* [PATCH] drivers: of: check input parameter name for __of_find_property @ 2015-09-11 13:44 Peng Fan 2015-09-15 15:56 ` Rob Herring 0 siblings, 1 reply; 5+ messages in thread From: Peng Fan @ 2015-09-11 13:44 UTC (permalink / raw) To: robh+dt; +Cc: devicetree, van.freenix, linux-kernel, Frank Rowand, Grant Likely Check input parameter 'name' for __of_find_property. If name is NULL, of_prop_cmp->strcasecmp may trigger panic. Signed-off-by: Peng Fan <van.freenix@gmail.com> Cc: Rob Herring <robh+dt@kernel.org> Cc: Frank Rowand <frowand.list@gmail.com> Cc: Grant Likely <grant.likely@linaro.org> --- drivers/of/base.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/of/base.c b/drivers/of/base.c index 8b5a187..e41436d 100644 --- a/drivers/of/base.c +++ b/drivers/of/base.c @@ -215,7 +215,7 @@ static struct property *__of_find_property(const struct device_node *np, { struct property *pp; - if (!np) + if (!np || !name) return NULL; for (pp = np->properties; pp; pp = pp->next) { -- 1.8.4 ^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH] drivers: of: check input parameter name for __of_find_property 2015-09-11 13:44 [PATCH] drivers: of: check input parameter name for __of_find_property Peng Fan @ 2015-09-15 15:56 ` Rob Herring 2015-09-16 0:16 ` Peng Fan 0 siblings, 1 reply; 5+ messages in thread From: Rob Herring @ 2015-09-15 15:56 UTC (permalink / raw) To: Peng Fan; +Cc: devicetree, linux-kernel, Frank Rowand, Grant Likely On 09/11/2015 08:44 AM, Peng Fan wrote: > Check input parameter 'name' for __of_find_property. If name is NULL, > of_prop_cmp->strcasecmp may trigger panic. Arguably that could be a feature. Do you have a usecase where name being NULL is valid and panicking is a problem? Rob > > Signed-off-by: Peng Fan <van.freenix@gmail.com> > Cc: Rob Herring <robh+dt@kernel.org> > Cc: Frank Rowand <frowand.list@gmail.com> > Cc: Grant Likely <grant.likely@linaro.org> > --- > drivers/of/base.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/of/base.c b/drivers/of/base.c > index 8b5a187..e41436d 100644 > --- a/drivers/of/base.c > +++ b/drivers/of/base.c > @@ -215,7 +215,7 @@ static struct property *__of_find_property(const struct device_node *np, > { > struct property *pp; > > - if (!np) > + if (!np || !name) > return NULL; > > for (pp = np->properties; pp; pp = pp->next) { > ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] drivers: of: check input parameter name for __of_find_property 2015-09-15 15:56 ` Rob Herring @ 2015-09-16 0:16 ` Peng Fan 2015-09-16 2:27 ` Rob Herring 0 siblings, 1 reply; 5+ messages in thread From: Peng Fan @ 2015-09-16 0:16 UTC (permalink / raw) To: Rob Herring; +Cc: devicetree, linux-kernel, Frank Rowand, Grant Likely Hi Rob, On Tue, Sep 15, 2015 at 10:56:28AM -0500, Rob Herring wrote: > On 09/11/2015 08:44 AM, Peng Fan wrote: > > Check input parameter 'name' for __of_find_property. If name is NULL, > > of_prop_cmp->strcasecmp may trigger panic. > > Arguably that could be a feature. Do you have a usecase where name being > NULL is valid and panicking is a problem? In drivers/pinctrl/devicetree.c 195 propname = kasprintf(GFP_KERNEL, "pinctrl-%d", state); 196 prop = of_find_property(np, propname, &size); 197 kfree(propname); 198 if (!prop) 199 break; If propname is NULL, of_find_property may trigger panic. Anyway propname should be checked before passing to of_find_property. I did not met panic message. I wrote this patch when I was reading the piece code. I think the name parameter should be checked before doing string compare. Regards, Peng. > > Rob > > > > > Signed-off-by: Peng Fan <van.freenix@gmail.com> > > Cc: Rob Herring <robh+dt@kernel.org> > > Cc: Frank Rowand <frowand.list@gmail.com> > > Cc: Grant Likely <grant.likely@linaro.org> > > --- > > drivers/of/base.c | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/drivers/of/base.c b/drivers/of/base.c > > index 8b5a187..e41436d 100644 > > --- a/drivers/of/base.c > > +++ b/drivers/of/base.c > > @@ -215,7 +215,7 @@ static struct property *__of_find_property(const struct device_node *np, > > { > > struct property *pp; > > > > - if (!np) > > + if (!np || !name) > > return NULL; > > > > for (pp = np->properties; pp; pp = pp->next) { > > > ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] drivers: of: check input parameter name for __of_find_property 2015-09-16 0:16 ` Peng Fan @ 2015-09-16 2:27 ` Rob Herring 2015-09-16 13:55 ` Peng Fan 0 siblings, 1 reply; 5+ messages in thread From: Rob Herring @ 2015-09-16 2:27 UTC (permalink / raw) To: Peng Fan; +Cc: devicetree, linux-kernel, Frank Rowand, Grant Likely On 09/15/2015 07:16 PM, Peng Fan wrote: > Hi Rob, > > On Tue, Sep 15, 2015 at 10:56:28AM -0500, Rob Herring wrote: >> On 09/11/2015 08:44 AM, Peng Fan wrote: >>> Check input parameter 'name' for __of_find_property. If name is NULL, >>> of_prop_cmp->strcasecmp may trigger panic. >> >> Arguably that could be a feature. Do you have a usecase where name being >> NULL is valid and panicking is a problem? > > In drivers/pinctrl/devicetree.c > > 195 propname = kasprintf(GFP_KERNEL, "pinctrl-%d", state); > 196 prop = of_find_property(np, propname, &size); > 197 kfree(propname); > 198 if (!prop) > 199 break; > > If propname is NULL, of_find_property may trigger panic. Anyway propname should be checked > before passing to of_find_property. propname will only be NULL if the memory allocation failed which gives you a big warning message. I don't think you would want to continue on in this case. So I'm inclined to leave this as is with passing NULL to of_find_property to always be an error and fatal. Rob > I did not met panic message. I wrote this patch when I was reading the piece code. > I think the name parameter should be checked before doing string compare. > > Regards, > Peng. > >> >> Rob >> >>> >>> Signed-off-by: Peng Fan <van.freenix@gmail.com> >>> Cc: Rob Herring <robh+dt@kernel.org> >>> Cc: Frank Rowand <frowand.list@gmail.com> >>> Cc: Grant Likely <grant.likely@linaro.org> >>> --- >>> drivers/of/base.c | 2 +- >>> 1 file changed, 1 insertion(+), 1 deletion(-) >>> >>> diff --git a/drivers/of/base.c b/drivers/of/base.c >>> index 8b5a187..e41436d 100644 >>> --- a/drivers/of/base.c >>> +++ b/drivers/of/base.c >>> @@ -215,7 +215,7 @@ static struct property *__of_find_property(const struct device_node *np, >>> { >>> struct property *pp; >>> >>> - if (!np) >>> + if (!np || !name) >>> return NULL; >>> >>> for (pp = np->properties; pp; pp = pp->next) { >>> >> ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] drivers: of: check input parameter name for __of_find_property 2015-09-16 2:27 ` Rob Herring @ 2015-09-16 13:55 ` Peng Fan 0 siblings, 0 replies; 5+ messages in thread From: Peng Fan @ 2015-09-16 13:55 UTC (permalink / raw) To: Rob Herring; +Cc: devicetree, linux-kernel, Frank Rowand, Grant Likely On Tue, Sep 15, 2015 at 09:27:59PM -0500, Rob Herring wrote: > On 09/15/2015 07:16 PM, Peng Fan wrote: > > Hi Rob, > > > > On Tue, Sep 15, 2015 at 10:56:28AM -0500, Rob Herring wrote: > >> On 09/11/2015 08:44 AM, Peng Fan wrote: > >>> Check input parameter 'name' for __of_find_property. If name is NULL, > >>> of_prop_cmp->strcasecmp may trigger panic. > >> > >> Arguably that could be a feature. Do you have a usecase where name being > >> NULL is valid and panicking is a problem? > > > > In drivers/pinctrl/devicetree.c > > > > 195 propname = kasprintf(GFP_KERNEL, "pinctrl-%d", state); > > 196 prop = of_find_property(np, propname, &size); > > 197 kfree(propname); > > 198 if (!prop) > > 199 break; > > > > If propname is NULL, of_find_property may trigger panic. Anyway propname should be checked > > before passing to of_find_property. > > propname will only be NULL if the memory allocation failed which gives > you a big warning message. I don't think you would want to continue on > in this case. > > So I'm inclined to leave this as is with passing NULL to > of_find_property to always be an error and fatal. of_find_property has EXPORT_SYMBOL, it maybe called in different places. And it should handle the case that name is NULL, right? If passing NULL to of_find_property is an error, then why check parameter 'np' like "if (!np)" in of_find_property? Regards, Peng. > > Rob > > > I did not met panic message. I wrote this patch when I was reading the piece code. > > I think the name parameter should be checked before doing string compare. > > > > Regards, > > Peng. > > > >> > >> Rob > >> > >>> > >>> Signed-off-by: Peng Fan <van.freenix@gmail.com> > >>> Cc: Rob Herring <robh+dt@kernel.org> > >>> Cc: Frank Rowand <frowand.list@gmail.com> > >>> Cc: Grant Likely <grant.likely@linaro.org> > >>> --- > >>> drivers/of/base.c | 2 +- > >>> 1 file changed, 1 insertion(+), 1 deletion(-) > >>> > >>> diff --git a/drivers/of/base.c b/drivers/of/base.c > >>> index 8b5a187..e41436d 100644 > >>> --- a/drivers/of/base.c > >>> +++ b/drivers/of/base.c > >>> @@ -215,7 +215,7 @@ static struct property *__of_find_property(const struct device_node *np, > >>> { > >>> struct property *pp; > >>> > >>> - if (!np) > >>> + if (!np || !name) > >>> return NULL; > >>> > >>> for (pp = np->properties; pp; pp = pp->next) { > >>> > >> > ^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2015-09-16 13:55 UTC | newest] Thread overview: 5+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2015-09-11 13:44 [PATCH] drivers: of: check input parameter name for __of_find_property Peng Fan 2015-09-15 15:56 ` Rob Herring 2015-09-16 0:16 ` Peng Fan 2015-09-16 2:27 ` Rob Herring 2015-09-16 13:55 ` Peng Fan
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox