[PATCH] pxafb: fix regression from d2a34c13e7ccec5d06eafd60e6f80ea531b34668

[PATCH] pxafb: fix regression from d2a34c13e7ccec5d06eafd60e6f80ea531b34668

[Martin Jansa]

fbi->dev->platform_data is void * and gcc does not allow to access
!fbi->dev->platform_data->smart_update without cast platform_data to
(struct pxafb_mach_info) or assignment as used in this patch.

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
---
 drivers/video/pxafb.c |    7 ++++++-
 1 files changed, 6 insertions(+), 1 deletions(-)

diff --git a/drivers/video/pxafb.c b/drivers/video/pxafb.c
index 415858b..241468c 100644
--- a/drivers/video/pxafb.c
+++ b/drivers/video/pxafb.c
@@ -1223,12 +1223,17 @@ static int pxafb_smart_thread(void *arg)
 	struct pxafb_info *fbi = arg;
 	struct pxafb_mach_info *inf;
 
-	if (!fbi || !fbi->dev->platform_data->smart_update) {
+	if (!fbi) {
 		pr_err("%s: not properly initialized, thread terminated\n",
 				__func__);
 		return -EINVAL;
 	}
 	inf = fbi->dev->platform_data;
+	if (!inf || !inf->smart_update) {
+		pr_err("%s: not properly initialized, thread terminated\n",
+				__func__);
+		return -EINVAL;
+	}
 
 	pr_debug("%s(): task starting\n", __func__);
 
-- 
1.6.6

Re: [PATCH] pxafb: fix regression from d2a34c13e7ccec5d06eafd60e6f80ea531b34668

[Andrew Morton]

On Mon, 28 Dec 2009 16:54:43 +0100
Martin Jansa <martin.jansa@gmail.com> wrote:

> fbi->dev->platform_data is void * and gcc does not allow to access
> !fbi->dev->platform_data->smart_update without cast platform_data to
> (struct pxafb_mach_info) or assignment as used in this patch.
> 
> Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
> ---
>  drivers/video/pxafb.c |    7 ++++++-
>  1 files changed, 6 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/video/pxafb.c b/drivers/video/pxafb.c
> index 415858b..241468c 100644
> --- a/drivers/video/pxafb.c
> +++ b/drivers/video/pxafb.c
> @@ -1223,12 +1223,17 @@ static int pxafb_smart_thread(void *arg)
>  	struct pxafb_info *fbi = arg;
>  	struct pxafb_mach_info *inf;
>  
> -	if (!fbi || !fbi->dev->platform_data->smart_update) {
> +	if (!fbi) {
>  		pr_err("%s: not properly initialized, thread terminated\n",
>  				__func__);
>  		return -EINVAL;
>  	}
>  	inf = fbi->dev->platform_data;
> +	if (!inf || !inf->smart_update) {
> +		pr_err("%s: not properly initialized, thread terminated\n",
> +				__func__);
> +		return -EINVAL;
> +	}
>  
>  	pr_debug("%s(): task starting\n", __func__);
>  

This got fixed differently in Eric's tree.  The patch is still in
linux-next, not mainline:

: commit da2c3f0ead336c04b4d1ad36ff42d8d264f44f65
: Author:     Eric Miao <eric.y.miao@gmail.com>
: AuthorDate: Sat Dec 26 16:25:18 2009 +0800
: Commit:     Eric Miao <eric.y.miao@gmail.com>
: CommitDate: Tue Dec 29 14:11:27 2009 +0800
: 
:     [ARM] pxafb: fix building issue of incorrect reference
:     
:     Commit "d2a34c1 drivers/video: Move dereference after NULL test" introduced
:     a build error of "fbi->dev->platform_data->smart_update" being unknown type
:     to the compiler, fix this by removing the unnecessary test of 'fbi'.
:     
:     Cc: Julia Lawall <julia@diku.dk>
:     Signed-off-by: Eric Miao <eric.y.miao@gmail.com>
: 
: diff --git a/drivers/video/pxafb.c b/drivers/video/pxafb.c
: index 415858b..825b665 100644
: --- a/drivers/video/pxafb.c
: +++ b/drivers/video/pxafb.c
: @@ -1221,9 +1221,9 @@ static void setup_smart_timing(struct pxafb_info *fbi,
:  static int pxafb_smart_thread(void *arg)
:  {
:  	struct pxafb_info *fbi = arg;
: -	struct pxafb_mach_info *inf;
: +	struct pxafb_mach_info *inf = fbi->dev->platform_data;
:  
: -	if (!fbi || !fbi->dev->platform_data->smart_update) {
: +	if (!inf->smart_update) {
:  		pr_err("%s: not properly initialized, thread terminated\n",
:  				__func__);
:  		return -EINVAL;
: 

That patch fixes the build, but it also eliminates the test for
fbi==NULL.  The changelog doesn't explain that change and I wonder if
it was intentional and correct?

Re: [PATCH] pxafb: fix regression from d2a34c13e7ccec5d06eafd60e6f80ea531b34668

[Eric Miao]

>
> That patch fixes the build, but it also eliminates the test for
> fbi==NULL.  The changelog doesn't explain that change and I wonder if
> it was intentional and correct?
>
>

fbi has been well tested in pxafb_probe() and then passed to pxafb_smart_init()
and to pxafb_smart_update(), it's just unnecessary to test it again and again.