Re: [PATCH] i2c driver fixes for 2.6.0

[Greg KH]

ChangeSet 1.1496.23.2, 2003/12/19 11:37:00-08:00, trini@xxxxxxxxxxxxxxxxxxx

[PATCH] I2C: make i2c-piix4 fix optional

On Thu, Dec 18, 2003 at 10:26:40AM -0800, Greg KH wrote:


 drivers/i2c/busses/i2c-piix4.c |   21 ++++++++++++++++++---
 1 files changed, 18 insertions(+), 3 deletions(-)


diff -Nru a/drivers/i2c/busses/i2c-piix4.c b/drivers/i2c/busses/i2c-piix4.c
--- a/drivers/i2c/busses/i2c-piix4.c	Tue Dec 30 12:29:37 2003
+++ b/drivers/i2c/busses/i2c-piix4.c	Tue Dec 30 12:29:37 2003
@@ -99,6 +99,13 @@
 		 "Forcibly enable the PIIX4 at the given address. "
 		 "EXTREMELY DANGEROUS!");
 
+/* If fix_hstcfg is set to anything different from 0, we reset one of the
+   registers to be a valid value. */
+static int fix_hstcfg = 0;
+MODULE_PARM(fix_hstcfg, "i");
+MODULE_PARM_DESC(fix_hstcfg,
+		"Fix config register. Needed on some boards (Force CPCI735).");
+
 static int piix4_transaction(void);
 
 
@@ -164,9 +171,17 @@
 	/* Some BIOS will set up the chipset incorrectly and leave a register
 	   in an undefined state (causing I2C to act very strangely). */
 	if (temp & 0x02) {
-		dev_info(&PIIX4_dev->dev, "Worked around buggy BIOS (I2C)\n");
-		temp = temp & 0xfd;
-		pci_write_config_byte(PIIX4_dev, SMBHSTCFG, temp);
+		if (fix_hstcfg) {
+			dev_info(&PIIX4_dev->dev, "Working around buggy BIOS "
+					"(I2C)\n");
+			temp &= 0xfd;
+			pci_write_config_byte(PIIX4_dev, SMBHSTCFG, temp);
+		} else {
+			dev_info(&PIIX4_dev->dev, "Unusual config register "
+					"value\n");
+			dev_info(&PIIX4_dev->dev, "Try using fix_hstcfg=1 if "
+					"you experience problems\n");
+		}
 	}
  
 	/* If force_addr is set, we program the new address here. Just to make

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/