From: Jeykumar Sankaran <jsanka-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
To: Sean Paul <sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
Cc: Sean Paul <seanpaul-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>,
linux-arm-msm-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
robdclark-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org,
freedreno-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org,
dri-devel-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
Subject: Re: [PATCH v2 15/24] drm/msm: dpu: Stop using encoder->crtc pointer
Date: Mon, 03 Dec 2018 11:43:43 -0800 [thread overview]
Message-ID: <a86861caa0c7b9a6a3615dcc8034d464@codeaurora.org> (raw)
In-Reply-To: <20181126215336.GE154160@art_vandelay>
On 2018-11-26 13:53, Sean Paul wrote:
> On Mon, Nov 19, 2018 at 12:03:53PM -0800, Jeykumar Sankaran wrote:
>> On 2018-11-16 13:14, Sean Paul wrote:
>> > On Fri, Nov 16, 2018 at 12:05:09PM -0800, Jeykumar Sankaran wrote:
>> > > On 2018-11-16 10:42, Sean Paul wrote:
>> > > > From: Sean Paul <seanpaul@chromium.org>
>> > > >
>> > > > It's for legacy drivers, for atomic drivers
> crtc->state->encoder_mask
>> > > > should be used to map encoder to crtc.
>> > > >
>> > > > Changes in v2:
>> > > > - None
>> > > >
>> > > > Signed-off-by: Sean Paul <seanpaul@chromium.org>
>> > > > ---
>> > > > drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c | 46
>> > ++++++++----------------
>> > > > drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 19 +++++++---
>> > > > 2 files changed, 29 insertions(+), 36 deletions(-)
>> > > >
>> > > > diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
>> > > > b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
>> > > > index 156f4c77ca44..a008a87a8113 100644
>> > > > --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
>> > > > +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
>> > > > @@ -284,9 +284,9 @@ enum dpu_intf_mode
> dpu_crtc_get_intf_mode(struct
>> > > > drm_crtc *crtc)
>> > > > return INTF_MODE_NONE;
>> > > > }
>> > > >
>> > > > - drm_for_each_encoder(encoder, crtc->dev)
>> > > > - if (encoder->crtc == crtc)
>> > > > - return dpu_encoder_get_intf_mode(encoder);
>> > > > + /* TODO: Returns the first INTF_MODE, could there be
> multiple
>> > > > values? */
>> > > > + drm_for_each_encoder_mask(encoder, crtc->dev,
>> > > > crtc->state->encoder_mask)
>> > > > + return dpu_encoder_get_intf_mode(encoder);
>> > > >
>> > > > return INTF_MODE_NONE;
>> > > > }
>> > > > @@ -551,13 +551,9 @@ static void dpu_crtc_atomic_begin(struct
> drm_crtc
>> > > > *crtc,
>> > > > spin_unlock_irqrestore(&dev->event_lock, flags);
>> > > > }
>> > > >
>> > > > - list_for_each_entry(encoder,
> &dev->mode_config.encoder_list, head)
>> > > > {
>> > > > - if (encoder->crtc != crtc)
>> > > > - continue;
>> > > > -
>> > > > - /* encoder will trigger pending mask now */
>> > > > + /* encoder will trigger pending mask now */
>> > > > + drm_for_each_encoder_mask(encoder, crtc->dev,
>> > > > crtc->state->encoder_mask)
>> > > > dpu_encoder_trigger_kickoff_pending(encoder);
>> > > > - }
>> > > >
>> > > > /*
>> > > > * If no mixers have been allocated in
> dpu_crtc_atomic_check(),
>> > > > @@ -704,7 +700,6 @@ static int
> _dpu_crtc_wait_for_frame_done(struct
>> > > > drm_crtc *crtc)
>> > > > void dpu_crtc_commit_kickoff(struct drm_crtc *crtc)
>> > > > {
>> > > > struct drm_encoder *encoder;
>> > > > - struct drm_device *dev = crtc->dev;
>> > > > struct dpu_crtc *dpu_crtc = to_dpu_crtc(crtc);
>> > > > struct dpu_kms *dpu_kms = _dpu_crtc_get_kms(crtc);
>> > > > struct dpu_crtc_state *cstate =
> to_dpu_crtc_state(crtc->state);
>> > > > @@ -720,16 +715,13 @@ void dpu_crtc_commit_kickoff(struct drm_crtc
>> > > > *crtc)
>> > > >
>> > > > DPU_ATRACE_BEGIN("crtc_commit");
>> > > >
>> > > > - list_for_each_entry(encoder,
> &dev->mode_config.encoder_list, head)
>> > > > {
>> > > > + /*
>> > > > + * Encoder will flush/start now, unless it has a tx
> pending. If
>> > > > so, it
>> > > > + * may delay and flush at an irq event (e.g. ppdone)
>> > > > + */
>> > > > + drm_for_each_encoder_mask(encoder, crtc->dev,
>> > > > + crtc->state->encoder_mask) {
>> > > > struct dpu_encoder_kickoff_params params = { 0 };
>> > > > -
>> > > > - if (encoder->crtc != crtc)
>> > > > - continue;
>> > > > -
>> > > > - /*
>> > > > - * Encoder will flush/start now, unless it has a
> tx
>> > > > pending.
