LWN.net Logo

KS2009: Regressions

KS2009: Regressions

Posted Oct 23, 2009 23:22 UTC (Fri) by jmspeex (subscriber, #51639)
Parent article: KS2009: Regressions

I think one problem is that older hardware doesn't seem to get tested. From my (limited) experience, what happens is that when a new machine gets released, Linux support is often not great because it doesn't know how to handle some of the new hardware. Then, things get better as the bugs are fixed. Then after a year or so, you start seeing regression as the machine becomes old and less tested. For example, my last two laptops both had suspend fully working at some point and both had it broken by some kernel update.


(Log in to post comments)

KS2009: Regressions

Posted Oct 24, 2009 11:19 UTC (Sat) by nix (subscriber, #2304) [Link]

That's when you report the bugs :))) I've kept some machines working for a
decade and more simply by reporting bugs when their old hardware breaks.
The only machine I've ever given up on was a box with an ancient Promise
ISA disk controller, which suddenly broke in 2.4.x versus 2.2.x. Of course
this was before git and bisection: these days I could bisect to find the
faulty commit and report it, but in those days I was pretty stuffed.

KS2009: Regressions

Posted Oct 24, 2009 18:01 UTC (Sat) by jmspeex (subscriber, #51639) [Link]

For the first laptop on which suspend got broken I was able to pinpoint that the problem happened between 2.6.12-rc5-git5 and 2.6.12-rc5-git6 after several months of testing (the problem occurred only after several days and it was my main work machine). This was by using the git "pre-releases" on kernel.org. After finding that, the response I got was simply "you need to learn git to further bisect otherwise we're not even looking at it" (I had already bisected to within one *day* of patches). At this point I just gave up. I *do* still report bugs for projects that treat reports better (gcc is one that comes to mind), but I'm no longer bothering with kernel bugs unless it *really* annoys me or there's a developer committed to putting some time into finding the problem instead of forever asking me to do more work without even looking at the code.

Copyright © 2013, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds