On Sat, Nov 22, 2003 at 06:40:48PM +0100, Jean Delvare wrote:
>
> > > This is the very same as Martin's (except that he seems to have USB
> > > disabled), so if there is a way to differenciate the two of them,
> > > it's not by using the PCI data. I think we'll have to disable the
> > > workaround by default. We might consider adding a module parameter
> > > to reenable it. If you can provide another way to automatically
> > > detect when that workaround is needed, we would implement it.
> >
> > What are the contents of the register on his board?
>
> How would I know? ;) I don't even know how to see this. Please explain,
> so that Martin can answer.
Okay, here's a patch against the current lm_sensors2 module that should
be able to get me started on this:
Index: i2c-piix4.c
===================================================================
RCS file: /home/cvs/lm_sensors2/kernel/busses/i2c-piix4.c,v
retrieving revision 1.59
diff -u -u -r1.59 i2c-piix4.c
--- i2c-piix4.c 21 Nov 2003 22:57:38 -0000 1.59
+++ i2c-piix4.c 26 Nov 2003 17:13:07 -0000
@@ -100,7 +100,7 @@
"Forcibly enable the PIIX4 at the given address. "
"EXTREMELY DANGEROUS!");
-static int fix_hstcfg = 0;
+static int fix_hstcfg = 1;
MODULE_PARM(fix_hstcfg, "i");
MODULE_PARM_DESC(fix_hstcfg,
"Fix config register. Needed on some boards (Force CPCI735).");
@@ -143,7 +143,6 @@
static int piix4_setup(struct pci_dev *PIIX4_dev, const struct pci_device_id *id)
{
int error_return = 0;
- unsigned char temp;
/* match up the function */
if (PCI_FUNC(PIIX4_dev->devfn) != id->driver_data)
@@ -182,15 +181,23 @@
error_return = -ENODEV;
goto END;
}
-
- pci_read_config_byte(PIIX4_dev, SMBHSTCFG, &temp);
-
+
/* Some BIOS will set up the chipset incorrectly and leave a register
- in an undefined state (causing I2C to act very strangely). */
- if (fix_hstcfg && (temp & 0x02)) {
- printk("Worked around buggy BIOS (I2C)\n");
- temp &= 0xfd;
- pci_write_config_byte(PIIX4_dev, SMBHSTCFG, temp);
+ in an undefined state (causing I2C to act very strangely). Others become
+ unhappy if you change this. */
+ if (fix_hstcfg) {
+ unsigned char temp;
+
+ pci_read_config_byte(PIIX4_dev, SMBHSTCFG, &temp);
+ if (temp & 0x02) {
+#if 0
+ printk("Worked around buggy BIOS (I2C)\n");
+ temp &= 0xfd;
+ pci_write_config_byte(PIIX4_dev, SMBHSTCFG, temp);
+#else
+ printk("temp: %x\n", temp);
+#endif
+ }
}
/* If force_addr is set, we program the new address here. Just to make
--
Tom Rini
http://gate.crashing.org/~trini/
