2.5.68 oops booting with initrd

2.5.68 oops booting with initrd

[Kevin P. Fleming]

Very small and simple kernel configuration (includes devfs, which is 
where this problem came from), using Etherboot to load it along with a 
small (768K) initrd.

Unable to handle kernel NULL pointer dereference at virtual address 00000000
  printing eip:
c0109835
*pde = 00000000
Oops: 0002 [#1]
CPU:    0
EIP:    0060:[<c0109835>]    Not tainted
EFLAGS: 00010002
EIP is at __down+0x55/0xc0
eax: 00000000   ebx: c021c190   ecx: c021c190   edx: c151de5c
esi: 00000286   edi: dfece040   ebp: c021c198   esp: c151de50
ds: 007b   es: 007b   ss: 0068
Process swapper (pid: 1, threadinfo=c151c000 task=dfece040)
Stack: 00000001 dfece040 c0114000 c021c198 00000000 63736900 69642f73 
00306373
        c154bd80 00000000 c154be6c dfcfe6c0 c01099cc c021c190 c021c180 
00000000
        c01772d9 c154be40 c01a292a dff228a0 c154bd80 dff294c0 c017722c 
c021c180
Call Trace:
  [<c0114000>] default_wake_function+0x0/0x20
  [<c01099cc>] __down_failed+0x8/0xc
  [<c01772d9>] .text.lock.util+0x55/0x7c
  [<c01a292a>] blk_unregister_region+0x9a/0xa0
  [<c017722c>] devfs_remove_partitions+0x3c/0x50
  [<c01687ed>] del_gendisk+0xfd/0x100
  [<c01a567f>] initrd_release+0x3f/0x70
  [<c014002b>] __fput+0x9b/0xa0
  [<c013e8bd>] filp_close+0x4d/0x80
  [<c013e940>] sys_close+0x50/0x60
  [<c010a98b>] syscall_call+0x7/0xb
  [<c026b6dd>] rd_load_image+0xbd/0x390
  [<c026be96>] initrd_load+0x56/0x80
  [<c026ae93>] prepare_namespace+0x33/0x120
  [<c010506a>] init+0x2a/0x140
  [<c0105040>] init+0x0/0x140
  [<c0108a3d>] kernel_thread_helper+0x5/0x18

Code: 89 10 89 44 24 10 ff 43 04 89 f6 8b 43 04 48 01 03 0f 98 c0
  <0>Kernel panic: Attempted to kill init!

.config is as follows:

CONFIG_X86=y
CONFIG_MMU=y
CONFIG_UID16=y
CONFIG_GENERIC_ISA_DMA=y
CONFIG_EXPERIMENTAL=y
CONFIG_SWAP=y
CONFIG_X86_PC=y
CONFIG_MK7=y
CONFIG_X86_CMPXCHG=y
CONFIG_X86_XADD=y
CONFIG_RWSEM_XCHGADD_ALGORITHM=y
CONFIG_X86_WP_WORKS_OK=y
CONFIG_X86_INVLPG=y
CONFIG_X86_BSWAP=y
CONFIG_X86_POPAD_OK=y
CONFIG_X86_GOOD_APIC=y
CONFIG_X86_USE_PPRO_CHECKSUM=y
CONFIG_X86_USE_3DNOW=y
CONFIG_X86_TSC=y
CONFIG_NOHIGHMEM=y
CONFIG_PCI=y
CONFIG_PCI_GODIRECT=y
CONFIG_PCI_DIRECT=y
CONFIG_PCI_NAMES=y
CONFIG_HOTPLUG=y
CONFIG_KCORE_ELF=y
CONFIG_BINFMT_ELF=y
CONFIG_BLK_DEV_RAM=y
CONFIG_BLK_DEV_INITRD=y
CONFIG_IDE=y
CONFIG_BLK_DEV_IDE=y
CONFIG_BLK_DEV_IDEDISK=y
CONFIG_IDEDISK_MULTI_MODE=y
CONFIG_BLK_DEV_IDECD=y
CONFIG_BLK_DEV_IDEFLOPPY=y
CONFIG_BLK_DEV_IDEPCI=y
CONFIG_IDEPCI_SHARE_IRQ=y
CONFIG_BLK_DEV_IDEDMA_PCI=y
CONFIG_IDEDMA_PCI_AUTO=y
CONFIG_BLK_DEV_IDEDMA=y
CONFIG_BLK_DEV_ADMA=y
CONFIG_IDEDMA_AUTO=y
CONFIG_SCSI=y
CONFIG_BLK_DEV_SD=y
CONFIG_MD=y
CONFIG_BLK_DEV_DM=y
CONFIG_INPUT=y
CONFIG_SOUND_GAMEPORT=y
CONFIG_SERIO=y
CONFIG_SERIO_I8042=y
CONFIG_INPUT_KEYBOARD=y
CONFIG_KEYBOARD_ATKBD=y
CONFIG_VT=y
CONFIG_VT_CONSOLE=y
CONFIG_HW_CONSOLE=y
CONFIG_SERIAL_8250=y
CONFIG_SERIAL_8250_CONSOLE=y
CONFIG_SERIAL_CORE=y
CONFIG_SERIAL_CORE_CONSOLE=y
CONFIG_FAT_FS=y
CONFIG_MSDOS_FS=y
CONFIG_VFAT_FS=y
CONFIG_PROC_FS=y
CONFIG_DEVFS_FS=y
CONFIG_DEVFS_MOUNT=y
CONFIG_TMPFS=y
CONFIG_RAMFS=y
CONFIG_MSDOS_PARTITION=y
CONFIG_NLS=y
CONFIG_VGA_CONSOLE=y
CONFIG_DUMMY_CONSOLE=y
CONFIG_USB=y
CONFIG_USB_DEVICEFS=y
CONFIG_USB_EHCI_HCD=y
CONFIG_USB_OHCI_HCD=y
CONFIG_USB_UHCI_HCD=y
CONFIG_USB_PRINTER=y
CONFIG_USB_STORAGE=y
CONFIG_DEBUG_KERNEL=y
CONFIG_KALLSYMS=y
CONFIG_X86_BIOS_REBOOT=y

