On Tue, 2014-01-14 at 12:13 +0100, Sylvain LÉVÊQUE wrote: > Hello > > > (But Sylvain, it would be great if you could run with earlyprintk too > > to confirm the output). > > I will try to do this tonight.
BTW, I got round to this and I see the same thing with earlyprintk as
reported before.
I've started bisecting and as you might have guessed the issue appeared
between v3.11 and v3.12-rc1. Although I've not 100% ruled out a .config
change (some of my initial attempts at bisect were a bit weird).
bisecting is slow progress, mostly because I have to remember to come
back every 10 mins to boot a new kernel, rather than wander off and get
distracted by something else...
> One question though: from a test point
> of view, do you agree that booting from a 3.12 or 3.13 kernel sent
> over tftp (and I assume pushed to RAM, seeing how fast it goes) is
> equivalent to have it flashed (a several minutes process)?
>
> I only tried the flashed way so far, but I'd rather go the tftp way if
> it's ok, to be a reboot away from a functional system.
FWIW I'm doing:
To create the image:
( devio "wl 0xe3a01c09,4" "wl 0xe38110c6,4" ; cat arch/arm/boot/zImage ) > x
sudo mkimage -A arm -T kernel -O linux -C none -a 0x8000 -e 0x8000 -d x
/srv/tftp/u419
To boot it:
setenv autoload no;dhcp;setenv bootargs ${bootargs} earlyprintk
tftp 0x800000 /u419;cp.b 0xf8400000 0xa00000 0x900000;bootm 0x800000
That gets it to either fail or to the point where the initramfs cannot
find the rootfs (because the driver is not present).
Ian.
signature.asc
Description: This is a digitally signed message part
