Patchwork [v2,1/1] drivers: net: davinci_cpdma: acknowledge interrupt properly

login
register
mail settings
Submitter Mugunthan V N
Date Feb. 14, 2013, 6:26 p.m.
Message ID <1360866406-17261-1-git-send-email-mugunthanvnm@ti.com>
Download mbox | patch
Permalink /patch/220478/
State Changes Requested
Delegated to: David Miller
Headers show

Comments

Mugunthan V N - Feb. 14, 2013, 6:26 p.m.
CPDMA interrupts are not properly acknowledged which leads to interrupt
storm, only cpdma interrupt 0 is acknowledged in Davinci CPDMA driver.
Changed cpdma_ctlr_eoi api to acknowledge 1 and 2 interrupts which are
used for rx and tx respectively.

Reported-by: Pantelis Antoniou <panto@antoniou-consulting.com>
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
---
Changes from Initial Version:
* Changed Tx and Rx EOI seperated so that Tx interrupt will not be cleared
  without processing any txpackets so that none of the interrupt are missed
  or cleared without processing.

 drivers/net/ethernet/ti/cpsw.c          |   25 ++++++++++++++++---------
 drivers/net/ethernet/ti/davinci_cpdma.c |    4 ++--
 drivers/net/ethernet/ti/davinci_cpdma.h |    9 ++++++++-
 3 files changed, 26 insertions(+), 12 deletions(-)
David Miller - Feb. 15, 2013, 8:05 p.m.
From: Mugunthan V N <mugunthanvnm@ti.com>
Date: Thu, 14 Feb 2013 23:56:46 +0530

> +enum {
> +	CPDMA_EOI_RX_THRESH = 0,
> +	CPDMA_EOI_RX,
> +	CPDMA_EOI_TX,
> +	CPDMA_EOI_MISC,
> +};

Do not use enumerations for hardware register values, which must be
exact, otherwise you are potentially going to succumb to the vagaries
of C language enumeration value assignment.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller - Feb. 15, 2013, 8:05 p.m.
Also please indicate, clearly, what tree your patch is targetted at.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Joe Perches - Feb. 15, 2013, 8:18 p.m.
On Fri, 2013-02-15 at 15:05 -0500, David Miller wrote:
> From: Mugunthan V N <mugunthanvnm@ti.com>
> Date: Thu, 14 Feb 2013 23:56:46 +0530
> 
> > +enum {
> > +     CPDMA_EOI_RX_THRESH = 0,
> > +     CPDMA_EOI_RX,
> > +     CPDMA_EOI_TX,
> > +     CPDMA_EOI_MISC,
> > +};
> 
> Do not use enumerations for hardware register values, which must be
> exact, otherwise you are potentially going to succumb to the vagaries
> of C language enumeration value assignment.

Vagaries?

In what way is c enumeration (6.7.2.2) vague?
char vs int?  Isn't smaller mostly better?

Concern about possible future reordering given an
insertion or deletion might be the only consideration
I could imagine.


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller - Feb. 15, 2013, 8:20 p.m.
From: Joe Perches <joe@perches.com>
Date: Fri, 15 Feb 2013 12:18:59 -0800

> On Fri, 2013-02-15 at 15:05 -0500, David Miller wrote:
>> From: Mugunthan V N <mugunthanvnm@ti.com>
>> Date: Thu, 14 Feb 2013 23:56:46 +0530
>> 
>> > +enum {
>> > +     CPDMA_EOI_RX_THRESH = 0,
>> > +     CPDMA_EOI_RX,
>> > +     CPDMA_EOI_TX,
>> > +     CPDMA_EOI_MISC,
>> > +};
>> 
>> Do not use enumerations for hardware register values, which must be
>> exact, otherwise you are potentially going to succumb to the vagaries
>> of C language enumeration value assignment.
> 
> Vagaries?
> 
> In what way is c enumeration (6.7.2.2) vague?
> char vs int?  Isn't smaller mostly better?
> 
> Concern about possible future reordering given an
> insertion or deletion might be the only consideration
> I could imagine.

Right.  That's why you should use explicit CPP defines for
register offsets and values.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Mugunthan V N - Feb. 17, 2013, 9:34 a.m.
On 2/16/2013 1:35 AM, David Miller wrote:
> From: Mugunthan V N <mugunthanvnm@ti.com>
> Date: Thu, 14 Feb 2013 23:56:46 +0530
>
>> +enum {
>> +	CPDMA_EOI_RX_THRESH = 0,
>> +	CPDMA_EOI_RX,
>> +	CPDMA_EOI_TX,
>> +	CPDMA_EOI_MISC,
>> +};
> Do not use enumerations for hardware register values, which must be
> exact, otherwise you are potentially going to succumb to the vagaries
> of C language enumeration value assignment.
Will change enum to define and submit next version patch.

Regards
Mugunthan V N
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Mugunthan V N - Feb. 17, 2013, 9:35 a.m.
On 2/16/2013 1:35 AM, David Miller wrote:
> Also please indicate, clearly, what tree your patch is targetted at.
Sorry for not mentioning the tree, will mention in the future patches.

