| Message ID | 20200415074034.175360-57-daniel.vetter@ffwll.ch (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | devm_drm_dev_alloc, v2 | expand |
On Wed, Apr 15, 2020 at 09:40:31AM +0200, Daniel Vetter wrote: > Since aspeed doesn't use devm_kzalloc anymore we can use the managed > mode config cleanup. > > Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> > Cc: Joel Stanley <joel@jms.id.au> > Cc: Andrew Jeffery <andrew@aj.id.au> > Cc: linux-aspeed@lists.ozlabs.org > Cc: linux-arm-kernel@lists.infradead.org Hmm, the helper function makes no sense, maybe embed it? One Q below. Whith Q addressed: Acked-by: Sam Ravnborg <sam@ravnborg.org> > --- > drivers/gpu/drm/aspeed/aspeed_gfx_drv.c | 11 ++++++----- > 1 file changed, 6 insertions(+), 5 deletions(-) > > diff --git a/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c b/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c > index 6b27242b9ee3..6e464b84a256 100644 > --- a/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c > +++ b/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c > @@ -63,15 +63,15 @@ static const struct drm_mode_config_funcs aspeed_gfx_mode_config_funcs = { > .atomic_commit = drm_atomic_helper_commit, > }; > > -static void aspeed_gfx_setup_mode_config(struct drm_device *drm) > +static int aspeed_gfx_setup_mode_config(struct drm_device *drm) > { > - drm_mode_config_init(drm); > - > drm->mode_config.min_width = 0; > drm->mode_config.min_height = 0; > drm->mode_config.max_width = 800; > drm->mode_config.max_height = 600; > drm->mode_config.funcs = &aspeed_gfx_mode_config_funcs; > + > + return drmm_mode_config_init(drm); I do not see anything that documents that it is OK to init min/max width/heigh not funcs before drmm_mode_config_init() is called. Maybe drmm_mode_config_init() gain an memset(drm->mode_config), and we loose all the assingments from before the call to init(). Also most (all?) other users of drmm_mode_config_init() set them after the call to drmm_mode_config_init(). So re-order here and then embed while you are touching the code again. Sam > } > > static irqreturn_t aspeed_gfx_irq_handler(int irq, void *data) > @@ -144,7 +144,9 @@ static int aspeed_gfx_load(struct drm_device *drm) > writel(0, priv->base + CRT_CTRL1); > writel(0, priv->base + CRT_CTRL2); > > - aspeed_gfx_setup_mode_config(drm); > + ret = aspeed_gfx_setup_mode_config(drm); > + if (ret < 0) > + return ret; > > ret = drm_vblank_init(drm, 1); > if (ret < 0) { > @@ -181,7 +183,6 @@ static int aspeed_gfx_load(struct drm_device *drm) > static void aspeed_gfx_unload(struct drm_device *drm) > { > drm_kms_helper_poll_fini(drm); > - drm_mode_config_cleanup(drm); > } > > DEFINE_DRM_GEM_CMA_FOPS(fops); > -- > 2.25.1 > > _______________________________________________ > dri-devel mailing list > dri-devel@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/dri-devel
On Fri, Apr 24, 2020 at 08:10:02PM +0200, Sam Ravnborg wrote: > On Wed, Apr 15, 2020 at 09:40:31AM +0200, Daniel Vetter wrote: > > Since aspeed doesn't use devm_kzalloc anymore we can use the managed > > mode config cleanup. > > > > Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> > > Cc: Joel Stanley <joel@jms.id.au> > > Cc: Andrew Jeffery <andrew@aj.id.au> > > Cc: linux-aspeed@lists.ozlabs.org > > Cc: linux-arm-kernel@lists.infradead.org > > Hmm, the helper function makes no sense, maybe embed it? > > One Q below. Whith Q addressed: > Acked-by: Sam Ravnborg <sam@ravnborg.org> > > > --- > > drivers/gpu/drm/aspeed/aspeed_gfx_drv.c | 11 ++++++----- > > 1 file changed, 6 insertions(+), 5 deletions(-) > > > > diff --git a/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c b/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c > > index 6b27242b9ee3..6e464b84a256 