Re: [PATCH] staging: rtl8192u: Remove an unnecessary NULL check

2019-05-22 Thread Nathan Chancellor
On Wed, May 22, 2019 at 10:04:18AM +0300, Dan Carpenter wrote:
> On Tue, May 21, 2019 at 10:42:21AM -0700, Nathan Chancellor wrote:
> > Clang warns:
> > 
> > drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:2663:47: warning:
> > address of array 'param->u.wpa_ie.data' will always evaluate to 'true'
> > [-Wpointer-bool-conversion]
> > (param->u.wpa_ie.len && !param->u.wpa_ie.data))
> > ~^~~~
> > 
> > This was exposed by commit deabe03523a7 ("Staging: rtl8192u: ieee80211:
> > Use !x in place of NULL comparisons") because we disable the warning
> > that would have pointed out the comparison against NULL is also false:
> > 
> 
> Heh.  Weird.  Why would people disable one and not the other?
> 
> regards,
> dan carpenter
> 

-Wtautological-compare has a lot of different sub-warnings under it,
one of which is the one shown. -Wno-tautological-compare turns off all
of those other sub-warnings. The reason that was done is there are quite
a few of them:

https://gist.github.com/nathanchance/3336adc6e796b57eadd53b106b96c569

https://clang.llvm.org/docs/DiagnosticsReference.html#wtautological-compare

It is probably worth looking into turning that on, I'm going to try to
do that as I have time.

Cheers,
Nathan
___
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel


Re: [PATCH] staging: rtl8192u: Remove an unnecessary NULL check

2019-05-22 Thread Dan Carpenter
On Tue, May 21, 2019 at 03:57:46PM -0700, Nick Desaulniers wrote:
> > > -   if (param->u.wpa_ie.len > MAX_WPA_IE_LEN ||
> > > -   (param->u.wpa_ie.len && !param->u.wpa_ie.data))
> >
> > Right so, the types in this expression:
> >
> > param: struct ieee_param*
> > param->u: *anonymous union*
> > param->u.wpa_ie: *anonymous struct*
> > param->u.wpa_ie.len: u32
> > param->u.wpa_ie.data: u8 [0]
> > as defined in drivers/staging/rtl8192u/ieee80211/ieee80211.h#L295
> > https://github.com/ClangBuiltLinux/linux/blob/9c7db5004280767566e91a33445bf93aa479ef02/drivers/staging/rtl8192u/ieee80211/ieee80211.h#L295-L322
> >
> > so this is a tricky case, because in general array members can never
> > themselves be NULL,


Unless they array was the first struct member, obviously.


> >  and usually I trust -Wpointer-bool-conversion, but
> > this is a special case because of the flexible array member:

Nah.  It's the same thing.  That patch is fine.

regards,
dan carpenter

___
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel


Re: [PATCH] staging: rtl8192u: Remove an unnecessary NULL check

2019-05-22 Thread Dan Carpenter
On Tue, May 21, 2019 at 10:42:21AM -0700, Nathan Chancellor wrote:
> Clang warns:
> 
> drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:2663:47: warning:
> address of array 'param->u.wpa_ie.data' will always evaluate to 'true'
> [-Wpointer-bool-conversion]
> (param->u.wpa_ie.len && !param->u.wpa_ie.data))
> ~^~~~
> 
> This was exposed by commit deabe03523a7 ("Staging: rtl8192u: ieee80211:
> Use !x in place of NULL comparisons") because we disable the warning
> that would have pointed out the comparison against NULL is also false:
> 

Heh.  Weird.  Why would people disable one and not the other?