Regards
Mugunthan V N
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/drivers/net/ethernet/ti/cpsw.c b/drivers/net/ethernet/ti/cpsw.c
index 4ceed6e..7e93df6 100644
--- a/drivers/net/ethernet/ti/cpsw.c
+++ b/drivers/net/ethernet/ti/cpsw.c
@@ -510,19 +510,21 @@  static int cpsw_poll(struct napi_struct *napi, int budget)
 	int			num_tx, num_rx;
 
 	num_tx = cpdma_chan_process(priv->txch, 128);
-	num_rx = cpdma_chan_process(priv->rxch, budget);
-
-	if (num_rx || num_tx)
-		cpsw_dbg(priv, intr, "poll %d rx, %d tx pkts\n",
-			 num_rx, num_tx);
+	if (num_tx)
+		cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_TX);
 
+	num_rx = cpdma_chan_process(priv->rxch, budget);
 	if (num_rx < budget) {
 		napi_complete(napi);
 		cpsw_intr_enable(priv);
-		cpdma_ctlr_eoi(priv->dma);
+		cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_RX);
 		cpsw_enable_irq(priv);
 	}
 
+	if (num_rx || num_tx)
+		cpsw_dbg(priv, intr, "poll %d rx, %d tx pkts\n",
+			 num_rx, num_tx);
+
 	return num_rx;
 }
 
@@ -835,7 +837,8 @@  static int cpsw_ndo_open(struct net_device *ndev)
 	cpdma_ctlr_start(priv->dma);
 	cpsw_intr_enable(priv);
 	napi_enable(&priv->napi);
-	cpdma_ctlr_eoi(priv->dma);
+	cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_RX);
+	cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_TX);
 
 	if (priv->data.dual_emac)
 		priv->slaves[priv->emac_port].open_stat = true;
@@ -1075,7 +1078,9 @@  static void cpsw_ndo_tx_timeout(struct net_device *ndev)
 	cpdma_chan_start(priv->txch);
 	cpdma_ctlr_int_ctrl(priv->dma, true);
 	cpsw_intr_enable(priv);
-	cpdma_ctlr_eoi(priv->dma);
+	cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_RX);
+	cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_TX);
+
 }
 
 static struct net_device_stats *cpsw_ndo_get_stats(struct net_device *ndev)
@@ -1094,7 +1099,9 @@  static void cpsw_ndo_poll_controller(struct net_device *ndev)
 	cpsw_interrupt(ndev->irq, priv);
 	cpdma_ctlr_int_ctrl(priv->dma, true);
 	cpsw_intr_enable(priv);
-	cpdma_ctlr_eoi(priv->dma);
+	cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_RX);
+	cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_TX);
+
 }
 #endif
 
diff --git a/drivers/net/ethernet/ti/davinci_cpdma.c b/drivers/net/ethernet/ti/davinci_cpdma.c
index 7d3bffd..68c3418 100644
--- a/drivers/net/ethernet/ti/davinci_cpdma.c
+++ b/drivers/net/ethernet/ti/davinci_cpdma.c
@@ -493,9 +493,9 @@  int cpdma_ctlr_int_ctrl(struct cpdma_ctlr *ctlr, bool enable)
 	return 0;
 }
 
-void cpdma_ctlr_eoi(struct cpdma_ctlr *ctlr)
+void cpdma_ctlr_eoi(struct cpdma_ctlr *ctlr, u32 value)
 {
-	dma_reg_write(ctlr, CPDMA_MACEOIVECTOR, 0);
+	dma_reg_write(ctlr, CPDMA_MACEOIVECTOR, value);
 }
 
 struct cpdma_chan *cpdma_chan_create(struct cpdma_ctlr *ctlr, int chan_num,
diff --git a/drivers/net/ethernet/ti/davinci_cpdma.h b/drivers/net/ethernet/ti/davinci_cpdma.h
index a97d6ab..ac65033 100644
--- a/drivers/net/ethernet/ti/davinci_cpdma.h
+++ b/drivers/net/ethernet/ti/davinci_cpdma.h
@@ -63,6 +63,13 @@  struct cpdma_chan_stats {
 	u32			teardown_dequeue;
 };
 
+enum {
+	CPDMA_EOI_RX_THRESH = 0,
+	CPDMA_EOI_RX,
+	CPDMA_EOI_TX,
+	CPDMA_EOI_MISC,
+};
+
 struct cpdma_ctlr;
 struct cpdma_chan;
 
@@ -88,7 +95,7 @@  int cpdma_chan_submit(struct cpdma_chan *chan, void *token, void *data,
 int cpdma_chan_process(struct cpdma_chan *chan, int quota);
 
 int cpdma_ctlr_int_ctrl(struct cpdma_ctlr *ctlr, bool enable);
-void cpdma_ctlr_eoi(struct cpdma_ctlr *ctlr);
+void cpdma_ctlr_eoi(struct cpdma_ctlr *ctlr, u32 value);
 int cpdma_chan_int_ctrl(struct cpdma_chan *chan, bool enable);
 bool cpdma_check_free_tx_desc(struct cpdma_chan *chan);