linux-pm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* re: PM / OPP: Add OPP sharing information to OPP library
@ 2015-08-10 16:40 Dan Carpenter
  2015-08-11  8:22 ` Viresh Kumar
  0 siblings, 1 reply; 2+ messages in thread
From: Dan Carpenter @ 2015-08-10 16:40 UTC (permalink / raw)
  To: viresh.kumar; +Cc: linux-pm

Hello Viresh Kumar,

The patch 064416586190: "PM / OPP: Add OPP sharing information to OPP
library" from Jul 29, 2015, leads to the following static checker
warning:

	drivers/base/power/opp.c:1341 _of_init_opp_table_v2()
	error: 'dev_opp' dereferencing possible ERR_PTR()

drivers/base/power/opp.c
  1328          }
  1329  
  1330          /* There should be one of more OPP defined */
  1331          if (WARN_ON(!count))
  1332                  goto put_opp_np;

Should we set "ret" here?

  1333  
  1334          if (!ret) {
  1335                  if (!dev_opp) {
                             ^^^^^^^
No need to test this, we tested at the start of the function.

  1336                          dev_opp = _find_device_opp(dev);
  1337                          if (WARN_ON(!dev_opp))

This should be checking for IS_ERR().  We probably want to set "ret =
PTR_ERR(dev_opp) as well.

  1338                                  goto put_opp_np;
  1339                  }
  1340  
  1341                  dev_opp->np = opp_np;
  1342                  dev_opp->shared_opp = of_property_read_bool(opp_np,
  1343                                                              "opp-shared");
  1344          } else {
  1345                  of_free_opp_table(dev);
  1346          }
  1347  
  1348  put_opp_np:
  1349          of_node_put(opp_np);
  1350  
  1351          return ret;
  1352  }

Checking for "!ret" on line 1334 is "success handling" style code.
Success handling leads to multiple indent levels and is more twisted and
confusing.  I generally prefer to keep my error handling paths separate
from the success path although in this case we call of_node_put() on
both paths, so maybe a shared exit path makes sense.

To me boils down to a question of, "How much is shared and how much is
different.  Also should we call of_free_opp_table() if _find_device_opp()
fails?"  If not then I would use a common exit path, otherwise I would
split them apart.

Shared:

	if (ret) {
		of_free_opp_table(dev);
		goto put_opp_npl;
	}

	dev_opp = _find_device_opp(dev);
	if (WARN_ON(IS_ERR(dev_opp))) {
		ret = PTR_ERR(dev_opp);
		goto put_opp_np;
	}

	dev_opp->np = opp_np;
	dev_opp->shared_opp = of_property_read_bool(opp_np, "opp-shared");

put_opp_np:
	of_node_put(opp_np);
	return ret;

Split apart:

	if (ret)
		goto free_opp_table;

	dev_opp = _find_device_opp(dev);
	if (WARN_ON(IS_ERR(dev_opp))) {
		ret = PTR_ERR(dev_opp);
		goto free_opp_table;
	}

	dev_opp->np = opp_np;
	dev_opp->shared_opp = of_property_read_bool(opp_np, "opp-shared");

	of_node_put(opp_np);
	return 0;

free_opp_table:
	of_free_opp_table(dev);
put_opp_np:
	of_node_put(opp_np);
	return ret;

regards,
dan carpenter

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: PM / OPP: Add OPP sharing information to OPP library
  2015-08-10 16:40 PM / OPP: Add OPP sharing information to OPP library Dan Carpenter
@ 2015-08-11  8:22 ` Viresh Kumar
  0 siblings, 0 replies; 2+ messages in thread
From: Viresh Kumar @ 2015-08-11  8:22 UTC (permalink / raw)
  To: Dan Carpenter; +Cc: linux-pm

Hi Dan,

Are the static checker scripts available (and easy to use) for others
as well? Maybe I can use them on code, before sending patches.

On 10-08-15, 19:40, Dan Carpenter wrote:
> Hello Viresh Kumar,
> 
> The patch 064416586190: "PM / OPP: Add OPP sharing information to OPP
> library" from Jul 29, 2015, leads to the following static checker
> warning:
> 
> 	drivers/base/power/opp.c:1341 _of_init_opp_table_v2()
> 	error: 'dev_opp' dereferencing possible ERR_PTR()
> 
> drivers/base/power/opp.c
>   1328          }
>   1329  
>   1330          /* There should be one of more OPP defined */
>   1331          if (WARN_ON(!count))
>   1332                  goto put_opp_np;
> 
> Should we set "ret" here?

Yes, one of my new patches is doing this.

>   1333  
>   1334          if (!ret) {
>   1335                  if (!dev_opp) {
>                              ^^^^^^^
> No need to test this, we tested at the start of the function.

Yes, I have killed this extra indentation level as well.

>   1336                          dev_opp = _find_device_opp(dev);
>   1337                          if (WARN_ON(!dev_opp))
> 
> This should be checking for IS_ERR().  We probably want to set "ret =
> PTR_ERR(dev_opp) as well.

Okay, that's new. Will fix.

>   1338                                  goto put_opp_np;
>   1339                  }
>   1340  
>   1341                  dev_opp->np = opp_np;
>   1342                  dev_opp->shared_opp = of_property_read_bool(opp_np,
>   1343                                                              "opp-shared");
>   1344          } else {
>   1345                  of_free_opp_table(dev);
>   1346          }
>   1347  
>   1348  put_opp_np:
>   1349          of_node_put(opp_np);
>   1350  
>   1351          return ret;
>   1352  }
> 
> Checking for "!ret" on line 1334 is "success handling" style code.
> Success handling leads to multiple indent levels and is more twisted and
> confusing.  I generally prefer to keep my error handling paths separate
> from the success path although in this case we call of_node_put() on
> both paths, so maybe a shared exit path makes sense.
> 
> To me boils down to a question of, "How much is shared and how much is
> different.  Also should we call of_free_opp_table() if _find_device_opp()
> fails?"  If not then I would use a common exit path, otherwise I would
> split them apart.
> 
> Shared:
> 
> 	if (ret) {
> 		of_free_opp_table(dev);
> 		goto put_opp_npl;
> 	}
> 
> 	dev_opp = _find_device_opp(dev);
> 	if (WARN_ON(IS_ERR(dev_opp))) {
> 		ret = PTR_ERR(dev_opp);
> 		goto put_opp_np;
> 	}
> 
> 	dev_opp->np = opp_np;
> 	dev_opp->shared_opp = of_property_read_bool(opp_np, "opp-shared");
> 
> put_opp_np:
> 	of_node_put(opp_np);
> 	return ret;
> 
> Split apart:
> 
> 	if (ret)
> 		goto free_opp_table;
> 
> 	dev_opp = _find_device_opp(dev);
> 	if (WARN_ON(IS_ERR(dev_opp))) {
> 		ret = PTR_ERR(dev_opp);
> 		goto free_opp_table;
> 	}
> 
> 	dev_opp->np = opp_np;
> 	dev_opp->shared_opp = of_property_read_bool(opp_np, "opp-shared");
> 
> 	of_node_put(opp_np);
> 	return 0;
> 
> free_opp_table:
> 	of_free_opp_table(dev);
> put_opp_np:
> 	of_node_put(opp_np);
> 	return ret;

Will check that as well.

-- 
viresh

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2015-08-11  8:22 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2015-08-10 16:40 PM / OPP: Add OPP sharing information to OPP library Dan Carpenter
2015-08-11  8:22 ` Viresh Kumar

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).