| Message ID | 1642732195-25349-1-git-send-email-quic_abhinavk@quicinc.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [RFC] drm: allow passing a real encoder object for wb connector | expand |
Hi Abhinav, Thank you for the patch. On Thu, Jan 20, 2022 at 06:29:55PM -0800, Abhinav Kumar wrote: > Instead of creating an internal encoder for the writeback > connector to satisfy DRM requirements, allow the clients > to pass a real encoder to it by changing the drm_writeback's > encoder to a pointer. > > If a real encoder is not passed, drm_writeback_connector_init > will internally allocate one. > > This will help the clients to manage the real encoder states > better as they will allocate and maintain the encoder. A writeback connector is a bit of a hack. It was implemented that way to minimize the extensions to the KMS userspace API for writeback support. There's no "encoder" there, as there's no real "connector" either. The only reason we register a drm_encoder in the writeback implementation is because encoders are exposed to userspace and are thus required (this is considered a historical mistake that we can't fix anymore). Why do you thus need to create a "real encoder" ? > Signed-off-by: Abhinav Kumar <quic_abhinavk@quicinc.com> > --- > drivers/gpu/drm/drm_writeback.c | 11 +++++++---- > include/drm/drm_writeback.h | 2 +- > 2 files changed, 8 insertions(+), 5 deletions(-) > > diff --git a/drivers/gpu/drm/drm_writeback.c b/drivers/gpu/drm/drm_writeback.c > index dccf4504..fdb7381 100644 > --- a/drivers/gpu/drm/drm_writeback.c > +++ b/drivers/gpu/drm/drm_writeback.c > @@ -189,8 +189,11 @@ int drm_writeback_connector_init(struct drm_device *dev, > if (IS_ERR(blob)) > return PTR_ERR(blob); > > - drm_encoder_helper_add(&wb_connector->encoder, enc_helper_funcs); > - ret = drm_encoder_init(dev, &wb_connector->encoder, > + /* allocate the internal drm encoder if a real one wasnt passed */ > + if (!wb_connector->encoder) > + wb_connector->encoder = devm_kzalloc(dev->dev, sizeof(struct drm_encoder), GFP_KERNEL); > + drm_encoder_helper_add(wb_connector->encoder, enc_helper_funcs); > + ret = drm_encoder_init(dev, wb_connector->encoder, > &drm_writeback_encoder_funcs, > DRM_MODE_ENCODER_VIRTUAL, NULL); > if (ret) > @@ -204,7 +207,7 @@ int drm_writeback_connector_init(struct drm_device *dev, > goto connector_fail; > > ret = drm_connector_attach_encoder(connector, > - &wb_connector->encoder); > + wb_connector->encoder); > if (ret) > goto attach_fail; > > @@ -233,7 +236,7 @@ int drm_writeback_connector_init(struct drm_device *dev, > attach_fail: > drm_connector_cleanup(connector); > connector_fail: > - drm_encoder_cleanup(&wb_connector->encoder); > + drm_encoder_cleanup(wb_connector->encoder); > fail: > drm_property_blob_put(blob); > return ret; > diff --git a/include/drm/drm_writeback.h b/include/drm/drm_writeback.h > index 9697d27..f0d8147 100644 > --- a/include/drm/drm_writeback.h > +++ b/include/drm/drm_writeback.h > @@ -31,7 +31,7 @@ struct drm_writeback_connector { > * by passing the @enc_funcs parameter to drm_writeback_connector_init() > * function. > */ > - struct drm_encoder encoder; > + struct drm_encoder *encoder; > > /** > * @pixel_formats_blob_ptr:
Hi Laurent Thanks for the response. On 1/20/2022 6:43 PM, Laurent Pinchart wrote: > Hi Abhinav, > > Thank you for the patch. > > On Thu, Jan 20, 2022 at 06:29:55PM -0800, Abhinav Kumar wrote: >> Instead of creating an internal encoder for the writeback >> connector to satisfy DRM requirements, allow the clients >> to pass a real encoder to it by changing the drm_writeback's >> encoder to a pointer. >> >> If a real encoder is not passed, drm_writeback_connector_init >> will internally allocate one. >> >> This will help the clients to manage the real encoder states >> better as they will allocate and maintain the encoder. > > A writeback connector is a bit of a hack. It was implemented that way