[lm-sensors] [PATCH 2.6] cleanup of i2c-nforce2

[lm-sensors] [PATCH 2.6] cleanup of i2c-nforce2

[Hans-Frieder Vogt]

Hi list,

attached is a small patch that removes unused code from i2c-nforce2 and 
adds a single debug message. The patch is against 2.6.13-rc3-mm1.
I have tested the patch with 2.6.13-rc3: compiles cleanly and works as 
without the patch (as expected).

Greg,
can you please include the patch into the next kernel release?
Thanks.

Signed-off-by: Hans-Frieder Vogt <hfvogt at arcor.de>

*--- linux-2.6.13-rc3-mm1.orig/drivers/i2c/busses/i2c-nforce2.c	2005-07-18 17:55:06.116509000 +0200*
*+++ linux-2.6.13-rc3-mm1/drivers/i2c/busses/i2c-nforce2.c	2005-07-23 14:37:19.176001249 +0200*
@@ -131,7 +131,6 @@
 	struct nforce2_smbus *smbus = adap->algo_data;
 	unsigned char protocol, pec, temp;
 	unsigned char len = 0; /* to keep the compiler quiet */
-	int timeout = 0;
 	int i;
 
 	protocol = (read_write = I2C_SMBUS_READ) ? NVIDIA_SMB_PRTCL_READ :
@@ -191,29 +190,10 @@
 		case I2C_SMBUS_PROC_CALL:
 			dev_err(&adap->dev, "I2C_SMBUS_PROC_CALL not supported!\n");
 			return -1;
-			/*
-			outb_p(command, NVIDIA_SMB_CMD);
-			outb_p(data->word, NVIDIA_SMB_DATA);
-			outb_p(data->word >> 8, NVIDIA_SMB_DATA + 1);
-			protocol = NVIDIA_SMB_PRTCL_PROC_CALL | pec;
-			read_write = I2C_SMBUS_READ;
-			break;
-			 */
 
 		case I2C_SMBUS_BLOCK_PROC_CALL:
 			dev_err(&adap->dev, "I2C_SMBUS_BLOCK_PROC_CALL not supported!\n");
 			return -1;
-			/*
-			protocol |= pec;
-			len = min_t(u8, data->block[0], 31);
-			outb_p(command, NVIDIA_SMB_CMD);
-			outb_p(len, NVIDIA_SMB_BCNT);
-			for (i = 0; i < len; i++)
-				outb_p(data->block[i + 1], NVIDIA_SMB_DATA + i);
-			protocol = NVIDIA_SMB_PRTCL_BLOCK_PROC_CALL | pec;
-			read_write = I2C_SMBUS_READ;
-			break;
-			*/
 
 		case I2C_SMBUS_WORD_DATA_PEC:
 		case I2C_SMBUS_BLOCK_DATA_PEC:
@@ -232,12 +212,6 @@
 
 	temp = inb_p(NVIDIA_SMB_STS);
 
-#if 0
-	do {
-		i2c_do_pause(1);
-		temp = inb_p(NVIDIA_SMB_STS);
-	} while (((temp & NVIDIA_SMB_STS_DONE) = 0) && (timeout++ < MAX_TIMEOUT));
-#endif
 	if (~temp & NVIDIA_SMB_STS_DONE) {
 		udelay(500);
 		temp = inb_p(NVIDIA_SMB_STS);
@@ -247,9 +221,10 @@
 		temp = inb_p(NVIDIA_SMB_STS);
 	}
 
-	if ((timeout >= MAX_TIMEOUT) || (~temp & NVIDIA_SMB_STS_DONE)
-		|| (temp & NVIDIA_SMB_STS_STATUS))
+	if ((~temp & NVIDIA_SMB_STS_DONE) || (temp & NVIDIA_SMB_STS_STATUS)) {
+		dev_dbg(&adap->dev, "SMBus Timeout! (0x%02x)\n", temp);
 		return -1;
+	}
 
 	if (read_write = I2C_SMBUS_WRITE)
 		return 0;




--
Hans-Frieder Vogt               e-mail: hfvogt <at> arcor .dot. de
                                        hfvogt <at> gmx .dot. net

[lm-sensors] Re: [PATCH 2.6] cleanup of i2c-nforce2

[Jean Delvare]

Hallo Hans-Frieder,

> attached is a small patch that removes unused code from i2c-nforce2
> and adds a single debug message. The patch is against 2.6.13-rc3-mm1.
> I have tested the patch with 2.6.13-rc3: compiles cleanly and works
> as without the patch (as expected).

Thanks for this cleanup patch, I like it. However, I would have a few
comments as I think we could clean up this driver even more.

1* The MAX_TIMEOUT constant doesn't seem to be used anywhere anymore, so
you could rip it.

2* There are a few more lines of code that are commented out:

265		/* case I2C_SMBUS_PROC_CALL: not supported */

270		/* case I2C_SMBUS_BLOCK_PROC_CALL: not supported */

288	    I2C_FUNC_SMBUS_BYTE_DATA | I2C_FUNC_SMBUS_WORD_DATA /* |
289	    I2C_FUNC_SMBUS_BLOCK_DATA */;

Maybe you can rip them away as well (but see point 4*.)

3* This comment should be updated:

126 /* Return -1 on error. See smbus.h for more information */

There is no file named smbus.h.

4* The driver isn't very clear about what SMBus transactions it
supports. nforce2_func reports that only quick, byte, byte data and word
data transfers are supported, but nforce2_access seems to implement
block and i2c block transfers. If the block implementations work, they
should be advertised by nforce2_func. If not, they should be removed.

5* In nforce2_access, you have several cases for dealing with the
various transfer types that the driver doesn't support. As users should
check for advertised functionalities before they call nforce2_access, I
wouldn't bother detailing the errors with different error messages. The
default case should be enough.

May we have a similar patch for the Linux 2.4 version of the driver as
found in the lm_sensors CVS repository?

Thanks,
-- 
Jean Delvare