public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Siddhesh Poyarekar <siddhesh@sourceware.org>
To: libc-alpha@sourceware.org
Subject: [PATCH v2] realloc: Return unchanged if request is within usable size
Date: Mon, 28 Nov 2022 12:26:46 -0500	[thread overview]
Message-ID: <20221128172646.244742-1-siddhesh@sourceware.org> (raw)
In-Reply-To: <20221125200948.2399996-1-siddhesh@sourceware.org>

If there is enough space in the chunk to satisfy the new size, return
the old pointer as is, thus avoiding any locks or reallocations.  The
only real place this has a benefit is in large chunks that tend to get
satisfied with mmap, since there is a large enough spare size (up to a
page) for it to matter.  For allocations on heap, the extra size is
typically barely a few bytes (up to 15) and it's unlikely that it would
make much difference in performance.

Also added a smoke test to ensure that the old pointer is returned
unchanged if the new size to realloc is within usable size of the old
pointer.

Signed-off-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
---
Changes from v1:
- Fixed up test as per review comments
- Do the realloc bypass only when the new size is within trim_threshold
  bytes of the usable size so that shrinking does not hold up memory.

 malloc/malloc.c      | 10 ++++++++++
 malloc/tst-realloc.c | 23 +++++++++++++++++++++++
 2 files changed, 33 insertions(+)

diff --git a/malloc/malloc.c b/malloc/malloc.c
index 2a61c8b5ee..ef8c794fb7 100644
--- a/malloc/malloc.c
+++ b/malloc/malloc.c
@@ -1100,6 +1100,8 @@ static void munmap_chunk(mchunkptr p);
 static mchunkptr mremap_chunk(mchunkptr p, size_t new_size);
 #endif
 
+static size_t musable (void *mem);
+
 /* ------------------ MMAP support ------------------  */
 
 
@@ -3396,6 +3398,14 @@ __libc_realloc (void *oldmem, size_t bytes)
   if (__glibc_unlikely (mtag_enabled))
     *(volatile char*) oldmem;
 
+  /* Return the chunk as is whenever possible, i.e. there's enough usable space
+     but not so much that we end up fragmenting the block.  We use the trim
+     threshold as the heuristic to decide the latter.  */
+  size_t usable = musable (oldmem);
+  if (bytes <= usable
+      && (unsigned long) (usable - bytes) <= mp_.trim_threshold)
+    return oldmem;
+
   /* chunk corresponding to oldmem */
   const mchunkptr oldp = mem2chunk (oldmem);
   /* its size */
diff --git a/malloc/tst-realloc.c b/malloc/tst-realloc.c
index 5eb62a770f..3b78a2420a 100644
--- a/malloc/tst-realloc.c
+++ b/malloc/tst-realloc.c
@@ -17,6 +17,7 @@
 
 #include <errno.h>
 #include <malloc.h>
+#include <stdint.h>
 #include <stdio.h>
 #include <string.h>
 #include <libc-diag.h>
@@ -142,6 +143,28 @@ do_test (void)
 
   free (p);
 
+  /* Smoke test to make sure that allocations do not move if they have enough
+     space to expand in the chunk.  */
+  for (size_t sz = 3; sz < 256 * 1024; sz += 2048)
+    {
+      p = realloc (NULL, sz);
+      if (p == NULL)
+	FAIL_EXIT1 ("realloc (NULL, %zu) returned NULL.", sz);
+      size_t newsz = malloc_usable_size (p);
+      printf ("size: %zu, usable size: %zu, extra: %zu\n",
+	      sz, newsz, newsz - sz);
+      uintptr_t oldp = (uintptr_t) p;
+      void *new_p = realloc (p, newsz);
+      if ((uintptr_t) new_p != oldp)
+	FAIL_EXIT1 ("Expanding (%zu bytes) to usable size (%zu) moved block",
+		    sz, newsz);
+      free (new_p);
+
+      /* We encountered a large enough extra size at least once.  */
+      if (newsz - sz > 1024)
+	break;
+    }
+
   return 0;
 }
 
-- 
2.38.1


  parent reply	other threads:[~2022-11-28 17:26 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-25 20:09 [PATCH] " Siddhesh Poyarekar
2022-11-25 20:40 ` Andreas Schwab
2022-11-28 14:23   ` Siddhesh Poyarekar
2022-11-28 14:29     ` Andreas Schwab
2022-11-28 17:26 ` Siddhesh Poyarekar [this message]
2022-12-06 22:33   ` [PATCH v2] " DJ Delorie

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20221128172646.244742-1-siddhesh@sourceware.org \
    --to=siddhesh@sourceware.org \
    --cc=libc-alpha@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).