armel/versatile: Can't kexec

[Loïc Minier]

Binary package hint: kexec-tools

Hi

Before updating kexec-tools to 2.0.1, it wouldn't work at all; I updated to the latest upstream and cherry-picked some patches, it works "more" but qemu/linux now hang:
qemu-system-arm -m 256 -drive file=rootfs.img,media=disk -M versatilepb -cpu cortex-a8 -kernel zImage -append 'rootwait root=/dev/sda S console=ttyAMA0,115200' -nographic
Uncompressing Linux.................................................................................................................................................................................................... done, booting the kernel.
[ 0.000000] Initializing cgroup subsys cpuset
[ 0.000000] Initializing cgroup subsys cpu
[ 0.000000] Linux version 2.6.32.6versatile2 (lool@fox) (gcc version 4.4.1 (GCC) ) #64 Sun Feb 7 20:05:51 CET 2010 ()
[ 0.000000] CPU: ARMv7 Processor [410fc080] revision 0 (ARMv7), cr=10c03c7f
[...]
[ 5.388074] VFS: Mounted root (ext3 filesystem) readonly on device 8:0.
[ 5.393392] devtmpfs: mounted
[ 5.395317] Freeing init memory: 152K
mountall: Could not connect to Plymouth
fsck from util-linux-ng 2.16
/dev/sda: clean, 13529/76912 files, 202983/307200 blocks

Give root password for maintenance
(or type Control-D to continue):
root@bee:~# cat /proc/cmdline
rootwait root=/dev/sda S console=ttyAMA0,115200
root@bee:~# kexec -l /root/zImage-1 --append="rootwait root=/dev/sda S console=ttyAMA0,115200"
root@bee:~# sync
root@bee:~# kexec -e
[ 96.628568] Starting new kernel
lsi_scsi: error: Unhandled writeb 0xbc = 0x6c
lsi_scsi: error: Unhandled writeb 0xbd = 0x7e
lsi_scsi: error: Unhandled writeb 0xbe = 0x71
lsi_scsi: error: Unhandled writeb 0xbf = 0xc7
lsi_scsi: error: Unhandled writeb 0x15 = 0x1
lsi_scsi: error: Unhandled writeb 0x18 = 0xff
lsi_scsi: error: Unhandled writeb 0x19 = 0xff
lsi_scsi: error: Unhandled writeb 0x20 = 0x0
lsi_scsi: error: Unhandled writeb 0x23 = 0x0
lsi_scsi: error: Unhandled writeb 0x27 = 0x0
lsi_scsi: error: Unhandled writeb 0xfc = 0x90
lsi_scsi: error: Unhandled writeb 0xfd = 0x23
lsi_scsi: error: Unhandled writeb 0xfe = 0x0
lsi_scsi: error: Unhandled writeb 0xff = 0x50
lsi_scsi: error: Immediate Arbritration not implemented
lsi_scsi: error: Unhandled writeb 0x9 = 0xff
lsi_scsi: error: Unhandled writeb 0x15 = 0x0
lsi_scsi: error: Unhandled writeb 0x18 = 0x0
lsi_scsi: error: Unhandled writeb 0x19 = 0x0
lsi_scsi: error: Unhandled writeb 0x20 = 0x0
lsi_scsi: error: Unhandled writeb 0x23 = 0x0
lsi_scsi: error: Unhandled writeb 0x27 = 0xc7
lsi_scsi: error: Unhandled writeb 0xec = 0x80
lsi_scsi: error: Unhandled writeb 0xed = 0x22
lsi_scsi: error: Unhandled writeb 0xee = 0x0
lsi_scsi: error: Unhandled writeb 0xef = 0x50
lsi_scsi: error: Unhandled writeb 0xf0 = 0x84
lsi_scsi: error: Unhandled writeb 0xf1 = 0xa
lsi_scsi: error: Unhandled writeb 0xf2 = 0x41
lsi_scsi: error: Unhandled writeb 0xf3 = 0xc0
lsi_scsi: error: Unhandled writeb 0xf4 = 0x93
lsi_scsi: error: Unhandled writeb 0xf5 = 0x1
lsi_scsi: error: Unhandled writeb 0xf6 = 0x0
lsi_scsi: error: Unhandled writeb 0xf7 = 0x60
lsi_scsi: error: Unhandled writeb 0xf8 = 0xff
lsi_scsi: error: Unhandled writeb 0xf9 = 0xff
lsi_scsi: error: Unhandled writeb 0xfa = 0xff
lsi_scsi: error: Unhandled writeb 0xfb = 0xff
lsi_scsi: error: Unhandled writeb 0xfc = 0xa4
lsi_scsi: error: Unhandled writeb 0xfd = 0x21
lsi_scsi: error: Unhandled writeb 0xfe = 0x0
lsi_scsi: error: Unhandled writeb 0xff = 0x50
lsi_scsi: error: Unhandled writeb 0x9 = 0x0
lsi_scsi: error: Unhandled writeb 0x15 = 0x7e
lsi_scsi: error: Unhandled writeb 0x18 = 0x2
lsi_scsi: error: Unhandled writeb 0x19 = 0xc
lsi_scsi: error: Destination ID does not match SSID
lsi_scsi: error: Unhandled writeb 0x9 = 0xc
qemu: hardware error: pl011_write: Bad offset fcc

