| Message ID | 1537962368-14019-1-git-send-email-zhongjiang@huawei.com (mailing list archive) |
|---|---|
| State | Rejected |
| Headers | show |
| Series | powerpc: Move a dereference below a NULL test | expand |
Le 26/09/2018 à 13:46, zhong jiang a écrit : > It is safe to move dereference below a NULL test. > > Signed-off-by: zhong jiang <zhongjiang@huawei.com> > --- > arch/powerpc/kernel/cacheinfo.c | 5 ++--- > 1 file changed, 2 insertions(+), 3 deletions(-) > > diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c > index a8f20e5..7f19714 100644 > --- a/arch/powerpc/kernel/cacheinfo.c > +++ b/arch/powerpc/kernel/cacheinfo.c > @@ -401,14 +401,13 @@ static struct cache *cache_lookup_or_instantiate(struct device_node *node, > struct cache *cache; > > cache = cache_lookup_by_node(node); > + if (!cache) > + cache = cache_do_one_devnode(node, level); But by doing this, you change the meaning of the following warning. Is that what you want ? In that case the text of the WARN_ONCE() should be changed, because the mismatch is not only on lookup now. Christophe > > WARN_ONCE(cache && cache->level != level, > "cache level mismatch on lookup (got %d, expected %d)\n", > cache->level, level); > > - if (!cache) > - cache = cache_do_one_devnode(node, level); > - > return cache; > } > >
On 2018/9/26 21:58, Christophe LEROY wrote: > > > Le 26/09/2018 à 13:46, zhong jiang a écrit : >> It is safe to move dereference below a NULL test. >> >> Signed-off-by: zhong jiang <zhongjiang@huawei.com> >> --- >> arch/powerpc/kernel/cacheinfo.c | 5 ++--- >> 1 file changed, 2 insertions(+), 3 deletions(-) >> >> diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c >> index a8f20e5..7f19714 100644 >> --- a/arch/powerpc/kernel/cacheinfo.c >> +++ b/arch/powerpc/kernel/cacheinfo.c >> @@ -401,14 +401,13 @@ static struct cache *cache_lookup_or_instantiate(struct device_node *node, >> struct cache *cache; >> cache = cache_lookup_by_node(node); >> + if (!cache) >> + cache = cache_do_one_devnode(node, level); > > But by doing this, you change the meaning of the following warning. Is that what you want ? In that case the text of the WARN_ONCE() should be changed, because the mismatch is not only on lookup now. > Yep, I forget the WARN_ONCE. I think we should just remove it. Thought? Thanks, zhong jiang > Christophe > >> WARN_ONCE(cache && cache->level != level, >> "cache level mismatch on lookup (got %d, expected %d)\n", >> cache->level, level); >> - if (!cache) >> - cache = cache_do_one_devnode(node, level); >> - >> return cache; >> } >> > > . >
On Wed, 26 Sep 2018 19:46:08 +0800 zhong jiang <zhongjiang@huawei.com> wrote: > It is safe to move dereference below a NULL test. > > Signed-off-by: zhong jiang <zhongjiang@huawei.com> > --- > arch/powerpc/kernel/cacheinfo.c | 5 ++--- > 1 file changed, 2 insertions(+), 3 deletions(-) > > diff --git a/arch/powerpc/kernel/cacheinfo.c > b/arch/powerpc/kernel/cacheinfo.c index a8f20e5..7f19714 100644 > --- a/arch/powerpc/kernel/cacheinfo.c > +++ b/arch/powerpc/kernel/cacheinfo.c > @@ -401,14 +401,13 @@ static struct cache > *cache_lookup_or_instantiate(struct device_node *node, struct cache > *cache; > cache = cache_lookup_by_node(node); > + if (!cache) > + cache = cache_do_one_devnode(node, level); > > WARN_ONCE(cache && cache->level != level, This has also null test so cache should be dereferenced only when non-null here. Thanks Michal
On 2018/9/26 22:22, Michal Suchánek wrote: > On Wed, 26 Sep 2018 19:46:08 +0800 > zhong jiang <zhongjiang@huawei.com> wrote: > >> It is safe to move dereference below a NULL test. >> >> Signed-off-by: zhong jiang <zhongjiang@huawei.com> >> --- >> arch/powerpc/kernel/cacheinfo.c | 5 ++--- >> 1 file changed, 2 insertions(+), 3 deletions(-) >> >> diff --git a/arch/powerpc/kernel/cacheinfo.c >> b/arch/powerpc/kernel/cacheinfo.c index a8f20e5..7f19714 100644 >> --- a/arch/powerpc/kernel/cacheinfo.c >> +++ b/arch/powerpc/kernel/cacheinfo.c >> @@ -401,14 +401,13 @@ static struct cache >> *cache_lookup_or_instantiate(struct device_node *node, struct cache >> *cache; >> cache = cache_lookup_by_node(node); >> + if (!cache) >> + cache = cache_do_one_devnode(node, level); >> >> WARN_ONCE(cache && cache->level != level, > This has also null test so cache should be dereferenced only when > non-null here. :-[ , you're right. I forget WARN_ONCE. please ignore the patch. Sincerely, zhong jiang > Thanks > > Michal > > . >
diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c index a8f20e5..7f19714 100644 --- a/arch/powerpc/kernel/cacheinfo.c +++ b/arch/powerpc/kernel/cacheinfo.c @@ -401,14 +401,13 @@ static struct cache *cache_lookup_or_instantiate(struct device_node *node, struct cache *cache; cache = cache_lookup_by_node(node); + if (!cache) + cache = cache_do_one_devnode(node, level); WARN_ONCE(cache && cache->level != level, "cache level mismatch on lookup (got %d, expected %d)\n", cache->level, level); - if (!cache) - cache = cache_do_one_devnode(node, level); - return cache; }
It is safe to move dereference below a NULL test. Signed-off-by: zhong jiang <zhongjiang@huawei.com> --- arch/powerpc/kernel/cacheinfo.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)