dmaengine: tegra210-adma: fix of_irq_get() error check

Message ID 20170730181051.733168837@cogentembedded.com
State New
Headers show

Commit Message

Sergei Shtylyov July 30, 2017, 6:10 p.m.
of_irq_get() may return 0 as well as negative error number on failure,
while the driver only checks for the negative values. The driver would then
call request_irq(0, ...) in tegra_adma_alloc_chan_resources() and never get
valid channel interrupt.

Check for 'tdc->irq <= 0' instead and return -ENXIO from the driver's probe
iff of_irq_get() returned 0.

Fixes: f46b195799b5 ("dmaengine: tegra-adma: Add support for Tegra210 ADMA")
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>

---
 drivers/dma/tegra210-adma.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)


--
To unsubscribe from this list: send the line "unsubscribe linux-tegra" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

Sergei Shtylyov July 30, 2017, 7:46 p.m. | #1
Forgot to mention that the patch is against the 'fixes' branch of Vinod's 
'slave-dma.git' repo.

--
To unsubscribe from this list: send the line "unsubscribe linux-tegra" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Thierry Reding July 31, 2017, 9:30 a.m. | #2
On Sun, Jul 30, 2017 at 09:10:44PM +0300, Sergei Shtylyov wrote:
> of_irq_get() may return 0 as well as negative error number on failure,
> while the driver only checks for the negative values. The driver would then
> call request_irq(0, ...) in tegra_adma_alloc_chan_resources() and never get
> valid channel interrupt.
> 
> Check for 'tdc->irq <= 0' instead and return -ENXIO from the driver's probe
> iff of_irq_get() returned 0.
> 
> Fixes: f46b195799b5 ("dmaengine: tegra-adma: Add support for Tegra210 ADMA")
> Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
> 
> ---
>  drivers/dma/tegra210-adma.c |    4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)

Yeah, that interface isn't very optimal. The problem with it, and we've
had this kind before, is that every driver ends up having to implement a
fallback for == 0 with the result of everyone returning a different
error code.

Perhaps a good long-term solution would be to fix of_irq_get() to only
return positive for valid interrupts and negative error codes, so that
nobody has to deal with == 0 anymore.

That's obviously going to be a fairly big undertaking, so I'm fine with
fixing up the individual callsites first.

> Index: slave-dma/drivers/dma/tegra210-adma.c
> ===================================================================
> --- slave-dma.orig/drivers/dma/tegra210-adma.c
> +++ slave-dma/drivers/dma/tegra210-adma.c
> @@ -717,8 +717,8 @@ static int tegra_adma_probe(struct platf
>  		tdc->chan_addr = tdma->base_addr + ADMA_CH_REG_OFFSET(i);
>  
>  		tdc->irq = of_irq_get(pdev->dev.of_node, i);
> -		if (tdc->irq < 0) {
> -			ret = tdc->irq;
> +		if (tdc->irq <= 0) {
> +			ret = tdc->irq ?: -ENXIO;
>  			goto irq_dispose;
>  		}
>  

I think in this particular case it would be better to just call
platform_get_irq(), which already implements the equivalent.

Thierry
Sergei Shtylyov July 31, 2017, 9:57 a.m. | #3
Hello!

On 7/31/2017 12:30 PM, Thierry Reding wrote:

>> of_irq_get() may return 0 as well as negative error number on failure,
>> while the driver only checks for the negative values. The driver would then
>> call request_irq(0, ...) in tegra_adma_alloc_chan_resources() and never get
>> valid channel interrupt.
>>
>> Check for 'tdc->irq <= 0' instead and return -ENXIO from the driver's probe
>> iff of_irq_get() returned 0.
>>
>> Fixes: f46b195799b5 ("dmaengine: tegra-adma: Add support for Tegra210 ADMA")
>> Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
>>
>> ---
>>   drivers/dma/tegra210-adma.c |    4 ++--
>>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> Yeah, that interface isn't very optimal. The problem with it, and we've
> had this kind before, is that every driver ends up having to implement a
> fallback for == 0 with the result of everyone returning a different
> error code.

    Agreed.

> Perhaps a good long-term solution would be to fix of_irq_get() to only
> return positive for valid interrupts and negative error codes, so that
> nobody has to deal with == 0 anymore.

     See http://marc.info/?t=146447243300002&r=1 for my attempt to do this 
back in 2016.

