[PATCH] core: devres: don't use devres code when CONFIG_DEVRES is not defined

[PATCH] core: devres: don't use devres code when CONFIG_DEVRES is not defined

[Angus Ainslie]

Put guards around the devres code so that it isn't compiled during the SPL.

Signed-off-by: Angus Ainslie <angus@akkea.ca>
---
 drivers/core/devres.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/core/devres.c b/drivers/core/devres.c
index 313ddc7089..92a237c64c 100644
--- a/drivers/core/devres.c
+++ b/drivers/core/devres.c
@@ -91,6 +91,7 @@ void *_devres_alloc(dr_release_t release, size_t size, gfp_t gfp)
 	return dr->data;
 }
 
+#if CONFIG_DEVRES
 void devres_free(void *res)
 {
 	if (res) {
@@ -257,7 +258,7 @@ void devres_get_stats(const struct udevice *dev, struct devres_stats *stats)
 	}
 }
 
-#endif
+#endif /* CONFIG_DEBUG_DEVRES */
 
 /*
  * Managed kmalloc/kfree
@@ -292,3 +293,4 @@ void devm_kfree(struct udevice *dev, void *p)
 	rc = devres_destroy(dev, devm_kmalloc_release, devm_kmalloc_match, p);
 	assert_noisy(!rc);
 }
+#endif /* CONFIG_DEVRES */
-- 
2.25.1

Re: [PATCH] core: devres: don't use devres code when CONFIG_DEVRES is not defined

[Simon Glass]

On Wed, 2 Feb 2022 at 16:16, Angus Ainslie <angus@akkea.ca> wrote:
>
> Put guards around the devres code so that it isn't compiled during the SPL.
>
> Signed-off-by: Angus Ainslie <angus@akkea.ca>
> ---
>  drivers/core/devres.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)

I think the correct fix here is to change the Makefile rule for this file to use

CONFIG_$(SPL_)DEVRES

instead of

CONFIG_DEVRES

Can you try it?

Regards,
Simon
[..]

Re: [PATCH] core: devres: don't use devres code when CONFIG_DEVRES is not defined

[Angus Ainslie]

Hi Simon,

On 2022-02-19 14:12, Simon Glass wrote:
> On Wed, 2 Feb 2022 at 16:16, Angus Ainslie <angus@akkea.ca> wrote:
>> 
>> Put guards around the devres code so that it isn't compiled during the 
>> SPL.
>> 
>> Signed-off-by: Angus Ainslie <angus@akkea.ca>
>> ---
>>  drivers/core/devres.c | 4 +++-
>>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> I think the correct fix here is to change the Makefile rule for this 
> file to use
> 
> CONFIG_$(SPL_)DEVRES
> 
> instead of
> 
> CONFIG_DEVRES
> 
> Can you try it?
> 

I think it will work that way and I did think about doing the patch that 
way. My concern was that it might break some boards because they are 
expecting DEVRES in the SPL.

I can rework the patch that way but not sure how I'd test which boards 
needed the CONFIG_SPL_DEVRES defined.

Thanks
Angus

> Regards,
> Simon
> [..]

Re: [PATCH] core: devres: don't use devres code when CONFIG_DEVRES is not defined

[Simon Glass]

Hi Angus,

On Mon, 21 Feb 2022 at 05:50, Angus Ainslie <angus@akkea.ca> wrote:
>
> Hi Simon,
>
> On 2022-02-19 14:12, Simon Glass wrote:
> > On Wed, 2 Feb 2022 at 16:16, Angus Ainslie <angus@akkea.ca> wrote:
> >>
> >> Put guards around the devres code so that it isn't compiled during the
> >> SPL.
> >>
> >> Signed-off-by: Angus Ainslie <angus@akkea.ca>
> >> ---
> >>  drivers/core/devres.c | 4 +++-
> >>  1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > I think the correct fix here is to change the Makefile rule for this
> > file to use
> >
> > CONFIG_$(SPL_)DEVRES
> >
> > instead of
> >
> > CONFIG_DEVRES
> >
> > Can you try it?
> >
>
> I think it will work that way and I did think about doing the patch that
> way. My concern was that it might break some boards because they are
> expecting DEVRES in the SPL.
>
> I can rework the patch that way but not sure how I'd test which boards
> needed the CONFIG_SPL_DEVRES defined.

I very much doubt any are, but the maintainers can test them as
needed. After all, freeing memory is not normally supported in SPL,
except with full malloc().

Regards,
Simon