* [PATCH] drm/mediatek: fix null pointer dereference
@ 2016-10-28 8:34 ` CK Hu
0 siblings, 0 replies; 6+ messages in thread
From: CK Hu @ 2016-10-28 8:34 UTC (permalink / raw)
To: linux-arm-kernel
Hi, Matthias:
Even though OVL HW would not be enabled before component_add() in
current design, your patch would be safe for any situation.
Acked-by CK Hu <ck.hu@mediatek.com>
Regards,
CK
On Wed, 2016-10-26 at 16:09 +0200, Matthias Brugger wrote:
> The probe function requests the interrupt before initializing
> the ddp component. Which leads to a null pointer dereference at boot.
> Fix this by requesting the interrput after all components got
> initialized properly.
>
> Fixes: 119f5173628a ("drm/mediatek: Add DRM Driver for Mediatek SoC
> MT8173.")
> Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com>
> ---
> drivers/gpu/drm/mediatek/mtk_disp_ovl.c | 14 +++++++-------
> 1 file changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/gpu/drm/mediatek/mtk_disp_ovl.c b/drivers/gpu/drm/mediatek/mtk_disp_ovl.c
> index 019b7ca..1e78159 100644
> --- a/drivers/gpu/drm/mediatek/mtk_disp_ovl.c
> +++ b/drivers/gpu/drm/mediatek/mtk_disp_ovl.c
> @@ -250,13 +250,6 @@ static int mtk_disp_ovl_probe(struct platform_device *pdev)
> if (irq < 0)
> return irq;
>
> - ret = devm_request_irq(dev, irq, mtk_disp_ovl_irq_handler,
> - IRQF_TRIGGER_NONE, dev_name(dev), priv);
> - if (ret < 0) {
> - dev_err(dev, "Failed to request irq %d: %d\n", irq, ret);
> - return ret;
> - }
> -
> comp_id = mtk_ddp_comp_get_id(dev->of_node, MTK_DISP_OVL);
> if (comp_id < 0) {
> dev_err(dev, "Failed to identify by alias: %d\n", comp_id);
> @@ -272,6 +265,13 @@ static int mtk_disp_ovl_probe(struct platform_device *pdev)
>
> platform_set_drvdata(pdev, priv);
>
> + ret = devm_request_irq(dev, irq, mtk_disp_ovl_irq_handler,
> + IRQF_TRIGGER_NONE, dev_name(dev), priv);
> + if (ret < 0) {
> + dev_err(dev, "Failed to request irq %d: %d\n", irq, ret);
> + return ret;
> + }
> +
> ret = component_add(dev, &mtk_disp_ovl_component_ops);
> if (ret)
> dev_err(dev, "Failed to add component: %d\n", ret);
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] drm/mediatek: fix null pointer dereference
@ 2016-10-28 8:34 ` CK Hu
0 siblings, 0 replies; 6+ messages in thread
From: CK Hu @ 2016-10-28 8:34 UTC (permalink / raw)
To: Matthias Brugger
Cc: p.zabel, airlied, dri-devel, linux-arm-kernel, linux-mediatek,
linux-kernel
Hi, Matthias:
Even though OVL HW would not be enabled before component_add() in
current design, your patch would be safe for any situation.
Acked-by CK Hu <ck.hu@mediatek.com>
Regards,
CK
On Wed, 2016-10-26 at 16:09 +0200, Matthias Brugger wrote:
> The probe function requests the interrupt before initializing
> the ddp component. Which leads to a null pointer dereference at boot.
> Fix this by requesting the interrput after all components got
> initialized properly.
>
> Fixes: 119f5173628a ("drm/mediatek: Add DRM Driver for Mediatek SoC
> MT8173.")
> Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com>
> ---
> drivers/gpu/drm/mediatek/mtk_disp_ovl.c | 14 +++++++-------
> 1 file changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/gpu/drm/mediatek/mtk_disp_ovl.c b/drivers/gpu/drm/mediatek/mtk_disp_ovl.c
> index 019b7ca..1e78159 100644
> --- a/drivers/gpu/drm/mediatek/mtk_disp_ovl.c
> +++ b/drivers/gpu/drm/mediatek/mtk_disp_ovl.c
> @@ -250,13 +250,6 @@ static int mtk_disp_ovl_probe(struct platform_device *pdev)
> if (irq < 0)
> return irq;
>
> - ret = devm_request_irq(dev, irq, mtk_disp_ovl_irq_handler,
> - IRQF_TRIGGER_NONE, dev_name(dev), priv);
> - if (ret < 0) {
> - dev_err(dev, "Failed to request irq %d: %d\n", irq, ret);
> - return ret;
> - }
> -
> comp_id = mtk_ddp_comp_get_id(dev->of_node, MTK_DISP_OVL);
> if (comp_id < 0) {
> dev_err(dev, "Failed to identify by alias: %d\n", comp_id);
> @@ -272,6 +265,13 @@ static int mtk_disp_ovl_probe(struct platform_device *pdev)
>
> platform_set_drvdata(pdev, priv);
>
> + ret = devm_request_irq(dev, irq, mtk_disp_ovl_irq_handler,
> + IRQF_TRIGGER_NONE, dev_name(dev), priv);
> + if (ret < 0) {
> + dev_err(dev, "Failed to request irq %d: %d\n", irq, ret);
> + return ret;
> + }
> +
> ret = component_add(dev, &mtk_disp_ovl_component_ops);
> if (ret)
> dev_err(dev, "Failed to add component: %d\n", ret);
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] drm/mediatek: fix null pointer dereference
2016-10-28 8:34 ` CK Hu
(?)
@ 2016-10-28 10:39 ` Matthias Brugger
-1 siblings, 0 replies; 6+ messages in thread
From: Matthias Brugger @ 2016-10-28 10:39 UTC (permalink / raw)
To: CK Hu
Cc: p.zabel, airlied, dri-devel, linux-arm-kernel, linux-mediatek,
linux-kernel
On 10/28/2016 10:34 AM, CK Hu wrote:
> Hi, Matthias:
>
> Even though OVL HW would not be enabled before component_add() in
> current design, your patch would be safe for any situation.
Maybe the FW I use left an interrupt pending before loading the kernel
and this leads to the case where we enter the handler before all data
structures are set up.
>
> Acked-by CK Hu <ck.hu@mediatek.com>
Thanks!
Matthias
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH] drm/mediatek: fix null pointer dereference
@ 2016-10-28 10:39 ` Matthias Brugger
0 siblings, 0 replies; 6+ messages in thread
From: Matthias Brugger @ 2016-10-28 10:39 UTC (permalink / raw)
To: linux-arm-kernel
On 10/28/2016 10:34 AM, CK Hu wrote:
> Hi, Matthias:
>
> Even though OVL HW would not be enabled before component_add() in
> current design, your patch would be safe for any situation.
Maybe the FW I use left an interrupt pending before loading the kernel
and this leads to the case where we enter the handler before all data
structures are set up.
>
> Acked-by CK Hu <ck.hu@mediatek.com>
Thanks!
Matthias
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] drm/mediatek: fix null pointer dereference
@ 2016-10-28 10:39 ` Matthias Brugger
0 siblings, 0 replies; 6+ messages in thread
From: Matthias Brugger @ 2016-10-28 10:39 UTC (permalink / raw)
To: CK Hu
Cc: airlied-cv59FeDIM0c, linux-kernel-u79uwXL29TY76Z2rM5mHXA,
dri-devel-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW,
linux-mediatek-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r,
p.zabel-bIcnvbaLZ9MEGnE8C9+IrQ,
linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
On 10/28/2016 10:34 AM, CK Hu wrote:
> Hi, Matthias:
>
> Even though OVL HW would not be enabled before component_add() in
> current design, your patch would be safe for any situation.
Maybe the FW I use left an interrupt pending before loading the kernel
and this leads to the case where we enter the handler before all data
structures are set up.
>
> Acked-by CK Hu <ck.hu-NuS5LvNUpcJWk0Htik3J/w@public.gmane.org>
Thanks!
Matthias
^ permalink raw reply [flat|nested] 6+ messages in thread