Re: [PATCH v3 4/4] fbdev: sh_mobile_lcdc: Make FB_DEVICE dependency optional

From
Helge Deller <deller@gmx.de>
Date
2026-01-03 09:59:52
Also in
dri-devel, linux-omap, lkml
Reply to
Chintan Patel, 2025-12-30
On 12/30/25 19:25, Chintan Patel wrote:

On 12/30/25 00:13, Helge Deller wrote:
quoted
* Chintan Patel [off-list ref]:
quoted
The sh_mobile_lcdc driver exposes overlay configuration via sysfs, but the
core driver does not require CONFIG_FB_DEVICE.

Make sysfs support optional by defining overlay_sysfs_groups conditionally
using PTR_IF(). The driver always sets .dev_groups, and the kernel
naturally skips NULL attribute groups while the code remains buildable
and type-checked.

Suggested-by: Helge Deller <deller@gmx.de>
Signed-off-by: Chintan Patel <redacted>
---
  drivers/video/fbdev/sh_mobile_lcdcfb.c | 12 +++++++++++-
  1 file changed, 11 insertions(+), 1 deletion(-)
diff --git a/drivers/video/fbdev/sh_mobile_lcdcfb.c b/drivers/video/fbdev/sh_mobile_lcdcfb.c
index dd950e4ab5ce..cb7ed1ff9165 100644
--- a/drivers/video/fbdev/sh_mobile_lcdcfb.c
+++ b/drivers/video/fbdev/sh_mobile_lcdcfb.c
@@ -1350,7 +1350,17 @@ static struct attribute *overlay_sysfs_attrs[] = {
      &dev_attr_overlay_rop3.attr,
      NULL,
  };
-ATTRIBUTE_GROUPS(overlay_sysfs);
Instead of replacing the ^ ATTRIBUTE_GROUPS() by the code below,
isn't it possible to just mark the overlay_sysfs_attrs[] array
_maybe_unused, and just do:
+ #ifdef CONFIG_FB_DEVICE
+ ATTRIBUTE_GROUPS(overlay_sysfs);
+ #endif

?
Hi Helge,

Yes, the __maybe_unused + #ifdef ATTRIBUTE_GROUPS() approach would work.

I went with the PTR_IF(IS_ENABLED()) pattern because Andy suggested
using PTR_IF() to conditionally include overlay_sysfs_group in
overlay_sysfs_groups, and to keep .dev_groups always populated while
letting the device core skip NULL groups. This avoids conditional
wiring via #ifdef and keeps the code type-checked without
CONFIG_FB_DEVICE.
If you still prefer the simpler #ifdef ATTRIBUTE_GROUPS() approach
for this driver, I can switch to that, but I wanted to follow Andy’s
guidance here.
I assume Andy will agree to my suggested approach, as it's cleaner
and avoids code bloat/duplication. Maybe you send out a v4 with my
suggested approach, then it's easier to judge... ?

Helge
Thread (20 messages) 20 messages, 6 authors, last reply 2026-01-06
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help