From: Yu Dai <yu.dai@intel.com>
To: Dave Gordon <david.s.gordon@intel.com>, intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH v4 2/6] drm/i915/guc: move guc_ring_doorbell() nearer to callsite
Date: Wed, 13 Apr 2016 10:52:13 -0700 [thread overview]
Message-ID: <570E874D.7000504@intel.com> (raw)
In-Reply-To: <1460049678-21918-3-git-send-email-david.s.gordon@intel.com>
LGTM.
Reviewed-by: Alex Dai <yu.dai@intel.com>
On 04/07/2016 10:21 AM, Dave Gordon wrote:
> Just code movement, no actual change to the function. This is in
> preparation for the next patch, which will reorganise all the other
> doorbell code, but doesn't change this function. So let's shuffle it
> down near its caller rather than leaving it mixed in with the setup
> code. Unlike the doorbell management code, this function is somewhat
> time-critical, so putting it near its caller may even yield a tiny
> performance improvement.
>
> Signed-off-by: Dave Gordon <david.s.gordon@intel.com>
> ---
> drivers/gpu/drm/i915/i915_guc_submission.c | 128 +++++++++++++++--------------
> 1 file changed, 67 insertions(+), 61 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_guc_submission.c b/drivers/gpu/drm/i915/i915_guc_submission.c
> index da86bdb..2171759 100644
> --- a/drivers/gpu/drm/i915/i915_guc_submission.c
> +++ b/drivers/gpu/drm/i915/i915_guc_submission.c
> @@ -190,67 +190,6 @@ static void guc_init_doorbell(struct intel_guc *guc,
> kunmap_atomic(base);
> }
>
> -static int guc_ring_doorbell(struct i915_guc_client *gc)
> -{
> - struct guc_process_desc *desc;
> - union guc_doorbell_qw db_cmp, db_exc, db_ret;
> - union guc_doorbell_qw *db;
> - void *base;
> - int attempt = 2, ret = -EAGAIN;
> -
> - base = kmap_atomic(i915_gem_object_get_page(gc->client_obj, 0));
> - desc = base + gc->proc_desc_offset;
> -
> - /* Update the tail so it is visible to GuC */
> - desc->tail = gc->wq_tail;
> -
> - /* current cookie */
> - db_cmp.db_status = GUC_DOORBELL_ENABLED;
> - db_cmp.cookie = gc->cookie;
> -
> - /* cookie to be updated */
> - db_exc.db_status = GUC_DOORBELL_ENABLED;
> - db_exc.cookie = gc->cookie + 1;
> - if (db_exc.cookie == 0)
> - db_exc.cookie = 1;
> -
> - /* pointer of current doorbell cacheline */
> - db = base + gc->doorbell_offset;
> -
> - while (attempt--) {
> - /* lets ring the doorbell */
> - db_ret.value_qw = atomic64_cmpxchg((atomic64_t *)db,
> - db_cmp.value_qw, db_exc.value_qw);
> -
> - /* if the exchange was successfully executed */
> - if (db_ret.value_qw == db_cmp.value_qw) {
> - /* db was successfully rung */
> - gc->cookie = db_exc.cookie;
> - ret = 0;
> - break;
> - }
> -
> - /* XXX: doorbell was lost and need to acquire it again */
> - if (db_ret.db_status == GUC_DOORBELL_DISABLED)
> - break;
> -
> - DRM_ERROR("Cookie mismatch. Expected %d, returned %d\n",
> - db_cmp.cookie, db_ret.cookie);
> -
> - /* update the cookie to newly read cookie from GuC */
> - db_cmp.cookie = db_ret.cookie;
> - db_exc.cookie = db_ret.cookie + 1;
> - if (db_exc.cookie == 0)
> - db_exc.cookie = 1;
> - }
> -
> - /* Finally, update the cached copy of the GuC's WQ head */
> - gc->wq_head = desc->head;
> -
> - kunmap_atomic(base);
> - return ret;
> -}
> -
> static void guc_disable_doorbell(struct intel_guc *guc,
> struct i915_guc_client *client)
> {
> @@ -471,6 +410,12 @@ static void guc_fini_ctx_desc(struct intel_guc *guc,
> sizeof(desc) * client->ctx_index);
> }
>
> +/*
> + * Everything above here is concerned with setup & teardown, and is
> + * therefore not part of the somewhat time-critical batch-submission
> + * path of i915_guc_submit() below.
> + */
> +
> int i915_guc_wq_check_space(struct i915_guc_client *gc)
> {
> struct guc_process_desc *desc;
> @@ -559,6 +504,67 @@ static int guc_add_workqueue_item(struct i915_guc_client *gc,
> return 0;
> }
>
> +static int guc_ring_doorbell(struct i915_guc_client *gc)
> +{
> + struct guc_process_desc *desc;
> + union guc_doorbell_qw db_cmp, db_exc, db_ret;
> + union guc_doorbell_qw *db;
> + void *base;
> + int attempt = 2, ret = -EAGAIN;
> +
> + base = kmap_atomic(i915_gem_object_get_page(gc->client_obj, 0));
> + desc = base + gc->proc_desc_offset;
> +
> + /* Update the tail so it is visible to GuC */
> + desc->tail = gc->wq_tail;
> +
> + /* current cookie */
> + db_cmp.db_status = GUC_DOORBELL_ENABLED;
> + db_cmp.cookie = gc->cookie;
> +
> + /* cookie to be updated */
> + db_exc.db_status = GUC_DOORBELL_ENABLED;
> + db_exc.cookie = gc->cookie + 1;
> + if (db_exc.cookie == 0)
> + db_exc.cookie = 1;
> +
> + /* pointer of current doorbell cacheline */
> + db = base + gc->doorbell_offset;
> +
> + while (attempt--) {
> + /* lets ring the doorbell */
> + db_ret.value_qw = atomic64_cmpxchg((atomic64_t *)db,
> + db_cmp.value_qw, db_exc.value_qw);
> +
> + /* if the exchange was successfully executed */
> + if (db_ret.value_qw == db_cmp.value_qw) {
> + /* db was successfully rung */
> + gc->cookie = db_exc.cookie;
> + ret = 0;
> + break;
> + }
> +
> + /* XXX: doorbell was lost and need to acquire it again */
> + if (db_ret.db_status == GUC_DOORBELL_DISABLED)
> + break;
> +
> + DRM_ERROR("Cookie mismatch. Expected %d, returned %d\n",
> + db_cmp.cookie, db_ret.cookie);
> +
> + /* update the cookie to newly read cookie from GuC */
> + db_cmp.cookie = db_ret.cookie;
> + db_exc.cookie = db_ret.cookie + 1;
> + if (db_exc.cookie == 0)
> + db_exc.cookie = 1;
> + }
> +
> + /* Finally, update the cached copy of the GuC's WQ head */
> + gc->wq_head = desc->head;
> +
> + kunmap_atomic(base);
> + return ret;
> +}
> +
> /**
> * i915_guc_submit() - Submit commands through GuC
> * @client: the guc client where commands will go through
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
next prev parent reply other threads:[~2016-04-13 17:53 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-04-07 17:21 [PATCH v4 0/6] Fixes and workarounds for GuC/doorbell setup Dave Gordon
2016-04-07 17:21 ` [PATCH v4 1/6] drm/i915/guc: add doorbell map to debugfs/i915_guc_info Dave Gordon
2016-04-13 17:51 ` Yu Dai
2016-04-07 17:21 ` [PATCH v4 2/6] drm/i915/guc: move guc_ring_doorbell() nearer to callsite Dave Gordon
2016-04-13 17:52 ` Yu Dai [this message]
2016-04-07 17:21 ` [PATCH v4 3/6] drm/i915/guc: refactor doorbell management code Dave Gordon
2016-04-07 21:26 ` Yu Dai
2016-04-12 7:30 ` Dave Gordon
2016-04-07 17:21 ` [PATCH v4 4/6] drm/i915/guc: (re)initialise doorbell h/w when enabling GuC submission Dave Gordon
2016-04-13 17:50 ` Yu Dai
2016-04-13 19:46 ` Dave Gordon
2016-04-13 20:13 ` Yu Dai
2016-04-20 15:19 ` Dave Gordon
2016-04-07 17:21 ` [PATCH v4 5/6] drm/i915/guc: disable GuC submission earlier during GuC (re)load Dave Gordon
2016-04-13 17:51 ` Yu Dai
2016-04-07 17:21 ` [PATCH v4 6/6] DO NOT MERGE: add enable_guc_loading parameter Dave Gordon
2016-04-08 7:59 ` ✗ Fi.CI.BAT: failure for Fixes and workarounds for GuC/doorbell setup Patchwork
2016-04-20 15:28 ` Dave Gordon
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=570E874D.7000504@intel.com \
--to=yu.dai@intel.com \
--cc=david.s.gordon@intel.com \
--cc=intel-gfx@lists.freedesktop.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