@matheusbsbs / matheus-bagliano / commits / b80741e5b9

stable-qsort: avoid using potentially unaligned access

Like in the previous patch for compat/qsort_s.c, remove the optimization of using an on-stack buffer to avoid small allocations. This ensures maximum alignment for the array elements and simplifies the code a bit. The performance impact for the current callers is unlikely to be noticeable: * compat/mingw.c::make_environment_block() uses ALLOC_ARRAY and ALLOC_GROW several times already, so another allocation of up to 1KB should not matter much. * diffcore-rename.c::diffcore_rename_extended() is called once per diff or twice per merge, and those require allocations for each object and more already. * merge-ort.c::detect_and_process_renames() is called once per merge. It's responsible for the two per-merge diffcore_rename_extended() calls mentioned above as well, though. So this is possibly the most impacted caller. Per-object allocations are likely to dwarf the additional small allocations in git_stable_qsort(), though. Signed-off-by: René Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>

René Scharfe committed Jan 8, 2022 at 00:30 UTC b80741e5b9c6b47a6f802ffdb0bba87357b35a9f

1 file changed +5 -11

stable-qsort.c

+5 -11

@@ -48,15 +48,9 @@ void git_stable_qsort(void *b, size_t n, size_t s,

48 int (*cmp)(const void *, const void *))

49 {

50 const size_t size = st_mult(n, s);

51 - char buf[1024];

52 -

53 - if (size < sizeof(buf)) {

54 - /* The temporary array fits on the small on-stack buffer. */

55 - msort_with_tmp(b, n, s, cmp, buf);

56 - } else {

57 - /* It's somewhat large, so malloc it. */

58 - char *tmp = xmalloc(size);

59 - msort_with_tmp(b, n, s, cmp, tmp);

60 - free(tmp);

61 - }

51 + char *tmp;

52 +

53 + tmp = xmalloc(size);

54 + msort_with_tmp(b, n, s, cmp, tmp);

55 + free(tmp);

56 }