bitmap_parselist has been evolved from a pretty simple idea for long and now lacks for refactoring. It is not structured, has nested loops and a set of opaque-named variables. All this leads to extremely hard understanding of the code. Once during the optimization of it I missed a scenario which leads to kernel hangup. Tetsuo Handa spotted this and found it simpler to rewrite the code instead fixing it. (Though, that attempt had some flaws.) https://lkml.org/lkml/2018/4/1/93
Things are more complicated than they may be because bitmap_parselist() is part of user interface, and its behavior should not change. In this patchset - __bitmap_parselist() is reworked (patches 2 and 3); - time measurement in test_bitmap_parselist switched to ktime_get (patch 4); - new tests introduced (patch 5), and - bitmap_parselist_user() testing enabled with the same testset as bitmap_parselist() (patch 6). Patch 1 is a fix and may be applied separately. V1: https://lkml.org/lkml/2018/12/23/50 v2: - use PTR_ERR() and ERR_PTR() where appropriate; - fix parser logic (last byte of string handling); - tests for bitmap_parselist_user() in patch 5. Yury Norov (4): bitmap_parselist: don't calculate length of the input string bitmap_parselist: move non-parser logic to helpers bitmap_parselist: rework input string parser lib/test_bitmap: switch test_bitmap_parselist to ktime_get() lib/test_bitmap: add testcases for bitmap_parselist lib/test_bitmap: add tests for bitmap_parselist_user lib/bitmap.c | 293 ++++++++++++++++++++++++++++++---------------- lib/test_bitmap.c | 67 +++++++++-- 2 files changed, 245 insertions(+), 115 deletions(-) -- 2.17.1