CPU #0:
R00=c060d838 R01=101f4160 R02=c0410a84 R03=60000193
R04=ffffffff R05=101f4084 R06=00000000 R07=00000000
R08=f2000000 R09=c0410aa4 R10=00000000 R11=c7717e6c
R12=00000c02 R13=101f3fcc R14=c0410a9c R15=c0410a60
PSR=60000193 -ZC- A svc32
zsh: abort (core dumped) qemu-system-arm -m 256 -drive file=rootfs.img,media=disk -M versatilepb -cpu

If I don't use serial console (ie without -nographic), qemu just hangs after "Starting new kernel", using 100% CPU.

Bye

[Loïc Minier (lool)]

Tried again with the kernel at http://people.canonical.com/~roc/kernel/kexec/ prepared by Bryan from the sources at http://kernel.ubuntu.com/git?p=roc/ubuntu-lucid.git;a=shortlog;h=refs/heads/kexec-versatile:

qemu-system-arm -m 256 -drive file=lucid.img,media=disk -M versatilepb -cpu cortex-a8 -kernel linux-image-2.6.32-16-versatile_2.6.32-16.24/boot/vmlinuz-2.6.32-16-versatile -append 'root=/dev/sda console=ttyAMA0,115200 S' -nographic

kexec -l vmlinuz-2.6.32-16-versatile --append="rootwait root=/dev/sda console=ttyAMA0,115200 S"
kexec -e

All I get is:
[ 87.854989] Starting new kernel

But at least it doesn't dump core anymore; apparently it avoids writing bogus data over the I/O space.

I also tried without rootwait, but I really get no more output after "Starting new kernel" despite the use of a serial console.

[Loïc Minier (lool)]

So booting the same kernel and kexec-loading a vmlinux instead (created from the vmlinuz bzImage) works!

I can't boot the vmlinux in qemu though.

[Loïc Minier (lool)]

What I meant in comment #2 was that I can't use qemu-system-arm -kernel vmlinux ..., this hangs (without even initializing the fb); if I boot qemu-system-arm with a vmlinuz and kexec the vmlinux, it works.

Summary:
qemu-system-arm: works with vmlinuz, not vmlinux
kexec: works with vmlinux, not vmlinuz

[Loïc Minier (lool)]

Closing the linux task since it works with the in archive kernel just fine (didn't actually test initramfs though).

[Loïc Minier (lool)]

So what remains to be fixed here is kexec on qemu/versatile when an using a vmlinuz; either it should properly compute the address (not sure what actually breaks when using a vmlinuz) or it should unconditionally unpack the vmlinuz into a vmlinux (to save time when kexec -e comes).

[Loïc Minier (lool)]

I've sent a heads up to the kexec@ list.