[patch] drm/msm/dsi: free first element on error

[patch] drm/msm/dsi: free first element on error

[Dan Carpenter]

We want to free msm_host->bus_clks[0] so the > should be >=.

Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>

diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
index 1fc07ce24686..239e79b39a45 100644
--- a/drivers/gpu/drm/msm/dsi/dsi_host.c
+++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
@@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
 
 	return 0;
 err:
-	for (; i > 0; i--)
+	for (; i >= 0; i--)
 		clk_disable_unprepare(msm_host->bus_clks[i]);
 
 	return ret;
_______________________________________________
Freedreno mailing list
Freedreno@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/freedreno

Re: [patch] drm/msm/dsi: free first element on error

[Jani Nikula]

On Thu, 16 Feb 2017, Dan Carpenter <dan.carpenter@oracle.com> wrote:
> We want to free msm_host->bus_clks[0] so the > should be >=.
>
> Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
>
> diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
> index 1fc07ce24686..239e79b39a45 100644
> --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
> +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
> @@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
>  
>  	return 0;
>  err:
> -	for (; i > 0; i--)
> +	for (; i >= 0; i--)
>  		clk_disable_unprepare(msm_host->bus_clks[i]);

By the looks of it this is also wrong. I didn't look at the functions,
but you probably don't want to unprepare something where prepare failed,
i.e. you want to -1 both the start and end offsets. Perhaps the right
fix is

	while (i--)
		clk_disable_unprepare(msm_host->bus_clks[i]);

which also seems to be widely used on error paths.

BR,
Jani.


>  
>  	return ret;
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

-- 
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [patch] drm/msm/dsi: free first element on error

[Dan Carpenter]

On Thu, Feb 16, 2017 at 01:27:47PM +0200, Jani Nikula wrote:
> On Thu, 16 Feb 2017, Dan Carpenter <dan.carpenter@oracle.com> wrote:
> > We want to free msm_host->bus_clks[0] so the > should be >=.
> >
> > Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
> > Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
> >
> > diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
> > index 1fc07ce24686..239e79b39a45 100644
> > --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
> > +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
> > @@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
> >  
> >  	return 0;
> >  err:
> > -	for (; i > 0; i--)
> > +	for (; i >= 0; i--)
> >  		clk_disable_unprepare(msm_host->bus_clks[i]);
> 
> By the looks of it this is also wrong. I didn't look at the functions,
> but you probably don't want to unprepare something where prepare failed,
> i.e. you want to -1 both the start and end offsets. Perhaps the right
> fix is
> 
> 	while (i--)
> 		clk_disable_unprepare(msm_host->bus_clks[i]);
> 
> which also seems to be widely used on error paths.
> 

Ah yeah.  You're right.  I'll resend.

regards,
dan carpenter

_______________________________________________
Freedreno mailing list
Freedreno@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/freedreno

Re: [patch] drm/msm/dsi: free first element on error

[walter harms]

Am 16.02.2017 12:53, schrieb Dan Carpenter:
> On Thu, Feb 16, 2017 at 01:27:47PM +0200, Jani Nikula wrote:
>> On Thu, 16 Feb 2017, Dan Carpenter <dan.carpenter@oracle.com> wrote:
>>> We want to free msm_host->bus_clks[0] so the > should be >=.
>>>
>>> Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
>>> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
>>>
>>> diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
>>> index 1fc07ce24686..239e79b39a45 100644
>>> --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
>>> +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
>>> @@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
>>>  
>>>  	return 0;
>>>  err:
>>> -	for (; i > 0; i--)
>>> +	for (; i >= 0; i--)
>>>  		clk_disable_unprepare(msm_host->bus_clks[i]);
>>
>> By the looks of it this is also wrong. I didn't look at the functions,
>> but you probably don't want to unprepare something where prepare failed,
>> i.e. you want to -1 both the start and end offsets. Perhaps the right
>> fix is
>>
>> 	while (i--)
>> 		clk_disable_unprepare(msm_host->bus_clks[i]);
>>
>> which also seems to be widely used on error paths.
>>
> 

We already know that programmers are bad in counting backwards ...

any chance to make that into a forward loop ?

re,
 wh

Re: [patch] drm/msm/dsi: free first element on error

[Rob Clark]

