Re: [PATCH V5 06/30] csky: Cache and TLB routines

[Guo Ren]

On Thu, Sep 27, 2018 at 09:08:59AM +0200, Peter Zijlstra wrote:
> On Thu, Sep 27, 2018 at 01:27:38PM +0800, Guo Ren wrote:
> > On Tue, Sep 25, 2018 at 09:24:07AM +0200, Peter Zijlstra wrote:
> > > On Mon, Sep 24, 2018 at 10:36:22PM +0800, Guo Ren wrote:
> > > > diff --git a/arch/csky/abiv1/inc/abi/cacheflush.h b/arch/csky/abiv1/inc/abi/cacheflush.h
> > > > new file mode 100644
> > > > index 0000000..f0de49c
> > > > --- /dev/null
> > > > +++ b/arch/csky/abiv1/inc/abi/cacheflush.h
> 
> > > > +#define flush_cache_range(mm,start,end)	cache_wbinv_range(start, end)
> > >                              ^^^ should be vma
> > Yes, I'll change it to:
> > #define flush_cache_range(mm,start,end)	cache_wbinv_all()
> > 
> > I'll improve it later after test.
> 
> That's not what I meant; I meant you need something like:
> 
> #define flush_cache_range(vma, start, end) cache_wbinv_range(start, end)
If you remove the tlb_start_vma in my tlb.h, I want to use cache_wbinv_all() is
more safe. And I'll improve it in future.

My cache_wbinv_range(start, end) won't care vma->mm's asid and they just use current
asid in mmu reg. If current_mm != vma->mm, then flush_cache_range will be broken.
Perhaps, I need improve flush_cache_range first ...

> 
> The first argument is a vma, not an mm.
Yes, vma! My fault :-P

> > > Also, I'll be shortly removing this:
> > > 
> > >   https://lkml.kernel.org/r/20180913092812.071989585@xxxxxxxxxxxxx
> > Ok, I'll follow the rules.
> 
> 
> > > > diff --git a/arch/csky/abiv2/inc/abi/cacheflush.h b/arch/csky/abiv2/inc/abi/cacheflush.h
> > > > new file mode 100644
> > > > index 0000000..756beb7
> > > > --- /dev/null
> > > > +++ b/arch/csky/abiv2/inc/abi/cacheflush.h
> > > > @@ -0,0 +1,40 @@
> 
> > > > +#define flush_cache_range(vma, start, end)	do { } while (0)
> > >                              ^^^ like here..
> > #define flush_cache_range(vma, start, end) \
> > 	do { \
> > 		if (vma->vm_flags & VM_EXEC) \
> > 			icache_inv_all(); \
> > 	}
> > 
> > Hmm ?
> > 
> 
> For v2, which IIUC is PIPT (as opposed to v1 which is VIPT), what you
> had was correct.
I'll improve icache_inv_all() with flush_range() in future.

> 
> I was merely pointing out that the flush_cache_range() definition was
> inconsitent between v1 and v2; v1 using @mm and v2 using @vma for the
> first argument.
Must be vma, you are right. And I know it's a vma in mind, but fault in
code. Thx for your review.

Best Regards
 Guo Ren