>> > > > - * If so, it may delay and flush at an irq event
> (e.g.
>> > > > ppdone)
>> > > > - */
>> > > > dpu_encoder_prepare_for_kickoff(encoder, ¶ms);
>> > > > }
>> > > >
>> > > > @@ -754,12 +746,8 @@ void dpu_crtc_commit_kickoff(struct drm_crtc
>> > *crtc)
>> > > >
>> > > > dpu_vbif_clear_errors(dpu_kms);
>> > > >
>> > > > - list_for_each_entry(encoder,
> &dev->mode_config.encoder_list, head)
>> > > > {
>> > > > - if (encoder->crtc != crtc)
>> > > > - continue;
>> > > > -
>> > > > + drm_for_each_encoder_mask(encoder, crtc->dev,
>> > > > crtc->state->encoder_mask)
>> > > > dpu_encoder_kickoff(encoder);
>> > > > - }
>> > > We wont be holding the modeset locks here (and in crtc_atomic_begin)
>> > > in
>> > the
>> > > display thread. Is
>> > > it safe to iterate over encoder_mask?
>> >
>> > Hmm, I'm not sure I follow. AFAICT, there are 2 callsites for
>> > dpu_crtc_commit_kickoff():
>> >
>> > 1- dpu_kms_encoder_enable() which is called via the
>> > encoder->funcs->enable
>> > hook
>> > 2- dpu_kms_commit() which is called in the
>> > mode_config->funcs->atomic_commit_tail
>>
>> atomic_commit_tail will be called from the WQ. Do we hold
>> the modeset locks in the WQ? I believe userspace thread will drop
>> the locks at the end of drm_mode_atomic_ioctl.
>
> Accessing state is safe in non-blocking commits due to stalling any
> subsequent
> commits until the previous one has completed. Check out the comment at:
>
> https://cgit.freedesktop.org/drm-tip/tree/drivers/gpu/drm/drm_atomic_helpe
> r.c#n1754
>
> Sean
>
Thanks for the link Sean!
Reviewed-by: Jeykumar Sankaran <jsanka@codeaurora.org>
>>
>> Thanks
>> >
>> > Both of these callsites will hold the modeset locks.
>> >
>> > Am I missing something?
>> >
>> > Thanks for your review,
>> >
>> > Sean
>> >
>> > > >
>> > > > end:
>> > > > reinit_completion(&dpu_crtc->frame_done_comp);
>> > > > @@ -883,11 +871,8 @@ static void dpu_crtc_disable(struct drm_crtc
>> > *crtc)
>> > > >
>> > > > dpu_core_perf_crtc_update(crtc, 0, true);
>> > > >
>> > > > - drm_for_each_encoder(encoder, crtc->dev) {
>> > > > - if (encoder->crtc != crtc)
>> > > > - continue;
>> > > > + drm_for_each_encoder_mask(encoder, crtc->dev,
>> > > > crtc->state->encoder_mask)
>> > > > dpu_encoder_register_frame_event_callback(encoder,
> NULL,
>> > > > NULL);
>> > > > - }
>> > > >
>> > > > memset(cstate->mixers, 0, sizeof(cstate->mixers));
>> > > > cstate->num_mixers = 0;
>> > > > @@ -922,12 +907,9 @@ static void dpu_crtc_enable(struct drm_crtc
>> > *crtc,
>> > > > DRM_DEBUG_KMS("crtc%d\n", crtc->base.id);
>> > > > dpu_crtc = to_dpu_crtc(crtc);
>> > > >
>> > > > - drm_for_each_encoder(encoder, crtc->dev) {
>> > > > - if (encoder->crtc != crtc)
>> > > > - continue;
>> > > > + drm_for_each_encoder_mask(encoder, crtc->dev,
>> > > > crtc->state->encoder_mask)
>> > > > dpu_encoder_register_frame_event_callback(encoder,
>> > > > dpu_crtc_frame_event_cb, (void
> *)crtc);
>> > > > - }
>> > > >
>> > > > mutex_lock(&dpu_crtc->crtc_lock);
>> > > > trace_dpu_crtc_enable(DRMID(crtc), true, dpu_crtc);
>> > > > diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
>> > > > b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
>> > > > index 1bec4540f3e1..64134d619748 100644
>> > > > --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
>> > > > +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
>> > > > @@ -320,7 +320,10 @@ static void dpu_kms_prepare_commit(struct
> msm_kms
>> > > > *kms,
>> > > > struct dpu_kms *dpu_kms;
>> > > > struct msm_drm_private *priv;
>> > > > struct drm_device *dev;
>> > > > + struct drm_crtc *crtc;
>> > > > + struct drm_crtc_state *crtc_state;
>> > > > struct drm_encoder *encoder;
>> > > > + int i;
>> > > >
