Fwd: Tegra 30 System hangs because of commit "pinctrl: move subsystem mutex to pinctrl_dev struct"

Olof Johansson olof at lixom.net
Thu May 9 12:13:08 EDT 2013


On Thu, May 9, 2013 at 6:25 AM, Grant Likely <grant.likely at secretlab.ca> wrote:
> On Mon, May 6, 2013 at 9:12 PM, Linus Walleij <linus.walleij at linaro.org> wrote:
>> On Mon, May 6, 2013 at 9:54 PM, Russell King - ARM Linux
>> <linux at arm.linux.org.uk> wrote:
>>
>>> My conclusion is that build and boot test systems just don't work in an
>>> open source environment as no one actively checks the results.  Why would
>>> they - maintainers already have enough to do with reading 6500+ emails a
>>> month on mailing lists.  Why bother going to look at a website as well
>>> which might give even more work.
>>
>> Point taken. I was in a discussion some time back with Greg and some
>> other people and we concluded that unless the system mails you when
>> something breaks it won't make a difference. And it mustn't flood you,
>> just be very precise.
>>
>> Actually the autobuilder at kernel.org helped me out but then stopped
>> building my trees at some point, but that was nice and the same thing
>> doing boots would be great.
>
> +1 on what has already been stated on this thread; An autobuilder does
> work if it emails you when something breaks without spamming you. I've
> also found the 0day build tester actually effective. It has caught
> things that I've missed and I immediately was able to go an fix.

Yes, a builder is quite useful. Several of us do our own builds as well.

When it comes to boot coverage, I'm not sure how successful a central
approach would be. It's something where I think Linaro could
contribute a lot for their member platforms, but I haven't seen much
engagement there yet.

Beyond there, I'd say the expectation is that each maintainer should
keep an eye on their own platforms. They _should_ try building and
booting linux-next on their platforms, to catch regressions early. And
definitely around now, I'd like to see people boot mainline on their
platforms to make sure we haven't added any severe regressions in the
3.10 merge window, etc. Some people are really good at this, others
less so.

What's the old saying, you can lead a horse to water but you can't
make him drink? You can build a boot test cluster but you can't make
anyone pay attention to it. :(

-Olof



More information about the linux-arm-kernel mailing list