mon, Jan 4, 2016 at 2:36 PM, Eric Dumazet wrote: >> > On Sun, 2016-01-03 at 15:22 -0800, Tom Herbert wrote: >> > \... >> >> +402: /* Length 2, align is 1, 3, or 5 */ >> >> + movb

Re: [PATCH net-next] net: Implement fast csum_partial for x86_64

2016-01-04 Thread Eric Dumazet
On Mon, 2016-01-04 at 15:34 -0800, Tom Herbert wrote: > On Mon, Jan 4, 2016 at 2:36 PM, Eric Dumazet wrote: > > On Sun, 2016-01-03 at 15:22 -0800, Tom Herbert wrote: > > \... > >> +402: /* Length 2, align is 1, 3, or 5 */ > >> + movb(%rdi), %al > >> + movb1(%rdi), %ah > > > > Looks

Re: [PATCH net-next] net: Implement fast csum_partial for x86_64

2016-01-04 Thread Hannes Frederic Sowa
On 05.01.2016 00:34, Tom Herbert wrote: On Mon, Jan 4, 2016 at 2:36 PM, Eric Dumazet wrote: On Sun, 2016-01-03 at 15:22 -0800, Tom Herbert wrote: \... +402: /* Length 2, align is 1, 3, or 5 */ + movb(%rdi), %al + movb1(%rdi), %ah Looks like a movw (%rdi),%ax Wouldn't that b

Re: [PATCH net-next] net: Implement fast csum_partial for x86_64

2016-01-04 Thread Tom Herbert
On Mon, Jan 4, 2016 at 2:36 PM, Eric Dumazet wrote: > On Sun, 2016-01-03 at 15:22 -0800, Tom Herbert wrote: > \... >> +402: /* Length 2, align is 1, 3, or 5 */ >> + movb(%rdi), %al >> + movb1(%rdi), %ah > > Looks like a movw (%rdi),%ax > Wouldn't that be an unaligned access? > > A

Re: [PATCH net-next] net: Implement fast csum_partial for x86_64

2016-01-04 Thread Eric Dumazet
On Sun, 2016-01-03 at 15:22 -0800, Tom Herbert wrote: \... > +402: /* Length 2, align is 1, 3, or 5 */ > + movb(%rdi), %al > + movb1(%rdi), %ah Looks like a movw (%rdi),%ax Also you probably should send this patch to x86 maintainers. -- To unsubscribe from this list: send the l

Re: [PATCH net-next] net: Implement fast csum_partial for x86_64

2016-01-04 Thread Hannes Frederic Sowa
On 04.01.2016 00:22, Tom Herbert wrote: Implement assembly routine for csum_partial for 64 bit x86. This primarily speeds up checksum calculation for smaller lengths such as those that are present when doing skb_postpull_rcsum when getting CHECKSUM_COMPLETE from device or after CHECKSUM_UNNECESSA

[PATCH net-next] net: Implement fast csum_partial for x86_64

2016-01-03 Thread Tom Herbert
Implement assembly routine for csum_partial for 64 bit x86. This primarily speeds up checksum calculation for smaller lengths such as those that are present when doing skb_postpull_rcsum when getting CHECKSUM_COMPLETE from device or after CHECKSUM_UNNECESSARY conversion. This implementation is sim