From: "Michael Chan" <[EMAIL PROTECTED]>
Date: Mon, 07 Aug 2006 20:05:36 -0700
> Actually, I can probably hide it in the TX_BUFFS_AVAIL() macro.
Sounds like a good idea.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordom
On Tue, 2006-08-08 at 12:51 +1000, Herbert Xu wrote:
> > 2. Eliminate the private tx_lock altogether and rely solely on
> > netif_tx_lock. This eliminates one spinlock in tg3_start_xmit()
> > when the ring is full.
>
> Why not get rid of the lock altogether? By making sure memory
> barriers are
From: "Michael Chan" <[EMAIL PROTECTED]>
Date: Mon, 07 Aug 2006 19:46:26 -0700
> 4. Make tx_cons and tx_prod volatile to guarantee that
> tg3_start_xmit() and tg3_tx() will see the latest ring indices.
Marking variables volatile is always an error.
If the locks and memory barriers are in the rig
On Mon, Aug 07, 2006 at 07:46:26PM -0700, Michael Chan wrote:
> Fix a subtle race condition between tg3_start_xmit() and tg3_tx()
> discovered by Herbert Xu <[EMAIL PROTECTED]>:
Thanks for the patch Michael.
> 2. Eliminate the private tx_lock altogether and rely solely on
> netif_tx_lock. This
Fix a subtle race condition between tg3_start_xmit() and tg3_tx()
discovered by Herbert Xu <[EMAIL PROTECTED]>:
CPU0CPU1
tg3_start_xmit()
if (tx_ring_full) {
tx_lock
tg3_tx()
