System hangs when FIMD registers are accessed to disable hardware overlays. This is because of the clocks which are not enabled before register access.
'Hardware overlay disable' is cleaned from the FIMD probe.
Signed-off-by: Rahul Sharma rahul.sharma@samsung.com --- Based on exynos-drm-next branch.
drivers/gpu/drm/exynos/exynos_drm_fimd.c | 17 ----------------- 1 file changed, 17 deletions(-)
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimd.c b/drivers/gpu/drm/exynos/exynos_drm_fimd.c index bd30d0c..2ec634f 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimd.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimd.c @@ -703,19 +703,6 @@ static void fimd_win_disable(struct exynos_drm_manager *mgr, int zpos) win_data->enabled = false; }
-static void fimd_clear_win(struct fimd_context *ctx, int win) -{ - writel(0, ctx->regs + WINCON(win)); - writel(0, ctx->regs + VIDOSD_A(win)); - writel(0, ctx->regs + VIDOSD_B(win)); - writel(0, ctx->regs + VIDOSD_C(win)); - - if (win == 1 || win == 2) - writel(0, ctx->regs + VIDOSD_D(win)); - - fimd_shadow_protect_win(ctx, win, false); -} - static void fimd_window_suspend(struct exynos_drm_manager *mgr) { struct fimd_context *ctx = mgr->ctx; @@ -898,16 +885,12 @@ static int fimd_bind(struct device *dev, struct device *master, void *data) { struct fimd_context *ctx = fimd_manager.ctx; struct drm_device *drm_dev = data; - int win;
fimd_mgr_initialize(&fimd_manager, drm_dev); exynos_drm_crtc_create(&fimd_manager); if (ctx->display) exynos_drm_create_enc_conn(drm_dev, ctx->display);
- for (win = 0; win < WINDOWS_NR; win++) - fimd_clear_win(ctx, win); - return 0;
}
Hi Rahul,
On 05/28/2014 08:11 AM, Rahul Sharma wrote:
System hangs when FIMD registers are accessed to disable hardware overlays. This is because of the clocks which are not enabled before register access.
'Hardware overlay disable' is cleaned from the FIMD probe.
Signed-off-by: Rahul Sharma rahul.sharma@samsung.com
This patch causes regression on some exynos4210-universal_c210 devices, everything works expect colors are incorrect - it seems blue component is very dark, almost black.
Regards Andrzej
Based on exynos-drm-next branch.
drivers/gpu/drm/exynos/exynos_drm_fimd.c | 17 ----------------- 1 file changed, 17 deletions(-)
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimd.c b/drivers/gpu/drm/exynos/exynos_drm_fimd.c index bd30d0c..2ec634f 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimd.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimd.c @@ -703,19 +703,6 @@ static void fimd_win_disable(struct exynos_drm_manager *mgr, int zpos) win_data->enabled = false; }
-static void fimd_clear_win(struct fimd_context *ctx, int win) -{
- writel(0, ctx->regs + WINCON(win));
- writel(0, ctx->regs + VIDOSD_A(win));
- writel(0, ctx->regs + VIDOSD_B(win));
- writel(0, ctx->regs + VIDOSD_C(win));
- if (win == 1 || win == 2)
writel(0, ctx->regs + VIDOSD_D(win));- fimd_shadow_protect_win(ctx, win, false);
-}
static void fimd_window_suspend(struct exynos_drm_manager *mgr) { struct fimd_context *ctx = mgr->ctx; @@ -898,16 +885,12 @@ static int fimd_bind(struct device *dev, struct device *master, void *data) { struct fimd_context *ctx = fimd_manager.ctx; struct drm_device *drm_dev = data;
int win;
fimd_mgr_initialize(&fimd_manager, drm_dev); exynos_drm_crtc_create(&fimd_manager); if (ctx->display) exynos_drm_create_enc_conn(drm_dev, ctx->display);
for (win = 0; win < WINDOWS_NR; win++)
fimd_clear_win(ctx, win);return 0;
}
On 2 June 2014 14:41, Andrzej Hajda a.hajda@samsung.com wrote:
Hi Rahul,
On 05/28/2014 08:11 AM, Rahul Sharma wrote:
System hangs when FIMD registers are accessed to disable hardware overlays. This is because of the clocks which are not enabled before register access.
'Hardware overlay disable' is cleaned from the FIMD probe.
Signed-off-by: Rahul Sharma rahul.sharma@samsung.com
This patch causes regression on some exynos4210-universal_c210 devices, everything works expect colors are incorrect - it seems blue component is very dark, almost black.
Oh.... Sorry for that. I did not see any problem on 5250/5420/5800. I do not have setup for 4210. Better we should revert this patch.
Would you please help me by verifying the following patch on 4210? This is an alternate solution to the same problem.
http://www.mail-archive.com/linux-samsung-soc@vger.kernel.org/msg31426.html
Thanks Andrej, for bringing it to notice.
Regards, Rahul Sharma
Regards Andrzej
Based on exynos-drm-next branch.
drivers/gpu/drm/exynos/exynos_drm_fimd.c | 17 ----------------- 1 file changed, 17 deletions(-)
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimd.c b/drivers/gpu/drm/exynos/exynos_drm_fimd.c index bd30d0c..2ec634f 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimd.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimd.c @@ -703,19 +703,6 @@ static void fimd_win_disable(struct exynos_drm_manager *mgr, int zpos) win_data->enabled = false; }
-static void fimd_clear_win(struct fimd_context *ctx, int win) -{
writel(0, ctx->regs + WINCON(win));writel(0, ctx->regs + VIDOSD_A(win));writel(0, ctx->regs + VIDOSD_B(win));writel(0, ctx->regs + VIDOSD_C(win));if (win == 1 || win == 2)writel(0, ctx->regs + VIDOSD_D(win));fimd_shadow_protect_win(ctx, win, false);-}
static void fimd_window_suspend(struct exynos_drm_manager *mgr) { struct fimd_context *ctx = mgr->ctx; @@ -898,16 +885,12 @@ static int fimd_bind(struct device *dev, struct device *master, void *data) { struct fimd_context *ctx = fimd_manager.ctx; struct drm_device *drm_dev = data;
int win; fimd_mgr_initialize(&fimd_manager, drm_dev); exynos_drm_crtc_create(&fimd_manager); if (ctx->display) exynos_drm_create_enc_conn(drm_dev, ctx->display);for (win = 0; win < WINDOWS_NR; win++)fimd_clear_win(ctx, win);return 0;}
Hi Rahul, Andrzej,
On 02.06.2014 11:42, Rahul Sharma wrote:
On 2 June 2014 14:41, Andrzej Hajda a.hajda@samsung.com wrote:
Hi Rahul,
On 05/28/2014 08:11 AM, Rahul Sharma wrote:
System hangs when FIMD registers are accessed to disable hardware overlays. This is because of the clocks which are not enabled before register access.
'Hardware overlay disable' is cleaned from the FIMD probe.
Signed-off-by: Rahul Sharma rahul.sharma@samsung.com
This patch causes regression on some exynos4210-universal_c210 devices, everything works expect colors are incorrect - it seems blue component is very dark, almost black.
Oh.... Sorry for that. I did not see any problem on 5250/5420/5800. I do not have setup for 4210. Better we should revert this patch.
Would you please help me by verifying the following patch on 4210? This is an alternate solution to the same problem.
http://www.mail-archive.com/linux-samsung-soc@vger.kernel.org/msg31426.html
Thanks Andrej, for bringing it to notice.
I don't see how this patch could introduce such regression, as all the affected registers seem to be properly reconfigured in fimd_win_commit() anyway.
IMHO instead of reverting the patch, this issue should be investigated and fixed properly.
Best regards, Tomasz
On 06/02/2014 12:11 PM, Tomasz Figa wrote:
Hi Rahul, Andrzej,
On 02.06.2014 11:42, Rahul Sharma wrote:
On 2 June 2014 14:41, Andrzej Hajda a.hajda@samsung.com wrote:
Hi Rahul,
On 05/28/2014 08:11 AM, Rahul Sharma wrote:
System hangs when FIMD registers are accessed to disable hardware overlays. This is because of the clocks which are not enabled before register access.
'Hardware overlay disable' is cleaned from the FIMD probe.
Signed-off-by: Rahul Sharma rahul.sharma@samsung.com
This patch causes regression on some exynos4210-universal_c210 devices, everything works expect colors are incorrect - it seems blue component is very dark, almost black.
Oh.... Sorry for that. I did not see any problem on 5250/5420/5800. I do not have setup for 4210. Better we should revert this patch.
Would you please help me by verifying the following patch on 4210? This is an alternate solution to the same problem.
http://www.mail-archive.com/linux-samsung-soc@vger.kernel.org/msg31426.html
Thanks Andrej, for bringing it to notice.
I don't see how this patch could introduce such regression, as all the affected registers seem to be properly reconfigured in fimd_win_commit() anyway.
IMHO instead of reverting the patch, this issue should be investigated and fixed properly.
Best regards, Tomasz
I am looking at the problem, it is quite strange as it happens only on one of two targets I have access to. Anyway it seems that something should be added to fimd initialization sequence if we want to remove hw accessing code from probe.
Regards Andrzej
On 06/02/2014 12:42 PM, Andrzej Hajda wrote:
On 06/02/2014 12:11 PM, Tomasz Figa wrote:
Hi Rahul, Andrzej,
On 02.06.2014 11:42, Rahul Sharma wrote:
On 2 June 2014 14:41, Andrzej Hajda a.hajda@samsung.com wrote:
Hi Rahul,
On 05/28/2014 08:11 AM, Rahul Sharma wrote:
System hangs when FIMD registers are accessed to disable hardware overlays. This is because of the clocks which are not enabled before register access.
'Hardware overlay disable' is cleaned from the FIMD probe.
Signed-off-by: Rahul Sharma rahul.sharma@samsung.com
This patch causes regression on some exynos4210-universal_c210 devices, everything works expect colors are incorrect - it seems blue component is very dark, almost black.
Oh.... Sorry for that. I did not see any problem on 5250/5420/5800. I do not have setup for 4210. Better we should revert this patch.
Would you please help me by verifying the following patch on 4210? This is an alternate solution to the same problem.
http://www.mail-archive.com/linux-samsung-soc@vger.kernel.org/msg31426.html
Thanks Andrej, for bringing it to notice.
I don't see how this patch could introduce such regression, as all the affected registers seem to be properly reconfigured in fimd_win_commit() anyway.
IMHO instead of reverting the patch, this issue should be investigated and fixed properly.
Best regards, Tomasz
I am looking at the problem, it is quite strange as it happens only on one of two targets I have access to. Anyway it seems that something should be added to fimd initialization sequence if we want to remove hw accessing code from probe.
The problem is that fimd does not clear unused windows. Simple patch which helps:
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimd.c b/drivers/gpu/drm/exynos/exynos_drm_fimd.c index 2ec634f..b58fce2 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimd.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimd.c @@ -741,6 +741,8 @@ static void fimd_apply(struct exynos_drm_manager *mgr) win_data = &ctx->win_data[i]; if (win_data->enabled) fimd_win_commit(mgr, i); + else + fimd_win_disable(mgr, i); }
fimd_commit(mgr);
But I am not fully familiar with window management code, so I do not know if it does not breaks other stuff.
Regards Andrzej
Regards Andrzej
On 2014년 06월 02일 20:48, Andrzej Hajda wrote:
On 06/02/2014 12:42 PM, Andrzej Hajda wrote:
On 06/02/2014 12:11 PM, Tomasz Figa wrote:
Hi Rahul, Andrzej,
On 02.06.2014 11:42, Rahul Sharma wrote:
On 2 June 2014 14:41, Andrzej Hajda a.hajda@samsung.com wrote:
Hi Rahul,
On 05/28/2014 08:11 AM, Rahul Sharma wrote:
System hangs when FIMD registers are accessed to disable hardware overlays. This is because of the clocks which are not enabled before register access.
'Hardware overlay disable' is cleaned from the FIMD probe.
Signed-off-by: Rahul Sharma rahul.sharma@samsung.com
This patch causes regression on some exynos4210-universal_c210 devices, everything works expect colors are incorrect - it seems blue component is very dark, almost black.
Oh.... Sorry for that. I did not see any problem on 5250/5420/5800. I do not have setup for 4210. Better we should revert this patch.
Would you please help me by verifying the following patch on 4210? This is an alternate solution to the same problem.
http://www.mail-archive.com/linux-samsung-soc@vger.kernel.org/msg31426.html
Thanks Andrej, for bringing it to notice.
I don't see how this patch could introduce such regression, as all the affected registers seem to be properly reconfigured in fimd_win_commit() anyway.
IMHO instead of reverting the patch, this issue should be investigated and fixed properly.
Best regards, Tomasz
I am looking at the problem, it is quite strange as it happens only on one of two targets I have access to. Anyway it seems that something should be added to fimd initialization sequence if we want to remove hw accessing code from probe.
The problem is that fimd does not clear unused windows. Simple patch which helps:
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimd.c b/drivers/gpu/drm/exynos/exynos_drm_fimd.c index 2ec634f..b58fce2 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimd.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimd.c @@ -741,6 +741,8 @@ static void fimd_apply(struct exynos_drm_manager *mgr) win_data = &ctx->win_data[i]; if (win_data->enabled) fimd_win_commit(mgr, i);
elsefimd_win_disable(mgr, i); } fimd_commit(mgr);But I am not fully familiar with window management code, so I do not know if it does not breaks other stuff.
It looks good to me. Can you post it?
Thanks, Inki Dae
Regards Andrzej
Regards Andrzej
The patch disables non-enabled HW windows on applying configuration, it will allow to clear windows enabled by bootloader.
Signed-off-by: Andrzej Hajda a.hajda@samsung.com --- drivers/gpu/drm/exynos/exynos_drm_fimd.c | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimd.c b/drivers/gpu/drm/exynos/exynos_drm_fimd.c index bb45ab2..33161ad 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimd.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimd.c @@ -741,6 +741,8 @@ static void fimd_apply(struct exynos_drm_manager *mgr) win_data = &ctx->win_data[i]; if (win_data->enabled) fimd_win_commit(mgr, i); + else + fimd_win_disable(mgr, i); }
fimd_commit(mgr);
Gently ping.
Regards Andrzej
On 06/09/2014 04:10 PM, Andrzej Hajda wrote:
The patch disables non-enabled HW windows on applying configuration, it will allow to clear windows enabled by bootloader.
Signed-off-by: Andrzej Hajda a.hajda@samsung.com
drivers/gpu/drm/exynos/exynos_drm_fimd.c | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimd.c b/drivers/gpu/drm/exynos/exynos_drm_fimd.c index bb45ab2..33161ad 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimd.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimd.c @@ -741,6 +741,8 @@ static void fimd_apply(struct exynos_drm_manager *mgr) win_data = &ctx->win_data[i]; if (win_data->enabled) fimd_win_commit(mgr, i);
elsefimd_win_disable(mgr, i);}
fimd_commit(mgr);
On 2014년 06월 23일 21:12, Andrzej Hajda wrote:
Gently ping.
Oops, sorry. Applied.
Thanks, Inki Dae
Regards Andrzej
On 06/09/2014 04:10 PM, Andrzej Hajda wrote:
The patch disables non-enabled HW windows on applying configuration, it will allow to clear windows enabled by bootloader.
Signed-off-by: Andrzej Hajda a.hajda@samsung.com
drivers/gpu/drm/exynos/exynos_drm_fimd.c | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimd.c b/drivers/gpu/drm/exynos/exynos_drm_fimd.c index bb45ab2..33161ad 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimd.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimd.c @@ -741,6 +741,8 @@ static void fimd_apply(struct exynos_drm_manager *mgr) win_data = &ctx->win_data[i]; if (win_data->enabled) fimd_win_commit(mgr, i);
elsefimd_win_disable(mgr, i);}
fimd_commit(mgr);
Hi Rahul,
On 06/02/2014 06:42 PM, Rahul Sharma wrote:
On 2 June 2014 14:41, Andrzej Hajda a.hajda@samsung.com wrote:
Hi Rahul,
On 05/28/2014 08:11 AM, Rahul Sharma wrote:
System hangs when FIMD registers are accessed to disable hardware overlays. This is because of the clocks which are not enabled before register access.
'Hardware overlay disable' is cleaned from the FIMD probe.
Signed-off-by: Rahul Sharma rahul.sharma@samsung.com
This patch causes regression on some exynos4210-universal_c210 devices, everything works expect colors are incorrect - it seems blue component is very dark, almost black.
Oh.... Sorry for that. I did not see any problem on 5250/5420/5800. I do not have setup for 4210. Better we should revert this patch.
Would you please help me by verifying the following patch on 4210? This is an alternate solution to the same problem.
http://www.mail-archive.com/linux-samsung-soc@vger.kernel.org/msg31426.html
In my humble opinion, the patch is required to modify like below:
+ ret = pm_runtime_get_sync(ctx->dev); + if (ret < 0) {
When the bootloader enabled FIMD to show logo(boot image) or animation, this function pm_runtime_get_sync() in fimd_bind() would return 1.
Thank you. Best regards YJ
Thanks Andrej, for bringing it to notice.
Regards, Rahul Sharma
Regards Andrzej
Based on exynos-drm-next branch.
drivers/gpu/drm/exynos/exynos_drm_fimd.c | 17 ----------------- 1 file changed, 17 deletions(-)
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimd.c b/drivers/gpu/drm/exynos/exynos_drm_fimd.c index bd30d0c..2ec634f 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimd.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimd.c @@ -703,19 +703,6 @@ static void fimd_win_disable(struct exynos_drm_manager *mgr, int zpos) win_data->enabled = false; }
-static void fimd_clear_win(struct fimd_context *ctx, int win) -{
writel(0, ctx->regs + WINCON(win));writel(0, ctx->regs + VIDOSD_A(win));writel(0, ctx->regs + VIDOSD_B(win));writel(0, ctx->regs + VIDOSD_C(win));if (win == 1 || win == 2)writel(0, ctx->regs + VIDOSD_D(win));fimd_shadow_protect_win(ctx, win, false);-}
- static void fimd_window_suspend(struct exynos_drm_manager *mgr) { struct fimd_context *ctx = mgr->ctx;
@@ -898,16 +885,12 @@ static int fimd_bind(struct device *dev, struct device *master, void *data) { struct fimd_context *ctx = fimd_manager.ctx; struct drm_device *drm_dev = data;
int win; fimd_mgr_initialize(&fimd_manager, drm_dev); exynos_drm_crtc_create(&fimd_manager); if (ctx->display) exynos_drm_create_enc_conn(drm_dev, ctx->display);for (win = 0; win < WINDOWS_NR; win++)fimd_clear_win(ctx, win);return 0;}
dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel
On 2014년 06월 02일 18:42, Rahul Sharma wrote:
On 2 June 2014 14:41, Andrzej Hajda a.hajda@samsung.com wrote:
Hi Rahul,
On 05/28/2014 08:11 AM, Rahul Sharma wrote:
System hangs when FIMD registers are accessed to disable hardware overlays. This is because of the clocks which are not enabled before register access.
'Hardware overlay disable' is cleaned from the FIMD probe.
Signed-off-by: Rahul Sharma rahul.sharma@samsung.com
This patch causes regression on some exynos4210-universal_c210 devices, everything works expect colors are incorrect - it seems blue component is very dark, almost black.
That issue would be because your boot-loader enabled a hardware overlay but kernel driver enables another hardware overlay, which would also be why it happens only on one target of your targets, and why it works correctly by cleaning hardware overlays at fimd probe.
Oh.... Sorry for that. I did not see any problem on 5250/5420/5800. I do not have setup for 4210. Better we should revert this patch.
No, it should work correctly as is because fimd overlays will be reconfigured at booting time by register_framebuffer() if console framebuffer is enabled. Otherwise, a master process, X or Weston, will reconfigure fimd overlays after kernel booting.
Thanks, Inki Dae
Would you please help me by verifying the following patch on 4210? This is an alternate solution to the same problem.
http://www.mail-archive.com/linux-samsung-soc@vger.kernel.org/msg31426.html
Thanks Andrej, for bringing it to notice.
Regards, Rahul Sharma
Regards Andrzej
Based on exynos-drm-next branch.
drivers/gpu/drm/exynos/exynos_drm_fimd.c | 17 ----------------- 1 file changed, 17 deletions(-)
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimd.c b/drivers/gpu/drm/exynos/exynos_drm_fimd.c index bd30d0c..2ec634f 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimd.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimd.c @@ -703,19 +703,6 @@ static void fimd_win_disable(struct exynos_drm_manager *mgr, int zpos) win_data->enabled = false; }
-static void fimd_clear_win(struct fimd_context *ctx, int win) -{
writel(0, ctx->regs + WINCON(win));writel(0, ctx->regs + VIDOSD_A(win));writel(0, ctx->regs + VIDOSD_B(win));writel(0, ctx->regs + VIDOSD_C(win));if (win == 1 || win == 2)writel(0, ctx->regs + VIDOSD_D(win));fimd_shadow_protect_win(ctx, win, false);-}
static void fimd_window_suspend(struct exynos_drm_manager *mgr) { struct fimd_context *ctx = mgr->ctx; @@ -898,16 +885,12 @@ static int fimd_bind(struct device *dev, struct device *master, void *data) { struct fimd_context *ctx = fimd_manager.ctx; struct drm_device *drm_dev = data;
int win; fimd_mgr_initialize(&fimd_manager, drm_dev); exynos_drm_crtc_create(&fimd_manager); if (ctx->display) exynos_drm_create_enc_conn(drm_dev, ctx->display);for (win = 0; win < WINDOWS_NR; win++)fimd_clear_win(ctx, win);return 0;}
dri-devel@lists.freedesktop.org
-
Andrzej Hajda -
Inki Dae -
Rahul Sharma -
Tomasz Figa -
YoungJun Cho