Skip to contentSkip to conversation

NS83820 2.6.0-test5 driver seems unstable on IA64

DISCUSSION Andi Kleen 2 participants 2 messages Sep 26, 2003 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

Andi Kleen

author
#1 lore raw
Quoted 9 lines · Manfred Spraul
Quoted 3 lines · David

Fine, then we should have something like an rx_copybreak scheme in the ns83820 driver too.

Is that really the right solution? Add a full-packet copy to every driver? IMHO the fastest solution would be to copy only the ip & tcp headers, and keep the rest as it is. And preferable in the network core, to avoid having to copy&paste that into every driver.

One problem is that you still have an unaligned->aligned copy to user space in recvmsg (the user buffer is usually aligned and the network payload will be unaligned). And that will be very slow.

-Andi

#2 lore raw
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line · Andi Kleen
Quoted 10 lines · Andi> Manfred Spraul <[email protected]> writes:
Quoted 10 lines
Quoted 2 lines · David

Fine, then we should have something like an rx_copybreak scheme in the ns83820 driver too.

Is that really the right solution? Add a full-packet copy to every driver? IMHO the fastest solution would be to copy only the ip & tcp headers, and keep the rest as it is. And preferable in the network core, to avoid having to copy&paste that into every driver.

Andi> One problem is that you still have an unaligned->aligned copy Andi> to user space in recvmsg (the user buffer is usually aligned Andi> and the network payload will be unaligned). And that will be Andi> very slow.

Why would that be? Slower, yes, but very slow?

	--david