On Wed, 2011-03-09 at 20:30 +0100, Loïc Minier wrote:
On Wed, Mar 09, 2011, Guilherme Salgado wrote:
My concern is that sfdisk -l returning non-zero may not be a very good indication that we should sleep. It will return non-zero on any error and I expect most errors to not go away after a 30 seconds sleep, in which case we'd be delaying l-m-c's exit unnecessarily. That may not be a big deal, though.
Ah I see what you mean, for instance if the MMC is definitely dead and wont ever come back, you'd like to be able to detect that.
I don't know what to test for though: we just wrote a blank partition table (parted mklabel msdos), and we're just trying sfdisk -l on it. What test did you have in mind?
I didn't have anything in mind; just wanted to know what was the cause of the delay to see if there's anything we could use to identify failure modes which can be resolved with just a sleep.