[PATCH 1/3] drm/tegra: Fix lockup on a use of staging API

[PATCH 1/3] drm/tegra: Fix lockup on a use of staging API

[Dmitry Osipenko]

Commit bdd2f9cd ("Don't leak kernel pointer to userspace") added a mutex
around staging IOCTL's, some of those mutexes are taken twice.

Fixes: bdd2f9cd10eb ("drm/tegra: Don't leak kernel pointer to userspace")
Signed-off-by: Dmitry Osipenko <digetx-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
---
 drivers/gpu/drm/tegra/drm.c | 20 ++++----------------
 1 file changed, 4 insertions(+), 16 deletions(-)

diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c
index ab2dfd4e4bd9..768750226452 100644
--- a/drivers/gpu/drm/tegra/drm.c
+++ b/drivers/gpu/drm/tegra/drm.c
@@ -430,18 +430,6 @@ int tegra_drm_submit(struct tegra_drm_context *context,
 
 
 #ifdef CONFIG_DRM_TEGRA_STAGING
-static struct tegra_drm_context *
-tegra_drm_file_get_context(struct tegra_drm_file *file, u32 id)
-{
-	struct tegra_drm_context *context;
-
-	mutex_lock(&file->lock);
-	context = idr_find(&file->contexts, id);
-	mutex_unlock(&file->lock);
-
-	return context;
-}
-
 static int tegra_gem_create(struct drm_device *drm, void *data,
 			    struct drm_file *file)
 {
@@ -585,7 +573,7 @@ static int tegra_close_channel(struct drm_device *drm, void *data,
 
 	mutex_lock(&fpriv->lock);
 
-	context = tegra_drm_file_get_context(fpriv, args->context);
+	context = idr_find(&fpriv->contexts, args->context);
 	if (!context) {
 		err = -EINVAL;
 		goto unlock;
@@ -610,7 +598,7 @@ static int tegra_get_syncpt(struct drm_device *drm, void *data,
 
 	mutex_lock(&fpriv->lock);
 
-	context = tegra_drm_file_get_context(fpriv, args->context);
+	context = idr_find(&fpriv->contexts, args->context);
 	if (!context) {
 		err = -ENODEV;
 		goto unlock;
@@ -639,7 +627,7 @@ static int tegra_submit(struct drm_device *drm, void *data,
 
 	mutex_lock(&fpriv->lock);
 
-	context = tegra_drm_file_get_context(fpriv, args->context);
+	context = idr_find(&fpriv->contexts, args->context);
 	if (!context) {
 		err = -ENODEV;
 		goto unlock;
@@ -664,7 +652,7 @@ static int tegra_get_syncpt_base(struct drm_device *drm, void *data,
 
 	mutex_lock(&fpriv->lock);
 
-	context = tegra_drm_file_get_context(fpriv, args->context);
+	context = idr_find(&fpriv->contexts, args->context);
 	if (!context) {
 		err = -ENODEV;
 		goto unlock;
-- 
2.12.2

[PATCH 2/3] drm/tegra: Correct idr_alloc() minimum id

[Dmitry Osipenko]

The start = 0 is invalid and causes weird CDMA channel timeouts, presumably
some memory misuse/corruption is going on.

Fixes: bdd2f9cd10eb ("drm/tegra: Don't leak kernel pointer to userspace")
Signed-off-by: Dmitry Osipenko <digetx-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
---
 drivers/gpu/drm/tegra/drm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c
index 768750226452..732c8d98044f 100644
--- a/drivers/gpu/drm/tegra/drm.c
+++ b/drivers/gpu/drm/tegra/drm.c
@@ -518,7 +518,7 @@ static int tegra_client_open(struct tegra_drm_file *fpriv,
 	if (err < 0)
 		return err;
 
-	err = idr_alloc(&fpriv->contexts, context, 0, 0, GFP_KERNEL);
+	err = idr_alloc(&fpriv->contexts, context, 1, 0, GFP_KERNEL);
 	if (err < 0) {
 		client->ops->close_channel(context);
 		return err;
-- 
2.12.2

Re: [PATCH 1/3] drm/tegra: Fix lockup on a use of staging API

[Mikko Perttunen]

Reviewed-by: Mikko Perttunen <mperttunen-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>

On 05/12/2017 10:00 PM, Dmitry Osipenko wrote:
> Commit bdd2f9cd ("Don't leak kernel pointer to userspace") added a mutex
> around staging IOCTL's, some of those mutexes are taken twice.
>
> Fixes: bdd2f9cd10eb ("drm/tegra: Don't leak kernel pointer to userspace")
> Signed-off-by: Dmitry Osipenko <digetx-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> ---
>  drivers/gpu/drm/tegra/drm.c | 20 ++++----------------
>  1 file changed, 4 insertions(+), 16 deletions(-)
>
> diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c
> index ab2dfd4e4bd9..768750226452 100644
> --- a/drivers/gpu/drm/tegra/drm.c
> +++ b/drivers/gpu/drm/tegra/drm.c
> @@ -430,18 +430,6 @@ int tegra_drm_submit(struct tegra_drm_context *context,
>
>
>  #ifdef CONFIG_DRM_TEGRA_STAGING
> -static struct tegra_drm_context *
> -tegra_drm_file_get_context(struct tegra_drm_file *file, u32 id)
> -{
> -	struct tegra_drm_context *context;
> -
> -	mutex_lock(&file->lock);
> -	context = idr_find(&file->contexts, id);
> -	mutex_unlock(&file->lock);
> -
> -	return context;
> -}
> -
>  static int tegra_gem_create(struct drm_device *drm, void *data,
>  			    struct drm_file *file)
>  {
> @@ -585,7 +573,7 @@ static int tegra_close_channel(struct drm_device *drm, void *data,
>
>  	mutex_lock(&fpriv->lock);
>
> -	context = tegra_drm_file_get_context(fpriv, args->context);
> +	context = idr_find(&fpriv->contexts, args->context);
>  	if (!context) {
>  		err = -EINVAL;
>  		goto unlock;
> @@ -610,7 +598,7 @@ static int tegra_get_syncpt(struct drm_device *drm, void *data,
>
>  	mutex_lock(&fpriv->lock);
>
> -	context = tegra_drm_file_get_context(fpriv, args->context);
> +	context = idr_find(&fpriv->contexts, args->context);
>  	if (!context) {
>  		err = -ENODEV;
>  		goto unlock;
> @@ -639,7 +627,7 @@ static int tegra_submit(struct drm_device *drm, void *data,
>
>  	mutex_lock(&fpriv->lock);
>
> -	context = tegra_drm_file_get_context(fpriv, args->context);
> +	context = idr_find(&fpriv->contexts, args->context);
>  	if (!context) {
>  		err = -ENODEV;
>  		goto unlock;
> @@ -664,7 +652,7 @@ static int tegra_get_syncpt_base(struct drm_device *drm, void *data,
>
>  	mutex_lock(&fpriv->lock);
>
> -	context = tegra_drm_file_get_context(fpriv, args->context);
> +	context = idr_find(&fpriv->contexts, args->context);
>  	if (!context) {
>  		err = -ENODEV;
>  		goto unlock;
>

Re: [PATCH 2/3] drm/tegra: Correct idr_alloc() minimum id

[Mikko Perttunen]

On 05/12/2017 10:00 PM, Dmitry Osipenko wrote:
> The start = 0 is invalid and causes weird CDMA channel timeouts, presumably
> some memory misuse/corruption is going on.

What makes you think start = 0 is invalid? I can't see anything pointing 
to that in the idr code and there are many users in the kernel passing 0 
as start.

>
> Fixes: bdd2f9cd10eb ("drm/tegra: Don't leak kernel pointer to userspace")
> Signed-off-by: Dmitry Osipenko <digetx-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> ---
>  drivers/gpu/drm/tegra/drm.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c
> index 768750226452..732c8d98044f 100644
> --- a/drivers/gpu/drm/tegra/drm.c
> +++ b/drivers/gpu/drm/tegra/drm.c
> @@ -518,7 +518,7 @@ static int tegra_client_open(struct tegra_drm_file *fpriv,
>  	if (err < 0)
>  		return err;
>
> -	err = idr_alloc(&fpriv->contexts, context, 0, 0, GFP_KERNEL);
> +	err = idr_alloc(&fpriv->contexts, context, 1, 0, GFP_KERNEL);
>  	if (err < 0) {
>  		client->ops->close_channel(context);
>  		return err;
>

Re: [PATCH 2/3] drm/tegra: Correct idr_alloc() minimum id

[Dmitry Osipenko]

On 14.05.2017 14:53, Mikko Perttunen wrote:
> On 05/12/2017 10:00 PM, Dmitry Osipenko wrote:
>> The start = 0 is invalid and causes weird CDMA channel timeouts, presumably
>> some memory misuse/corruption is going on.
> 
> What makes you think start = 0 is invalid? I can't see anything pointing to that
> in the idr code and there are many users in the kernel passing 0 as start.
> 

Well, I can't see either. You are right that there are quite many others with 0
as a start, the 1 probably just masks the bug.

-- 
Dmitry

Re: [PATCH 2/3] drm/tegra: Correct idr_alloc() minimum id

[Dmitry Osipenko]

On 14.05.2017 16:02, Dmitry Osipenko wrote:
> On 14.05.2017 14:53, Mikko Perttunen wrote:
>> On 05/12/2017 10:00 PM, Dmitry Osipenko wrote:
>>> The start = 0 is invalid and causes weird CDMA channel timeouts, presumably
>>> some memory misuse/corruption is going on.
>>
>> What makes you think start = 0 is invalid? I can't see anything pointing to that
>> in the idr code and there are many users in the kernel passing 0 as start.
>>
> 
> Well, I can't see either. You are right that there are quite many others with 0
> as a start, the 1 probably just masks the bug.
> 

Finally, I found the root of the issue. The job->client is set to the context ID
in the tegra_drm_submit() and the host1x_cdma sets client ID to 0 to mark CDMA
job timeout timer as already armed. I'll send V2 with a new commit description.

-- 
Dmitry