On Thu, Feb 16, 2017 at 7:03 AM, walter harms <wharms@bfs.de> wrote:
>
>
> Am 16.02.2017 12:53, schrieb Dan Carpenter:
>> On Thu, Feb 16, 2017 at 01:27:47PM +0200, Jani Nikula wrote:
>>> On Thu, 16 Feb 2017, Dan Carpenter <dan.carpenter@oracle.com> wrote:
>>>> We want to free msm_host->bus_clks[0] so the > should be >=.
>>>>
>>>> Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
>>>> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
>>>>
>>>> diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
>>>> index 1fc07ce24686..239e79b39a45 100644
>>>> --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
>>>> +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
>>>> @@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
>>>>
>>>>     return 0;
>>>>  err:
>>>> -   for (; i > 0; i--)
>>>> +   for (; i >= 0; i--)
>>>>             clk_disable_unprepare(msm_host->bus_clks[i]);
>>>
>>> By the looks of it this is also wrong. I didn't look at the functions,
>>> but you probably don't want to unprepare something where prepare failed,
>>> i.e. you want to -1 both the start and end offsets. Perhaps the right
>>> fix is
>>>
>>>      while (i--)
>>>              clk_disable_unprepare(msm_host->bus_clks[i]);
>>>
>>> which also seems to be widely used on error paths.
>>>
>>
>
> We already know that programmers are bad in counting backwards ...
>
> any chance to make that into a forward loop ?
>

well, this *is* a common pattern.  And in some cases you actually do
need to disable clks in reverse order.  So meh, I think while (i--)
approach is fine

BR,
-R

Re: [patch] drm/msm/dsi: free first element on error

[Jani Nikula]

On Thu, 16 Feb 2017, walter harms <wharms@bfs.de> wrote:
> Am 16.02.2017 12:53, schrieb Dan Carpenter:
>> On Thu, Feb 16, 2017 at 01:27:47PM +0200, Jani Nikula wrote:
>>> On Thu, 16 Feb 2017, Dan Carpenter <dan.carpenter@oracle.com> wrote:
>>>> We want to free msm_host->bus_clks[0] so the > should be >=.
>>>>
>>>> Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
>>>> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
>>>>
>>>> diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
>>>> index 1fc07ce24686..239e79b39a45 100644
>>>> --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
>>>> +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
>>>> @@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
>>>>  
>>>>  	return 0;
>>>>  err:
>>>> -	for (; i > 0; i--)
>>>> +	for (; i >= 0; i--)
>>>>  		clk_disable_unprepare(msm_host->bus_clks[i]);
>>>
>>> By the looks of it this is also wrong. I didn't look at the functions,
>>> but you probably don't want to unprepare something where prepare failed,
>>> i.e. you want to -1 both the start and end offsets. Perhaps the right
>>> fix is
>>>
>>> 	while (i--)
>>> 		clk_disable_unprepare(msm_host->bus_clks[i]);
>>>
>>> which also seems to be widely used on error paths.
>>>
>> 
>
> We already know that programmers are bad in counting backwards ...
>
> any chance to make that into a forward loop ?

In most cases I'd agree with you. But I see that this while (i--) is
becoming somewhat of a pattern for error paths (grep for it), and I
think following patterns like this is more important. After a while, you
don't have to think about counting when you see it.

Besides, it's generally preferred to cleanup in the reverse order of
init, so a forward counting loop would require two variables here.

BR,
Jani.



>
> re,
>  wh
>
>

-- 
Jani Nikula, Intel Open Source Technology Center

[patch v2] drm/msm/dsi: free first element on error

[Dan Carpenter]

We're off by one here.  We free something that wasn't allocated and we
don't free msm_host->bus_clks[].

Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>

diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
index 1fc07ce24686..4fa32296162e 100644
--- a/drivers/gpu/drm/msm/dsi/dsi_host.c
+++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
@@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
 
 	return 0;
 err:
-	for (; i > 0; i--)
+	while (i--)
 		clk_disable_unprepare(msm_host->bus_clks[i]);
 
 	return ret;

Re: [patch v2] drm/msm/dsi: free first element on error

[Rob Clark]

On Thu, Feb 16, 2017 at 7:00 AM, Dan Carpenter <dan.carpenter@oracle.com> wrote:
> We're off by one here.  We free something that wasn't allocated and we
> don't free msm_host->bus_clks[].
>
> Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>

Thanks

Reviewed-by: Rob Clark <robdclark@gmail.com>

>
> diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
> index 1fc07ce24686..4fa32296162e 100644
> --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
> +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
> @@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
>
>         return 0;
>  err:
> -       for (; i > 0; i--)
> +       while (i--)
>                 clk_disable_unprepare(msm_host->bus_clks[i]);
>
>         return ret;

Re: [patch v2] drm/msm/dsi: free first element on error

[Jani Nikula]

