From: Andy Gospodarek <gospo@redhat.com>
Subject: [RHEL5.1 PATCH] forcedeth: missed 2 chunks from 0.60 backport
Date: Mon, 13 Aug 2007 16:47:21 -0400
Bugzilla: 252034
Message-Id: <20070813204719.GH19875@gospo.rdu.redhat.com>
Changelog: [net] forcedeth: optimize the tx data path



I somehow missed this small chunk from my recent forcedeth backport, so
it needs to be added.  Dropping these calls simply removes calls that
should ensure that the writel calls completed successfully, but
apparently that isn't needed.  Without this it seems like we aren't
actually trying to optimize the code path as hoped by the following
commit (most of which was included):

commit 445583b89d71b48cf8c64e26acc5a710248feed7
Author: Ayaz Abdulla <aabdulla@nvidia.com>

    forcedeth: tx data path optimization

    This patch optimizes the tx data paths and cleans up the code (removes
    vlan from descr1/2 since only valid for desc3, changes to make code
    easier to read, etc).

    Signed-Off-By: Ayaz Abdulla <aabdulla@nvidia.com>

    Signed-off-by: Jeff Garzik <jeff@garzik.org>

This resolves BZ 252034.

---

 forcedeth.c |    2 --
 1 files changed, 2 deletions(-)

--- linux-2.6.18.x86_64/drivers/net/forcedeth.c.orig
+++ linux-2.6.18.x86_64/drivers/net/forcedeth.c
@@ -1722,7 +1722,6 @@ static int nv_start_xmit(struct sk_buff 
 
 	dev->trans_start = jiffies;
 	writel(NVREG_TXRXCTL_KICK|np->txrxctl_bits, get_hwbase(dev) + NvRegTxRxControl);
-	pci_push(get_hwbase(dev));
 	return NETDEV_TX_OK;
 }
 
@@ -1851,7 +1850,6 @@ static int nv_start_xmit_optimized(struc
 
 	dev->trans_start = jiffies;
 	writel(NVREG_TXRXCTL_KICK|np->txrxctl_bits, get_hwbase(dev) + NvRegTxRxControl);
-	pci_push(get_hwbase(dev));
 	return NETDEV_TX_OK;
 }