**** BEGIN LOGGING AT Mon Dec 17 02:59:58 2012
Dec 17 09:14:37 <nareshbhat> ant_work: ping
Dec 17 09:33:57 <nareshbhat> ant_work: I have a question.  on my cortex-a15 target after executing the following command the target is rebooted # kexec /boot/uImage --dtb=vexpress.dtb --command-line="console=ttyAMA0,38400n8" suppose if I try to pass following switch cases "-f/-l/-e" with kexec command says "Starting new kernel" I don't understand how the switch cases makes the difference
Dec 17 10:30:15 <ant_work> nareshbhat: hi
Dec 17 10:30:40 <nareshbhat> gi
Dec 17 10:30:44 <nareshbhat> hi
Dec 17 10:30:47 <ant_work> you can do in 2 steps or in one
Dec 17 10:30:55 <ant_work> 1)     kexec -l kernel-image --append=command-line-options --initrd=initrd-image
Dec 17 10:31:02 <ant_work> 2) kexec -e
Dec 17 10:31:14 <ant_work> or using -f (force)
Dec 17 10:31:30 <ant_work>     kexec -f kernel-image --append=command-line-options --initrd=initrd-image
Dec 17 10:31:53 <nareshbhat> yes, for all the above switch cases
Dec 17 10:31:56 <nareshbhat> le=ttyAMA0,38400n8":~# kexec -l vmlinux --dtb=vexpress.dtb --command-line="conso
Dec 17 10:31:56 <nareshbhat> root@arm-cortex-a15:~# kexec -e
Dec 17 10:31:56 <nareshbhat> Starting new kernel
Dec 17 10:32:32 <nareshbhat> that's it..but if I don't pass the above switch case the target will reboot..I mean for the following command
Dec 17 10:32:41 <ant_work> see man kexec:
Dec 17 10:32:42 <ant_work> It's also possible to invoke kexec without an option parameter. In that case, kexec loads the specified kernel and then invokes shutdown(1).
Dec 17 10:32:57 <nareshbhat> kexec /boot/uImage-3.4.22.arm-cortex-a15-rt32 --dtb=vexpress.dtb --command-line="console=ttyAMA0,38400n8"
Dec 17 10:33:29 <nareshbhat> or kexec /boot/vmlinux-3.4.22.arm-cortex-a15-rt32 --dtb=vexpress.dtb --command-line="console=ttyAMA0,38400n8"
Dec 17 10:34:01 <nareshbhat> is that "reboot" is the expected behavior ?
Dec 17 10:34:10 <ant_work> "If the shutdown scripts of your Linux distribution support kexec-based rebooting, they then call kexec -e just before actually rebooting the machine. That way, the machine does a clean shutdown including all shutdown scripts. "
Dec 17 10:34:19 <ant_work> http://linux.die.net/man/8/kexec
Dec 17 10:37:20 <ant_work> as we use it typically in initramfs, there is anything to unmount
Dec 17 10:37:47 <ant_work> so kexecboot just does -l and -e
Dec 17 10:38:22 <nareshbhat> ok
Dec 17 10:38:54 <ant_work> so it's ok if it shows "Starting new kernel" and hopefully boots
Dec 17 10:39:22 <ant_work> you should see "Uncompressing linux" at first
Dec 17 10:41:51 <nareshbhat> yes, I understand ..but I am not able to see either of the above messages  on console..the command just issues reboot
Dec 17 10:44:50 <ant_work> well, it should boot the new kernel. you mean hard-reboot, from bootloader?
Dec 17 10:48:46 <nareshbhat> yes
Dec 17 10:51:46 <ant_work> do you have a 100% working kernel to test ?
Dec 17 10:52:23 <ant_work> I've never seen it reboot, broken kernels typically hangs afte kexec
Dec 17 10:53:35 <nareshbhat> I mean the target will reboot after execution of # kexec /boot/uImage-3.4.22.arm-cortex-a15-rt32 --dtb=vexpress.dtb --command-line="console=ttyAMA0,38400n8"
Dec 17 10:54:09 <nareshbhat> but the board is booted with the same kernel image uImage-3.4.22.arm-cortex-a15-rt32
Dec 17 10:55:08 <ant_work> I don't know what your distro does about 'reboot' scripts
Dec 17 10:55:19 <ant_work> pls just use -l and -e or -f
Dec 17 10:55:26 <nareshbhat> ok
Dec 17 10:59:15 <ant_work> note that you have typically to unmount the filesystems by hand if you're kexecing from shell
Dec 17 10:59:33 <ant_work> again, we use kexecboot in the initramfs so it's easier
Dec 17 10:59:45 <nareshbhat> ok
Dec 17 11:50:09 <nareshbhat> do I need to pass any kernel command line parameters for the host kernel ?
Dec 17 11:51:14 <ant_work> no, the host kernel just need to have kexec enabled
Dec 17 11:53:07 <nareshbhat> root@arm-cortex-a15:~# kexec -l vmlinux
Dec 17 11:53:07 <nareshbhat> Modified cmdline: root=/dev/nfs
Dec 17 11:53:07 <nareshbhat> Unable to find /proc/device-tree//chosen/linux,stdout-path, printing from purgatory is diabled
Dec 17 11:53:50 <nareshbhat> what's this the warning "Unable to find /proc/device-tree//chosen/linux,stdout-path, printing from purgatory is diabled" something related to console messages ?