100644 > > --- a/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c > > +++ b/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c > > @@ -63,15 +63,15 @@ static const struct drm_mode_config_funcs aspeed_gfx_mode_config_funcs = { > > .atomic_commit = drm_atomic_helper_commit, > > }; > > > > -static void aspeed_gfx_setup_mode_config(struct drm_device *drm) > > +static int aspeed_gfx_setup_mode_config(struct drm_device *drm) > > { > > - drm_mode_config_init(drm); > > - > > drm->mode_config.min_width = 0; > > drm->mode_config.min_height = 0; > > drm->mode_config.max_width = 800; > > drm->mode_config.max_height = 600; > > drm->mode_config.funcs = &aspeed_gfx_mode_config_funcs; > > + > > + return drmm_mode_config_init(drm); > > I do not see anything that documents that it is OK to init min/max > width/heigh not funcs before drmm_mode_config_init() is called. > Maybe drmm_mode_config_init() gain an memset(drm->mode_config), > and we loose all the assingments from before the call to init(). > > Also most (all?) other users of drmm_mode_config_init() > set them after the call to drmm_mode_config_init(). > So re-order here and then embed while you are touching the code again. Only reason I've done it like this is that it saves a few lines of diff compared to other options. Wrt calling stuff the wrong way round: We pretty much assume throughout that structures are allocated with kzalloc, none of our _init() functions in drm have a memset. We'd break the world if we start doing memset() in random _init() functions I think. Also the main aspeed_gfx_load() function is quite long already, smashing more random stuff in there won't help it's readability. Anyway I don't care, if you insist I'm happy to repaint this in whatever color choice you deem best :-) Cheers, Daniel > > Sam > > > } > > > > static irqreturn_t aspeed_gfx_irq_handler(int irq, void *data) > > @@ -144,7 +144,9 @@ static int aspeed_gfx_load(struct drm_device *drm) > > writel(0, priv->base + CRT_CTRL1); > > writel(0, priv->base + CRT_CTRL2); > > > > - aspeed_gfx_setup_mode_config(drm); > > + ret = aspeed_gfx_setup_mode_config(drm); > > + if (ret < 0) > > + return ret; > > > > ret = drm_vblank_init(drm, 1); > > if (ret < 0) { > > @@ -181,7 +183,6 @@ static int aspeed_gfx_load(struct drm_device *drm) > > static void aspeed_gfx_unload(struct drm_device *drm) > > { > > drm_kms_helper_poll_fini(drm); > > - drm_mode_config_cleanup(drm); > > } > > > > DEFINE_DRM_GEM_CMA_FOPS(fops); > > -- > > 2.25.1 > > > > _______________________________________________ > > dri-devel mailing list > > dri-devel@lists.freedesktop.org > > https://lists.freedesktop.org/mailman/listinfo/dri-devel
On Tue, Apr 28, 2020 at 04:12:21PM +0200, Daniel Vetter wrote: > On Fri, Apr 24, 2020 at 08:10:02PM +0200, Sam Ravnborg wrote: > > On Wed, Apr 15, 2020 at 09:40:31AM +0200, Daniel Vetter wrote: > > > Since aspeed doesn't use devm_kzalloc anymore we can use the managed > > > mode config cleanup. > > > > > > Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> > > > Cc: Joel Stanley <joel@jms.id.au> > > > Cc: Andrew Jeffery <andrew@aj.id.au> > > > Cc: linux-aspeed@lists.ozlabs.org > > > Cc: linux-arm-kernel@lists.infradead.org > > > > Hmm, the helper function makes no sense, maybe embed it? > > > > One Q below. Whith Q addressed: > > Acked-by: Sam Ravnborg <sam@ravnborg.org> > > > > > --- > > > drivers/gpu/drm/aspeed/aspeed_gfx_drv.c | 11 ++++++----- > > > 1 file changed, 6 insertions(+), 5 deletions(-) > > > > > > diff --git a/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c b/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c > > > index 6b27242b9ee3..6e464b84a256 100644 > > > --- a/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c > > > +++ b/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c > > > @@ -63,15 +63,15 @@ static const struct drm_mode_config_funcs aspeed_gfx_mode_config_funcs = { > > > .atomic_commit = drm_atomic_helper_commit, > > > }; > > > > > > -static void aspeed_gfx_setup_mode_config(struct drm_device *drm) > > > +static int aspeed_gfx_setup_mode_config(struct drm_device *drm) > > > { > > > - drm_mode_config_init(drm); > > > - > > > drm->mode_config.min_width = 0; > > > drm->mode_config.min_height = 0; > > > drm->mode_config.max_width = 800; > > > drm->mode_config.max_height = 600; > > > drm->mode_config.funcs = &aspeed_gfx_mode_config_funcs; > > > + > > > + return drmm_mode_config_init(drm); > > > > I do not see anything that documents that it is OK to init min/max > > width/heigh not funcs before drmm_mode_config_init() is called. > > Maybe drmm_mode_config_init() gain an memset(drm->mode_config), > > and we loose all the assingments from before the call to init(). > > > > Also most (all?) other users of drmm_mode_config_init() > > set them after the call to drmm_mode_config_init(). > > So re-order here and then embed while you are touching the code again. > > Only reason I've done it like this is that it saves a few lines of diff > compared to other options. > > Wrt calling stuff the wrong way round: We pretty much assume throughout > that structures are allocated with kzalloc, none of our _init() functions > in drm have a memset. We'd break the world if we start doing memset() in > random _init() functions I think. > > Also the main aspeed_gfx_load() function is quite long already, smashing > more random stuff in there won't help it's readability. > > Anyway I don't care, if you insist I'm happy to repaint this in whatever > color choice you deem best :-) From the principle of least suprises, you should at least call init and then set min_width and friends. This is easy to do in the helper, so easy to avoid the inlining I suggested. Sam
diff --git a/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c b/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c index 6b27242b9ee3..6e464b84a256 100644 --- a/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c +++ b/drivers/gpu/drm/aspeed/aspeed_gfx_drv.c @@ -63,15 +63,15 @@ static const struct drm_mode_config_funcs aspeed_gfx_mode_config_funcs = { .atomic_commit = drm_atomic_helper_commit, }; -static void aspeed_gfx_setup_mode_config(struct drm_device *drm) +static int aspeed_gfx_setup_mode_config(struct drm_device *drm) { - drm_mode_config_init(drm); - drm->mode_config.min_width = 0; drm->mode_config.min_height = 0; drm->mode_config.max_width = 800; drm->mode_config.max_height = 600; drm->mode_config.funcs = &aspeed_gfx_mode_config_funcs; + + return drmm_mode_config_init(drm); } static irqreturn_t aspeed_gfx_irq_handler(int irq, void *data) @@ -144,7 +144,9 @@ static int aspeed_gfx_load(struct drm_device *drm) writel(0, priv->base + CRT_CTRL1); writel(0, priv->base + CRT_CTRL2); - aspeed_gfx_setup_mode_config(drm); + ret = aspeed_gfx_setup_mode_config(drm); + if (ret < 0) + return ret; ret = drm_vblank_init(drm, 1); if (ret < 0) { @@ -181,7 +183,6 @@ static int aspeed_gfx_load(struct drm_device *drm) static void aspeed_gfx_unload(struct drm_device *drm) { drm_kms_helper_poll_fini(drm); - drm_mode_config_cleanup(drm); } DEFINE_DRM_GEM_CMA_FOPS(fops);
Since aspeed doesn't use devm_kzalloc anymore we can use the managed mode config cleanup. Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Joel Stanley <joel@jms.id.au> Cc: Andrew Jeffery <andrew@aj.id.au> Cc: linux-aspeed@lists.ozlabs.org Cc: linux-arm-kernel@lists.infradead.org --- drivers/gpu/drm/aspeed/aspeed_gfx_drv.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)