More potential problems:
When the ATA device is reset via the IDE control reg, the Winbond doc.
says that the controller is also reset to 8 bits fixed-timing i/os.
I've added some "hackish" code to the do_reset1 function in
drivers/block/ide.c that looks like this:
#if OK_TO_RESET_CONTROLLER
if (!IDE_CONTROL_REG) {
__restore_flags(flags);
return;
}
/*
* Note that we also set nIEN while resetting the device,
* to mask unwanted interrupts from the interface during the reset.
* However, due to the design of PC hardware, this will cause an
* immediate interrupt due to the edge transition it produces.
* This single interrupt gives us a "fast poll" for drives that
* recover from reset very quickly, saving us the first 50ms wait time.
*/
OUT_BYTE(drive->ctl|6,IDE_CONTROL_REG); /* set SRST and nIEN */
udelay(10); /* more than enough time */
OUT_BYTE(drive->ctl|2,IDE_CONTROL_REG); /* clear SRST, leave nIEN */
udelay(10); /* more than enough time */
hwgroup->poll_timeout = jiffies + WAIT_WORSTCASE;
ide_set_handler (drive, &reset_pollfunc, HZ/20);
+#ifdef CONFIG_BLK_DEV_SL82C105
+ /* Winbond need a reset of the controller or secondary port will be
+ set to 8 bits i/o which is definitely not what we want !
+ */
+ if (hwif->pci_devid == DEVID_W82C105)
+ ide_init_sl82c105(hwif);
+#endif
#endif /* OK_TO_RESET_CONTROLLER */
__restore_flags (flags); /* local CPU only */
}
The ide_init_sl82c105() must be previously declared extern, and this
works since ide_init_sl82c105 currently does the magic of writing 0x33 in
the config register at 0x40.
I'm still having problems with a disk which is very slow to power up and
doesn't show up at all if I don't wait around 20-30secs before letting
the bios scan the IDE (a caviar). I'm trying to figure out a way to
detect that there's indeed a disk which is not ready.
--
E-Mail: <mailto:[EMAIL PROTECTED]>
BenH. Web : <http://calvaweb.calvacom.fr/bh40/>
unsubscribe: body of `unsubscribe linux-arm' to [EMAIL PROTECTED]
