Re: Status of kernel-janitors?

On Mon, 30 Mar 2026, Linus Probert wrote:

> Hello,
>
> I've been a long time builder and fiddler of kernel source. Running
> self-compiled kernels and pulling in patches before official release
> etc. Bog standard for most of you I assume.
>
> I'd love to dedicate some spare to help the project. Like so many
> others.
>
> The kernel-janoitors "project" does pop up in various resources such as
> kernelnewbies and I think it might have been mentioned in one of the
> Linux Foundations free courses. But links to potential homepages seem to
> be dead and resources are sometimes dated.
>
> I read somewhere that this has mostly evolved into janitorial work in
> drivers/staging. TODOs were mentioned.
>
> So the actual question. In my mind janitorial code work is a good place
> to dip ones toes when starting out. Is this the right mailing list to
> ask about such things? Any suggestions to find an area where ones
> assistance would be welcome?
>
> I'm aware of checkpatch.pl but for many drivers diverging from these
> rules seem intentional and just blindly fixing checkpatch warnings are
> not helpful to the maintainer.
> Open source is a lot of work so the goal would be to be helpful while
> not stealing too much attention from concerned parties.

  I mention this once in a while ... *many* years ago, I wrote some
admittedly hacky scripts that scanned the kernel source tree looking
for what I considered "obvious" cleanups. One of those cleanups was to
abbreviate the numerous calculations of the length of an array
sprinkled throughout the source code, so I wrote a script that can be
run from the top of the kernel source tree and can take an argument of
which subdirectory to examine, looking for a particular regular
expression that I can barely recognize anymore:

  #!/bin/sh
  DIR=${1-.}
  grep -Er "sizeof ?\(?([^\)]+)\)? ?/ ?sizeof ?\(?.*\1.*" ${DIR}

For example, if I run:

  $ arraysize.sh drivers/gpu/drm

I get the output:

drivers/gpu/drm/xe/xe_guc_hxg_helpers.h:#define hxg_sizeof(T)	(sizeof(T) / sizeof(u32) + BUILD_BUG_ON_ZERO(sizeof(T) % sizeof(u32)))
drivers/gpu/drm/nouveau/nvif/fifo.c:	a->m.count = sizeof(a->v) / sizeof(a->v.runlists);
drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.c:#define NUM_ELEMENTS(a) (sizeof(a) / sizeof((a)[0]))
drivers/gpu/drm/amd/display/dc/mpc/dcn20/dcn20_mpc.c:#define NUM_ELEMENTS(a) (sizeof(a) / sizeof((a)[0]))
drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp_cm.c:#define NUM_ELEMENTS(a) (sizeof(a) / sizeof((a)[0]))
drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp_cm.c:	int arr_size = sizeof(dpp_input_csc_matrix)/sizeof(struct dpp_input_csc_matrix);
drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c:	int arr_size = sizeof(dpp_input_csc_matrix)/sizeof(struct dpp_input_csc_matrix);
drivers/gpu/drm/amd/display/dc/dpp/dcn20/dcn20_dpp_cm.c:	int arr_size = sizeof(dpp_input_csc_matrix)/sizeof(struct dpp_input_csc_matrix);
drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c:#define NUM_ELEMENTS(a) (sizeof(a) / sizeof((a)[0]))
drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c:	int arr_size = sizeof(dpp_input_csc_matrix) / sizeof(struct dpp_input_csc_matrix);
drivers/gpu/drm/amd/display/dc/dce110/dce110_opp_csc_v.c:	int arr_size = sizeof(input_csc_matrix)/sizeof(struct input_csc_matrix);
drivers/gpu/drm/amd/display/dc/dml/dsc/rc_calc_fpu.c:	table_size = sizeof(qp_table_##mode##_##bpc##bpc_##max)/sizeof(*qp_table_##mode##_##bpc##bpc_##max); \
drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c:	for (k = 0; k < sizeof(wb_arb_params->cli_watermark)/sizeof(wb_arb_params->cli_watermark[0]); k++) {
drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c:#define NUM_ELEMENTS(a) (sizeof(a) / sizeof((a)[0]))
drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.c:#define NUM_ELEMENTS(a) (sizeof(a) / sizeof((a)[0]))

Note the number of places in that subsystem which calculate the length
of an array; much of that can be abbreviated by now taking advantage
of the kernel header file include/linux/array_size.h (only relevant
line shown here):

#define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) +
__must_be_array(arr))

That would seem to be obvious janitor work -- simplifying code a
subsystem at a time (do *not* try to submit a single patch that
covers the entire source tree).

  I have other examples, I should clean them up and post them.

rday