linux-riscv.lists.infradead.org archive mirror
 help / color / mirror / Atom feed
From: palmer@sifive.com (Palmer Dabbelt)
To: linux-riscv@lists.infradead.org
Subject: [RFC 0/2] RISC-V: A proposal to add vendor-specific code
Date: Thu, 01 Nov 2018 10:50:04 -0700 (PDT)	[thread overview]
Message-ID: <mhng-22ded2c8-7f87-4acc-b017-627e369cf874@palmer-si-x1c4> (raw)
In-Reply-To: <20181101005541.GA25604@andestech.com>

On Wed, 31 Oct 2018 17:55:42 PDT (-0700), alankao at andestech.com wrote:
> On Wed, Oct 31, 2018 at 07:17:45AM -0700, Christoph Hellwig wrote:
>> On Wed, Oct 31, 2018 at 04:46:10PM +0530, Anup Patel wrote:
>> > I agree that we need a place for vendor-specific ISA extensions and
>> > having vendor-specific directories is also good.
>>
>> The only sensible answer is that we should not allow vendor specific
>> extensions in the kernel at all.  ...
>
> How can this even be possible if a extension includes an extra register
> set as some domain-specific context?  In such a case, kernel should
> at least process the context during any context switch, just like how it
> deals with the FP context.

Ya, I think there are cases where vendor-specific extensions are going to be 
necessary to handle within the kernel.  Right now the only one I can think of 
is the performance counter stuff, where we explicitly allow vendor-specific 
counters as part of the ISA spec.

For stateful extensions, we currently have a standard mechanism where the XS 
bits get set in sstatus and the actual save/restore code is hidden behind an 
SBI call.  That call doesn't currently exist, but if we just go ahead and add 
one it should be easy to support this from within Linux.  We'll need to figure 
out how to enable these custom extensions from userspace, but that seems 
tractable as well.  We'll probably also want some fast-path for the V extension 
(and any other stateful standard extensions), but I think as long as the V 
extension adds a quick check for dirtiness then it's not a big deal.

Do you guys have stateful extensions?  We're trying really hard to avoid them 
at SiFive because they're a huge headache, so unless there's a compelling base 
of software using one I don't want to go add support if we can avoid it.

WARNING: multiple messages have this Message-ID (diff)
From: Palmer Dabbelt <palmer@sifive.com>
To: alankao@andestech.com
Cc: zong@andestech.com, aou@eecs.berkeley.edu,
	Arnd Bergmann <arnd@arndb.de>,
	greentime@andestech.com, anup@brainfault.org,
	linux-kernel@vger.kernel.org,
	Christoph Hellwig <hch@infradead.org>,
	vincentc@andestech.com, linux-riscv@lists.infradead.org,
	deanbo422@gmail.com
Subject: Re: [RFC 0/2] RISC-V: A proposal to add vendor-specific code
Date: Thu, 01 Nov 2018 10:50:04 -0700 (PDT)	[thread overview]
Message-ID: <mhng-22ded2c8-7f87-4acc-b017-627e369cf874@palmer-si-x1c4> (raw)
Message-ID: <20181101175004.Btuwgn5MuKG3hhugpSMLKRYlB8vEGMaZ_1QEgYAPhkc@z> (raw)
In-Reply-To: <20181101005541.GA25604@andestech.com>

On Wed, 31 Oct 2018 17:55:42 PDT (-0700), alankao@andestech.com wrote:
> On Wed, Oct 31, 2018 at 07:17:45AM -0700, Christoph Hellwig wrote:
>> On Wed, Oct 31, 2018 at 04:46:10PM +0530, Anup Patel wrote:
>> > I agree that we need a place for vendor-specific ISA extensions and
>> > having vendor-specific directories is also good.
>>
>> The only sensible answer is that we should not allow vendor specific
>> extensions in the kernel at all.  ...
>
> How can this even be possible if a extension includes an extra register
> set as some domain-specific context?  In such a case, kernel should
> at least process the context during any context switch, just like how it
> deals with the FP context.

