[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAAhV-H42_dNyfYHDpdmPM=kG=VTF8xtPHMn-ojWHNkLnTNVDiQ@mail.gmail.com>
Date: Tue, 18 Oct 2022 17:36:45 +0800
From: Huacai Chen <chenhuacai@...nel.org>
To: David Laight <David.Laight@...lab.com>
Cc: WANG Xuerui <kernel@...0n.name>,
Huacai Chen <chenhuacai@...ngson.cn>,
"loongarch@...ts.linux.dev" <loongarch@...ts.linux.dev>,
Xuefeng Li <lixuefeng@...ngson.cn>,
Tiezhu Yang <yangtiezhu@...ngson.cn>,
Guo Ren <guoren@...nel.org>,
Jiaxun Yang <jiaxun.yang@...goat.com>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH V2] LoongArch: Add unaligned access support
On Tue, Oct 18, 2022 at 3:48 PM David Laight <David.Laight@...lab.com> wrote:
>
> From: Huacai Chen
> > Sent: 18 October 2022 08:33
> ...
> > > What about my more structured approach in another reply that avoids the
> > > huge else-if conditions? Both the terrible line wraps and codegen could
> > > be avoided.
> ...
> > OK, let me try.
>
> I suspect you can mask out some 'operand size' bits from the
> instructions - instead of checking each opcode.
>
> I'm also pretty sure you can't assume the FP register are live.
> If a read from userspace faults then there can be a full
> process switch - so by the time you try to write to the
> FP registers they no longer belong to the current process.
>
> It might be safer and simpler to just enforce the FP
> registers be saved and then act on the save area.
> I'd guess they get restored in the 'return to userspace'
> code.
Good catch, will be fixed in V4.
Huacai
>
> David
>
> -
> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
> Registration No: 1397386 (Wales)
Powered by blists - more mailing lists