Re: /boot partition changes when it should not
On Tue, 9 Mar 2010 14:08:34 -0500 (EST), Karl Klinger wrote:
> Clive McBarton wrote:
>>
>> Good point, that is probably important. ext3.
>>
>
> From: http://www.mjmwired.net/kernel/Documentation/filesystems/ext3.txt
>
> ro Mount filesystem read only. Note that ext3 will replay
> the journal (and thus write to the partition) even when
> mounted "read only". Mount options "ro,noload" can be
> used to prevent writes to the filesystem.
>
> There is really no need for a journal on /boot. Why not use ext2?
Good observation, Karl. But if there have been no changes to the disk
during its last mount, there should be nothing in the journal that
causes any writes to be done. He has already verified that a
umount / mount sequence using the ro option does not change the
checksum. The most likely causes are a "dirty shutdown" (i.e.
no clean umount during shutdown) or alterations made by the boot
loader. For example, the boot loader may be updating the mount
count or updating the "last referenced date/time", if there is such
a field in the filesystem, for the kernel image or the initial RAM
disk image.
Many filesystems have a "last referenced date/time"
field for a file, which gets updated even if the file itself is accessed
read-only. This information is needed in support of archival software.
For example, you may write a rule in your archival software something
like, "if a file has not been referenced in the last 30 days, then
archive it to tape". You can't write rules like that unless the filesystem
keeps track of when a file was last referenced. If the filesystem is
mounted read-only, then this field may not be maintained. But the
boot loader doesn't know that Linux is going to mount the filesystem
read-only. And it may be maintaining that field when it reads
the kernel image or the initial RAM disk image.
Reply to: