Message ID | 1347610743-20189-1-git-send-email-srinivas.kandagatla@st.com (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
On Fri, 2012-09-14 at 09:19 +0100, Srinivas KANDAGATLA wrote: > From: Srinivas Kandagatla <srinivas.kandagatla@st.com> > > As follow-up to "dt: introduce of_get_child_by_name to get child node by > name." patch, This patch removes some of the code duplication in the > driver by replacing it with of_get_child_by_name instead. > > Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@st.com> > --- > arch/powerpc/platforms/powernv/opal.c | 6 ++---- > 1 files changed, 2 insertions(+), 4 deletions(-) > > diff --git a/arch/powerpc/platforms/powernv/opal.c b/arch/powerpc/platforms/powernv/opal.c > index aaa0dba..6dfb8af 100644 > --- a/arch/powerpc/platforms/powernv/opal.c > +++ b/arch/powerpc/platforms/powernv/opal.c > @@ -294,11 +294,9 @@ static int __init opal_init(void) > consoles = of_node_get(opal_node); > > /* Register serial ports */ > - for_each_child_of_node(consoles, np) { > - if (strcmp(np->name, "serial")) > - continue; > + np = of_get_child_by_name(consoles, "serial"); > + if (np) > of_platform_device_create(np, NULL, NULL); > - } > of_node_put(consoles); You've introduced a refcounting bug here. The return value of of_get_child_by_name() has its refcount elevated (via of_node_get()), so you have to drop that reference with of_node_put(). The old code was safe because it only used np inside the loop, and the loop logic deals with dropping the refcount for you. And yes of_platform_device_create() takes a reference for itself. cheers
On 17/09/12 03:07, Michael Ellerman wrote: > On Fri, 2012-09-14 at 09:19 +0100, Srinivas KANDAGATLA wrote: >> From: Srinivas Kandagatla <srinivas.kandagatla@st.com> >> >> As follow-up to "dt: introduce of_get_child_by_name to get child node by >> name." patch, This patch removes some of the code duplication in the >> driver by replacing it with of_get_child_by_name instead. >> >> Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@st.com> >> --- >> arch/powerpc/platforms/powernv/opal.c | 6 ++---- >> 1 files changed, 2 insertions(+), 4 deletions(-) >> >> diff --git a/arch/powerpc/platforms/powernv/opal.c b/arch/powerpc/platforms/powernv/opal.c >> index aaa0dba..6dfb8af 100644 >> --- a/arch/powerpc/platforms/powernv/opal.c >> +++ b/arch/powerpc/platforms/powernv/opal.c >> @@ -294,11 +294,9 @@ static int __init opal_init(void) >> consoles = of_node_get(opal_node); >> >> /* Register serial ports */ >> - for_each_child_of_node(consoles, np) { >> - if (strcmp(np->name, "serial")) >> - continue; >> + np = of_get_child_by_name(consoles, "serial"); >> + if (np) >> of_platform_device_create(np, NULL, NULL); >> - } >> of_node_put(consoles); > You've introduced a refcounting bug here. > > The return value of of_get_child_by_name() has its refcount elevated > (via of_node_get()), so you have to drop that reference with > of_node_put(). I agree.. Will take care of it in v3 patch-series. > The old code was safe because it only used np inside the loop, and the > loop logic deals with dropping the refcount for you. > > And yes of_platform_device_create() takes a reference for itself. > > cheers > > >
diff --git a/arch/powerpc/platforms/powernv/opal.c b/arch/powerpc/platforms/powernv/opal.c index aaa0dba..6dfb8af 100644 --- a/arch/powerpc/platforms/powernv/opal.c +++ b/arch/powerpc/platforms/powernv/opal.c @@ -294,11 +294,9 @@ static int __init opal_init(void) consoles = of_node_get(opal_node); /* Register serial ports */ - for_each_child_of_node(consoles, np) { - if (strcmp(np->name, "serial")) - continue; + np = of_get_child_by_name(consoles, "serial"); + if (np) of_platform_device_create(np, NULL, NULL); - } of_node_put(consoles); /* Find all OPAL interrupts and request them */