Long boot delay "Mounting Root Filesystem"

[Alf-Ivar Holm]

David Abrahams <dave at boost-consulting.com> writes:

> I just upgraded the kernel of my Thinkpad T60p from linux-386 to
> linux-686-smp, and suddenly there is a long delay at boot in the
> "Mounting Root Filesystem" step.  The root filesystem is an LVM
> logical volume on an SATA drive.  If I go back to the 386 kernel (same
> version) everything is fine -- well it takes 4-5 seconds to wake up at
> that step, which could be a manifestation of the same problem, but it
> still seems tolerably fast.
>
> With linux-686-smp, the disk sees a couple short bursts of activity
> during the beginning of this period, but nothing really proceeds for a
> good 40 seconds--sometimes longer--and usually not unless I press some
> keys on my keyboard and add some trackpad/button activity. Eventually
> it will wake up and continue if I give it physical stimulus, but it is
> really annoying to have to endure such a slow boot process just to
> make my machine run faster after it boots ;-)

I have similar problems with my Fujitsu Siemens S7020D with a
linux-686 (non-SMP) kernel, but have not tried much of the stimuli
approach, I have rather tried the nervous stare approach, where
according to my /var/log/messages* it seems quite constant difference
the last boots at what seems like 160 seconds, if my interpretation of
the time stamp column is correct (this is from last boot, which can be
seen with dmesg(8) as well):

[17179570.460000] pnp: PnP ACPI: found 14 devices
[17179570.460000] PnPBIOS: Disabled by ACPI PNP
[17179570.460000] PCI: Using ACPI for IRQ routing
[17179570.460000] PCI: If a device doesn't work, try "pci=routeirq".  If it helps, post a report
[17179570.460000] PCI: Cannot allocate resource region 7 of bridge 0000:00:1c.1
[17179570.460000] PCI: Cannot allocate resource region 8 of bridge 0000:00:1c.1
[17179570.460000] PCI: Cannot allocate resource region 9 of bridge 0000:00:1c.1
[17179570.476000] PCI: Ignore bogus resource 6 [0:0] of 0000:00:02.0

Where the last line turns up after the long wait, and 476000 - 460000
= 16000 some unit.  I get the same output if you switch to console 1
during the boot process, where the last line arrives after a lot of
staring.

I haven't had the time to do any further exploration of the issue, but
I don't remember that it was like this at an earlier Dapper stage, but
it may have been after I switched to the 686 kernel, which I had
accidentally forgot to do.  (I've also forgot to extend my log
rotation scheme, so I've lost that date.)  I have added another 1GB
RAM after that so I'm not going back to 386.

Sorry about not providing any help, just an AOL.  One from the hip
though: you could try with the non-SMP version, and see if that
narrows it down to a SMP problem or a 686 problem.

	Affi