On Thu, 16 Feb 2017, Rob Clark <robdclark@gmail.com> wrote:
> On Thu, Feb 16, 2017 at 7:00 AM, Dan Carpenter <dan.carpenter@oracle.com> wrote:
>> We're off by one here.  We free something that wasn't allocated and we
>> don't free msm_host->bus_clks[].
>>
>> Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
>> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
>
> Thanks
>
> Reviewed-by: Rob Clark <robdclark@gmail.com>

And for good measure,

Reviewed-by: Jani Nikula <jani.nikula@intel.com>


>
>>
>> diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
>> index 1fc07ce24686..4fa32296162e 100644
>> --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
>> +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
>> @@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
>>
>>         return 0;
>>  err:
>> -       for (; i > 0; i--)
>> +       while (i--)
>>                 clk_disable_unprepare(msm_host->bus_clks[i]);
>>
>>         return ret;

-- 
Jani Nikula, Intel Open Source Technology Center

Re: [patch v2] drm/msm/dsi: free first element on error

[Daniel Vetter]

On Thu, Feb 16, 2017 at 02:27:08PM +0200, Jani Nikula wrote:
> On Thu, 16 Feb 2017, Rob Clark <robdclark@gmail.com> wrote:
> > On Thu, Feb 16, 2017 at 7:00 AM, Dan Carpenter <dan.carpenter@oracle.com> wrote:
> >> We're off by one here.  We free something that wasn't allocated and we
> >> don't free msm_host->bus_clks[].
> >>
> >> Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
> >> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
> >
> > Thanks
> >
> > Reviewed-by: Rob Clark <robdclark@gmail.com>
> 
> And for good measure,
> 
> Reviewed-by: Jani Nikula <jani.nikula@intel.com>

So 2 r-b from maintainers, no one said he'll apply. This isn't really
great coordination :-) Note that drm-misc-next is always open, so you
could always smash it in there, irrespective of merge window state. hint,
hint ...
-Daniel

> 
> 
> >
> >>
> >> diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
> >> index 1fc07ce24686..4fa32296162e 100644
> >> --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
> >> +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
> >> @@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
> >>
> >>         return 0;
> >>  err:
> >> -       for (; i > 0; i--)
> >> +       while (i--)
> >>                 clk_disable_unprepare(msm_host->bus_clks[i]);
> >>
> >>         return ret;
> 
> -- 
> Jani Nikula, Intel Open Source Technology Center
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [patch v2] drm/msm/dsi: free first element on error

[Jani Nikula]

On Sun, 26 Feb 2017, Daniel Vetter <daniel@ffwll.ch> wrote:
> On Thu, Feb 16, 2017 at 02:27:08PM +0200, Jani Nikula wrote:
>> On Thu, 16 Feb 2017, Rob Clark <robdclark@gmail.com> wrote:
>> > On Thu, Feb 16, 2017 at 7:00 AM, Dan Carpenter <dan.carpenter@oracle.com> wrote:
>> >> We're off by one here.  We free something that wasn't allocated and we
>> >> don't free msm_host->bus_clks[].
>> >>
>> >> Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
>> >> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
>> >
>> > Thanks
>> >
>> > Reviewed-by: Rob Clark <robdclark@gmail.com>
>> 
>> And for good measure,
>> 
>> Reviewed-by: Jani Nikula <jani.nikula@intel.com>
>
> So 2 r-b from maintainers, no one said he'll apply. This isn't really
> great coordination :-) Note that drm-misc-next is always open, so you
> could always smash it in there, irrespective of merge window state. hint,
> hint ...

Admittedly I shied away from touching drm/msm.

BR,
Jani.

> -Daniel
>
>> 
>> 
>> >
>> >>
>> >> diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
>> >> index 1fc07ce24686..4fa32296162e 100644
>> >> --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
>> >> +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
>> >> @@ -437,7 +437,7 @@ static int dsi_bus_clk_enable(struct msm_dsi_host *msm_host)
>> >>
>> >>         return 0;
>> >>  err:
>> >> -       for (; i > 0; i--)
>> >> +       while (i--)
>> >>                 clk_disable_unprepare(msm_host->bus_clks[i]);
>> >>
>> >>         return ret;
>> 
>> -- 
>> Jani Nikula, Intel Open Source Technology Center
>> _______________________________________________
>> dri-devel mailing list
>> dri-devel@lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/dri-devel

-- 
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [patch v2] drm/msm/dsi: free first element on error

[Daniel Vetter]

