Re: [PATCH] x86: work around clang IAS bug referencing __force_order

[Arvind Sankar]

On Sat, Aug 22, 2020 at 08:17:32PM +0200, Miguel Ojeda wrote:
> On Sat, Aug 22, 2020 at 11:52 AM Sedat Dilek <sedat.dilek@xxxxxxxxx> wrote:
> >
> > I am asking myself who is using such ancient compilers?
> 
> There are many users/companies using older versions of compilers,
> kernels and everything. GCC <= 4.9 will still be used/supported (by
> third parties) for a handful of years at least.
> 
> However, the important question is whether those users/companies care
> about running the latest kernels. Many of those definitely do not want
> to touch their kernel either. For those that do, there are several
> longterms to pick from that still support 4.9, as well as other
> workarounds.
> 
> Thus I am usually in favor of raising the minimum whenever new hacks
> are required to be added. On the other hand, we already raised the
> version twice this year and it is not clear to me what is the minimum
> version we would need to go for to ensure this does not bite us.
> 
> > If this is a real problem with GCC version <= 5, so can this be moved
> > to a GCC specific include header-file?
> > Thinking of include/linux/compiler-gcc.h or
> > include/linux/compiler_types.h with a GCC-VERSION check?
> 
> That would be better if it can be done, yes.
> 
> Cheers,
> Miguel

The fix landed in gcc 6.5, 7.3 and 8.1. The bug is presumably quite
difficult to actually trigger. As a sample data point, I verified that
7.1 vs 7.1+fix have no differences on 32-bit and 64-bit x86 defconfigs,
on current mainline.

Assuming we don't want to risk removing force_order, I'd suggest
- make it an input/output operand, so it enforces ordering fully.
- either restrict it to gcc < 8, or just provide a proper definition in
  some file (maybe arch/x86/kernel/cpu/common.c)?

Thanks.