>> > > > if (!kms)
>> > > > return;
>> > > > @@ -332,9 +335,13 @@ static void dpu_kms_prepare_commit(struct
> msm_kms
>> > > > *kms,
>> > > > priv = dev->dev_private;
>> > > > pm_runtime_get_sync(&dpu_kms->pdev->dev);
>> > > >
>> > > > - list_for_each_entry(encoder,
> &dev->mode_config.encoder_list, head)
>> > > > - if (encoder->crtc != NULL)
>> > > > + /* Call prepare_commit for all affected encoders */
>> > > > + for_each_new_crtc_in_state(state, crtc, crtc_state, i) {
>> > > > + drm_for_each_encoder_mask(encoder, crtc->dev,
>> > > > +
> crtc_state->encoder_mask) {
>> > > > dpu_encoder_prepare_commit(encoder);
>> > > > + }
>> > > > + }
>> > > > }
>> > > >
>> > > > /*
>> > > > @@ -344,13 +351,17 @@ static void dpu_kms_prepare_commit(struct
>> > msm_kms
>> > > > *kms,
>> > > > void dpu_kms_encoder_enable(struct drm_encoder *encoder)
>> > > > {
>> > > > const struct drm_encoder_helper_funcs *funcs =
>> > > > encoder->helper_private;
>> > > > - struct drm_crtc *crtc = encoder->crtc;
>> > > > + struct drm_device *dev = encoder->dev;
>> > > > + struct drm_crtc *crtc;
>> > > >
>> > > > /* Forward this enable call to the commit hook */
>> > > > if (funcs && funcs->commit)
>> > > > funcs->commit(encoder);
>> > > >
>> > > > - if (crtc && crtc->state->active) {
>> > > > + drm_for_each_crtc(crtc, dev) {
>> > > > + if (!(crtc->state->encoder_mask &
>> > > > drm_encoder_mask(encoder)))
>> > > > + continue;
>> > > > +
>> > > > trace_dpu_kms_enc_enable(DRMID(crtc));
>> > > > dpu_crtc_commit_kickoff(crtc);
>> > > > }
>> > >
>> > > --
>> > > Jeykumar S
>>
>> --
>> Jeykumar S
--
Jeykumar S
_______________________________________________
Freedreno mailing list
Freedreno@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/freedreno
next prev parent reply other threads:[~2018-12-03 19:43 UTC|newest]
Thread overview: 52+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-11-16 18:42 [PATCH v2 00/24] drm/msm: Various dpu locking and legacy cleanups Sean Paul
2018-11-16 18:42 ` [PATCH v2 01/24] drm/msm: dpu: Remove dpu_power_handle_get_dbus_name() Sean Paul
2018-11-16 18:42 ` [PATCH v2 03/24] drm/msm: dpu: Remove dpu_power_client Sean Paul
2018-11-16 18:42 ` [PATCH v2 04/24] drm/msm: dpu: Don't use power_event for vbif_init_memtypes Sean Paul
[not found] ` <20181116184238.170034-5-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-12-03 19:49 ` Jeykumar Sankaran
[not found] ` <20181116184238.170034-1-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-16 18:42 ` [PATCH v2 02/24] drm/msm: dpu: Remove unused trace_dpu_perf_update_bus() Sean Paul
2018-11-16 18:42 ` [PATCH v2 05/24] drm/msm: dpu: Handle crtc pm_runtime_resume() directly Sean Paul
[not found] ` <20181116184238.170034-6-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-19 19:41 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 06/24] drm/msm: dpu: Remove power_handle from core_perf Sean Paul
2018-11-16 18:42 ` [PATCH v2 07/24] drm/msm: dpu: Include dpu_io_util.h directly in dpu_kms.h Sean Paul
2018-11-16 18:42 ` [PATCH v2 08/24] drm/msm: dpu: Move DPU_POWER_HANDLE_DBUS_ID to core_perf Sean Paul
2018-11-16 18:42 ` [PATCH v2 10/24] drm/msm: dpu: Fix typo in dpu_encoder Sean Paul
2018-11-16 18:42 ` [PATCH v2 12/24] drm/msm: dpu: Move crtc runtime resume to encoder Sean Paul
[not found] ` <20181116184238.170034-13-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-12-03 19:52 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 13/24] drm/msm: dpu: Don't drop locks in crtc_vblank_enable Sean Paul
2018-11-19 19:43 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 14/24] drm/msm: dpu: Grab the modeset locks in frame_event Sean Paul
2018-11-16 20:02 ` Jeykumar Sankaran
