Re: why panic(9) ?

From: Boris Kochergin <spawk_at_acm.poly.edu>
Date: Tue, 11 Jan 2011 15:50:07 -0500
On 01/11/11 15:37, C. P. Ghost wrote:
> On Tue, Jan 11, 2011 at 9:32 PM, Xin LI<delphij_at_delphij.net>  wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA256
>>
>> On 01/11/11 12:11, David DEMELIER wrote:
>>> Yes, why this function exists? There is no way to solve a problem
>>> without panic'ing? Is panic really needed? Imagine someone working on
>> [...]
>>
>> Panic is used to stop the kernel in an aggressive way when data damage
>> is detected and the damage is already beyond what the kernel can recover
>> from.
>>
>> The kernel can and should be made more robust but no, I don't think we
>> can totally eliminate panic().
> Exactly. One area where the kernel should be made more robust
> is UFS with disappearing disks (e.g. USB mounted file systems,
> or, as recently happened here with a loose external SATA cable).
> Panicing here is REALLY annoying. ;-)

Getting slightly off-topic here, but... there was progress made on this 
front a while ago. You can reliably detach at least USB storage with a 
mounted MSDOSFS or UFS filesystem without soft updates and not risk a 
system panic. There will be a panic if soft updates are enabled on UFS, 
however, at least as of my last test in 2010.

-Boris

>> Cheers,
>> - --
>> Xin LI<delphij_at_delphij.net>      http://www.delphij.net/
>> FreeBSD - The Power to Serve!          Live free or die
> -cpghost.
>
Received on Tue Jan 11 2011 - 19:56:44 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:10 UTC