[PATCH] ADS5121 Fix: put dummy byte to enable fixing EOF bug to first place in order not to break some drivers\nUse ALARM interrupt to avoid waiting for data to come in

Subsystems: spi subsystem, the rest

STALE6444d

7 messages, 3 authors, 2008-12-11 · open the first message on its own page

[PATCH] ADS5121 Fix: put dummy byte to enable fixing EOF bug to first place in order not to break some drivers\nUse ALARM interrupt to avoid waiting for data to come in

From: Matteo Fortini <hidden>
Date: 2008-12-10 18:50:52

 From d500e922b750a2bea554d32d8f12937f4da9c80a Mon Sep 17 00:00:00 2001
From: Matteo Fortini <redacted>
Date: Wed, 10 Dec 2008 19:33:16 +0100
Subject: [PATCH] Fix: put dummy byte to enable fixing EOF bug to first 
place in order not to break some drivers
 Use ALARM interrupt to avoid waiting for data to come in


Signed-off-by: Matteo Fortini <redacted>
---
 drivers/spi/mpc512x_psc_spi.c |   53 
++++++++++++++++++++++++-----------------
 1 files changed, 31 insertions(+), 22 deletions(-)
diff --git a/drivers/spi/mpc512x_psc_spi.c b/drivers/spi/mpc512x_psc_spi.c
index 1fd7ad4..41faa49 100644
--- a/drivers/spi/mpc512x_psc_spi.c
+++ b/drivers/spi/mpc512x_psc_spi.c
@@ -136,6 +136,7 @@ static void mpc512x_psc_spi_deactivate_cs(struct 
spi_device *spi)
                    (spi->mode & SPI_CS_HIGH) ? 1 : 0);
 }
 
+
 /*
  * Current MPC5121's have a bug in the SS logic that requires setting
  * the EOF flag on the next to last byte instead of the last
@@ -155,7 +156,7 @@ static int mpc512x_psc_spi_transfer_rxtx(struct 
spi_device *spi,
 
     if (!tx_buf && !rx_buf && t->len)
         return -EINVAL;
-
+    
     /*
      * zero out Mode register 2
      * From the ref man:
@@ -175,6 +176,7 @@ static int mpc512x_psc_spi_transfer_rxtx(struct 
spi_device *spi,
         u8 data;
         size_t fifosz;
         int rxcount;
+        int txcount;
 
         /*
          * The number of bytes that can be sent at a time
@@ -183,41 +185,49 @@ static int mpc512x_psc_spi_transfer_rxtx(struct 
spi_device *spi,
         fifosz = MPC512x_PSC_FIFO_SZ(in_be32(&fifo->txsz));
         count = min(fifosz, len);
 
+        txcount = 0;
+        /*
+         * Insert a dummy byte before a message of len 1 to make it at 
least 2 bytes long
+         * to be able to set EOF correctly
+         */
+        if (t->len == 1) {
+            out_8(&fifo->txdata_8, 0);
+            txcount++;
+        }
         for (i = count; i > 0; i--) {
             if (len == EOFBYTE || t->len == 1)
                 setbits32(&fifo->txcmd, MPC512x_PSC_FIFO_EOF);
             data = tx_buf ? *tx_buf++ : 0;
             out_8(&fifo->txdata_8, data);
-            if (t->len == 1)
-                out_8(&fifo->txdata_8, 0);
+            txcount++;
             len--;
         }
 
         INIT_COMPLETION(mps->done);
 
-        /* interrupt on tx fifo empty */
-        out_be32(&fifo->txisr, MPC512x_PSC_FIFO_EMPTY);
-        out_be32(&fifo->tximr, MPC512x_PSC_FIFO_EMPTY);
+        /* Enable FIFO_ALARM interrupts for rx_fifo */
+        out_be32(&fifo->rxalarm, txcount);
+        out_be32(&fifo->rxisr, MPC512x_PSC_FIFO_ALARM);
+        out_be32(&fifo->rximr, MPC512x_PSC_FIFO_ALARM);
 
-        /* enable transmiter/receiver */
-        out_8(&psc->command, MPC52xx_PSC_TX_ENABLE | 
MPC52xx_PSC_RX_ENABLE);
-
-        wait_for_completion(&mps->done);
+        /* Disable tx_fifo interrupts */
+        out_be32(&fifo->txisr, 0xffffffff);
+        out_be32(&fifo->tximr, 0);
 
-        mdelay(1);
+        out_8(&psc->command, MPC52xx_PSC_TX_ENABLE | 
MPC52xx_PSC_RX_ENABLE);
 
-        /* rx fifo should have count bytes in it */
-        rxcount = in_be32(&fifo->rxcnt);
-        if (rxcount != count)
-            mdelay(1);
+        wait_for_completion (&mps->done);
 
         rxcount = in_be32(&fifo->rxcnt);
-        if (rxcount != count && t->len != 1)
+        if (rxcount != txcount)
             printk(KERN_WARNING "expected %d bytes in rx fifo "
-                       "but got %d\n", count, rxcount);
-
+                    "but got %d\n", txcount, rxcount);
         rxcount = min(rxcount, count);
         {
+            /* Throw away possible initial dummy byte */
+            if (t->len == 1) {
+                (void)in_8(&fifo->rxdata_8);
+            }
             for (i = rxcount; i > 0; i--) {
                 data = in_8(&fifo->rxdata_8);
                 if (rx_buf)
@@ -418,10 +428,9 @@ static irqreturn_t mpc512x_psc_spi_isr(int irq, 
void *dev_id)
     struct mpc512x_psc_fifo __iomem *fifo = mps->fifo;
 
     /* clear interrupt and wake up the work queue */
-    if (in_be32(&fifo->txisr)
-        & in_be32(&fifo->tximr) & MPC512x_PSC_FIFO_EMPTY) {
-        out_be32(&fifo->txisr, MPC512x_PSC_FIFO_EMPTY);
-        out_be32(&fifo->tximr, 0);
+    if (in_be32(&fifo->rxisr) & in_be32(&fifo->rximr) & 
MPC512x_PSC_FIFO_ALARM) {
+        out_be32(&fifo->rxisr, MPC512x_PSC_FIFO_ALARM);
+        out_be32(&fifo->rximr, 0);
         complete(&mps->done);
         return IRQ_HANDLED;
     }
-- 
1.5.4.3

Re: [PATCH] ADS5121 Fix: put dummy byte to enable fixing EOF bug to first place in order not to break some drivers\nUse ALARM interrupt to avoid waiting for data to come in

From: Stephen Rothwell <hidden>
Date: 2008-12-10 23:06:50

Hi Matteo,

On Wed, 10 Dec 2008 19:49:58 +0100 Matteo Fortini [off-list ref] wrote:
 From d500e922b750a2bea554d32d8f12937f4da9c80a Mon Sep 17 00:00:00 2001
From: Matteo Fortini <redacted>
Date: Wed, 10 Dec 2008 19:33:16 +0100
Subject: [PATCH] Fix: put dummy byte to enable fixing EOF bug to first 
place in order not to break some drivers
 Use ALARM interrupt to avoid waiting for data to come in
Please don't make unrelated cleanups in patches - it makes it very hard
for the reviewers to do their work.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

Re: [PATCH] ADS5121 Fix: put dummy byte to enable fixing EOF bug to first place in order not to break some drivers\nUse ALARM interrupt to avoid waiting for data to come in

From: Josh Boyer <hidden>
Date: 2008-12-11 14:35:27

On Wed, 10 Dec 2008 19:49:58 +0100
Matteo Fortini [off-list ref] wrote:
 From d500e922b750a2bea554d32d8f12937f4da9c80a Mon Sep 17 00:00:00 2001
From: Matteo Fortini <redacted>
Date: Wed, 10 Dec 2008 19:33:16 +0100
Subject: [PATCH] Fix: put dummy byte to enable fixing EOF bug to first 
place in order not to break some drivers
 Use ALARM interrupt to avoid waiting for data to come in


Signed-off-by: Matteo Fortini <redacted>
---
 drivers/spi/mpc512x_psc_spi.c |   53 
++++++++++++++++++++++++-----------------
 1 files changed, 31 insertions(+), 22 deletions(-)
This patch is whitespace corrupted.

josh

[RESEND] [PATCH] ADS5121 Fix: put dummy byte to enable fixing EOF bug to first place in order not to break some drivers\nUse ALARM interrupt to avoid waiting for data to come in

From: Matteo Fortini <hidden>
Date: 2008-12-11 18:40:14

This patch is to avoid breaking some drivers, in my case the ADS7846 
touchscreen one, which use 1 char messages.
If you put the dummy byte after the 1 char message, you get part of the 
answer to the message in the rxbuf of the message, which is thrown away.

