Re: [PATCH 3.16 000/132] 3.16.74-rc1 review

[Ben Hutchings]

On Sun, 2019-09-22 at 21:26 +0200, Miguel Ojeda wrote:
> On Sun, Sep 22, 2019 at 9:04 PM Ben Hutchings <ben@xxxxxxxxxxxxxxx> wrote:
> > It looks like this is triggered by you switching arm builds from gcc 8
> > to 9, rather than by any code change.
> > 
> > Does it actually make sense to try to support building Linux 3.16 with
> > gcc 9?  If so, I suppose I'll need to add:
> > 
> > commit edc966de8725f9186cc9358214da89d335f0e0bd
> > Author: Miguel Ojeda <miguel.ojeda.sandonis@xxxxxxxxx>
> > Date:   Fri Aug 2 12:37:56 2019 +0200
> > 
> >     Backport minimal compiler_attributes.h to support GCC 9
> > 
> > commit a6e60d84989fa0e91db7f236eda40453b0e44afa
> > Author: Miguel Ojeda <miguel.ojeda.sandonis@xxxxxxxxx>
> > Date:   Sat Jan 19 20:59:34 2019 +0100
> > 
> >     include/linux/module.h: copy __init/__exit attrs to init/cleanup_module
> 
> Yeah, those should fix it.

A week or two back I tried building 3.16 for x86_64 with gcc 8, which
produced some warnings but did succeed (and I know Guenter successfully
build-tests 3.16 with gcc 8 for many architectures).  However, the
kernel didn't boot on a test system, while the same code built with gcc
4.9 (if I remember correctly) did boot.

While I'm not about to remove support for gcc 8, this makes me think
that there are some not-so-obvious fixes required to make 3.16 properly
compatible with recent gcc versions.  So I would rather not continue
adding superficial support for them, that may lead to people wasting
time building broken kernels.

Ben.

-- 
Ben Hutchings
Theory and practice are closer in theory than in practice - John Levine

Attachment: signature.asc
Description: This is a digitally signed message part