[gentoo-dev] checkfs error

[gentoo-dev] checkfs error

[Jani Monoses]

Hi
for a while checkfs has been starting sulogin after cheking the fs
(ext3) even though it shouldn't have as filesystem error were corrected
At the bottom of checkfs when checking for the return status of fsck
it goes to sulogin if retval == 1 whereas it shouldn't thats errors
corrected (see man fsck)
so the test which says -gt 1 should be changed to -ge 1 at the bottom 
of checkfs because that's OK too, and exit normally.
latest baselayout and ext2fs. The error started appearing after an
update of the system where both baselayout and e2fstools were updates

                elif [ "${retval}" -ge 1 -a "${retval}" -lt 4 ]
Jani

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] checkfs error

[Martin Schlemmer]

[-- Attachment #1: Type: text/plain, Size: 939 bytes --]

On Wed, 2003-10-15 at 07:53, Jani Monoses wrote:
> Hi
> for a while checkfs has been starting sulogin after cheking the fs
> (ext3) even though it shouldn't have as filesystem error were corrected
> At the bottom of checkfs when checking for the return status of fsck
> it goes to sulogin if retval == 1 whereas it shouldn't thats errors
> corrected (see man fsck)
> so the test which says -gt 1 should be changed to -ge 1 at the bottom 
> of checkfs because that's OK too, and exit normally.
> latest baselayout and ext2fs. The error started appearing after an
> update of the system where both baselayout and e2fstools were updates
> 
>                 elif [ "${retval}" -ge 1 -a "${retval}" -lt 4 ]

You are using an old version, or did not update your check* scripts,
as this was fixed in July (?) already.

-- 

Martin Schlemmer
Gentoo Linux Developer, Desktop/System Team Developer
Cape Town, South Africa



[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

[gentoo-dev] Re: checkfs error

[Charlie C]

Martin Schlemmer wrote:
> On Wed, 2003-10-15 at 07:53, Jani Monoses wrote:
> ...
>> You are using an old version, or did not update your check* scripts,
>> as this was fixed in July (?) already.

That's true, but the new version is (i think) worse.
Now, it appears, return values of 1,2 or 3 do nothing, even though both 
2 and 3 should cause a reboot (unless I'm missing something).
All the rest cause a sulogin, and then ... no reboot. This seems to be 
taking a few chances with the users filesystems.

man fsck says to me that we must do a bitwise and between 2 and the 
return value to determine whether a reboot is required. i.e.:
   if  [  $((retval & 2)) != 0 ] ; then
       ewarn "fsck requires a reboot"
       rebootrequired="yes"
   fi
And common sense tells me that it would be prudent to do a reboot after 
any manual fsck-ing, when retval > 3

I've got a heavily modified version of the checkfs script(excerpt 
above). Having had some "heart-in-the-mouth" moments with the previous 
version, it's one of the things I keep an eye on...perhaps I'll have to 
submit it to bugzilla.

Charlie






--
gentoo-dev@gentoo.org mailing list