Linux kernel 6.12 SLOC

22nd November 2024

I don't believe SLOC (Source Lines of Code) is a particularly great indicator of complexity or quality (Please, please never use SLOC as a KPI!). But it's a useful "finger in the air" way to get a sense of the size of a project and its composition (which modules and languages contribute the most code).

I ran David Wheeler's sloccount against Linux kernel 6.12 and got the following results.

Interesting takeaways

I'm not a kernel developer (in fact, it's been years since I had to compile my own kernel), so for those familiar with the codebase none of this will be shocking, but I found the following stats quite interesting:

  • There are around 26m SLOC.
  • Around 70% of the code comes from drivers.
  • 98% of the code base is C.
  • There's way more Python (61k SLOC) than I expected.
  • There's about as much shell code as I would expect (133k SLOC).

Raw output

SLOC    Directory       SLOC-by-Language (Sorted)
18548831 drivers         ansic=18539246,asm=4567,yacc=1679,python=1485,perl=792,lex=771,sh=291
1777035 arch            ansic=1503474,asm=259266,perl=12102,sh=1426,awk=762,sed=5
1205419 sound           ansic=1205419
1166669 fs              ansic=1166669
998041  tools           ansic=810638,sh=122122,python=51806,perl=4881,asm=4672,yacc=1707,cpp=1141,lex=991,awk=58,ruby=25
942831  net             ansic=942831
795503  include         ansic=793056,cpp=2447
319187  kernel          ansic=319130,sh=57
180838  lib             ansic=180662,perl=123,sh=34,awk=13,asm=6
126769  mm              ansic=126769
89850   crypto          ansic=89850
77226   security        ansic=77226
68266   scripts         ansic=32842,perl=18123,sh=5859,python=5757,cpp=2305,yacc=1537,lex=1177,awk=606,asm=37,sed=23
43111   block           ansic=43111
28013   samples         ansic=25705,sh=2308
15834   io_uring        ansic=15834
7343    virt            ansic=7343
6875    ipc             ansic=6875
4925    Documentation   python=2046,perl=1571,sh=943,ansic=237,awk=128
3082    init            ansic=3082
948     usr             ansic=609,sh=196,perl=131,asm=12
754     certs           ansic=674,asm=55,awk=25
221     rust            ansic=221
0       LICENSES        (none)
0       top_dir         (none)

Totals grouped by language (dominant language first):
ansic:     25891503 (98.05%)
asm:         268615 (1.02%)
sh:          133236 (0.50%)
python:       61094 (0.23%)
perl:         37723 (0.14%)
cpp:           5893 (0.02%)
yacc:          4923 (0.02%)
lex:           2939 (0.01%)
awk:           1592 (0.01%)
sed:             28 (0.00%)
ruby:            25 (0.00%)



Total Physical Source Lines of Code (SLOC)                = 26,407,571
Development Effort Estimate, Person-Years (Person-Months) = 8,787.09 (105,445.03)
 (Basic COCOMO model, Person-Months = 2.4 * (KSLOC**1.05))
Schedule Estimate, Years (Months)                         = 16.89 (202.62)
 (Basic COCOMO model, Months = 2.5 * (person-months**0.38))
Estimated Average Number of Developers (Effort/Schedule)  = 520.40
Total Estimated Cost to Develop                           = $ 1,187,015,808
 (average salary = $56,286/year, overhead = 2.40).