| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Fri, 13 Apr 2007 01:26:11 -0500
From: Kim Phillips <kim.phillips@...escale.com>
To: netdev@...r.kernel.org
Cc: Jeff Garzik <jgarzik@...ox.com>
Subject: [PATCH v2 3/7] ucc_geth: NAPI-related bug fixes
From: Michael Reiss <michael.f.reiss@...escale.com>
Based partly on the gianfar driver, this patch fixes several
bugs which were causing NAPI to be completely unusable.
* An IRQ is still needed in NAPI, to kick off NAPI task,
and for Tx processing. Request the IRQ.
* If rx_work_limit = 0 we are not complete.
* While running Rx NAPI processing we must mask Rx events,
including Rx busy.
* ucc_geth_rx function does not need a lock.
Could lead to deadlock in NAPI case.
* There's no need to loop reading ucce multiple times in the ISR,
so while adding the call to schedule NAPI which was not there,
simplify the event processing into if-else format.
* Rx Busy now kicks off NAPI processing, while still
being counted as an error.
Signed-off-by: Michael Reiss <michael.f.reiss@...escale.com>
Signed-off-by: Michael Barkowski <michael.barkowski@...escale.com>
Signed-off-by: Kim Phillips <kim.phillips@...escale.com>
---
drivers/net/ucc_geth.c | 98 ++++++++++++++++++++++++++++-------------------
drivers/net/ucc_geth.h | 3 +
2 files changed, 61 insertions(+), 40 deletions(-)
diff --git a/drivers/net/ucc_geth.c b/drivers/net/ucc_geth.c
index d93cfde..60be1e7 100644
--- a/drivers/net/ucc_geth.c
+++ b/drivers/net/ucc_geth.c
@@ -3416,7 +3416,6 @@ static int ucc_geth_rx(struct ucc_geth_private *ugeth, u8 rxQ, int rx_work_limit
ugeth_vdbg("%s: IN", __FUNCTION__);
- spin_lock(&ugeth->lock);
/* collect received buffers */
bd = ugeth->rxBd[rxQ];
@@ -3464,7 +3463,6 @@ static int ucc_geth_rx(struct ucc_geth_private *ugeth, u8 rxQ, int rx_work_limit
skb = get_new_skb(ugeth, bd);
if (!skb) {
ugeth_warn("%s: No Rx Data Buffer", __FUNCTION__);
- spin_unlock(&ugeth->lock);
ugeth->stats.rx_dropped++;
break;
}
@@ -3485,7 +3483,6 @@ static int ucc_geth_rx(struct ucc_geth_private *ugeth, u8 rxQ, int rx_work_limit
}
ugeth->rxBd[rxQ] = bd;
- spin_unlock(&ugeth->lock);
return howmany;
}
@@ -3537,23 +3534,38 @@ static int ucc_geth_tx(struct net_device *dev, u8 txQ)
static int ucc_geth_poll(struct net_device *dev, int *budget)
{
struct ucc_geth_private *ugeth = netdev_priv(dev);
+ struct ucc_geth_info *ug_info;
+ struct ucc_fast_private *uccf;
int howmany;
- int rx_work_limit = *budget;
- u8 rxQ = 0;
+ u8 i;
+ int rx_work_limit;
+ register u32 uccm;
+ ug_info = ugeth->ug_info;
+
+ rx_work_limit = *budget;
if (rx_work_limit > dev->quota)
rx_work_limit = dev->quota;
- howmany = ucc_geth_rx(ugeth, rxQ, rx_work_limit);
+ howmany = 0;
+
+ for (i = 0; i < ug_info->numQueuesRx; i++) {
+ howmany += ucc_geth_rx(ugeth, i, rx_work_limit);
+ }
dev->quota -= howmany;
rx_work_limit -= howmany;
*budget -= howmany;
- if (rx_work_limit >= 0)
+ if (rx_work_limit > 0) {
netif_rx_complete(dev);
+ uccf = ugeth->uccf;
+ uccm = in_be32(uccf->p_uccm);
+ uccm |= UCCE_RX_EVENTS;
+ out_be32(uccf->p_uccm, uccm);
+ }
- return (rx_work_limit < 0) ? 1 : 0;
+ return (rx_work_limit > 0) ? 0 : 1;
}
#endif /* CONFIG_UGETH_NAPI */
@@ -3563,10 +3575,13 @@ static irqreturn_t ucc_geth_irq_handler(int irq, void *info)
struct ucc_geth_private *ugeth = netdev_priv(dev);
struct ucc_fast_private *uccf;
struct ucc_geth_info *ug_info;
- register u32 ucce = 0;
- register u32 bit_mask = UCCE_RXBF_SINGLE_MASK;
- register u32 tx_mask = UCCE_TXBF_SINGLE_MASK;
- register u8 i;
+ register u32 ucce;
+ register u32 uccm;
+#ifndef CONFIG_UGETH_NAPI
+ register u32 rx_mask;
+#endif
+ register u32 tx_mask;
+ u8 i;
ugeth_vdbg("%s: IN", __FUNCTION__);
@@ -3576,48 +3591,53 @@ static irqreturn_t ucc_geth_irq_handler(int irq, void *info)
uccf = ugeth->uccf;
ug_info = ugeth->ug_info;
- do {
- ucce |= (u32) (in_be32(uccf->p_ucce) & in_be32(uccf->p_uccm));
-
- /* clear event bits for next time */
- /* Side effect here is to mask ucce variable
- for future processing below. */
- out_be32(uccf->p_ucce, ucce); /* Clear with ones,
- but only bits in UCCM */
-
- /* We ignore Tx interrupts because Tx confirmation is
- done inside Tx routine */
+ /* read and clear events */
+ ucce = (u32) in_be32(uccf->p_ucce);
+ uccm = (u32) in_be32(uccf->p_uccm);
+ ucce &= uccm;
+ out_be32(uccf->p_ucce, ucce);
+ /* check for receive events that require processing */
+ if (ucce & UCCE_RX_EVENTS) {
+#ifdef CONFIG_UGETH_NAPI
+ if (netif_rx_schedule_prep(dev)) {
+ uccm &= ~UCCE_RX_EVENTS;
+ out_be32(uccf->p_uccm, uccm);
+ __netif_rx_schedule(dev);
+ }
+#else
+ rx_mask = UCCE_RXBF_SINGLE_MASK;
for (i = 0; i < ug_info->numQueuesRx; i++) {
- if (ucce & bit_mask)
- ucc_geth_rx(ugeth, i,
- (int)ugeth->ug_info->
- bdRingLenRx[i]);
- ucce &= ~bit_mask;
- bit_mask <<= 1;
+ if (ucce & rx_mask)
+ ucc_geth_rx(ugeth, i, (int)ugeth->ug_info->bdRingLenRx[i]);
+ ucce &= ~rx_mask;
+ rx_mask <<= 1;
}
+#endif /* CONFIG_UGETH_NAPI */
+ }
+ /* Tx event processing */
+ if (ucce & UCCE_TX_EVENTS) {
+ spin_lock(&ugeth->lock);
+ tx_mask = UCCE_TXBF_SINGLE_MASK;
for (i = 0; i < ug_info->numQueuesTx; i++) {
if (ucce & tx_mask)
ucc_geth_tx(dev, i);
ucce &= ~tx_mask;
tx_mask <<= 1;
}
+ spin_unlock(&ugeth->lock);
+ }
- /* Exceptions */
+ /* Errors and other events */
+ if (ucce & UCCE_OTHER) {
if (ucce & UCCE_BSY) {
- ugeth_vdbg("Got BUSY irq!!!!");
ugeth->stats.rx_errors++;
- ucce &= ~UCCE_BSY;
}
- if (ucce & UCCE_OTHER) {
- ugeth_vdbg("Got frame with error (ucce - 0x%08x)!!!!",
- ucce);
- ugeth->stats.rx_errors++;
- ucce &= ~ucce;
+ if (ucce & UCCE_TXE) {
+ ugeth->stats.tx_errors++;
}
}
- while (ucce);
return IRQ_HANDLED;
}
@@ -3677,7 +3697,6 @@ static int ucc_geth_open(struct net_device *dev)
phy_start(ugeth->phydev);
-#ifndef CONFIG_UGETH_NAPI
err =
request_irq(ugeth->ug_info->uf_info.irq, ucc_geth_irq_handler, 0,
"UCC Geth", dev);
@@ -3687,7 +3706,6 @@ static int ucc_geth_open(struct net_device *dev)
ucc_geth_stop(ugeth);
return err;
}
-#endif /* CONFIG_UGETH_NAPI */
err = ugeth_enable(ugeth, COMM_DIR_RX_AND_TX);
if (err) {
diff --git a/drivers/net/ucc_geth.h b/drivers/net/ucc_geth.h
index 6e97c20..7cf3dbc 100644
--- a/drivers/net/ucc_geth.h
+++ b/drivers/net/ucc_geth.h
@@ -205,6 +205,9 @@ struct ucc_geth {
#define UCCE_OTHER (UCCE_SCAR | UCCE_GRA | UCCE_CBPR | UCCE_BSY |\
UCCE_RXC | UCCE_TXC | UCCE_TXE)
+#define UCCE_RX_EVENTS (UCCE_RXF | UCCE_BSY)
+#define UCCE_TX_EVENTS (UCCE_TXB | UCCE_TXE)
+
/* UCC GETH UPSMR (Protocol Specific Mode Register) */
#define UPSMR_ECM 0x04000000 /* Enable CAM
Miss or
--
1.5.0.1
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists