Martmn Coco wrote:
> Hi there,
>
> We are beginning to do some tests with Compact Flash IDE adapters and
> OpenBSD 3.8.
>
> We installed the OpenBSD 3.8 using a SanDisk 1.0GB CompactFlash on a
> Pentium 4 (dmesg at the end of this message). The installation finished
> flawlessly. But when booting, it seems to take ages to boot. The last
> time we checked, it took about 55 minutes for it to finish booting. Once
> it has booted, all the speed issues seem to disappear.
whoa.
Flash isn't as fast as disk...but..not 55 minutes!
Where is it spending its time?
> We went through the BIOS to find anything related to PIO or DMA, but
> found nothing suitable.
Nah. I run OpenBSD on lots of machines without DMA, boot time is hardly
any different.
> We tried the very same card with a VIA Chipset and it worked like a
> charm, we couldn't tell the difference from booting from a normal HD.
ok, good media, good install. Good test. :)
> Any input on this will be greatly appreciated :)
>
> Thanks,
> Martmn.
>
> I attach the dmesg of the machine that seems to be having problems when
> booting:
>
> OpenBSD 3.8 (GENERIC) #138: Sat Sep 10 15:41:37 MDT 2005
> [EMAIL PROTECTED]:/usr/src/sys/arch/i386/compile/GENERIC
> cpu0: Intel(R) Pentium(R) 4 CPU 2.40GHz ("GenuineIntel" 686-class) 2.42 GHz
...
> pciide0 at pci0 dev 31 function 1 "Intel 82801EB/ER IDE" rev 0x02: DMA,
> channel 0 configured to compatibility, channel 1 configured to compatibility
> pciide0: channel 0 disabled (no drives)
^^^^^^^^^
> wd0 at pciide0 channel 1 drive 0: <SanDisk SDCFB-1024>
> wd0: 1-sector PIO, LBA, 977MB, 2001888 sectors
> wd0(pciide0:1:0): using PIO mode 4, DMA mode 2
...
I see one oddity and another POSSIBLE explanation...
The oddity is you have the flash on the SECOND disk channel. That
should work, but a buggy BIOS might get in the way.
The other POSSIBLE explanation is really a stretch, but it is so good
and explains things so well (fortunately, you didn't give details of
what part of the boot process took the time :), I gotta mention it:
I see you have a P4. Could the heat sink have fallen off/not been
mounted properly? Supposedly, the P4 will slow itself down when it
overheats. IF the heat sink were not on at all (or a tiny air gap
existed), the thing would probably reach critical temp within a couple
seconds of power-on, and slow to an absolute crawl. The kernel is
loaded by the BIOS, so until the kernel was completely loaded. At that
point, OpenBSD would be halting the processor when it was idle, and it
would probably stay cool enough to keep running at respectable speed.
Yeah, that's a wacko explanation, but it fits the facts so far (I think.
I live in a P4-free house, so I can't test this theory). I fixed a P3
machine over the phone that did the P3 version of the same problem
(started to boot, then froze, as P3's hang, rather than go glacial).
Blew a good service call by doing that. :)
Assuming those two ideas are not worth they electrons they were written
on, next test would be to try an ordinary HD in this machine.
Next thing I'd like to see is a running commentary on what's on the
screen at, say, every five or ten minute intervals, so we can get some
idea where the slow-down is, and what is going on in the machine at each
point. Booting is fairly complicated, a combination of ROM, boot
loaders, OS and hardware...lots of places for things to go wrong.
However, never heard of this one before...
