On Thu, 09 Jun 2011 22:28:15 +0200, Paul Menzel wrote:
>
> One more question. Is `rm -rf volatile.cache` actually a noop since it
> does not exist due to the prior check?
>
It looks to me like it is there to clear the cache when the script is called in
an 'update' operation - I think we should
Dear James,
Am Mittwoch, den 08.06.2011, 01:19 + schrieb James Limbouris:
> > Date: Tue, 07 Jun 2011 12:13:14 +0200
> > From: Paul Menzel
[…]
> > Am Dienstag, den 07.06.2011, 17:35 +0800 schrieb James Limbouris:
> > > When power is lost before a new volatile.cache has been flushed to disk,
> Date: Tue, 07 Jun 2011 12:13:14 +0200
> From: Paul Menzel
> Subject: Re: [oe] [PATCH v2] initscripts: Modify
> 'populate-volatile.sh' to atomically update 'volatile.cache'
> To: openembedded-devel@lists.openembedded.org
> Message-ID: <1307441594
Dear James,
thank you for the updated patch.
Am Dienstag, den 07.06.2011, 17:35 +0800 schrieb James Limbouris:
> When power is lost before a new volatile.cache has been flushed to disk,
> it can be corrupted, leaving the volatiles uncreated and causing software
> failure.
So the device is stil
When power is lost before a new volatile.cache has been flushed to disk,
it can be corrupted, leaving the volatiles uncreated and causing software
failure.
Using a temporary file on the same volume and renaming it avoids this issue.
The file is sync'd before the rename to avoid problems with newer