Dec 17 12:01:52 <ant_work> no, seems another issue with kernel
Dec 17 12:02:03 <ant_work> I think you can let th ekernel recreate the dt
Dec 17 12:02:19 <nareshbhat> yep it has created
Dec 17 12:02:58 <nareshbhat> As suggested by Daniel I have enabled the CONFIG_PROC_DEVICETREE"
Dec 17 12:03:27 <nareshbhat> root@arm-cortex-a15:~# ls /proc/device-tree/
Dec 17 12:03:27 <nareshbhat> #address-cells  dma@7ff00000                   model
Dec 17 12:03:27 <nareshbhat> #size-cells     hdlcd@2b000000                 motherboard
Dec 17 12:03:27 <nareshbhat> aliases         interrupt-controller@2c001000  name
Dec 17 12:03:27 <nareshbhat> arm,hbi         interrupt-parent               pmu
Dec 17 12:03:27 <nareshbhat> chosen          memory-controller@2b0a0000     timer
Dec 17 12:03:28 <nareshbhat> compatible      memory-controller@7ffd0000     wdt@2a490000
Dec 17 12:03:28 <nareshbhat> cpus            memory@80000000
Dec 17 12:03:29 <nareshbhat> root@arm-cortex-a15:~#
Dec 17 12:03:43 <nareshbhat> which is created under /proc
Dec 17 12:08:04 <ant_work> found on google:
Dec 17 12:08:04 <ant_work> * Test for /proc/device-tree/chosen/linux,stdout-path rather than the
Dec 17 12:08:04 <ant_work>     non-existent /proc/device-tree/options/chosen/linux,stdout-path.
Dec 17 12:08:36 <ant_work> hm,, no, that was Ubuntu
Dec 17 12:09:00 <nareshbhat> can you give me link ? is that something like bug /
Dec 17 12:09:37 <ant_work> it's too old, 2007
Dec 17 12:09:42 <nareshbhat> ok
Dec 17 12:10:37 <ant_work> this code comes from kexec-tools
Dec 17 12:12:23 <nareshbhat> yes
Dec 17 12:16:07 * nareshbhat probably Dec 18th to 25th week I will be on vacation. It may not be possible to progress anything on KEXEC
Dec 17 12:16:58 <nareshbhat> actually I would love to do atleast using ATAGs is possible
Dec 17 12:17:05 <ant_work> are you using kexec from git master?
Dec 17 12:17:41 <ant_work> afais dt for arm has been fixed recently
Dec 17 12:18:47 <ant_work> see, we use old armv5 boards without dt and the atag (default) method works even with older kexec-tools
Dec 17 12:19:48 <nareshbhat> yes, I am using the latest kexec-tools.  I have ARM v7 versatile express target
Dec 17 12:20:21 <nareshbhat> on the target the host kernel boots using DTB file
Dec 17 12:36:37 <ant_work> I really can't help. Maybe you need a more recent kernel
Dec 17 12:40:26 <ant_work> I see kernel+kexec were ok for versatile in 2011
Dec 17 12:40:44 <ant_work> http://lists.infradead.org/pipermail/linux-arm-kernel/2011-June/053252.html
Dec 17 12:41:03 <ant_work> tested by Will Deacon
Dec 17 12:42:17 <nareshbhat> its ok I am just discussing with you
Dec 17 12:42:28 <nareshbhat> thanks for the information
Dec 17 12:42:46 <ant_work> you see, we were hoping kexecboot does work on armv6/7
Dec 17 12:43:14 <ant_work> it seems it did before devicetree implementation
Dec 17 12:43:48 <nareshbhat> yep, if this is booted I will definitely test kexecboot also :)
Dec 17 12:43:58 <ant_work> ;)
Dec 17 13:10:30 <nareshbhat> humm..looks like I don't have the "MMU disabling code and kexec fixes" do you know which GIT repo maintained ?
Dec 17 15:00:48 <ant_work> I fear you should track the mailing list: the patch could be superseded and could have been replaced
Dec 17 15:07:49 <nareshbhat> ok
Dec 17 15:11:19 <ant_work> nareshbhat: the good news are, JAy7 is expected to test kexecboot on pandaboard (armv7)
Dec 17 15:11:54 <ant_work> he had problems importing the board, though
Dec 17 15:13:33 <nareshbhat> oh cool
Dec 17 15:14:00 <nareshbhat> customs issue ? even I had the same while importing V2P
Dec 17 15:14:05 <nareshbhat> :)
Dec 17 15:15:42 <ant_work> shipping issues I'd say :/
Dec 17 15:31:57 <nareshbhat> Looks like I need to pull these patches http://www.spinics.net/lists/arm-kernel/msg137568.html
Dec 17 15:34:30 <Jay7> shipping issues because customs law issues I'd say :)
Dec 17 15:40:27 <nareshbhat> yep
Dec 17 15:57:56 <ant_work> hi Jay7
Dec 17 15:58:25 <Jay7> ant_work: hi Andrea
Dec 17 15:58:35 <ant_work> nareshbhat: strange that the patches are not applied after more than one year...
Dec 17 15:58:51 <ant_work> and coming from Arm Co. ..
Dec 17 16:04:12 <ant_work> Jay7: any contact from zeldin ?
Dec 17 16:04:20 <Jay7> ant_work: nope
Dec 17 16:04:59 <ant_work> I don't know what was the offset problem he's fixing
Dec 17 16:05:30 <ant_work> I just don't see it, even on resized vga fb
**** ENDING LOGGING AT Tue Dec 18 02:59:58 2012