On Tue, 2012-06-12 at 12:22 +0200, [email protected] wrote: > Hi there, > > I apologise in advance, I'm new to kernel devel/deployment so my questions > might seem a bit simple. > I've encountered another problem using the kernel.boot command. > > The install part seems to be okay. Then mkinitramfs gets called and it > creates a file /boot/initrd-autotest but it does complain about a missing > config file: > > [stdout] INSTALL sound/usb/usx2y/snd-usb-usx2y.ko > 06/12 10:33:12 DEBUG|base_utils:0114| [stdout] DEPMOD 3.4.0-rc1-heca+ > 06/12 10:33:27 INFO | kernel:0010| --- START kernel.mkinitrd --- > 06/12 10:33:27 DEBUG|base_utils:0249| Detected OS vendor: Ubuntu > 06/12 10:33:27 DEBUG|base_utils:0077| Running '/usr/sbin/mkinitramfs -o > /boot/initrd-autotest 3.4.0-rc1-heca+' > 06/12 10:33:27 ERROR|base_utils:0114| [stderr] grep: > /boot/config-3.4.0-rc1-heca+: No such file or directory > 06/12 10:34:02 INFO | kernel:0016| --- END kernel.mkinitrd --- > > I'm not sure if that is a problem or not.
On Fedora, we use dracut instead of initramfs, hence I did not see that yet. Autotest saves the config under /boot/config-autotest, so the action course here is: 1) Find out whether mkinitramfs accepts an alternate config file through a command line switch. Ok, just made the search and seems it doesn't http://www.digipedia.pl/man/doc/view/mkinitramfs.8/ 2) We can work around this problem by copying the config-autotest file to the name mkinitramfs expects it to be... Will you please try this crude workaround and see if it improves the situation? If that is the case, we can write a slightly better version of it. diff --git a/client/kernel.py b/client/kernel.py index 6b1e090..642b082 100644 --- a/client/kernel.py +++ b/client/kernel.py @@ -471,6 +471,7 @@ class kernel(BootableKernel): cmd = '/usr/sbin/mkinitramfs' else: raise error.TestError('No Debian initrd builder') + utils.system('cp /boot/config.autotest /boot/config.%s' % version) utils.system('%s %s -o %s %s' % (cmd, args, initrd, version)) else: raise error.TestError('Unsupported vendor %s' % vendor) > However, the next error in my logs seems potentially more serious: > > 06/12 10:34:03 INFO | job:0210| GOOD build kernel.mkinitrd > timestamp=1339493642 localtime=Jun 12 10:34:02 > 06/12 10:34:03 INFO | kernel:0016| --- END kernel.install --- > 06/12 10:34:03 INFO | job:0210| GOOD build kernel.install > timestamp=1339493643 localtime=Jun 12 10:34:03 > 06/12 10:34:03 DEBUG| base_job:0349| Persistent state client.steps now set > to [([], 'job.end_reboot_and_verify', [1339493643, '3.4.0-rc1-heca+::#12 SMP > Mon Jun 11 18:35:13 BST 2012', 'build', []], {})] > 06/12 10:34:03 ERROR|logging_ma:0584| grubby: kernel not found > 06/12 10:34:03 ERROR|logging_ma:0584| grubby: kernel not found > 06/12 10:34:03 ERROR|logging_ma:0584| grubby: kernel not found > 06/12 10:34:03 ERROR|logging_ma:0584| grubby: kernel not found > 06/12 10:34:03 ERROR|logging_ma:0584| grubby: kernel not found > 06/12 10:34:03 ERROR|logging_ma:0584| grubby: kernel not found > 06/12 10:34:03 DEBUG| boottool:1136| No kernel found for title "autotest". > Assuming no entry exists, and emulating boottool(.pl) behavior and being > silent about it. > > When I run grubby --info=/boot/vmlinux-autotest, it gives a similar error, > i.e.: > > Grubby: kernel not found. > Any ideas why the kernel is not getting picked up? It is likely to be because the initramfs wasn't found... I'm assuming you're using the latest autotest master, since we've found some pretty nasty bugs in the kernel handling code before 0.14.1. > Thanks, > Steve _______________________________________________ Autotest mailing list [email protected] http://test.kernel.org/cgi-bin/mailman/listinfo/autotest
