Re: [Swsusp-devel] Re: swsusp problems [was Re: Your opinion on the merge?]

From: Michael Frank
Date: Wed Mar 24 2004 - 01:27:55 EST

On Wed, 24 Mar 2004 17:04:58 +1200, Nigel Cunningham <ncunningham@xxxxxxxxxxxxxxxxxxxxx> wrote:


On Wed, 2004-03-24 at 16:52, Dmitry Torokhov wrote:
OK, so you have an error condition on your CD. Does it prevent suspend from
completing? If not then I probably would not care about it till later when
I see it in syslog... I remember that the one thing that Pat complained
most often is your love for "panic"ing instead of trying to recover. In that

As easy, as clumsy.

case you understandably want as many preceding messages as possible left
intact on the screen to diagnose the problem. If error recovery is ever done
then some eye-candy probably won't hurt.

Error messages should be handled on a seperate VT eliminating the issue.

Suspend2 is capable of aborting (there are four panics for dire
situations; for the record swsusp.c has 14). I haven't tried to simulate
a media error, so I'm not perfectly sure it would abort okay,

I'll do more testing there in due course.

How are bad blocks on a swap partition handled by the vm?

but it
wouldn't take too much to fix any issues.

Which reminds me of the "failed to read a chunk" message, the guys who reported
it got all quiet after telling them to do more badblocks testing without diskcaching or
using dd to write random data and read them back, so likely was caused by
media problems.

Here we need more detailed error messages including the driver output and the
screen should be switched to a text VT so messages are visible. Also as the
error will cause resume to fail the system should be halted in this case.

IMO seperate message VT will eliminate all interference issues and further modularization
by keeping the eye candy seperate.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at