[not found] ` <0405e429330ba1de19d96a6918a6daae-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
2018-11-16 21:18 ` Sean Paul
[not found] ` <20181116184238.170034-15-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-30 22:00 ` [PATCH v3 " Sean Paul
2018-12-03 19:53 ` Jeykumar Sankaran
[not found] ` <20181130220011.39644-1-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-12-03 21:45 ` Daniel Vetter
2018-11-16 18:42 ` [PATCH v2 15/24] drm/msm: dpu: Stop using encoder->crtc pointer Sean Paul
[not found] ` <20181116184238.170034-16-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-16 20:05 ` Jeykumar Sankaran
[not found] ` <580ced340265a3c45b31836291bfbb4c-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
2018-11-16 21:14 ` Sean Paul
2018-11-19 20:03 ` Jeykumar Sankaran
2018-11-26 21:53 ` Sean Paul
2018-12-03 19:43 ` Jeykumar Sankaran [this message]
2018-11-16 18:42 ` [PATCH v2 16/24] drm/msm: dpu: Add modeset lock checks where applicable Sean Paul
2018-11-16 20:05 ` Jeykumar Sankaran
2019-10-09 22:20 ` Daniel Vetter
2019-10-10 13:13 ` Sean Paul
2018-11-16 18:42 ` [PATCH v2 17/24] drm/msm: dpu: Move pm_runtime_(get|put) from vblank_enable Sean Paul
2018-11-19 20:09 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 18/24] drm/msm: dpu: Remove crtc_lock from setup_mixers Sean Paul
[not found] ` <20181116184238.170034-19-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-19 20:09 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 20/24] drm/msm: dpu: Use atomic_disable for dpu_crtc_disable Sean Paul
[not found] ` <20181116184238.170034-21-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-19 20:12 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 21/24] drm/msm: dpu: Don't bother checking ->enabled in dpu_crtc_vblank Sean Paul
[not found] ` <20181116184238.170034-22-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-19 20:13 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 23/24] drm/msm: dpu: Remove vblank_requested flag from dpu_crtc Sean Paul
[not found] ` <20181116184238.170034-24-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-19 20:14 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 24/24] drm/msm: dpu: Remove crtc_lock Sean Paul
[not found] ` <20181116184238.170034-25-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-19 20:14 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 09/24] drm/msm: dpu: Remove dpu_power_handle Sean Paul
2018-11-16 18:42 ` [PATCH v2 11/24] drm/msm: dpu: Add ->enabled to dpu_encoder_virt Sean Paul
2018-12-03 19:51 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 19/24] drm/msm: dpu: Remove vblank_callback from encoder Sean Paul
[not found] ` <20181116184238.170034-20-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-19 20:11 ` Jeykumar Sankaran
2018-11-16 18:42 ` [PATCH v2 22/24] drm/msm: dpu: Separate crtc assignment from vblank enable Sean Paul
[not found] ` <20181116184238.170034-23-sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org>
2018-11-19 20:13 ` Jeykumar Sankaran
2018-12-04 15:15 ` [PATCH v2 00/24] drm/msm: Various dpu locking and legacy cleanups Sean Paul
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=a86861caa0c7b9a6a3615dcc8034d464@codeaurora.org \
--to=jsanka-sgv2jx0feol9jmxxk+q4oq@public.gmane.org \
--cc=dri-devel-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
--cc=freedreno-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
--cc=linux-arm-msm-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=robdclark-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org \
--cc=sean-p7yTbzM4H96eqtR555YLDQ@public.gmane.org \
--cc=seanpaul-F7+t8E8rja9g9hUCZPvPmw@public.gmane.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).