The solution is to put the dummy byte before the message, so that the 
slave doesn't respond.

It also optimizes the interrupt handling, by using the alarm function of 
the FIFO, to wait until the rx FIFO has received enough bytes, instead 
of waiting until the tx FIFO is empty.



Signed-off-by: Matteo Fortini <redacted>
---
 drivers/spi/mpc512x_psc_spi.c |   50 
+++++++++++++++++++++++-----------------
 1 files changed, 29 insertions(+), 21 deletions(-)
diff --git a/drivers/spi/mpc512x_psc_spi.c b/drivers/spi/mpc512x_psc_spi.c
index 1fd7ad4..eace796 100644
--- a/drivers/spi/mpc512x_psc_spi.c
+++ b/drivers/spi/mpc512x_psc_spi.c
@@ -175,6 +175,7 @@ static int mpc512x_psc_spi_transfer_rxtx(struct 
spi_device *spi,
         u8 data;
         size_t fifosz;
         int rxcount;
+        int txcount;
 
         /*
          * The number of bytes that can be sent at a time
@@ -183,41 +184,49 @@ static int mpc512x_psc_spi_transfer_rxtx(struct 
spi_device *spi,
         fifosz = MPC512x_PSC_FIFO_SZ(in_be32(&fifo->txsz));
         count = min(fifosz, len);
 
+        txcount = 0;
+        /*
+         * Insert a dummy byte before a message of len 1 to make it at 
least 2 bytes long
+         * to be able to set EOF correctly
+         */
+        if (t->len == 1) {
+            out_8(&fifo->txdata_8, 0);
+            txcount++;
+        }
         for (i = count; i > 0; i--) {
             if (len == EOFBYTE || t->len == 1)
                 setbits32(&fifo->txcmd, MPC512x_PSC_FIFO_EOF);
             data = tx_buf ? *tx_buf++ : 0;
             out_8(&fifo->txdata_8, data);
-            if (t->len == 1)
-                out_8(&fifo->txdata_8, 0);
+            txcount++;
             len--;
         }
 
         INIT_COMPLETION(mps->done);
 
-        /* interrupt on tx fifo empty */
-        out_be32(&fifo->txisr, MPC512x_PSC_FIFO_EMPTY);
-        out_be32(&fifo->tximr, MPC512x_PSC_FIFO_EMPTY);
-
-        /* enable transmiter/receiver */
-        out_8(&psc->command, MPC52xx_PSC_TX_ENABLE | 
MPC52xx_PSC_RX_ENABLE);
+        /* Enable FIFO_ALARM interrupts for rx_fifo */
+        out_be32(&fifo->rxalarm, txcount);
+        out_be32(&fifo->rxisr, MPC512x_PSC_FIFO_ALARM);
+        out_be32(&fifo->rximr, MPC512x_PSC_FIFO_ALARM);
 
-        wait_for_completion(&mps->done);
+        /* Disable tx_fifo interrupts */
+        out_be32(&fifo->txisr, 0xffffffff);
+        out_be32(&fifo->tximr, 0);
 
-        mdelay(1);
+        out_8(&psc->command, MPC52xx_PSC_TX_ENABLE | 
MPC52xx_PSC_RX_ENABLE);
 
-        /* rx fifo should have count bytes in it */
-        rxcount = in_be32(&fifo->rxcnt);
-        if (rxcount != count)
-            mdelay(1);
+        wait_for_completion (&mps->done);
 
         rxcount = in_be32(&fifo->rxcnt);
-        if (rxcount != count && t->len != 1)
+        if (rxcount != txcount)
             printk(KERN_WARNING "expected %d bytes in rx fifo "
-                       "but got %d\n", count, rxcount);
-
+                "but got %d\n", txcount, rxcount);
         rxcount = min(rxcount, count);
         {
+            /* Throw away possible initial dummy byte */
+            if (t->len == 1) {
+                (void)in_8(&fifo->rxdata_8);
+            }
             for (i = rxcount; i > 0; i--) {
                 data = in_8(&fifo->rxdata_8);
                 if (rx_buf)
@@ -418,10 +427,9 @@ static irqreturn_t mpc512x_psc_spi_isr(int irq, 
void *dev_id)
     struct mpc512x_psc_fifo __iomem *fifo = mps->fifo;
 
