[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

RE: dkms: Build with an alternative like compiler rather than gcc



> -----Original Message-----
> From: Sedat Dilek [mailto:sedat.dilek@gmail.com]
> Sent: Monday, July 9, 2018 9:41 AM
> To: Limonciello, Mario
> Cc: pere@hungry.com; dapal@debian.org; aron@debian.org;
> iuculano@debian.org; pkg-dkms-maint@lists.alioth.debian.org; debian-
> kernel@lists.debian.org
> Subject: Re: dkms: Build with an alternative like compiler rather than gcc
> 
> On Mon, Jul 9, 2018 at 4:37 PM,  <Mario.Limonciello@dell.com> wrote:
> >> -----Original Message-----
> >> From: Petter Reinholdtsen [mailto:pere@hungry.com]
> >> Sent: Monday, July 9, 2018 9:32 AM
> >> To: sedat.dilek@gmail.com; David Paleino; Aron Xu; Limonciello, Mario;
> Giuseppe
> >> Iuculano
> >> Cc: Dynamic Kernel Modules Support Team; debian-kernel@lists.debian.org
> >> Subject: Re: dkms: Build with an alternative like compiler rather than gcc
> >>
> >>
> >> [Sedat Dilek]
> >> > As a workaround I have symlinked my mycompiler wrapper-script to
> >> > /usr/bin/gcc.  That works.
> >> >
> >> > What is the recommended way to do this correctly?
> >>
> >> Perhaps you should use dpkg-divert to ensure your replaced gcc is not
> >> lost in a future package upgrade?
> >>
> >> The DKMS team in Debian need active members.  I'm not one of them. :(
> >>
> >> --
> >
> > I think this is an upstream problem not at all specific to Debian's implementation.
> > I would recommend to bring this discussion upstream to get it fixed there.
> >
> > I believe DKMS unsets CC before starting build, but maybe that's wrong to do
> > in this instance.
> >
> 
> Personally, I do not think it is an upstream problem.
> Unsetting CC before DKMS starts the build would explain the behaviour/handling.
> 

How exactly does clang get used by kernel build?  It's by setting CC variable right?
You can try to remove this line to see if it fixes the problem (it should).
https://github.com/dell/dkms/blob/master/dkms#L3515

The question would be where that variable should be ignored and where it shouldn't.
Which I believe is an upstream discussion not a Debian discussion.

Reply to: