Re: [PATCH 2/3] of: Simplify of_dma_set_restricted_buffer() to use of_for_each_phandle()

[Chen-Yu Tsai]

On Thu, Mar 27, 2025 at 2:53 AM Rob Herring <robh@xxxxxxxxxx> wrote:
>
> On Wed, Mar 26, 2025 at 1:44 AM Chen-Yu Tsai <wens@xxxxxxxxxx> wrote:
> >
> > Hi,
> >
> > On Tue, Mar 18, 2025 at 7:29 AM Rob Herring (Arm) <robh@xxxxxxxxxx> wrote:
> > >
> > > Simplify of_dma_set_restricted_buffer() by using of_property_present()
> > > and of_for_each_phandle() iterator.
> > >
> > > Signed-off-by: Rob Herring (Arm) <robh@xxxxxxxxxx>
> > > ---
> > >  drivers/of/device.c | 34 +++++++++++++---------------------
> > >  1 file changed, 13 insertions(+), 21 deletions(-)
> > >
> > > diff --git a/drivers/of/device.c b/drivers/of/device.c
> > > index edf3be197265..bb4a47d58249 100644
> > > --- a/drivers/of/device.c
> > > +++ b/drivers/of/device.c
> > > @@ -35,44 +35,36 @@ EXPORT_SYMBOL(of_match_device);
> > >  static void
> > >  of_dma_set_restricted_buffer(struct device *dev, struct device_node *np)
> > >  {
> > > -       struct device_node *node, *of_node = dev->of_node;
> > > -       int count, i;
> > > +       struct device_node *of_node = dev->of_node;
> > > +       struct of_phandle_iterator it;
> > > +       int rc, i = 0;
> > >
> > >         if (!IS_ENABLED(CONFIG_DMA_RESTRICTED_POOL))
> > >                 return;
> > >
> > > -       count = of_property_count_elems_of_size(of_node, "memory-region",
> > > -                                               sizeof(u32));
> > >         /*
> > >          * If dev->of_node doesn't exist or doesn't contain memory-region, try
> > >          * the OF node having DMA configuration.
> > >          */
> > > -       if (count <= 0) {
> > > +       if (!of_property_present(of_node, "memory-region"))
> > >                 of_node = np;
> > > -               count = of_property_count_elems_of_size(
> > > -                       of_node, "memory-region", sizeof(u32));
> > > -       }
> > >
> > > -       for (i = 0; i < count; i++) {
> > > -               node = of_parse_phandle(of_node, "memory-region", i);
> > > +       of_for_each_phandle(&it, rc, of_node, "memory-region", NULL, 0) {
> > >                 /*
> > >                  * There might be multiple memory regions, but only one
> > >                  * restricted-dma-pool region is allowed.
> > >                  */
> > > -               if (of_device_is_compatible(node, "restricted-dma-pool") &&
> > > -                   of_device_is_available(node)) {
> > > -                       of_node_put(node);
> > > -                       break;
> > > +               if (of_device_is_compatible(it.node, "restricted-dma-pool") &&
> > > +                   of_device_is_available(it.node)) {
> > > +                       if (!of_reserved_mem_device_init_by_idx(dev, of_node, i)) {
> > > +                               of_node_put(it.node);
> > > +                               return;
> > > +                       }
> > >                 }
> > > -               of_node_put(node);
> > > +               i++;
> > >         }
> > >
> > > -       /*
> > > -        * Attempt to initialize a restricted-dma-pool region if one was found.
> > > -        * Note that count can hold a negative error code.
> > > -        */
> > > -       if (i < count && of_reserved_mem_device_init_by_idx(dev, of_node, i))
> > > -               dev_warn(dev, "failed to initialise \"restricted-dma-pool\" memory node\n");
> > > +       dev_warn(dev, "failed to initialise \"restricted-dma-pool\" memory node\n");
> >
> > This changes the behavior. Before this patch, it was:
> >
> >     if a restricted dma pool was found, but initializing it failed, print
> >     a warning.
> >
> > Whereas now it has become:
> >
> >      print a warning unless a restricted dma pool was found and successfully
> >      initialized.
> >
> > This change causes the kernel to print out the warning for devices that
> > don't even do DMA:
>
> Thanks. I fixed it up to only warn if i is non-zero.

Not sure if that matches the old behavior though? A node could have
memory-regions for shared dma pools but not restricted dma pools,
and i would be non-zero.

IMO the warning should be in the "else" branch of

    if (!of_reserved_mem_device_init_by_idx(dev, of_node, i))


ChenYu