Thread: [SSI] Problems booting 0.5.1
Brought to you by:
brucewalker,
rogertsang
From: Asaph Z. <as...@cr...> - 2001-08-23 22:00:05
|
I am having trouble getting 0.5.1 to boot. I am using a pretty vanilla PC, with RedHat 7.1 and the SSI stuff. I can get as far along as the "Starting Init" message but immediately thereafter I get a kernel panic: "Unable to spawn INIT process!" Any ideas? BTW, on my system /usr was mounted on a different partition than / so the supplied mkinitrd.ssi didn't work. I can write a generic patch for this if there is interest (right now I just have this hard coded in for my machine). Thanks! Asaph |
From: Asaph Z. <as...@cr...> - 2001-08-24 21:08:42
|
I made some progress, and actually got it to boot, but I am not entirely happy with how. I investigated the: "Unable to spawn INIT process!" problem and determined that the cause was that there was a call in cluster/ssi/vproc/nsc_initproc.c to kernel_thread_with_pid() with the 4th argument (pid) = 1, but adding a kprintf to sys_clone_with_pid() in arch/i386/kernel/process.c revealed that doing: pid = regs.edi; gets a wrong value in pid, instead of 1 we get some large negative number. Oddly enough, adding a kprintf to kernel_thread_with_pid() made this problem go away and I was able to boot. Ideas on what's going on here greatly appreciated. Asaph |
From: David B. Z. <dav...@co...> - 2002-04-20 02:36:11
|
A preliminary version of our Cluster Filesystem has been checked in. This version has only been tested on client @ server. In order to try it mount a non-root filesystem by calling the mount() system call directly with MS_CFS flag passed (use the actual filesystem type of the partition (i.e. ext2)). The following features are NOT present: readdir file locking remote mounting async reading async writing Verification on Alpha processors Many operations have been tried successfully (i.e. mount lookup read write umount create rename link symlink readlink followlink revalidate). Monday I plan on starting work on the following: Test CI kernel build and boot (due to some minor CI changes) Making CFS configurable with CONFIG_CFS Add readdir Add remote mounting Test all the operations remote Verify coherency between two nodes -- David B. Zafman | Compaq Computer Corporation Linux Kernel Developer | Open SSI Clustering Project mailto:dav...@co... | http://www.compaq.com "Thus spake the master programmer: When you have learned to snatch the error code from the trap frame, it will be time for you to leave." |
From: Aneesh K. K.V <ane...@di...> - 2002-04-22 06:00:17
|
On Sat, 2002-04-20 at 08:00, David B. Zafman wrote: > > A preliminary version of our Cluster Filesystem has been checked in. > This version has only been tested on client @ server. In order to try it > mount a non-root filesystem by calling the mount() system call > directly with MS_CFS flag passed (use the actual filesystem type of the > partition (i.e. ext2)). The following features are NOT present: > > readdir > file locking > remote mounting > async reading > async writing > Verification on Alpha processors I will check out the code today and try to verify it on an Alpha machine. -aneesh |
From: Aneesh K. K.V <ane...@di...> - 2002-04-24 12:47:33
|
On Sat, 2002-04-20 at 08:00, David B. Zafman wrote: > > A preliminary version of our Cluster Filesystem has been checked in. > This version has only been tested on client @ server. In order to try it > mount a non-root filesystem by calling the mount() system call > directly with MS_CFS flag passed (use the actual filesystem type of the > partition (i.e. ext2)). The following features are NOT present: > > readdir > file locking > remote mounting > async reading > async writing > Verification on Alpha processors > I verified the lasted CVS code base on alpha. Every thing ( as far as I know ) works fine. -aneesh |