Re: [PATCH 1/5] remoteproc: stm32: Call of_node_put() on iteration error

From: Mathieu Poirier
Date: Tue Mar 21 2023 - 17:33:00 EST


On Tue, Mar 21, 2023 at 10:00:03AM +0100, Arnaud POULIQUEN wrote:
> Hi Mathieu,
>
> On 3/20/23 23:18, Mathieu Poirier wrote:
> > Function of_phandle_iterator_next() calls of_node_put() on the last
> > device_node it iterated over, but when the loop exits prematurely it has
> > to be called explicitly>
> > Fixes: 13140de09cc2 ("remoteproc: stm32: add an ST stm32_rproc driver")
> > Cc: stable@xxxxxxxxxxxxxxx
> > Signed-off-by: Mathieu Poirier <mathieu.poirier@xxxxxxxxxx>
> > ---
> > drivers/remoteproc/stm32_rproc.c | 6 +++++-
> > 1 file changed, 5 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/remoteproc/stm32_rproc.c b/drivers/remoteproc/stm32_rproc.c
> > index 7d782ed9e589..23c1690b8d73 100644
> > --- a/drivers/remoteproc/stm32_rproc.c
> > +++ b/drivers/remoteproc/stm32_rproc.c
> > @@ -223,11 +223,13 @@ static int stm32_rproc_prepare(struct rproc *rproc)
> > while (of_phandle_iterator_next(&it) == 0) {
> > rmem = of_reserved_mem_lookup(it.node);
> > if (!rmem) {
> > + of_node_put(it.node);
> > dev_err(dev, "unable to acquire memory-region\n");
> > return -EINVAL;
> > }
> >
> > if (stm32_rproc_pa_to_da(rproc, rmem->base, &da) < 0) {
> > + of_node_put(it.node);
> > dev_err(dev, "memory region not valid %pa\n",
> > &rmem->base);
> > return -EINVAL;
> > @@ -254,8 +256,10 @@ static int stm32_rproc_prepare(struct rproc *rproc)
> > it.node->name);
> > }
> >
> > - if (!mem)
> > + if (!mem) {
> > + of_node_put(it.node);
> > return -ENOMEM;
> > + }
>
> Good catch!
>
> Looking in code I don't see that we call of_node_put() when we release the
> carveouts.
> Please tell me if I'm wrong but look to me that we should also call of_node_put()
> in mem->release() op, in drivers.
>

Are you referring to entry->release(), which for stm32 is
stm32_rproc_mem_release(), in rproc_resource_cleanup()?

If so then no, it is not needed since of_phandle_iterator_next() calls
of_node_put() on the previous device_node with each iteration.

Otherwise I fail to understand the question and will ask you to clarify.

> This one remains valid.
> reviewed-by: Arnaud Pouliquen <arnaud.pouliquen@xxxxxxxxxxx>
>

Ok

> Thanks,
> Arnaud
>
>
> >
> > rproc_add_carveout(rproc, mem);
> > index++;