Rethinking -i option

[hakavlad]

You may fix -i option instead of removing it.

Look at #234 (comment)

[hakavlad]

Other way:

• do not change -i option to allow it to work with old kernels;
• add the new (fixed) option (-I?) to work with new kernels (5.9+).

[hakavlad]

see also

	/*
	 * Special case OOM_SCORE_ADJ_MIN for all others scale the
	 * badness value into [0, 2000] range which we have been
	 * exporting for a long time so userspace might depend on it.
	 */
	if (badness != LONG_MIN)
		points = (1000 + badness * 1000 / (long)totalpages) * 2 / 3;

• https://github.com/torvalds/linux/blob/v5.9/fs/proc/base.c#L558
• https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9066e5cfb73cdbcdbb49e87999482ab615e9fc76

[hakavlad]

compare (5.8 vs 5.9)

	points = oom_badness(task, totalpages) * 1000 / totalpages;

https://github.com/torvalds/linux/blob/v5.8/fs/proc/base.c#L555

	if (badness != LONG_MIN)
		points = (1000 + badness * 1000 / (long)totalpages) * 2 / 3;

https://github.com/torvalds/linux/blob/v5.9/fs/proc/base.c#L562

[rfjakob]

I think to keep this option, it should work well with both new & old kernels ( as explained in 7ebc455 ). This means you need kernel version detection. And I'm not sure anybody but me even uses the -i flag.