Subtle interactions in the embedded world - what bugs can teach us [LWN.net]

Subtle interactions in the embedded world - what bugs can teach us

Posted Feb 23, 2012 7:39 UTC (Thu) by kleptog (subscriber, #1183)
Parent article: Subtle interactions in the embedded world - what bugs can teach us

Thanks for the interesting article. This is a situation I think almost unique to embedded systems: you have an interface (the hardware) which is essentially well known but unchangeable. Most unchangeable interfaces are also opaque.

I must admit too that I like fixing bugs more than writing new code. I just wish the bugs wouldn't appear on a production system just before I was about to go home. You learn more from bugs, but it's generally true that you learn more from mistakes than successes.

to post comments

Not unique to embedded

Posted Feb 23, 2012 17:30 UTC (Thu) by hollis (subscriber, #6768) [Link]

This is definitely not unique to embedded systems, and really "Well known but unchangeable" is a better description of an x86 PC than an embedded system.

In embedded systems the hardware is often not well known. Maybe it's a brand new component, or maybe due to the variety of the ecosystem it's your first personal encounter with it.

Also, embedded hardware is more malleable -- there's freedom to select alternate components (even ones that don't conform to decades of PC tradition), or with in-house designs you can at least complain to the hardware designer.

Subtle interactions in the embedded world - what bugs can teach us

Posted Feb 23, 2012 18:12 UTC (Thu) by iabervon (subscriber, #722) [Link]

I personally learned a lot about the MSI interrupt mechanism for PCI and Linux interrupt handling in general by debugging a server which would start ignoring its hard drive interrupt if it didn't use the hard drive for more than a couple of hours.