2.4.20: ideN=... option stopped working

[Richard Kettlewell]

I have an inherited ancient PC with a weird IDE configuration that
requires ide1=0x170,0x376,14 specified on the kernel command line in
order to boot.  This worked fine in 2.4.18 but not in 2.4.20.  I
haven't checked 2.4.19.

After some investigation I found the relevant change and tried
reverting it.  The resulting kernel works for me.  Here's the diff:

========================================================================
--- drivers/ide/ide.c~	Thu Nov 28 23:53:13 2002
+++ drivers/ide/ide.c	Sat Dec 21 18:25:59 2002
@@ -2456,7 +2456,6 @@
 	memcpy(hwif->io_ports, hwif->hw.io_ports, sizeof(hwif->hw.io_ports));
 	hwif->irq = hw->irq;
 	hwif->noprobe = 0;
-	hwif->chipset = hw->chipset;
 
 	if (!initializing) {
 		ide_probe_module();
========================================================================

During my investigation I added some printk calls which make it
clearer what ide_setup objects to when it jumps to bad_option.
Perhaps these would be of more general use, diff below.

ttfn/rjk

========================================================================
--- drivers/ide/ide.c~	Thu Nov 28 23:53:13 2002
+++ drivers/ide/ide.c	Sat Dec 21 18:25:59 2002
@@ -3427,12 +3426,16 @@
 		 * Cryptic check to ensure chipset not already set for hwif:
 		 */
 		if (i > 0 || i <= -11) {			/* is parameter a chipset name? */
-			if (hwif->chipset != ide_unknown)
+			if (hwif->chipset != ide_unknown) {
+				printk(" -- chipset already specified");
 				goto bad_option;	/* chipset already specified */
+			}
 			if (i <= -11 && i != -18 && hw != 0)
 				goto bad_hwif;		/* chipset drivers are for "ide0=" only */
-			if (i <= -11 && i != -18 && ide_hwifs[hw+1].chipset != ide_unknown)
+			if (i <= -11 && i != -18 && ide_hwifs[hw+1].chipset != ide_unknown) {
+				printk(" -- chipset for 2nd port already specified");
 				goto bad_option;	/* chipset for 2nd port already specified */
+			}
 			printk("\n");
 		}