From: Fernando L. V. C. <fer...@in...> - 2006-02-17 10:23:17
|
Hi Jeong-Sang, I have a machine with specs very similar to yours so I will test whether mkdump works as expected there with the config file you sent me. Unfortunately I will not have access to that machine until Monday. I will let you know then and hope we can fix the problem in your environment. Sorry for the trouble. Best regards, Fernando On Thu, 2006-02-16 at 21:12 +0900, Jeong-Sang Ohn wrote: > Hi Fernando, > > First of all, I'm really appreciating your help. Thank You! :D > > I'll put my answers below. > > > On 2/16/06, Fernando Luis Vazquez Cao <fer...@in...> > wrote: > Hi Jeong-Sang, > > Thank you for the prompt reply. > > On Thu, 2006-02-16 at 18:36 +0900, Jeong-Sang Ohn wrote: > > > But after patching the kernel with the mkdump > > > patch(minik-patch-2.6.9-1.2), the > > > kernel won't boot. (Also tried to boot with this > kernel stand-alone, > > > but it > > > won't boot either.)Cannot see anything printing. > And after a few secs, > > > the > > > system reboots. > I have a hunch that what is happening is something like this: > > - The second kernel boots normally but nothing is sent to the > serial > console because it is not configured properly. (see below) > ----> Oh, sorry. That was a typo... the console parameter part is > actually, > "console=ttyS1,115200n8" > And it's working fine on the unpatched 2.6.9 kernel. > > > - The second kernel is not able to find the specified dump > device (see > below) and, consequently, reboots the machine (this is the > default > behavior when the device is not found). > ----> my dump device is /dev/hde2 (which is connected via PCI > IDE controller)[root@255_63_63_DPSA root]# fdisk -l > Disk /dev/hde: 40.0 GB, 40007761920 bytes > 16 heads, 63 sectors/track, 77520 cylinders > Units = cylinders of 1008 * 512 = 516096 bytes > > Device Boot Start End Blocks Id System > /dev/hde1 * 1 203 101872+ 83 Linux > /dev/hde2 203 73460 36921840 83 Linux > /dev/hde3 73461 77520 2046120 82 Linux swap > [root@255_63_63_DPSA root]# ls -al /dev/hde2 > brw-rw---- 1 root disk 33, 2 Jan 25 > 23:30 /dev/hde2 > > > /dev/hde2 is currently an empty partition made for the mkdump > test. > > And even though this part got wrong, I think I should still > see some messages > during the initialization of the mini-kernel. > > it looks like this (load is complete & haven't wrote > '2' to /proc/mkexec/stats yet): > > [ root@255_63_63_DPSA root]# cat /proc/mkexec/mem > > 16 > This is a valid value. > > > [root@255_63_63_DPSA root]# cat /proc/mkexec/dumpdev > > 0x2100002 > Are you sure this is a correct values? Could you show the > output of ls > -l /dev/your_dump_device? > > > [root@255_63_63_DPSA root]# > cat /proc/mkexec/parameter > > console=console=ttyS1,115200n8 lpj=2392060 no-hlt > > mkdump_reboot=halt dump_delay=0 > Regarding the console settings, you want either > "console=tty0 console=ttyS1,115200n8" > or > "console=ttyS1,115200n8". > > "console=console=ttyS1,115200n8" seems to be invalid. > > > [root@255_63_63_DPSA root]# cat /proc/mkexec/path > > /boot/vmlinux.bin > I assume this was copied from > "arch/i386/boot/compressed/vmlinux.bin". > Correct? > ----> Yes. that's the one I've used. > > > [root@255_63_63_DPSA root]# cat /proc/mkexec/stats > > 1 > This looks good. > > Could you make the aforementioned changes and try again? > Thanks in advance. > > Regards, > > Fernando > > > Well, in conclusion I can't find anything to change yet. > Could the relatively large RAM size(8GB) be a problem? > > And, I've also tested the mkdump on an Intel Pentium3 PC (256MB > RAM) in the same manner and it just works fine. > > Maybe my kernel configuration could inform you a bit. > I'll attatch my .config file for the mini-kernel. > > Thank you. > > |