public inbox for intel-gfx@lists.freedesktop.org
 help / color / mirror / Atom feed
From: Dave Gordon <david.s.gordon@intel.com>
To: Sagar Arun Kamble <sagar.a.kamble@intel.com>, animesh.manna@intel.com
Cc: intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH] drm/i915/skl: replace csr_mutex by completion in csr firmware loading
Date: Thu, 04 Jun 2015 15:36:32 +0100	[thread overview]
Message-ID: <55706270.5060404@intel.com> (raw)
In-Reply-To: <1433397575.17509.222.camel@sagar-desktop>

On 04/06/15 06:59, Sagar Arun Kamble wrote:
> 
> Hi Daniel,
> 
> We already are grabbing RPM reference before start of DMC FW load and
> release post load completion.
> 
> DC5/6 can happen without Runtime PM as well. So we need to wait for CSR
> FW load for some time once we disable PW2.
> 
> Having completion instead of csr.lock+csr.state is  correct thing to do.
> 
> Thanks,
> Sagar
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

Hi guys,

we have a new unified loader for GuC and HuC, which we intend should be
usable for any other microcontrollers that might be added in future. So
it would be good to consider using it for the DMC/CSR/etc.

The most-recently posted version was Alex's on 2015-04-29, but we're
just about to re-post an updated patchset for command submission via the
GuC, which will include the both unified firmware loading module and the
GuC-specific part of the loader.

Please have a look at these patches and let me know whether there's
anything that would help with using the unified code for your uC(s).

http://lists.freedesktop.org/archives/intel-gfx/2015-April/065575.html
http://lists.freedesktop.org/archives/intel-gfx/2015-April/065583.html

One thing that may not be obvious from the posted patches is that the
uC-specific code can override the way the f/w image is saved e.g. if the
loaded file image contains sections that are only needed once, they need
not be saved; more generally, if the way that the data in the image file
is organised doesn't match the way the data is used, the loader can
shuffle the content around for the convenience of the final consumer.
(We needed that at one stage, as the GuC firmware format didn't match
the constraints of the GuC DMA engine; but the format has been fixed
since, so the data-shuffler is not part of the public patch sequence).

Thanks,
Dave
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

  reply	other threads:[~2015-06-04 14:37 UTC|newest]

Thread overview: 42+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-05-21 10:19 [PATCH] drm/i915/skl: replace csr_mutex by completion in csr firmware loading Animesh Manna
2015-05-21 12:11 ` Daniel Vetter
2015-05-21 17:05   ` Animesh Manna
2015-05-21 21:29     ` Daniel Vetter
2015-06-04  5:59       ` Sagar Arun Kamble
2015-06-04 14:36         ` Dave Gordon [this message]
2015-06-15 10:07           ` Daniel Vetter
2015-06-15 18:41             ` Dave Gordon
2015-06-15 10:02         ` Daniel Vetter
2015-06-29  8:39 ` Daniel Vetter
2015-07-07 12:40   ` [PATCH] drm/i915: Resign firmware loading for dmc Animesh Manna
2015-07-07 13:18     ` Daniel Vetter
2015-07-08 14:24       ` [PATCH 0/6] Redesign the dmc firmware loading Animesh Manna
2015-07-08 14:24         ` [PATCH 1/6] drm/i915/gen9: Removed csr-lock and csr-state Animesh Manna
2015-07-09 18:17           ` Daniel Vetter
2015-07-08 14:24         ` [PATCH 2/6] drm/i915/gen9: Added a async work for fw-loading and dc5/dc6 programming Animesh Manna
2015-07-08 14:24         ` [PATCH 3/6] drm/i915/gen9: Replaced request_firmware_nowait() by request_firmware() Animesh Manna
2015-07-09 18:24           ` Daniel Vetter
2015-07-08 14:24         ` [PATCH 4/6] drm/i915/gen9: Added dmc_present flag to check firmware loading status Animesh Manna
2015-07-09 18:19           ` Daniel Vetter
2015-07-08 14:24         ` [PATCH 5/6] drm/i915/skl: Removed assert for csr-fw-loading during disabling dc6 Animesh Manna
2015-07-08 14:24         ` [PATCH 6/6] drm/i915/gen9: Corrected the sanity check of mmio address range for csr Animesh Manna
2015-07-08 19:39           ` shuang.he
2015-07-09 17:32           ` Daniel Vetter
2015-07-09 20:04         ` [PATCH 01/12] drm/i915: use correct power domain for csr loading Daniel Vetter
2015-07-09 20:04           ` [PATCH 02/12] drm/i915: Only allow rpm on gen9+ with dmc loaded Daniel Vetter
2015-07-10  8:20             ` Animesh Manna
2015-07-10 16:44               ` Daniel Vetter
2015-07-09 20:04           ` [PATCH 03/12] drm/i915: move assert_csr_loaded into intel_rpm.c Daniel Vetter
2015-07-09 20:04           ` [PATCH 04/12] drm/i915: Remove csr.state, csr_lock and related code Daniel Vetter
2015-07-09 20:04           ` [PATCH 05/12] drm/i915: Align line continuations in intel_csr.c Daniel Vetter
2015-07-09 20:04           ` [PATCH 06/12] drm/i915: Simplify csr loading failure printing Daniel Vetter
2015-07-09 20:04           ` [PATCH 07/12] drm/i915/csr: extract parse_csr_fw Daniel Vetter
2015-07-09 20:04           ` [PATCH 08/12] drm/i915: Don't try to load garbage dmc firmware on resume Daniel Vetter
2015-07-09 20:04           ` [PATCH 09/12] drm/i915: Use dev_priv in csr functions Daniel Vetter
2015-07-09 20:04           ` [PATCH 10/12] drm/i915: Use request_firmware and our own async work Daniel Vetter
2015-07-09 20:04           ` [PATCH 11/12] drm/i915: Use flush_work to synchronize with dmc loader Daniel Vetter
2015-07-09 20:04           ` [PATCH 12/12] drm/i915/csr: Simplify stepping computation Daniel Vetter
2015-07-11  9:22             ` shuang.he
2015-07-10  8:12           ` [PATCH 01/12] drm/i915: use correct power domain for csr loading Animesh Manna
2015-07-10 16:46             ` Daniel Vetter
2015-07-08 10:31     ` [PATCH] drm/i915: Resign firmware loading for dmc shuang.he

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=55706270.5060404@intel.com \
    --to=david.s.gordon@intel.com \
    --cc=animesh.manna@intel.com \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=sagar.a.kamble@intel.com \
    /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