Bug 1823

Summary:	Bad root= causes kernel panic without recourse
Product:	File System	Reporter:	Brian M Hunt (bmh_ca)
Component:	Other	Assignee:	fs_other
Status:	REJECTED WILL_NOT_FIX
Severity:	low	CC:	nacc
Priority:	P2
Hardware:	i386
OS:	Linux
Kernel Version:	2.6.0	Subsystem:
Regression:	---	Bisected commit-id:

Description Brian M Hunt 2004-01-09 10:08:54 UTC

Kernel 2.6.0

Steps to reproduce:
If you boot with a bad root= option to the kernel, the kernel will panic.

Preferable alternatives to this would inquire for a valid root partition, rather
than panic. Or, better still, inspect the available system devices for
potentially valid root partitions and prompt the user to select one of them.

I think in almost all contexts, some sort of failsafe and user interaction prior
to a panic would be preferable.

Comment 1 Zurd 2004-10-02 00:18:22 UTC

That is just the brightest idea I ever saw this month.  We can't even count how
much people is getting the dreaded "unable to mount root fs" error for a kernel
panic.  Especially with 2.6 (google reports almost 3000 links).

A new alternative for troubleshooting this error would be much appreciated. 
More error message at least, or even better a choice to select something, but
not a kernel panic.

(p.s. I can't change the severity of this bug, but if you are the owner of the
post or a moderator, please upgrade it at least to normal, not low)

Thank you

Comment 2 Brian M Hunt 2004-10-02 05:51:25 UTC

Modifying Severity, per Zurd's comment.

Comment 3 Nishanth Aravamudan 2006-05-10 19:12:11 UTC

FWIW, Andy Whitcroft posted a resolution to part of this quite recently,
http://marc.theaimsgroup.com/?l=linux-kernel&m=114667485101796&w=2, and Jan
Engelhardt responded with the other half. I will see if I can find their BugMe
IDs and add them to this bug. If they want to reference this bug # in subsequent
posts, it might help in the effort for inclusion.

I'm going to drop the priority back to low, though, it's not a bug, it's a
request for a new feature, really.

Thanks,
Nish

Comment 4 Nishanth Aravamudan 2006-05-11 09:11:13 UTC

Apparently Andy's patch is in -mm. In any case, it's rather unreasonable to
offer user input at this point in the boot process. Andy's (or Jan's) patch gets
it as close as it seems like it's likely too, where the user can at least see
what they did wrong.

Thanks, Nish