Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20180628104927.9130-1-greentime@andestech.com> <20180629084615.GA6236@guoren>
In-Reply-To: <20180629084615.GA6236@guoren>
From:   Greentime Hu <green.hu@gmail.com>
Date:   Mon, 2 Jul 2018 16:08:44 +0800
Message-ID: <CAEbi=3dj4mXVSOBq4g_jaGp0oXTZ2Ai9CU=ZP+cbxpYQXepJiA@mail.gmail.com>
Subject: Re: [PATCH 1/2] nds32: To implement these icache invalidation APIs
 since nds32 cores don't snoop data cache. This issue is found by Guo Ren.
 Based on the Documentation/core-api/cachetlb.rst and it says:
To:     Guo Ren <ren_guo@c-sky.com>
Cc:     Greentime <greentime@andestech.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Guo Ren <ren_guo@c-sky.com> =E6=96=BC 2018=E5=B9=B46=E6=9C=8829=E6=97=A5 =
=E9=80=B1=E4=BA=94 =E4=B8=8B=E5=8D=884:46=E5=AF=AB=E9=81=93=EF=BC=9A
>
> On Thu, Jun 28, 2018 at 06:49:27PM +0800, Greentime Hu wrote:
> > +void flush_icache_page(struct vm_area_struct *vma, struct page *page)
> > +{
> > +     unsigned long flags;
> > +     local_irq_save(flags);
> > +     cpu_cache_wbinval_page((unsigned long)page_address(page),
> > +                            vma->vm_flags & VM_EXEC);
> > +     local_irq_restore(flags);
> > +}
> > +EXPORT_SYMBOL(flush_icache_page);
> > +
> I'm afraid that the page_address(page) will return NULL for non-mapped pa=
ge.
>
> So I use kmap_atomic/kunmap_atomic here.
>
> ref: https://github.com/c-sky/csky-linux/blob/master/arch/csky/abiv2/cach=
eflush.c
>
> > +void flush_icache_user_range(struct vm_area_struct *vma, struct page *=
page,
> > +                          unsigned long addr, int len)
> > +{
> > +     unsigned long _addr =3D (unsigned long) page_address(page) + (add=
r & ~PAGE_MASK);
> > +     flush_icache_range(_addr, _addr + len);
> > +}
>
> The same as above.

Thank you, Ren.
I will prepare the next version patch  to use
kmap_atomic/kunmap_atomic to fix this issue.