Re: 2.5.68 oops booting with initrd

[Kevin P. Fleming]

Kevin P. Fleming wrote:

> Very small and simple kernel configuration (includes devfs, which is 
> where this problem came from), using Etherboot to load it along with a 
> small (768K) initrd.
> 

After sleeping on it I realized this problem may have been caused by 
neglecting to compile ext2 into the kernel (the initrd is an ext2 
image). Unfortunately, that is not the solution, the oops still occurs.

Re: 2.5.68 oops booting with initrd

[Herbert Xu]

Kevin P. Fleming <kpfleming@cox.net> wrote:
> Very small and simple kernel configuration (includes devfs, which is 
> where this problem came from), using Etherboot to load it along with a 
> small (768K) initrd.

> Call Trace:
>  [<c0114000>] default_wake_function+0x0/0x20
>  [<c01099cc>] __down_failed+0x8/0xc
>  [<c01772d9>] .text.lock.util+0x55/0x7c

This should fix it.
-- 
Debian GNU/Linux 3.0 is out! ( http://www.debian.org/ )
Email:  Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
--
Index: fs/partitions/check.c
===================================================================
RCS file: /home/gondolin/herbert/src/CVS/debian/kernel-source-2.5/fs/partitions/check.c,v
retrieving revision 1.1.1.8
retrieving revision 1.6
diff -u -r1.1.1.8 -r1.6
--- fs/partitions/check.c	7 Apr 2003 17:32:48 -0000	1.1.1.8
+++ fs/partitions/check.c	13 Apr 2003 00:15:54 -0000	1.6
@@ -414,7 +414,8 @@
 	unlink_gendisk(disk);
 	disk_stat_set_all(disk, 0);
 	disk->stamp = disk->stamp_idle = 0;
-	devfs_remove_partitions(disk);
+	if (disk->minors != 1)
+		devfs_remove_partitions(disk);
 	if (disk->driverfs_dev) {
 		sysfs_remove_link(&disk->kobj, "device");
 		sysfs_remove_link(&disk->driverfs_dev->kobj, "block");

Re: 2.5.68 oops booting with initrd

[Kevin P. Fleming]

Herbert Xu wrote:
> Kevin P. Fleming <kpfleming@cox.net> wrote:
> 
>>Very small and simple kernel configuration (includes devfs, which is 
>>where this problem came from), using Etherboot to load it along with a 
>>small (768K) initrd.
> 
> 
>>Call Trace:
>> [<c0114000>] default_wake_function+0x0/0x20
>> [<c01099cc>] __down_failed+0x8/0xc
>> [<c01772d9>] .text.lock.util+0x55/0x7c
> 
> 
> This should fix it.

The patch you supplied (check disk->minors != 1 before calling 
devfs_remove_partitions) did not apply do 2.5.68; the code in 
fs/partitions/check.c has changed since your base version. However, 
hand-editing to do the same check has solved the problem.

I'll Christoph supply a patch, since he's the one that's been working in 
that whole devfs area.

Re: 2.5.68 oops booting with initrd

[Christoph Hellwig]

On Mon, Apr 21, 2003 at 08:50:17AM -0700, Kevin P. Fleming wrote:
> The patch you supplied (check disk->minors != 1 before calling 
> devfs_remove_partitions) did not apply do 2.5.68; the code in 
> fs/partitions/check.c has changed since your base version. However, 
> hand-editing to do the same check has solved the problem.
> 
> I'll Christoph supply a patch, since he's the one that's been working in 
> that whole devfs area.

And 2.5.58-bk1 is completly different again but should have fixed it.

Make sure you apply the following diff when using devfs, though:


--- 1.1/fs/partitions/devfs.c	Sat Apr 19 20:57:36 2003
+++ edited/fs/partitions/devfs.c	Mon Apr 21 17:11:33 2003
@@ -81,7 +81,7 @@
 {
 	char dirname[64], symlink[16];
 
-	if (disk->devfs_name[0] != '\0')
+	if (disk->devfs_name[0] == '\0')
 		sprintf(disk->devfs_name, "%s/disc%d", disk->disk_name,
 				disk->first_minor >> disk->minor_shift);