regards,
dan carpenter

___
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel


Re: [PATCH] staging: rtl8192u: Remove an unnecessary NULL check

2019-05-21 Thread Nick Desaulniers
On Tue, May 21, 2019 at 10:42 AM Nathan Chancellor
 wrote:
>
> Clang warns:
>
> drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:2663:47: warning:
> address of array 'param->u.wpa_ie.data' will always evaluate to 'true'
> [-Wpointer-bool-conversion]
> (param->u.wpa_ie.len && !param->u.wpa_ie.data))
> ~^~~~
>
> This was exposed by commit deabe03523a7 ("Staging: rtl8192u: ieee80211:
> Use !x in place of NULL comparisons") because we disable the warning
> that would have pointed out the comparison against NULL is also false:
>
> drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:2663:46: warning:
> comparison of array 'param->u.wpa_ie.data' equal to a null pointer is
> always false [-Wtautological-pointer-compare]
> (param->u.wpa_ie.len && param->u.wpa_ie.data == NULL))
> ^~~~
>
> Remove it so clang no longer warns.
>
> Link: https://github.com/ClangBuiltLinux/linux/issues/487
> Signed-off-by: Nathan Chancellor 
> ---
>  drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c 
> b/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c
> index f38f9d8b78bb..e0da0900a4f7 100644
> --- a/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c
> +++ b/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c
> @@ -2659,8 +2659,7 @@ static int ieee80211_wpa_set_wpa_ie(struct 
> ieee80211_device *ieee,
>  {
> u8 *buf;
>
> -   if (param->u.wpa_ie.len > MAX_WPA_IE_LEN ||
> -   (param->u.wpa_ie.len && !param->u.wpa_ie.data))

Right so, the types in this expression:

param: struct ieee_param*
param->u: *anonymous union*
param->u.wpa_ie: *anonymous struct*
param->u.wpa_ie.len: u32
param->u.wpa_ie.data: u8 [0]
as defined in drivers/staging/rtl8192u/ieee80211/ieee80211.h#L295
https://github.com/ClangBuiltLinux/linux/blob/9c7db5004280767566e91a33445bf93aa479ef02/drivers/staging/rtl8192u/ieee80211/ieee80211.h#L295-L322

so this is a tricky case, because in general array members can never
themselves be NULL, and usually I trust -Wpointer-bool-conversion, but
this is a special case because of the flexible array member:
https://en.wikipedia.org/wiki/Flexible_array_member. (It seems that
having the 0 in the length explicitly was pre-c99 GNU extension:
https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html). I wonder if
-Wtautological-pointer-compare applies to Flexible Array Members or
not (Richard, do you know)?  In general, you'd be setting
param->u.wpa_ie to the return value of a dynamic memory allocation,
not param->u.wpa_ie.data, so this check is fishy to me.

> +   if (param->u.wpa_ie.len > MAX_WPA_IE_LEN)
> return -EINVAL;
>
> if (param->u.wpa_ie.len) {
> --
> 2.22.0.rc1
>


-- 
Thanks,
~Nick Desaulniers
___
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel


[PATCH] staging: rtl8192u: Remove an unnecessary NULL check

2019-05-21 Thread Nathan Chancellor
Clang warns:

drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:2663:47: warning:
address of array 'param->u.wpa_ie.data' will always evaluate to 'true'
[-Wpointer-bool-conversion]
(param->u.wpa_ie.len && !param->u.wpa_ie.data))
~^~~~

This was exposed by commit deabe03523a7 ("Staging: rtl8192u: ieee80211:
Use !x in place of NULL comparisons") because we disable the warning
that would have pointed out the comparison against NULL is also false:

drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:2663:46: warning:
comparison of array 'param->u.wpa_ie.data' equal to a null pointer is
always false [-Wtautological-pointer-compare]
(param->u.wpa_ie.len && param->u.wpa_ie.data == NULL))
^~~~

Remove it so clang no longer warns.

Link: https://github.com/ClangBuiltLinux/linux/issues/487
Signed-off-by: Nathan Chancellor 
---
 drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c 
b/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c
index f38f9d8b78bb..e0da0900a4f7 100644
--- a/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c
+++ b/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c
@@ -2659,8 +2659,7 @@ static int ieee80211_wpa_set_wpa_ie(struct 
ieee80211_device *ieee,
 {
u8 *buf;
 
-   if (param->u.wpa_ie.len > MAX_WPA_IE_LEN ||
-   (param->u.wpa_ie.len && !param->u.wpa_ie.data))
+   if (param->u.wpa_ie.len > MAX_WPA_IE_LEN)
return -EINVAL;
 
if (param->u.wpa_ie.len) {
-- 
2.22.0.rc1

___
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel