From: Krzysztof Kozlowski <hidden> Date: 2016-01-26 00:42:02
In fimc_md_parse_port_node() endpoint node is get with of_get_next_child()
but it is not put on error path.
Fixes: 56fa1a6a6a7d ("[media] s5p-fimc: Change the driver directory name to exynos4-is")
Signed-off-by: Krzysztof Kozlowski <redacted>
---
Not tested on hardware, only built+static checkers.
---
drivers/media/platform/exynos4-is/media-dev.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
From: Krzysztof Kozlowski <hidden> Date: 2016-01-26 00:42:07
In fimc_md_parse_port_node() remote port parent node is get with
of_graph_get_remote_port_parent() but it is not put on error path.
Fixes: fa91f1056f17 ("[media] exynos4-is: Add support for asynchronous subdevices registration")
Signed-off-by: Krzysztof Kozlowski <redacted>
---
Not tested on hardware, only built+static checkers.
---
drivers/media/platform/exynos4-is/media-dev.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
From: Javier Martinez Canillas <hidden> Date: 2016-03-07 14:15:54
Hello Krzysztof,
On 01/25/2016 09:41 PM, Krzysztof Kozlowski wrote:
quoted hunk
In fimc_md_parse_port_node() endpoint node is get with of_get_next_child()
but it is not put on error path.
Fixes: 56fa1a6a6a7d ("[media] s5p-fimc: Change the driver directory name to exynos4-is")
Signed-off-by: Krzysztof Kozlowski <redacted>
---
Not tested on hardware, only built+static checkers.
---
drivers/media/platform/exynos4-is/media-dev.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
From: Javier Martinez Canillas <hidden> Date: 2016-03-07 14:18:07
Hello Krzysztof,
On 01/25/2016 09:41 PM, Krzysztof Kozlowski wrote:
quoted hunk
In fimc_md_parse_port_node() remote port parent node is get with
of_graph_get_remote_port_parent() but it is not put on error path.
Fixes: fa91f1056f17 ("[media] exynos4-is: Add support for asynchronous subdevices registration")
Signed-off-by: Krzysztof Kozlowski <redacted>
---
Not tested on hardware, only built+static checkers.
---
drivers/media/platform/exynos4-is/media-dev.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
On 01/26/2016 01:41 AM, Krzysztof Kozlowski wrote:
In fimc_md_parse_port_node() endpoint node is get with of_get_next_child()
but it is not put on error path.
"is get" doesn't sound right to me, how about rephrasing this to:
"In fimc_md_parse_port_node() reference count of the endpoint node
"is incremented by of_get_next_child() but it is not decremented
on error path."
quoted hunk
Fixes: 56fa1a6a6a7d ("[media] s5p-fimc: Change the driver directory name to exynos4-is")
Signed-off-by: Krzysztof Kozlowski <redacted>
---
Not tested on hardware, only built+static checkers.
---
drivers/media/platform/exynos4-is/media-dev.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
@@ -339,8 +339,10 @@ static int fimc_md_parse_port_node(struct fimc_md *fmd,return0;v4l2_of_parse_endpoint(ep,&endpoint);-if(WARN_ON(endpoint.base.port==0)||index>=FIMC_MAX_SENSORS)+if(WARN_ON(endpoint.base.port==0)||index>=FIMC_MAX_SENSORS){+of_node_put(ep);return-EINVAL;+}
Thanks for the patch, it looks correct but it doesn't apply cleanly
due to patches already in media master branch [1]. Could you refresh
this patch and resend?
Also I don't quite like multiple calls to of_node_put(), how about
doing something like this instead:
---------------------------8<----------------------------------
@@ -397,18 +397,19 @@ static int fimc_md_parse_port_node(struct fimc_md *fmd,return0;ret=v4l2_of_parse_endpoint(ep,&endpoint);-if(ret){-of_node_put(ep);-returnret;+if(!ret){+if(WARN_ON(endpoint.base.port==0)||+index>=FIMC_MAX_SENSORS){+ret=-EINVAL;+}else{+pd->mux_id=(endpoint.base.port-1)&0x1;+rem=of_graph_get_remote_port_parent(ep);+}}--if(WARN_ON(endpoint.base.port==0)||index>=FIMC_MAX_SENSORS)-return-EINVAL;--pd->mux_id=(endpoint.base.port-1)&0x1;--rem=of_graph_get_remote_port_parent(ep);of_node_put(ep);+if(ret<0)+returnret;+if(rem==NULL){v4l2_info(&fmd->v4l2_dev,"Remote device at %s not found\n",ep->full_name);---------------------------8<----------------------------------
On 01/26/2016 01:41 AM, Krzysztof Kozlowski wrote:
In fimc_md_parse_port_node() remote port parent node is get with
of_graph_get_remote_port_parent() but it is not put on error path.
Fixes: fa91f1056f17 ("[media] exynos4-is: Add support for
asynchronous subdevices registration")
Signed-off-by: Krzysztof Kozlowski <redacted>
Thanks, patch applied with s/is get/is acquired/.
quoted hunk
---
Not tested on hardware, only built+static checkers.
---
drivers/media/platform/exynos4-is/media-dev.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
From: Krzysztof Kozlowski <hidden> Date: 2016-03-13 23:51:10
On 11.03.2016 21:30, Sylwester Nawrocki wrote:
On 01/26/2016 01:41 AM, Krzysztof Kozlowski wrote:
quoted
In fimc_md_parse_port_node() endpoint node is get with of_get_next_child()
but it is not put on error path.
"is get" doesn't sound right to me, how about rephrasing this to:
"In fimc_md_parse_port_node() reference count of the endpoint node
"is incremented by of_get_next_child() but it is not decremented
on error path."
quoted
Fixes: 56fa1a6a6a7d ("[media] s5p-fimc: Change the driver directory name to exynos4-is")
Signed-off-by: Krzysztof Kozlowski <redacted>
---
Not tested on hardware, only built+static checkers.
---
drivers/media/platform/exynos4-is/media-dev.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
@@ -339,8 +339,10 @@ static int fimc_md_parse_port_node(struct fimc_md *fmd,return0;v4l2_of_parse_endpoint(ep,&endpoint);-if(WARN_ON(endpoint.base.port==0)||index>=FIMC_MAX_SENSORS)+if(WARN_ON(endpoint.base.port==0)||index>=FIMC_MAX_SENSORS){+of_node_put(ep);return-EINVAL;+}
Thanks for the patch, it looks correct but it doesn't apply cleanly
due to patches already in media master branch [1]. Could you refresh
this patch and resend?
Also I don't quite like multiple calls to of_node_put(), how about
doing something like this instead:
How about sending your patch then with my reported-by?
Best regards,
Krzysztof