to > minimize the extensions to the KMS userspace API for writeback support. > There's no "encoder" there, as there's no real "connector" either. The > only reason we register a drm_encoder in the writeback implementation is > because encoders are exposed to userspace and are thus required (this is > considered a historical mistake that we can't fix anymore). Why do you > thus need to create a "real encoder" ? On some hardware, it is possible that the writeback encoder is shared. That is, in terms of hardware resources, we can only mutually drive either the physical interface or the writeback one. In that case, it would be better that drm_writeback accepts the real encoder that is being used instead of allocating a dummy one internally. Moreover, the drm_writeback_connector_init() does already accept passing our own enc_helper_funcs to perform necessary checks on the internal encoder. These hooks are provided to perform various operations on the encoder to fit the respective needs. In that case why shouldnt the writeback have its own real encoder? Thanks Abhinav > >> Signed-off-by: Abhinav Kumar <quic_abhinavk@quicinc.com> >> --- >> drivers/gpu/drm/drm_writeback.c | 11 +++++++---- >> include/drm/drm_writeback.h | 2 +- >> 2 files changed, 8 insertions(+), 5 deletions(-) >> >> diff --git a/drivers/gpu/drm/drm_writeback.c b/drivers/gpu/drm/drm_writeback.c >> index dccf4504..fdb7381 100644 >> --- a/drivers/gpu/drm/drm_writeback.c >> +++ b/drivers/gpu/drm/drm_writeback.c >> @@ -189,8 +189,11 @@ int drm_writeback_connector_init(struct drm_device *dev, >> if (IS_ERR(blob)) >> return PTR_ERR(blob); >> >> - drm_encoder_helper_add(&wb_connector->encoder, enc_helper_funcs); >> - ret = drm_encoder_init(dev, &wb_connector->encoder, >> + /* allocate the internal drm encoder if a real one wasnt passed */ >> + if (!wb_connector->encoder) >> + wb_connector->encoder = devm_kzalloc(dev->dev, sizeof(struct drm_encoder), GFP_KERNEL); >> + drm_encoder_helper_add(wb_connector->encoder, enc_helper_funcs); >> + ret = drm_encoder_init(dev, wb_connector->encoder, >> &drm_writeback_encoder_funcs, >> DRM_MODE_ENCODER_VIRTUAL, NULL); >> if (ret) >> @@ -204,7 +207,7 @@ int drm_writeback_connector_init(struct drm_device *dev, >> goto connector_fail; >> >> ret = drm_connector_attach_encoder(connector, >> - &wb_connector->encoder); >> + wb_connector->encoder); >> if (ret) >> goto attach_fail; >> >> @@ -233,7 +236,7 @@ int drm_writeback_connector_init(struct drm_device *dev, >> attach_fail: >> drm_connector_cleanup(connector); >> connector_fail: >> - drm_encoder_cleanup(&wb_connector->encoder); >> + drm_encoder_cleanup(wb_connector->encoder); >> fail: >> drm_property_blob_put(blob); >> return ret; >> diff --git a/include/drm/drm_writeback.h b/include/drm/drm_writeback.h >> index 9697d27..f0d8147 100644 >> --- a/include/drm/drm_writeback.h >> +++ b/include/drm/drm_writeback.h >> @@ -31,7 +31,7 @@ struct drm_writeback_connector { >> * by passing the @enc_funcs parameter to drm_writeback_connector_init() >> * function. >> */ >> - struct drm_encoder encoder; >> + struct drm_encoder *encoder; >> >> /** >> * @pixel_formats_blob_ptr: >
On Thu, 20 Jan 2022, Abhinav Kumar <quic_abhinavk@quicinc.com> wrote: > Instead of creating an internal encoder for the writeback > connector to satisfy DRM requirements, allow the clients > to pass a real encoder to it by changing the drm_writeback's > encoder to a pointer. > > If a real encoder is not passed, drm_writeback_connector_init > will internally allocate one. > > This will help the clients to manage the real encoder states > better as they will allocate and maintain the encoder. See also the thread starting at [1], and please try to coordinate. I don't know what the end result should be like, I'm just saying please collaborate instead of racing to get one set of changes in. BR, Jani. [1] https://patchwork.freedesktop.org/patch/msgid/20220111101801.28310-1-suraj.kandpal@intel.com > > Signed-off-by: Abhinav Kumar <quic_abhinavk@quicinc.com> > --- > drivers/gpu/drm/drm_writeback.c | 11 +++++++---- > include/drm/drm_writeback.h | 2 +- > 2 files changed, 8 insertions(+), 5 deletions(-) > > diff --git a/drivers/gpu/drm/drm_writeback.c b/drivers/gpu/drm/drm_writeback.c > index dccf4504..fdb7381 100644 > --- a/drivers/gpu/drm/drm_writeback.c > +++ b/drivers/gpu/drm/drm_writeback.c > @@ -189,8 +189,11 @@ int drm_writeback_connector_init(struct drm_device *dev, > if (IS_ERR(blob)) > return PTR_ERR(blob); > > - drm_encoder_helper_add(&wb_connector->encoder, enc_helper_funcs); > - ret = drm_encoder_init(dev, &wb_connector->encoder, > + /* allocate the internal drm encoder if a real one wasnt passed */ > + if (!wb_connector->encoder) > + wb_connector->encoder = devm_kzalloc(dev->dev, sizeof(struct drm_encoder), GFP_KERNEL); > + drm_encoder_helper_add(wb_connector->encoder, enc_helper_funcs); > + ret = drm_encoder_init(dev, wb_connector->encoder, > &drm_writeback_encoder_funcs, > DRM_MODE_ENCODER_VIRTUAL, NULL); > if (ret) > @@ -204,7 +207,7 @@ int drm_writeback_connector_init(struct drm_device *dev, > goto connector_fail; > > ret = drm_connector_attach_encoder(connector, > - &wb_connector->encoder); > + wb_connector->encoder); > if (ret) > goto attach_fail; > > @@ -233,7 +236,7 @@ int drm_writeback_connector_init(struct drm_device *dev, > attach_fail: > drm_connector_cleanup(connector); > connector_fail: > - drm_encoder_cleanup(&wb_connector->encoder); > + drm_encoder_cleanup(wb_connector->encoder); > fail: > drm_property_blob_put(blob); > return ret; > diff --git a/include/drm/drm_writeback.h b/include/drm/drm_writeback.h > index 9697d27..f0d8147 100644 > --- a/include/drm/drm_writeback.h > +++ b/include/drm/drm_writeback.h > @@ -31,7 +31,7 @@ struct drm_writeback_connector { > * by passing the @enc_funcs parameter to drm_writeback_connector_init() > * function. > */ > - struct drm_encoder encoder; > + struct drm_encoder *encoder; > > /** > * @pixel_formats_blob_ptr:
Hi Jani On 1/21/2022 1:17 AM, Jani Nikula wrote: > On Thu, 20 Jan 2022, Abhinav Kumar <quic_abhinavk@quicinc.com> wrote: >> Instead of creating an internal encoder for the writeback >> connector to satisfy DRM requirements, allow the clients >> to pass a real encoder to it by changing the drm_writeback's >> encoder to a pointer. >> >> If a real encoder is not passed, drm_writeback_connector_init >> will internally allocate one. >> >> This will help the clients to manage the real encoder states >> better as they will allocate and maintain the encoder. > > See also the thread starting at [1], and please try to coordinate. > > I don't know what the end result should be like, I'm just saying please > collaborate instead of racing to get one set of changes in. > > BR, > Jani. > > > [1] https://patchwork.freedesktop.org/patch/msgid/20220111101801.28310-1-suraj.kandpal@intel.com > Thanks for pointing to this thread. Since https://patchwork.freedesktop.org/patch/469090/ has been posted earlier and is more complete in terms of handling other vendor changes, we can continue on that one. But I dont see any comments on that one yet. Hi Laurent In that case can you please check the https://patchwork.freedesktop.org/patch/469090/ thread , we can continue our discussion there. We also have the same issue too. Our encoder also maintains its own struct drm_encoder. Thanks Abhinav >> >> Signed-off-by: Abhinav Kumar <quic_abhinavk@quicinc.com> >> --- >> drivers/gpu/drm/drm_writeback.c | 11 +++++++---- >> include/drm/drm_writeback.h | 2 +- >> 2 files changed, 8 insertions(+), 5 deletions(-) >> >> diff --git a/drivers/gpu/drm/drm_writeback.c b/drivers/gpu/drm/drm_writeback.c >> index dccf4504..fdb7381 100644 >> --- a/drivers/gpu/drm/drm_writeback.c >> +++ b/drivers/gpu/drm/drm_writeback.c >> @@ -189,8 +189,11 @@ int drm_writeback_connector_init(struct