     /* clear interrupt and wake up the work queue */
-    if (in_be32(&fifo->txisr)
-        & in_be32(&fifo->tximr) & MPC512x_PSC_FIFO_EMPTY) {
-        out_be32(&fifo->txisr, MPC512x_PSC_FIFO_EMPTY);
-        out_be32(&fifo->tximr, 0);
+    if (in_be32(&fifo->rxisr) & in_be32(&fifo->rximr) & 
MPC512x_PSC_FIFO_ALARM) {
+        out_be32(&fifo->rxisr, MPC512x_PSC_FIFO_ALARM);
+        out_be32(&fifo->rximr, 0);
         complete(&mps->done);
         return IRQ_HANDLED;
     }
-- 
1.5.4.3

Re: [RESEND] [PATCH] ADS5121 Fix: put dummy byte to enable fixing EOF bug to first place in order not to break some drivers\nUse ALARM interrupt to avoid waiting for data to come in

From: Josh Boyer <hidden>
Date: 2008-12-11 18:46:45

On Thu, 11 Dec 2008 19:39:21 +0100
Matteo Fortini [off-list ref] wrote:
This patch is to avoid breaking some drivers, in my case the ADS7846 
touchscreen one, which use 1 char messages.
If you put the dummy byte after the 1 char message, you get part of the 
answer to the message in the rxbuf of the message, which is thrown away.

The solution is to put the dummy byte before the message, so that the 
slave doesn't respond.

It also optimizes the interrupt handling, by using the alarm function of 
the FIFO, to wait until the rx FIFO has received enough bytes, instead 
of waiting until the tx FIFO is empty.



Signed-off-by: Matteo Fortini <redacted>
---
 drivers/spi/mpc512x_psc_spi.c |   50 
+++++++++++++++++++++++-----------------
 1 files changed, 29 insertions(+), 21 deletions(-)
This still seems to be whitespace corrupted.  Also, that file doesn't
even exist in any upstream kernel I can find.  What tree did you diff
this from?

josh

Re: [RESEND] [PATCH] ADS5121 Fix: put dummy byte to enable fixing EOF bug to first place in order not to break some drivers\nUse ALARM interrupt to avoid waiting for data to come in

From: Josh Boyer <hidden>
Date: 2008-12-11 22:06:41

On Thu, 11 Dec 2008 22:57:56 +0100
Matteo Fortini [off-list ref] wrote:
I rechecked it and I should have taken away all the only-whitespace changes.
No, I mean the patch is corrupted.  Your mailer has taken tabs and
turned them into spaces.  Thunderbird sucks.
The diffs are from the ads5121 branch from Denx.
OK.  You should send this to Denx then.

josh

Re: [RESEND] [PATCH] ADS5121 Fix: put dummy byte to enable fixing EOF bug to first place in order not to break some drivers\nUse ALARM interrupt to avoid waiting for data to come in

From: Matteo Fortini <hidden>
Date: 2008-12-11 22:07:20

I rechecked it and I should have taken away all the only-whitespace changes.

The diffs are from the ads5121 branch from Denx.

Regards,
M


Josh Boyer ha scritto:
On Thu, 11 Dec 2008 19:39:21 +0100
Matteo Fortini [off-list ref] wrote:

  
quoted
This patch is to avoid breaking some drivers, in my case the ADS7846 
touchscreen one, which use 1 char messages.
If you put the dummy byte after the 1 char message, you get part of the 
answer to the message in the rxbuf of the message, which is thrown away.

The solution is to put the dummy byte before the message, so that the 
slave doesn't respond.

It also optimizes the interrupt handling, by using the alarm function of 
the FIFO, to wait until the rx FIFO has received enough bytes, instead 
of waiting until the tx FIFO is empty.



Signed-off-by: Matteo Fortini <redacted>
---
 drivers/spi/mpc512x_psc_spi.c |   50 
+++++++++++++++++++++++-----------------
 1 files changed, 29 insertions(+), 21 deletions(-)
    
This still seems to be whitespace corrupted.  Also, that file doesn't
even exist in any upstream kernel I can find.  What tree did you diff
this from?

josh
_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@ozlabs.org
https://ozlabs.org/mailman/listinfo/linuxppc-dev

  
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help