On Mon, Sep 07, 2009 at 12:36:08AM +0200, Piotr Kaczuba wrote:
> W dniu 06.09.2009 13:25, Bastian Blank pisze:
>> On Sat, Sep 05, 2009 at 04:20:55PM +0200, Piotr Kaczuba wrote:
>>> The postinst script must run "telinit u" so that init re-executes itself.
>>> Otherwise, on reboot, the old libc won't be properly deleted.
>>
>> Please show your test results. This would mean that the kernel is not
>> able to properly deallocate inodes in use by init.
>>
>>> See also
>>> bug #25444.
>>
>> This bug is over 10 years and many, many kernel and glibc releases old.
>> Please explain why the behaviour is still the same.
> [...]
>
> No sophisticated test results are needed, really. Just do an upgrade of
> libc and see what happens on reboot. Init is still using the old libc
> and the read-only remount of / fails with "mount: / is busy".
That is expected as there are writes pending. However you have to show
that this is a real problem.
> I really can't explain you why the behaviour is still the same.
The mentioned bug shows a different problem.
> Anyway, before removing "telinit u" from
> the postinst script, you should have checked yourself that it's not
> needed any more.
No. You have to show that it is a bug in the first place.
Bastian
--
"That unit is a woman."
"A mass of conflicting impulses."
-- Spock and Nomad, "The Changeling", stardate 3541.9
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
