Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20200922091505.471-1-zhenzhong.duan@gmail.com>
 <20200922162901.GA3421308@ZenIV.linux.org.uk> <CAJF2gTSMWc_=j1NKCTXqhLj7cmSB_A3dYB7nL4F7H3jqT+u38A@mail.gmail.com>
 <20200923002315.GC3421308@ZenIV.linux.org.uk> <CAJF2gTSU4e4yU63z1q502SeuTf2m2BKaD0yZ0deFj0TkiVupFg@mail.gmail.com>
 <20200923045231.GH3421308@ZenIV.linux.org.uk>
In-Reply-To: <20200923045231.GH3421308@ZenIV.linux.org.uk>
From:   Guo Ren <guoren@kernel.org>
Date:   Fri, 25 Sep 2020 14:33:53 +0800
Message-ID: <CAJF2gTTb1QN=BmbtcqkNPpUzGa1fQsQtMM2pjrRtNOVxdkkWkQ@mail.gmail.com>
Subject: Re: [PATCH] csky: Fix a size determination in gpr_get()
To:     Al Viro <viro@zeniv.linux.org.uk>
Cc:     Zhenzhong Duan <zhenzhong.duan@gmail.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        linux-csky@vger.kernel.org, Oleg Nesterov <oleg@redhat.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Sep 23, 2020 at 12:52 PM Al Viro <viro@zeniv.linux.org.uk> wrote:
>
> On Wed, Sep 23, 2020 at 10:37:31AM +0800, Guo Ren wrote:
>
> > > What's going on there?  The mapping is really weird - assuming
> > > you had v0..v31 in the first 32 elements of regs->vr[], you
> > > end up with
> > >
> > > v0 v1 v2 v3 v2 v3 v6 v7 v4 v5 v10 v11 v6 v7 v14 v15
> > > v8 v9 v18 v19 v10 v11 v22 v23 v12 v13 v26 v27 v14 v15 v30 v31
> > >
> > > in the beginning of the output.  Assuming it is the intended
> > > behaviour, it's probably worth some comments...
> > FPU & VDSP use the same regs. 32 FPU regs' width is 64b and 16 VDSP
> > regs' width is 128b.
> >
> > vr[0], vr[1] = fp[0] & vr[0] vr[1], vr[2], vr[3] = vdsp reg[0]
> > ...
> > vr[60], vr[61] = fp[15] & vr[60] vr[61], vr[62], vr[63] = vdsp reg[15]
> > vr[64], vr[65] = fp[16]
> > vr[66], vr[67] = fp[17]
> > ...
> > vr[94], vr[95] = fp[31]
> >
> > Yeah, this is confusing and I'll add a comment later.
>
> Umm...  It would help if you described these 3 layouts:
>         1) kernel-side with VDSP
With VDSP: we use vdsp ld/st instructions to access first 16
128bit-regs and use fpu ld/st instructions to access last 16
64bit-regs.

>         2) userland (identical to (1)?)
Identical to 1.

>         3) kernel-side without VDSP
Without VDSP: we use fpu ld/st instructions to access last 32 64bit-regs.

So, there are 96 32bit-vr[] for the struct. And with VDSP or not will
got different storage format.
With VDSP:
vr128[16] // contain first fp64[16]
fp64[16]; // second fp64[16]

Without VDSP:
fp64[32]
no-use for the reset vr[]

> Still confused...
>
> PS: my apologies re commit message - I left a note to myself when doing
> that series and then forgot about it ;-/
>
> Anyway, which tree should it go through?  In any case, that fix is
Thx for your job, and pushing to linus with Zhenzhong Duan's advice.

> Acked-by: Al Viro <viro@zeniv.linux.org.uk>
> and I can take it through vfs.git or you guys can pick in csky tree;
> up to you.

--
Best Regards
 Guo Ren

ML: https://lore.kernel.org/linux-csky/