lkml.org 
[lkml]   [2001]   [Oct]   [27]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectBug in fs.h preventing boot from ramdisk
Date
***** BUG REPORT *****

Summary: Kernel 2.4.10 crashes when used to boot from floppy disk
Description: I created a two-disk set to boot from (boot/root)
The second disk contains a compressed root-fs.
I am using LILO as boot-loader.
Right after the prompt for the root-floppy and
hitting Enter after inserting the floppy, the
kernel crashes due to a nil pointer dereference.
Keywords: boot, floppy
Version : 2.4.10
Rest: not applicable since error already debugged as follows


Kernel 2.4.10 should crash with a cernel panic in Swapper if you supply the
appropriate boot parameters via LILO:

boot = /dev/fd0
install=/boot/boot.b
backup=/dev/null
compact
verbose=1

image= /boot/vmlinuz
root = /dev/fd0
label = Bootdisk
read-write
append = "load_ramdisk=1 prompt_ramdisk=1"
vga = normal

The problem lives in include/linux/fs.h. The ramdisk driver (drivers/
block/rd.c) calls for read oerations into generic_file_read()
(mm/filemap.c)
which in turn calls do_generic_file_read(). The last instruction in
do_generic_file_read() should update inode's access time. To update access
times, a macro, UPDATE_ATIME, is used, which is just a wrapper for
void update_atime (struct inode *inode), which lives in fs/inode.c. And
there is the bug. Before actually updating access time, update_atime()
checks if it is legal to change atime using several macros from fs.h.

These macros originally had a sanity-check, which tests if inode's
pointer to its superblock is not NULL. This sanity-check has been removed
(since which version I cannot tell). Normally, this pointer will never
be NULL, but in case of initial loading of a root image into a ramdisk,
there is just an empty inode created, without a superblock. Here are
the macros:

// These are the macro definitions from 2.2.13 which have an additional
// sanity-check (inode)->i_sb && ...
// in case of booting and loading a compressed ramdisk there is no
superblock !!!
// So the new macros crash miserably. Maybe, it was intended that the new
// function init_special_inode should provide a superblock ??
#define __IS_FLG(inode,flg) (((inode)->i_sb && (inode)->i_sb->s_flags &
(flg)) \
|| (inode)->i_flags & (flg))
#define IS_RDONLY(inode) (((inode)->i_sb) && ((inode)->i_sb->s_flags &
MS_RDONLY))

/* These are the Macros as defined for 2.4.10
#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))

#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
*/

I cannot tell, wether this is due to a failed attempt of optimization or
one just forgot to create an empty superblock in the newly introduced
function init_special_inode().

Can you please fix this for new kernel versions ?

Klaus Buehler
buehler@btbnet.de

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:11    [W:0.030 / U:1.112 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site