2015-04-01 08:08:33

by Russell King - ARM Linux

[permalink] [raw]
Subject: Re: linux-next: Tree for Mar 31 (build failures and culprits)

On Tue, Mar 31, 2015 at 07:17:36PM -0700, Guenter Roeck wrote:
> On 03/31/2015 04:07 PM, Greg Kroah-Hartman wrote:
> >You need 3.17 kernel headers to have memfd_create, not much the kdbus
> >test code can do about that. You might want to update the kernel
> >headers for these build boxes.
> >
>
> Seems there are brand-new requirements for kernel builds. The toolchain determines
> the headers used, not the native kernel (which, FWIW, is 3.19). You might want
> to document somewhere that "allmodconfig" and possibly "allyesconfig" now only
> builds if the toolchain is based on 3.17+ kernel headers.
>
> I think I'll just drop those allmodconfig builds instead, or find a means to drop
> the samples/kdbus builds. Sorry, my time is limited. Building toolchains costs a
> lot of time, which I simply don't have. Keep in mind that I am doing all this on
> my own spare time. You are putting a lot of burden on people doing test builds.

You _can_ avoid it by seeding those configurations. You place the
options you want to force to a specific value in a file, and then
do:

make allmodconfig KCONFIG_ALLCONFIG=seed-config-file

So, to turn off samples:

CONFIG_SAMPLES=n

in seed-config-file. Hence, you can still do an allmodconfig build but
without building the samples subdir.

--
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.


2015-04-01 12:24:24

by Guenter Roeck

[permalink] [raw]
Subject: Re: linux-next: Tree for Mar 31 (build failures and culprits)

On 04/01/2015 01:08 AM, Russell King - ARM Linux wrote:
> On Tue, Mar 31, 2015 at 07:17:36PM -0700, Guenter Roeck wrote:
>> On 03/31/2015 04:07 PM, Greg Kroah-Hartman wrote:
>>> You need 3.17 kernel headers to have memfd_create, not much the kdbus
>>> test code can do about that. You might want to update the kernel
>>> headers for these build boxes.
>>>
>>
>> Seems there are brand-new requirements for kernel builds. The toolchain determines
>> the headers used, not the native kernel (which, FWIW, is 3.19). You might want
>> to document somewhere that "allmodconfig" and possibly "allyesconfig" now only
>> builds if the toolchain is based on 3.17+ kernel headers.
>>
>> I think I'll just drop those allmodconfig builds instead, or find a means to drop
>> the samples/kdbus builds. Sorry, my time is limited. Building toolchains costs a
>> lot of time, which I simply don't have. Keep in mind that I am doing all this on
>> my own spare time. You are putting a lot of burden on people doing test builds.
>
> You _can_ avoid it by seeding those configurations. You place the
> options you want to force to a specific value in a file, and then
> do:
>
> make allmodconfig KCONFIG_ALLCONFIG=seed-config-file
>
> So, to turn off samples:
>
> CONFIG_SAMPLES=n
>
> in seed-config-file. Hence, you can still do an allmodconfig build but
> without building the samples subdir.
>
I am doing it more complicated (so far), by using sed to manipulate the config file
after it was created, but that is better - I'll switch to that method.

Thanks!

Guenter