On Mon, Feb 27, 2017 at 11:18 AM, Jani Nikula
<jani.nikula@linux.intel.com> wrote:
> On Sun, 26 Feb 2017, Daniel Vetter <daniel@ffwll.ch> wrote:
>> On Thu, Feb 16, 2017 at 02:27:08PM +0200, Jani Nikula wrote:
>>> On Thu, 16 Feb 2017, Rob Clark <robdclark@gmail.com> wrote:
>>> > On Thu, Feb 16, 2017 at 7:00 AM, Dan Carpenter <dan.carpenter@oracle.com> wrote:
>>> >> We're off by one here.  We free something that wasn't allocated and we
>>> >> don't free msm_host->bus_clks[].
>>> >>
>>> >> Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
>>> >> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
>>> >
>>> > Thanks
>>> >
>>> > Reviewed-by: Rob Clark <robdclark@gmail.com>
>>>
>>> And for good measure,
>>>
>>> Reviewed-by: Jani Nikula <jani.nikula@intel.com>
>>
>> So 2 r-b from maintainers, no one said he'll apply. This isn't really
>> great coordination :-) Note that drm-misc-next is always open, so you
>> could always smash it in there, irrespective of merge window state. hint,
>> hint ...
>
> Admittedly I shied away from touching drm/msm.

Well that's kinda my point, we have a pile of maintainers who could
push this, which means if no one says they do, the patch will likely
get lost. Especially if the main maintainer (Rob here) smashes an r-b
onto a patch it's super confusing, at least to me.

I guess this is a downside to having lots of committers, and I started
to stumble over this in a bunch of places. I think as a rule we should
always state when we plan to or have merged a patch, and if it's just
an r-b assume it's lost ... At least that's how I deal with core
refactorings touching drivers, otherwise we'd probably never get them
all landed.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
_______________________________________________
Freedreno mailing list
Freedreno@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/freedreno

Re: [patch v2] drm/msm/dsi: free first element on error

[Rob Clark]

On Mon, Feb 27, 2017 at 5:25 AM, Daniel Vetter <daniel@ffwll.ch> wrote:
> On Mon, Feb 27, 2017 at 11:18 AM, Jani Nikula
> <jani.nikula@linux.intel.com> wrote:
>> On Sun, 26 Feb 2017, Daniel Vetter <daniel@ffwll.ch> wrote:
>>> On Thu, Feb 16, 2017 at 02:27:08PM +0200, Jani Nikula wrote:
>>>> On Thu, 16 Feb 2017, Rob Clark <robdclark@gmail.com> wrote:
>>>> > On Thu, Feb 16, 2017 at 7:00 AM, Dan Carpenter <dan.carpenter@oracle.com> wrote:
>>>> >> We're off by one here.  We free something that wasn't allocated and we
>>>> >> don't free msm_host->bus_clks[].
>>>> >>
>>>> >> Fixes: 6e0eb52eba9e ("drm/msm/dsi: Parse bus clocks from a list")
>>>> >> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
>>>> >
>>>> > Thanks
>>>> >
>>>> > Reviewed-by: Rob Clark <robdclark@gmail.com>
>>>>
>>>> And for good measure,
>>>>
>>>> Reviewed-by: Jani Nikula <jani.nikula@intel.com>
>>>
>>> So 2 r-b from maintainers, no one said he'll apply. This isn't really
>>> great coordination :-) Note that drm-misc-next is always open, so you
>>> could always smash it in there, irrespective of merge window state. hint,
>>> hint ...
>>
>> Admittedly I shied away from touching drm/msm.
>
> Well that's kinda my point, we have a pile of maintainers who could
> push this, which means if no one says they do, the patch will likely
> get lost. Especially if the main maintainer (Rob here) smashes an r-b
> onto a patch it's super confusing, at least to me.

the r-b was in case you wanted to pick it up in drm-misc (with the
usual backup plan of sending it in an msm-fixes pull req after the
first rc or two).

I don't mind pushing small fixes to drm-misc instead, since I
generally don't have a lot of 'em

BR,
-R

> I guess this is a downside to having lots of committers, and I started
> to stumble over this in a bunch of places. I think as a rule we should
> always state when we plan to or have merged a patch, and if it's just
> an r-b assume it's lost ... At least that's how I deal with core
> refactorings touching drivers, otherwise we'd probably never get them
> all landed.
> -Daniel
> --
> Daniel Vetter
> Software Engineer, Intel Corporation
> +41 (0) 79 365 57 48 - http://blog.ffwll.ch
_______________________________________________
Freedreno mailing list
Freedreno@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/freedreno