drm_device *dev, >> if (IS_ERR(blob)) >> return PTR_ERR(blob); >> >> - drm_encoder_helper_add(&wb_connector->encoder, enc_helper_funcs); >> - ret = drm_encoder_init(dev, &wb_connector->encoder, >> + /* allocate the internal drm encoder if a real one wasnt passed */ >> + if (!wb_connector->encoder) >> + wb_connector->encoder = devm_kzalloc(dev->dev, sizeof(struct drm_encoder), GFP_KERNEL); >> + drm_encoder_helper_add(wb_connector->encoder, enc_helper_funcs); >> + ret = drm_encoder_init(dev, wb_connector->encoder, >> &drm_writeback_encoder_funcs, >> DRM_MODE_ENCODER_VIRTUAL, NULL); >> if (ret) >> @@ -204,7 +207,7 @@ int drm_writeback_connector_init(struct drm_device *dev, >> goto connector_fail; >> >> ret = drm_connector_attach_encoder(connector, >> - &wb_connector->encoder); >> + wb_connector->encoder); >> if (ret) >> goto attach_fail; >> >> @@ -233,7 +236,7 @@ int drm_writeback_connector_init(struct drm_device *dev, >> attach_fail: >> drm_connector_cleanup(connector); >> connector_fail: >> - drm_encoder_cleanup(&wb_connector->encoder); >> + drm_encoder_cleanup(wb_connector->encoder); >> fail: >> drm_property_blob_put(blob); >> return ret; >> diff --git a/include/drm/drm_writeback.h b/include/drm/drm_writeback.h >> index 9697d27..f0d8147 100644 >> --- a/include/drm/drm_writeback.h >> +++ b/include/drm/drm_writeback.h >> @@ -31,7 +31,7 @@ struct drm_writeback_connector { >> * by passing the @enc_funcs parameter to drm_writeback_connector_init() >> * function. >> */ >> - struct drm_encoder encoder; >> + struct drm_encoder *encoder; >> >> /** >> * @pixel_formats_blob_ptr: >
diff --git a/drivers/gpu/drm/drm_writeback.c b/drivers/gpu/drm/drm_writeback.c index dccf4504..fdb7381 100644 --- a/drivers/gpu/drm/drm_writeback.c +++ b/drivers/gpu/drm/drm_writeback.c @@ -189,8 +189,11 @@ int drm_writeback_connector_init(struct drm_device *dev, if (IS_ERR(blob)) return PTR_ERR(blob); - drm_encoder_helper_add(&wb_connector->encoder, enc_helper_funcs); - ret = drm_encoder_init(dev, &wb_connector->encoder, + /* allocate the internal drm encoder if a real one wasnt passed */ + if (!wb_connector->encoder) + wb_connector->encoder = devm_kzalloc(dev->dev, sizeof(struct drm_encoder), GFP_KERNEL); + drm_encoder_helper_add(wb_connector->encoder, enc_helper_funcs); + ret = drm_encoder_init(dev, wb_connector->encoder, &drm_writeback_encoder_funcs, DRM_MODE_ENCODER_VIRTUAL, NULL); if (ret) @@ -204,7 +207,7 @@ int drm_writeback_connector_init(struct drm_device *dev, goto connector_fail; ret = drm_connector_attach_encoder(connector, - &wb_connector->encoder); + wb_connector->encoder); if (ret) goto attach_fail; @@ -233,7 +236,7 @@ int drm_writeback_connector_init(struct drm_device *dev, attach_fail: drm_connector_cleanup(connector); connector_fail: - drm_encoder_cleanup(&wb_connector->encoder); + drm_encoder_cleanup(wb_connector->encoder); fail: drm_property_blob_put(blob); return ret; diff --git a/include/drm/drm_writeback.h b/include/drm/drm_writeback.h index 9697d27..f0d8147 100644 --- a/include/drm/drm_writeback.h +++ b/include/drm/drm_writeback.h @@ -31,7 +31,7 @@ struct drm_writeback_connector { * by passing the @enc_funcs parameter to drm_writeback_connector_init() * function. */ - struct drm_encoder encoder; + struct drm_encoder *encoder; /** * @pixel_formats_blob_ptr:
Instead of creating an internal encoder for the writeback connector to satisfy DRM requirements, allow the clients to pass a real encoder to it by changing the drm_writeback's encoder to a pointer. If a real encoder is not passed, drm_writeback_connector_init will internally allocate one. This will help the clients to manage the real encoder states better as they will allocate and maintain the encoder. Signed-off-by: Abhinav Kumar <quic_abhinavk@quicinc.com> --- drivers/gpu/drm/drm_writeback.c | 11 +++++++---- include/drm/drm_writeback.h | 2 +- 2 files changed, 8 insertions(+), 5 deletions(-)