> That's obviously going to be a fairly big undertaking, so I'm fine with
> fixing up the individual callsites first.
> 
>> Index: slave-dma/drivers/dma/tegra210-adma.c
>> ===================================================================
>> --- slave-dma.orig/drivers/dma/tegra210-adma.c
>> +++ slave-dma/drivers/dma/tegra210-adma.c
>> @@ -717,8 +717,8 @@ static int tegra_adma_probe(struct platf
>>   		tdc->chan_addr = tdma->base_addr + ADMA_CH_REG_OFFSET(i);
>>   
>>   		tdc->irq = of_irq_get(pdev->dev.of_node, i);
>> -		if (tdc->irq < 0) {
>> -			ret = tdc->irq;
>> +		if (tdc->irq <= 0) {
>> +			ret = tdc->irq ?: -ENXIO;
>>   			goto irq_dispose;
>>   		}
>>   
> 
> I think in this particular case it would be better to just call
> platform_get_irq(), which already implements the equivalent.

    OK, I'll try looking into that...

> Thierry

MBR, Sergei
--
To unsubscribe from this list: send the line "unsubscribe linux-tegra" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Sergei Shtylyov July 31, 2017, 12:59 p.m. | #4
On 07/31/2017 12:57 PM, Sergei Shtylyov wrote:

>>> of_irq_get() may return 0 as well as negative error number on failure,
>>> while the driver only checks for the negative values. The driver would then
>>> call request_irq(0, ...) in tegra_adma_alloc_chan_resources() and never get
>>> valid channel interrupt.
>>>
>>> Check for 'tdc->irq <= 0' instead and return -ENXIO from the driver's probe
>>> iff of_irq_get() returned 0.
>>>
>>> Fixes: f46b195799b5 ("dmaengine: tegra-adma: Add support for Tegra210 ADMA")
>>> Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
[...]

>>> Index: slave-dma/drivers/dma/tegra210-adma.c
>>> ===================================================================
>>> --- slave-dma.orig/drivers/dma/tegra210-adma.c
>>> +++ slave-dma/drivers/dma/tegra210-adma.c
>>> @@ -717,8 +717,8 @@ static int tegra_adma_probe(struct platf
>>>           tdc->chan_addr = tdma->base_addr + ADMA_CH_REG_OFFSET(i);
>>>             tdc->irq = of_irq_get(pdev->dev.of_node, i);
>>> -        if (tdc->irq < 0) {
>>> -            ret = tdc->irq;
>>> +        if (tdc->irq <= 0) {
>>> +            ret = tdc->irq ?: -ENXIO;
>>>               goto irq_dispose;
>>>           }
>>>
>>
>> I think in this particular case it would be better to just call
>> platform_get_irq(), which already implements the equivalent.

    Not that platform_get_irq() was no better (if not worse indeed, as it 
could return 0 both on failure ad success), until I fixed it back in 2016:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e330b9a6bb35dc7097a4f02cb1ae7b6f96df92af

>
>    OK, I'll try looking into that...

    I've looked at the commit history and no, I don't agree to use 
platform_get_irq(): my platform_get_irq() fix landed in mainline later than 
this driver, so fixing this bug in stable kernels would ensue an extra 
dependency...