Ya, I think there are cases where vendor-specific extensions are going to be 
necessary to handle within the kernel.  Right now the only one I can think of 
is the performance counter stuff, where we explicitly allow vendor-specific 
counters as part of the ISA spec.

For stateful extensions, we currently have a standard mechanism where the XS 
bits get set in sstatus and the actual save/restore code is hidden behind an 
SBI call.  That call doesn't currently exist, but if we just go ahead and add 
one it should be easy to support this from within Linux.  We'll need to figure 
out how to enable these custom extensions from userspace, but that seems 
tractable as well.  We'll probably also want some fast-path for the V extension 
(and any other stateful standard extensions), but I think as long as the V 
extension adds a quick check for dirtiness then it's not a big deal.

Do you guys have stateful extensions?  We're trying really hard to avoid them 
at SiFive because they're a huge headache, so unless there's a compelling base 
of software using one I don't want to go add support if we can avoid it.

_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

  parent reply	other threads:[~2018-11-01 17:50 UTC|newest]

Thread overview: 47+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-10-31 10:35 [RFC 0/2] RISC-V: A proposal to add vendor-specific code Vincent Chen
2018-10-31 10:35 ` Vincent Chen
2018-10-31 10:35 ` [RFC 1/2] RISC-V: An infrastructure " Vincent Chen
2018-10-31 10:35   ` Vincent Chen
2018-10-31 10:35 ` [RFC 2/2] RISC-V: make dma_map_ops work without cache coherent agent Vincent Chen
2018-10-31 10:35   ` Vincent Chen
2018-10-31 11:16 ` [RFC 0/2] RISC-V: A proposal to add vendor-specific code Anup Patel
2018-10-31 11:16   ` Anup Patel
2018-10-31 11:45   ` Arnd Bergmann
2018-10-31 11:45     ` Arnd Bergmann
2018-10-31 14:17   ` Christoph Hellwig
2018-10-31 14:17     ` Christoph Hellwig
2018-11-01  0:55     ` Alan Kao
2018-11-01  0:55       ` Alan Kao
2018-11-01 17:50       ` Palmer Dabbelt [this message]
2018-11-01 17:50         ` Palmer Dabbelt
2018-11-02  0:41         ` Alan Kao
2018-11-02  0:41           ` Alan Kao
2018-10-31 17:27   ` Palmer Dabbelt
2018-10-31 17:27     ` Palmer Dabbelt
2018-10-31 19:17     ` Olof Johansson
2018-10-31 19:17       ` Olof Johansson
2018-11-01 17:48     ` Karsten Merker
2018-11-05  6:58       ` Vincent Chen
2018-11-05  6:58         ` Vincent Chen
2018-11-05  7:05         ` Christoph Hellwig
2018-11-05  7:05           ` Christoph Hellwig
2018-11-05  8:52           ` Arnd Bergmann
2018-11-05  8:52             ` Arnd Bergmann
2018-11-05  9:08             ` Christoph Hellwig
2018-11-05  9:08               ` Christoph Hellwig
2018-11-05 13:51               ` Arnd Bergmann
2018-11-05 13:51                 ` Arnd Bergmann
2018-11-06  6:59                 ` Christoph Hellwig
2018-11-06  6:59                   ` Christoph Hellwig
2018-11-06 23:45             ` Palmer Dabbelt
2018-11-06 23:45               ` Palmer Dabbelt
2018-11-07  9:51               ` Arnd Bergmann
2018-11-07  9:51                 ` Arnd Bergmann
2018-11-06 23:45         ` Palmer Dabbelt
2018-11-06 23:45           ` Palmer Dabbelt
2018-11-08  2:43           ` Vincent Chen
2018-11-08  2:43             ` Vincent Chen
2018-11-05 19:39 ` Nick Kossifidis
2018-11-05 19:39   ` Nick Kossifidis
2018-11-06  6:56   ` Christoph Hellwig
2018-11-06  6:56     ` Christoph Hellwig

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=mhng-22ded2c8-7f87-4acc-b017-627e369cf874@palmer-si-x1c4 \
    --to=palmer@sifive.com \
    --cc=linux-riscv@lists.infradead.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).