IDE-Problems 1.3.8x-9x

a9002123@unet.univie.ac.at
Sat, 4 May 1996 12:51:22 +0200 (MET DST)


During the last weeks, for the first time I tried to use newer
kernels, namely .83-.91. There seems to be a problem with my second
IDE card: Until version .91 (approximately), the kernel sometimes failed to
correctly recognize the drives connected to the second adapter;
instead of reporting

Apr 19 02:03:23 cgvie kernel: hda: WDC AC2540H, 515MB w/128kB Cache, LBA, CHS=10
48/16/63
Apr 19 02:03:23 cgvie kernel: ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
Apr 19 02:03:23 cgvie kernel: hdc: FX001DE, ATAPI CDROM drive
Apr 19 02:03:23 cgvie kernel: hdd: ST3391A-XR, 325MB w/120kB Cache, CHS=768/14/6
2
Apr 19 02:03:23 cgvie kernel: ide1 at 0x170-0x177,0x376 on irq 15

it claimed something like this (Linux 1.3.89):

Apr 17 00:32:24 cgvie kernel: hda: WDC AC2540H, 515MB w/128kB Cache, LBA, CHS=10
48/16/63
Apr 17 00:32:24 cgvie kernel: ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
Apr 17 00:32:24 cgvie kernel: hdc: FX001DE, ATAPI CDROM drive
Apr 17 00:32:24 cgvie kernel: hdd: SP3^Q9^QA^AXP, 325MB w/40kB Cache, CHS=768/14
/62
Apr 17 00:32:24 cgvie kernel: ide1 at 0x170-0x177,0x376 on irq 15

The latest version I tried didn't recognize the two drives on ide1 at
all; on the other hand, kernel 1.2.13 (which I usually use) always
works, even with quite aggressive hdparm-settings.

The first IDE card (which always seems to work) is a VLB multi-I/O card
with a Winbond chipset (btw.: the IDE driver seems not to offer special
Winbond support - is ist ok if I do not set any option for this
chipset?). The second IDE card is a standard ISA multi-I/O card;
there is a "UMC UM82C865" chip. Of course, this chip isn't mentioned
in the documentation either.

Since this strange behaviour seems to be the reason of a file system
corruption on Apr 24 ("Apr 24 22:54:05 cgvie kernel: EXT2-fs error
(device 22/65): ext2_check_descriptors: Block bitmap for group 0 not in
group (block 0)!"), I am now afraid to use any 1.3-kernel as long as
I do not know the reason of the problem.

BTW, noautotune for hdc and hdd had no noticeable influence on the
number of failures in recognizing hdd.

Greetings,

Christian

-- 
Christian Gottschall       Gebraucht der Zeit, sie geht so schnell von
Kleine Pfarrgasse 5/5   hinnen, doch Ordnung lehrt euch Zeit gewinnen.
A-1020 Wien                     Mein teurer Freund, ich rat euch drum,
Tel./Fax (+43-1) 214-47-39                   zuerst Collegium Logicum!