>> Thierry

MBR, Sergei

--
To unsubscribe from this list: send the line "unsubscribe linux-tegra" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Thierry Reding July 31, 2017, 4:56 p.m. | #5
On Mon, Jul 31, 2017 at 03:59:35PM +0300, Sergei Shtylyov wrote:
> On 07/31/2017 12:57 PM, Sergei Shtylyov wrote:
> 
> > > > of_irq_get() may return 0 as well as negative error number on failure,
> > > > while the driver only checks for the negative values. The driver would then
> > > > call request_irq(0, ...) in tegra_adma_alloc_chan_resources() and never get
> > > > valid channel interrupt.
> > > > 
> > > > Check for 'tdc->irq <= 0' instead and return -ENXIO from the driver's probe
> > > > iff of_irq_get() returned 0.
> > > > 
> > > > Fixes: f46b195799b5 ("dmaengine: tegra-adma: Add support for Tegra210 ADMA")
> > > > Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
> [...]
> 
> > > > Index: slave-dma/drivers/dma/tegra210-adma.c
> > > > ===================================================================
> > > > --- slave-dma.orig/drivers/dma/tegra210-adma.c
> > > > +++ slave-dma/drivers/dma/tegra210-adma.c
> > > > @@ -717,8 +717,8 @@ static int tegra_adma_probe(struct platf
> > > >           tdc->chan_addr = tdma->base_addr + ADMA_CH_REG_OFFSET(i);
> > > >             tdc->irq = of_irq_get(pdev->dev.of_node, i);
> > > > -        if (tdc->irq < 0) {
> > > > -            ret = tdc->irq;
> > > > +        if (tdc->irq <= 0) {
> > > > +            ret = tdc->irq ?: -ENXIO;
> > > >               goto irq_dispose;
> > > >           }
> > > > 
> > > 
> > > I think in this particular case it would be better to just call
> > > platform_get_irq(), which already implements the equivalent.
> 
>    Not that platform_get_irq() was no better (if not worse indeed, as it
> could return 0 both on failure ad success), until I fixed it back in 2016:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e330b9a6bb35dc7097a4f02cb1ae7b6f96df92af
> 
> > 
> >    OK, I'll try looking into that...
> 
>    I've looked at the commit history and no, I don't agree to use
> platform_get_irq(): my platform_get_irq() fix landed in mainline later than
> this driver, so fixing this bug in stable kernels would ensue an extra
> dependency...

Well, surely the platform_get_irq() function should also be fixed in
stable kernels, so that all callers can be fixed at the same time.

That said, this does fix a bug, and it fixes it in the least invasive
way, and the damage I'm complaining about is preexisting, so:

Acked-by: Thierry Reding <treding@nvidia.com>
Sergei Shtylyov July 31, 2017, 5:07 p.m. | #6
On 07/31/2017 07:56 PM, Thierry Reding wrote:

>>>>> of_irq_get() may return 0 as well as negative error number on failure,
>>>>> while the driver only checks for the negative values. The driver would then
>>>>> call request_irq(0, ...) in tegra_adma_alloc_chan_resources() and never get
>>>>> valid channel interrupt.
>>>>>
>>>>> Check for 'tdc->irq <= 0' instead and return -ENXIO from the driver's probe
>>>>> iff of_irq_get() returned 0.
>>>>>
>>>>> Fixes: f46b195799b5 ("dmaengine: tegra-adma: Add support for Tegra210 ADMA")
>>>>> Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
>> [...]
>>
>>>>> Index: slave-dma/drivers/dma/tegra210-adma.c
>>>>> ===================================================================
>>>>> --- slave-dma.orig/drivers/dma/tegra210-adma.c
>>>>> +++ slave-dma/drivers/dma/tegra210-adma.c
>>>>> @@ -717,8 +717,8 @@ static int tegra_adma_probe(struct platf
>>>>>            tdc->chan_addr = tdma->base_addr + ADMA_CH_REG_OFFSET(i);
>>>>>              tdc->irq = of_irq_get(pdev->dev.of_node, i);
>>>>> -        if (tdc->irq < 0) {
>>>>> -            ret = tdc->irq;
>>>>> +        if (tdc->irq <= 0) {
>>>>> +            ret = tdc->irq ?: -ENXIO;
>>>>>                goto irq_dispose;
>>>>>            }
>>>>>
>>>>
>>>> I think in this particular case it would be better to just call
>>>> platform_get_irq(), which already implements the equivalent.

>>     Not that platform_get_irq() was no better (if not worse indeed, as it

    s/Not/Note/.

>> could return 0 both on failure ad success), until I fixed it back in 2016:

    s/ad/and/.

>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e330b9a6bb35dc7097a4f02cb1ae7b6f96df92af
>>
>>>     OK, I'll try looking into that...
>>
>>     I've looked at the commit history and no, I don't agree to use
>> platform_get_irq(): my platform_get_irq() fix landed in mainline later than
>> this driver, so fixing this bug in stable kernels would ensue an extra
>> dependency...
> 
> Well, surely the platform_get_irq() function should also be fixed in
> stable kernels, so that all callers can be fixed at the same time.

    It has been fixed in 3.16.y, 4.4.y and 4.8.y so far. I haven't received 
other stable mails yet.

> That said, this does fix a bug, and it fixes it in the least invasive
> way, and the damage I'm complaining about is preexisting, so:
> 
> Acked-by: Thierry Reding <treding@nvidia.com>

    Thank you!

MBR, Sergei
--
To unsubscribe from this list: send the line "unsubscribe linux-tegra" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Jon Hunter July 31, 2017, 8:11 p.m. | #7
On 31/07/17 17:56, Thierry Reding wrote:
> * PGP Signed by an unknown key
> 
> On Mon, Jul 31, 2017 at 03:59:35PM +0300, Sergei Shtylyov wrote:
>> On 07/31/2017 12:57 PM, Sergei Shtylyov wrote:
>>
>>>>> of_irq_get() may return 0 as well as negative error number on failure,
>>>>> while the driver only checks for the negative values. The driver would then
>>>>> call request_irq(0, ...) in tegra_adma_alloc_chan_resources() and never get
>>>>> valid channel interrupt.
>>>>>
>>>>> Check for 'tdc->irq <= 0' instead and return -ENXIO from the driver's probe
>>>>> iff of_irq_get() returned 0.
>>>>>
>>>>> Fixes: f46b195799b5 ("dmaengine: tegra-adma: Add support for Tegra210 ADMA")
>>>>> Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
>> [...]
>>
>>>>> Index: slave-dma/drivers/dma/tegra210-adma.c
>>>>> ===================================================================
>>>>> --- slave-dma.orig/drivers/dma/tegra210-adma.c
>>>>> +++ slave-dma/drivers/dma/tegra210-adma.c
>>>>> @@ -717,8 +717,8 @@ static int tegra_adma_probe(struct platf
>>>>>           tdc->chan_addr = tdma->base_addr + ADMA_CH_REG_OFFSET(i);
>>>>>             tdc->irq = of_irq_get(pdev->dev.of_node, i);
>>>>> -        if (tdc->irq < 0) {
>>>>> -            ret = tdc->irq;
>>>>> +        if (tdc->irq <= 0) {
>>>>> +            ret = tdc->irq ?: -ENXIO;
>>>>>               goto irq_dispose;
>>>>>           }
>>>>>
>>>>
>>>> I think in this particular case it would be better to just call
>>>> platform_get_irq(), which already implements the equivalent.
>>
>>    Not that platform_get_irq() was no better (if not worse indeed, as it
>> could return 0 both on failure ad success), until I fixed it back in 2016:
>>
>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e330b9a6bb35dc7097a4f02cb1ae7b6f96df92af
>>
>>>
>>>    OK, I'll try looking into that...
>>
>>    I've looked at the commit history and no, I don't agree to use
>> platform_get_irq(): my platform_get_irq() fix landed in mainline later than
>> this driver, so fixing this bug in stable kernels would ensue an extra
>> dependency...
> 
> Well, surely the platform_get_irq() function should also be fixed in
> stable kernels, so that all callers can be fixed at the same time.
> 
> That said, this does fix a bug, and it fixes it in the least invasive
> way, and the damage I'm complaining about is preexisting, so:
> 
> Acked-by: Thierry Reding <treding@nvidia.com>

FWIW ...

Acked-by: Jon Hunter <jonathanh@nvidia.com>

Thanks!
Jon
Vinod Koul Aug. 9, 2017, 6:10 a.m. | #8
On Sun, Jul 30, 2017 at 09:10:44PM +0300, Sergei Shtylyov wrote:
> of_irq_get() may return 0 as well as negative error number on failure,
> while the driver only checks for the negative values. The driver would then
> call request_irq(0, ...) in tegra_adma_alloc_chan_resources() and never get
> valid channel interrupt.
> 
> Check for 'tdc->irq <= 0' instead and return -ENXIO from the driver's probe
> iff of_irq_get() returned 0.

Applied, thanks

Patch

Index: slave-dma/drivers/dma/tegra210-adma.c
===================================================================
--- slave-dma.orig/drivers/dma/tegra210-adma.c
+++ slave-dma/drivers/dma/tegra210-adma.c
@@ -717,8 +717,8 @@  static int tegra_adma_probe(struct platf
 		tdc->chan_addr = tdma->base_addr + ADMA_CH_REG_OFFSET(i);
 
 		tdc->irq = of_irq_get(pdev->dev.of_node, i);
-		if (tdc->irq < 0) {
-			ret = tdc->irq;
+		if (tdc->irq <= 0) {
+			ret = tdc->irq ?: -ENXIO;
 			goto irq_dispose;
 		}