Hello, I'm new here, please don't shot me up to Mars ;-)
I use the 2.2.16 Kernel with one HD (now IBM DTLA307030 30,7GB HD) on primary HDC
and an ordinary IDE-CD-ROM on secondary HDC;
DMA is disabled and Enhanced IDE is compiled into kernel...
After a short time the kernel display the message
hda: lost interrupt
and after a reboot the disk isn't accessible (the pc has to be switched off!);
When i use an IBM DJNA-352500 (25GB) everything is fine, so i think there could be a big
problem in handling large disks... (no matter if i use LBA or Normal in BIOS!)
The same problem came up with a WD 307AA (30,7 GB, too!), so i think we can't blame the disks!
Additional Patches to kernel:
1) utime-Patch, the kernel has now an on demand precision of 1 usec!
jiffies aren't changed and behave as before (I can't take it back, because our application needs it!)
2) raw-device-patch, I checked this one and it doesn't affect the IDE-stuff (it works on top of it)
Here's the Hardware list:
MB: Tyan Titan Turbo Intel TX-chipset (BAT-form)
CPU: 233 MHz Pentium MMX (SINGLE)
RAM: 128 MB SDRAM
CDROM: Goldstar ATAPI (on secondary IDE-Port /dev/hdc)
NIC: 3c509B (PNP ISA)
Framegrabber: Zoran ZR36057 PCI (german production)
additional 4 port ISA serial adapter card (interrupts not enabled)
ATI AGP-Card 8MB (no interrupts)
Harddisks (both with the mentioned error):
a) IBM Deskstar DTLA-307030 (LBA : 3737 / 255 / 63 / 30,7 GB)
b) Western Digital WD307AA (LBA : 3739 / 255 / 63 / 30,7 GB)
The system is VERY heavy loaded and we write and read directly to the device;
but the error occurs as well with the /dev/raw as with the /dev/hda device
With the /dev/hda and the old ide interface driver the virtual memory stuff runs out of memory
and isn't able to reset itself (kernel OOPS!). I'm sorry to say that the buffer cache is
useless for heavy loaded machines with many disk transfers per second.
Perhaps somebody could fix this, but i'm now interested in E-IDE-fix.
Perhaps some sleeps in (E)IDE-code are based on the 10ms scheduler of "normal"
Linux on Pentium machines...
The ide-driver in kernel 2.0.36 worked without problems with the utime-patch!
But the same software configuration works fine with an IBM DJNA-352500 (25GB) so i can't
imagine that our software does something wrong.
We use a combination of llseek, read and write operations to get the data to or from the disk.
Someone an idea? Or something i could try to get closer to this problem?
Regards
Carsten
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Aug 15 2000 - 21:00:22 EST