---
Alternatively, you could remove the nowadays-useless
"if-before-kfree" test and leave the sometimes-redundant
assignment as-is: unconditional.
drivers/net/wireless/ath9k/main.c | 3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/drivers/net/wireless/ath9k/main.c b/drivers/net/wireless/ath9k/main.c
index 2888778..21bef29 100644
--- a/drivers/net/wireless/ath9k/main.c
+++ b/drivers/net/wireless/ath9k/main.c
@@ -1056,9 +1056,10 @@ void ath_tx_complete(struct ath_softc *sc, struct sk_buff *skb,
if (tx_info->flags & IEEE80211_TX_CTL_NO_ACK ||
tx_info->flags & IEEE80211_TX_STAT_TX_FILTERED) {
/* free driver's private data area of tx_info */
- if (tx_info->driver_data[0] != NULL)
+ if (tx_info->driver_data[0] != NULL) {
kfree(tx_info->driver_data[0]);
tx_info->driver_data[0] = NULL;
+ }
}
if (tx_status->flags & ATH_TX_BAR) {
--
1.6.0.90.g436ed
Hi Jim
> Alternatively, you could remove the nowadays-useless
> "if-before-kfree" test and leave the sometimes-redundant
> assignment as-is: unconditional.
actually the later one is preferred. Just remove the NULL check.
Regards
Marcel
---
Marcel Holtmann <[email protected]> wrote:
> Hi Jim
>
>> Alternatively, you could remove the nowadays-useless
>> "if-before-kfree" test and leave the sometimes-redundant
>> assignment as-is: unconditional.
>
> actually the later one is preferred. Just remove the NULL check.
I prefer it, too, but the indentation suggested that
adding braces would match the author's intent.
In case anyone is interested, I have been maintaining a patch
that removes *all* useless if-before-kfree tests. I have not
posted it because such changes have a reputation for provoking
flames on this list, but now you've provoked me ;-)
http://meyering.net/code/remove-useless-if-before-kfree.patch
74 files changed, 127 insertions(+), 250 deletions(-)
Here's the alternative patch:
drivers/net/wireless/ath9k/main.c | 5 ++---
1 files changed, 2 insertions(+), 3 deletions(-)
diff --git a/drivers/net/wireless/ath9k/main.c b/drivers/net/wireless/ath9k/main.c
index 2888778..6eb7ad7 100644
--- a/drivers/net/wireless/ath9k/main.c
+++ b/drivers/net/wireless/ath9k/main.c
@@ -1056,9 +1056,8 @@ void ath_tx_complete(struct ath_softc *sc, struct sk_buff *skb,
if (tx_info->flags & IEEE80211_TX_CTL_NO_ACK ||
tx_info->flags & IEEE80211_TX_STAT_TX_FILTERED) {
/* free driver's private data area of tx_info */
- if (tx_info->driver_data[0] != NULL)
- kfree(tx_info->driver_data[0]);
- tx_info->driver_data[0] = NULL;
+ kfree(tx_info->driver_data[0]);
+ tx_info->driver_data[0] = NULL;
}
if (tx_status->flags & ATH_TX_BAR) {
--
1.6.0.90.g436ed
Hi Jim,
> >> Alternatively, you could remove the nowadays-useless
> >> "if-before-kfree" test and leave the sometimes-redundant
> >> assignment as-is: unconditional.
> >
> > actually the later one is preferred. Just remove the NULL check.
>
> I prefer it, too, but the indentation suggested that
> adding braces would match the author's intent.
>
> In case anyone is interested, I have been maintaining a patch
> that removes *all* useless if-before-kfree tests. I have not
> posted it because such changes have a reputation for provoking
> flames on this list, but now you've provoked me ;-)
>
> http://meyering.net/code/remove-useless-if-before-kfree.patch
> 74 files changed, 127 insertions(+), 250 deletions(-)
I am all for this patch.
Even the Glibc free() function will do an "if-before-free" :)
Regards
Marcel
Marcel Holtmann <[email protected]> wrote:
>> >> Alternatively, you could remove the nowadays-useless
>> >> "if-before-kfree" test and leave the sometimes-redundant
>> >> assignment as-is: unconditional.
>> >
>> > actually the later one is preferred. Just remove the NULL check.
>>
>> I prefer it, too, but the indentation suggested that
>> adding braces would match the author's intent.
>>
>> In case anyone is interested, I have been maintaining a patch
>> that removes *all* useless if-before-kfree tests. I have not
>> posted it because such changes have a reputation for provoking
>> flames on this list, but now you've provoked me ;-)
>>
>> http://meyering.net/code/remove-useless-if-before-kfree.patch
>> 74 files changed, 127 insertions(+), 250 deletions(-)
>
> I am all for this patch.
>
> Even the Glibc free() function will do an "if-before-free" :)
Hi Marcel,
Thanks for the support.
Yes, with glibc, free(NULL) is a no-op, and POSIX has required that
behavior for long enough that *all* implementations of free on reasonable
porting targets are NULL-safe these days. There has been plenty of
discussion as I've helped a few projects clean up in this regard, e.g.,
http://thread.gmane.org/gmane.comp.version-control.git/74187
http://thread.gmane.org/gmane.emacs.devel/98144
http://thread.gmane.org/gmane.comp.lib.gnulib.bugs/12712
Here are most of the projects that have endured this janitorial work:
git
emacs
glibc
gnulib
coreutils
freeIPA
libvirt
util-linux-ng
idutils
There's even a tool to help ensure that no new useless if-before-kfree
tests sneak back in later.
http://git.sv.gnu.org/gitweb/?p=gnulib.git;a=blob_plain;f=build-aux/useless-if-before-free