On Tue, Nov 21, 2006 at 03:58:53PM -0800, Judith Lebzelter wrote:
> > gsi_irq_sharing(gsi) {
> > BUG_ON(gsi >= NR_IRQ_VECTORS);
> > }
> >
> > Now interesting thing to find out is howcome this gsi became greater than
> > NR_IRQ_VECTORS? Some corruption?
> >
> > - Looks like you are using a UP second kernel. You can try building second
> > kernel as SMP kernel and pass maxcpus=1 option while loading the second
> > kernel using kexec -p.
> >
> > kexec -p /boot/vmlinuz-2.6.18-kdump --args-linux --console-serial
> > --append="root=/dev/sda3 init 3 irqpoll console=ttyS0,115200,
> > earlyprintk=ttyS0,115200 maxcpus=1"
> >
> > This will give some idea whether this problem exists with second kernels
> > being SMP.
>
> Thank you, this kernel booted. :)
>
Excellent.
Going through the logs you uploaded, I think you can't boot a UP kernel
on your machine at all. So can you please compile your second kernel with
CONFIG_PHYSICAL_ADDR=0x100000 (1MB) and then try to boot it through
your boot loader (i think lilo). I think it will fail at the same place.
And reason for that seems to be becaues for UP kernels NR_IRQ_VECTORS
is defined as 32*NR_CPUS. So for UP kernel it is 32 and we are trying
to use gsi/irq higher than that (38).
GSI 38 sharing vector 0xB9 and IRQ 38
GSI 48 sharing vector 0xC1 and IRQ 48
GSI 49 sharing vector 0xC9 and IRQ 49
I am not sure why I can't reproduce this problem. I am still looking
into it.
If you can't boot UP kernel through lilo, can you report this issue
to LKML. Somebody knowing this code more can give it a thought.
Thanks
Vivek
_______________________________________________
fastboot mailing list
[email protected]
https://lists.osdl.org/mailman/listinfo/fastboot
