diff options
| author | Alex Gonzalez <alex.gonzalez@digi.com> | 2011-08-09 16:37:38 +0200 |
|---|---|---|
| committer | Alex Gonzalez <alex.gonzalez@digi.com> | 2011-08-09 16:37:38 +0200 |
| commit | f8493d7c13d83d99c300403dbe075555d806b510 (patch) | |
| tree | 764267b3d4510f89eebf58accdeb55f29c02a6c2 /drivers/net | |
| parent | f132ad28c29d66fe729837ab2d9b72ed9528f452 (diff) | |
mx28: Improved FEC patch for the packets bigger than MTU issue.
Freescale support confirmed that the delay is needed to allow for the
slow MX28 bus to DMA the data.
This empirically improved patch gives better performance than a single
delay.
Also, I confirmed with the SarOS team they have a similar patch in the
same place.
Signed-off-by: Alex Gonzalez <alex.gonzalez@digi.com>
Diffstat (limited to 'drivers/net')
| -rw-r--r-- | drivers/net/fec.c | 13 |
1 files changed, 11 insertions, 2 deletions
diff --git a/drivers/net/fec.c b/drivers/net/fec.c index 9077b81c71b0..69ab4ec7f990 100644 --- a/drivers/net/fec.c +++ b/drivers/net/fec.c @@ -250,6 +250,7 @@ fec_enet_start_xmit(struct sk_buff *skb, struct net_device *dev) unsigned short status; unsigned long estatus; unsigned long flags; + int i = 0; if (!fep->link) { /* Link is down or autonegotiation is in progress. */ @@ -325,10 +326,18 @@ fec_enet_start_xmit(struct sk_buff *skb, struct net_device *dev) | BD_ENET_TX_LAST | BD_ENET_TX_TC); bdp->cbd_sc = status; +#ifdef CONFIG_ARCH_MXS + /* Trigger transmission start */ + writel(0, fep->hwp + FEC_X_DES_ACTIVE); + while ( i++ < 5 ) { + while (readl(fep->hwp + FEC_X_DES_ACTIVE) == 1) + continue; + udelay(10); + writel(0, fep->hwp + FEC_X_DES_ACTIVE); + } +#else /* Trigger transmission start */ writel(0, fep->hwp + FEC_X_DES_ACTIVE); -#ifdef CONFIG_ARCH_MXS - udelay(100); #endif /* If this was the last BD in the ring, start at the beginning again. */ |