Hi everyone I'm using a vanilla-2.4.14 kernel, patched with the sgi 2.4.14-xfs patch, then patched with the 2.4.10 crypto-api patch. Things work great as long as I stay on my hard drives. I create an aes-encrypted ext2-container, it works fine no matter where I move it to. Then I burn it to CD-RW. I losetup and mount the CD, and it still works. Then I try to access a file (using ee), and get the following: kernel: Unable to handle kernel NULL pointer dereference at virtual address 00000000 kernel: printing eip: kernel: c01b9fbe kernel: *pde = 00000000 kernel: Oops: 0000 kernel: CPU: 0 kernel: EIP: 0010:[cdrom_pc_intr+462/464] Not tainted kernel: EFLAGS: 00010202 kernel: eax: cff2e200 ebx: 00000034 ecx: 0000000d edx: 0000000d kernel: esi: 00000000 edi: c02e9718 ebp: c02e9600 esp: c873fcd8 kernel: ds: 0018 es: 0018 ss: 0018 kernel: Process ee (pid: 840, stackpage=c873f000) kernel: Stack: 00000800 00000004 cff2e200 00000059 00000202 cff2e200 00000000 00000000 kernel: 0000000d 00000000 c02e9718 c02e9200 00000000 00000400 00000000 c02e9000 kernel: c01ba1e7 cff2e200 00000800 cff2e200 cff34370 c142d560 cff34320 00000000 Call Trace: [cdrom_write_intr+119/624] [intel_i830_insert_entries+48/160] [direct_check_left+4/32] [sys_gettimeofday+79/160] [dentry_open+184/400] kernel: [get_chrfops+209/224] [ext2_make_empty+199/272] [load_inode_bitmap+267/480] [chrdev_open+26/80] [get_empty_filp+83/256] [write_unlocked_buffers+4/64] kernel: [swap_out+176/1152] [unmap_fixup+101/304] [ext2_new_inode+331/1312] [load_inode_bitmap+156/480] [unmap_fixup+261/304] [build_mmap_rb+36/96] kernel: [invalidate_inode_pages+27/144] [truncate_list_pages+293/352] [truncate_list_pages+92/352] [sys_ftruncate+14/336] [do_signal+451/668] kernel: Code: f3 a5 f6 c3 02 74 02 66 a5 f6 c3 01 74 01 a4 0f b7 90 92 00 The stack trace looks to me as if it's file system stuff, not crypto, that goes wrong, and in fact this also happens when I use an unencrypted container, but it must have something to do with the loop device. Has anyone had similar problems when working with CD's? I browsed through the mailing list archives and didn't find anything. BTW, is there a loop device mailing list? I am aware that this mailing list is concerned with crypto loop, and I guess my problem is only partially related to it (though I don't know yet whether this also happens with the vanilla-kernel loop device - will recompile without crypto patch applied and see if it still oops' as soon as I find some time) I'd be grateful for any pointers. Pascal -- Two men walked into a bar. The third one ducked.
Attachment:
pgp00078.pgp
Description: PGP signature
