Re: [PATCH 1/4] cobalt/x86: adapt fpu code with Linux kernel upstream

[Fino Meng <fino.meng@linux.intel.com>]

On Fri, Sep 18, 2020 at 03:04:06PM +0200, Jan Kiszka wrote:
> <no commit message>
> 
> When I'm asking questions on the code, there was a need for a commit
> message.
> 
> On 18.09.20 08:29, Fino Meng wrote:
> > Signed-off-by: Fino Meng <fino.meng@linux.intel.com>
> > Signed-off-by: Mingliang Hu <mingliang.hu@intel.com>
> > ---
> >   kernel/cobalt/arch/x86/thread.c | 6 +++++-
> >   1 file changed, 5 insertions(+), 1 deletion(-)
> > 
> > diff --git a/kernel/cobalt/arch/x86/thread.c b/kernel/cobalt/arch/x86/thread.c
> > index ba807ac1e..c33ad832f 100644
> > --- a/kernel/cobalt/arch/x86/thread.c
> > +++ b/kernel/cobalt/arch/x86/thread.c
> > @@ -26,6 +26,10 @@
> >   #include <asm/mmu_context.h>
> >   #include <asm/processor.h>
> > +#ifdef IPIPE_X86_FPU_EAGER
> > +extern void fpu__initialize(struct fpu *fpu);
> 
> This will go along with ipipe for 5.4 patching that function to be
> non-static again, right?

looks like should use "void fpstate_init(union fpregs_state *state)" to
replace "void fpu__initialize(struct fpu *fpu)", the former is still a
API in header file.

> > +#endif
> > +
> >   static struct kmem_cache *xstate_cache;
> >   #ifdef IPIPE_X86_FPU_EAGER
> > @@ -472,7 +476,7 @@ void xnarch_leave_root(struct xnthread *root)
> >   #if LINUX_VERSION_CODE >= KERNEL_VERSION(4,14,0)
> >   	/* restore current's fpregs */
> >   	__cpu_invalidate_fpregs_state();
> > -	switch_fpu_finish(&current->thread.fpu, smp_processor_id());
> > +	switch_fpu_finish(&current->thread.fpu);
> 
> That is not correct for < 5.2 (5f409e20b794).

preparing V2 patches while fixing the smokey tests.

BR fino

> >   #else
> >   	/* mark current thread as not owning the FPU anymore */
> >   	if (fpregs_active())
> > 
> 
> Jan
> 
> -- 
> Siemens AG, Corporate Technology, CT RDA IOT SES-DE
> Corporate Competence Center Embedded Linux