Re: [PATCH v2 0/5] notifier error injection

From: Akinobu Mita
Date: Fri Jul 22 2011 - 00:49:24 EST

2011/7/22 Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>:
> On Mon, 18 Jul 2011 10:16:01 +0900
> Akinobu Mita <akinobu.mita@xxxxxxxxx> wrote:
>> This provides the ability to inject artifical errors to the following
>> notifier chain callbacks. It is useful to test the error handling of
>> notifier call chain failures.
> That all looks very nice, but I wonder how many people are actually
> using these things.  The injection framework itself doesn't seem to
> have had a lot of uptake.
> I wonder if we could improve things by adding an easy-to-run testing
> script which identifies all the available error-injection inputs,
> exercises them and then reports on the result?
> Such a script would logically reside under ./tests/fault-injection/,
> but we still don't have a tests/ directory.  Which perhaps tells us
> something ;)

Sounds goood. I'll create shell script that detects hot-pluggable CPU
or memory block and runs quick tests.

BTW, the patchset has been rewritten to be kernel modules rather than
initializing at late_initcall()s. The reason for switching to kernel
modules is to provide an easy way to specify notifier priority as
module parameter. I'll send newer patchset soon.
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