linux-kernel - Re: [PATCH] Introduce softpanic

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 25 Jan 2008 12:16:57 -0500
From:	Valdis.Kletnieks@...edu
To:	Jiri Kosina <jikos@...os.cz>
Cc:	Bodo Eggert <7eggert@....de>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] Introduce softpanic

On Fri, 25 Jan 2008 16:26:45 +0100, Jiri Kosina said:
> On Fri, 25 Jan 2008, Bodo Eggert wrote:
> 
> > Enabling this option changes a hard panic on boot errors to a
> > soft panic, which does not stop the system completely.
> > You can still scroll the screen and read the messages.
> 
> Hi Bodo,
> 
> if you are willing to put some effort into the ability to scroll the 
> console after panic, it might be worthwile to look making the keyboard 
> driver work even after panic (i.e. poll the keyboard, rather than relying 
> on interrupts).
> 
> That has significantly smaller potential of breaking because the system is 
> expected to be dead after panic (watchdogs, broken data structures in 
> memory, etc).

I think Bodo is trying to address one specific (but quite common) corner
case, where we currently panic() only because there's nothing else that
makes sense - if we can't find the root fs, we're hosed.  But (barring *other*
oops/etc issues), we're still up and running, our data structures are intact,
and there's no real reason we can't use such kernel services as we want (such
as the screen scroller) to help the sysadmin figure out what he misconfigured.

If we were trying to do a *general case* handler for post-panic, yes, we'd
want to do the polled keyboard and all that....

Content of type "application/pgp-signature" skipped