linuxkpi: Handle bin attributes in sysfs attribute groups

[dumbbell]

Summary

For instance, this is used by DRM drivers to declare the EDID property of an GPU output connector:

sysctl -b sys.device.drmn1.card0.card0-DP-1.edid | edid-decode

...
Block 0, Base EDID:
  EDID Structure Version & Revision: 1.4
  Vendor & Product Identification:
    Manufacturer: SAM
    Model: 29814
    Serial Number: 810635354 (0x3051505a)
    Made in: week 15 of 2025
...

Sponsored by: The FreeBSD Foundation

[jrm]

Thanks for adding this. I updated my laptop over the weekend, and drm-kmod failed to build with

/home/jrm/scm/freebsd/drm-kmod/drivers/gpu/drm/drm_sysfs.c:300:3: error: field designator 'bin_attrs' does not refer to any field in type 'const struct attribute_group'
  300 |         .bin_attrs = connector_bin_attrs,
      |         ~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 error generated.

This patch fixes the build error.

[jrm]

I foolishly tried to come up with a patch to fix the build before checking here first. There were a few differences, but I don't really know what I'm doing here.

1. To match the layout of Linux struct attribute_group, I also added is_bin_visible. It's clearly not required to fix the build, but is that helpful for anything that may initialize is_bin_visible in the future?
2. Do we need anything in sysfs_remove_group()?

[emaste]

I think this is a minimal, sufficient change. is_bin_visible could be added later if it becomes necessary. I don't think sysfs_remove_group needs anything, in that it doesn't do anything with grp->attrs now, but probably sysfs_unmerge_group does.

[emaste]

sys/compat/linuxkpi/common/include/linux/sysfs.h

449

We're iterating over grp->attrs here, so grp->bin_attrs seems missing.

[bz]

@dumbbell any updates? @emaste @jrm can we fix this ourselves and commit it if needed? A month of broken master seems unhelpful.

Fixes: https://github.com/freebsd/drm-kmod/issues/416

[wulf]

sys/compat/linuxkpi/common/include/linux/sysfs.h

449

And attr != NULL check added previously seems missing too

[emaste]

I've reverted d99f16276e06736ceff9ea3ae5a1ac09135bcfdc for now. Let's get this change fixed up and in to src and we can then reapply it.

[bz]

In D55176#1274177, @emaste wrote:

I've reverted d99f16276e06736ceff9ea3ae5a1ac09135bcfdc for now. Let's get this change fixed up and in to src and we can then reapply it.

Thanks