about summary refs log tree commit diff
path: root/manual
diff options
context:
space:
mode:
authorAdhemerval Zanella <adhemerval.zanella@linaro.org>2023-10-03 09:22:50 -0300
committerAdhemerval Zanella <adhemerval.zanella@linaro.org>2023-10-31 14:18:05 -0300
commit03bf8357e8291857a435afcc3048e0b697b6cc04 (patch)
treed4f97004d712dd066458d15f9be599fc964d5d7b /manual
parent274a46c9b25ab733a1fb9fb1497f1beecae30193 (diff)
downloadglibc-03bf8357e8291857a435afcc3048e0b697b6cc04.tar.gz
glibc-03bf8357e8291857a435afcc3048e0b697b6cc04.tar.xz
glibc-03bf8357e8291857a435afcc3048e0b697b6cc04.zip
stdlib: Remove use of mergesort on qsort (BZ 21719)
This patch removes the mergesort optimization on qsort implementation
and uses the introsort instead.  The mergesort implementation has some
issues:

  - It is as-safe only for certain types sizes (if total size is less
    than 1 KB with large element sizes also forcing memory allocation)
    which contradicts the function documentation.  Although not required
    by the C standard, it is preferable and doable to have an O(1) space
    implementation.

  - The malloc for certain element size and element number adds
    arbitrary latency (might even be worse if malloc is interposed).

  - To avoid trigger swap from memory allocation the implementation
    relies on system information that might be virtualized (for instance
    VMs with overcommit memory) which might lead to potentially use of
    swap even if system advertise more memory than actually has.  The
    check also have the downside of issuing syscalls where none is
    expected (although only once per execution).

  - The mergesort is suboptimal on an already sorted array (BZ#21719).

The introsort implementation is already optimized to use constant extra
space (due to the limit of total number of elements from maximum VM
size) and thus can be used to avoid the malloc usage issues.

Resulting performance is slower due the usage of qsort, specially in the
worst-case scenario (partialy or sorted arrays) and due the fact
mergesort uses a slight improved swap operations.

This change also renders the BZ#21719 fix unrequired (since it is meant
to fix the sorted input performance degradation for mergesort).  The
manual is also updated to indicate the function is now async-cancel
safe.

Checked on x86_64-linux-gnu.
Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
Diffstat (limited to 'manual')
-rw-r--r--manual/argp.texi2
-rw-r--r--manual/locale.texi3
-rw-r--r--manual/search.texi7
3 files changed, 5 insertions, 7 deletions
diff --git a/manual/argp.texi b/manual/argp.texi
index 0023441812..b77ad68285 100644
--- a/manual/argp.texi
+++ b/manual/argp.texi
@@ -735,7 +735,7 @@ for options, bad phase of the moon, etc.
 @c  hol_set_group ok
 @c   hol_find_entry ok
 @c  hol_sort @mtslocale @acucorrupt
-@c   qsort dup @acucorrupt
+@c   qsort dup
 @c    hol_entry_qcmp @mtslocale
 @c     hol_entry_cmp @mtslocale
 @c      group_cmp ok
diff --git a/manual/locale.texi b/manual/locale.texi
index 720e0ca952..f6afa5dc44 100644
--- a/manual/locale.texi
+++ b/manual/locale.texi
@@ -253,7 +253,7 @@ The symbols in this section are defined in the header file @file{locale.h}.
 @c    calculate_head_size ok
 @c    __munmap ok
 @c    compute_hashval ok
-@c    qsort dup @acucorrupt
+@c    qsort dup
 @c     rangecmp ok
 @c    malloc @ascuheap @acsmem
 @c    strdup @ascuheap @acsmem
@@ -275,7 +275,6 @@ The symbols in this section are defined in the header file @file{locale.h}.
 @c      realloc @ascuheap @acsmem
 @c     realloc @ascuheap @acsmem
 @c     fclose @ascuheap @asulock @acsmem @acsfd @aculock
-@c     qsort @ascuheap @acsmem
 @c      alias_compare dup
 @c    libc_lock_unlock @aculock
 @c   _nl_explode_name @ascuheap @acsmem
diff --git a/manual/search.texi b/manual/search.texi
index 5691bf2f2b..a550858478 100644
--- a/manual/search.texi
+++ b/manual/search.texi
@@ -159,7 +159,7 @@ To sort an array using an arbitrary comparison function, use the
 
 @deftypefun void qsort (void *@var{array}, size_t @var{count}, size_t @var{size}, comparison_fn_t @var{compare})
 @standards{ISO, stdlib.h}
-@safety{@prelim{}@mtsafe{}@assafe{}@acunsafe{@acucorrupt{}}}
+@safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}}
 The @code{qsort} function sorts the array @var{array}.  The array
 contains @var{count} elements, each of which is of size @var{size}.
 
@@ -199,9 +199,8 @@ Functions}):
 The @code{qsort} function derives its name from the fact that it was
 originally implemented using the ``quick sort'' algorithm.
 
-The implementation of @code{qsort} in this library might not be an
-in-place sort and might thereby use an extra amount of memory to store
-the array.
+The implementation of @code{qsort} in this library is an in-place sort
+and uses a constant extra space (allocated on the stack).
 @end deftypefun
 
 @node Search/Sort Example