public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
To: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Cc: 'GNU C Library' <libc-alpha@sourceware.org>,
	Cupertino Miranda <cupertino.miranda@oracle.com>
Subject: [PATCH v5 1/1] Created tunable to force small pages on stack allocation.
Date: Thu, 13 Apr 2023 15:43:25 +0000	[thread overview]
Message-ID: <PAWPR08MB89827ECD77A70FB1D3B81B6983989@PAWPR08MB8982.eurprd08.prod.outlook.com> (raw)

Hi Adhemerval,

I agree doing this automatically sounds like a better solution.
However:

+static __always_inline int
+advise_thp (void *mem, size_t size, size_t guardsize)
+{
+  enum malloc_thp_mode_t thpmode = __malloc_thp_mode ();
+  if (thpmode != malloc_thp_mode_always)
+    return 0;
+
+  unsigned long int thpsize = __malloc_default_thp_pagesize ();
+  if ((uintptr_t) mem % thpsize != 0
+      || size % thpsize != 0
+      || (size - guardsize) % thpsize != 0)
+    return 0;

Isn't the last part always true currently given the guard page size is based on
the standard page size? IIRC the issue was the mmap succeeds but the guard
page is taken from the original mmap which then causes the decomposition.

So you'd need something like:

|| guardsize % thpsize == 0)

Ie. we return without the madvise if the size and alignment is wrong for a huge
page or it is correct and the guardsize is a multiple of a huge page (in which
case it shouldn't decompose). 

+  return __madvise (mem, size, MADV_NOHUGEPAGE);
+}

Cheers,
Wilco

             reply	other threads:[~2023-04-13 15:43 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-04-13 15:43 Wilco Dijkstra [this message]
2023-04-13 16:23 ` Cupertino Miranda
2023-04-13 17:48   ` Adhemerval Zanella Netto
2023-04-14 11:28     ` Cupertino Miranda
2023-04-14 13:24       ` Wilco Dijkstra
2023-04-14 14:49         ` Cupertino Miranda
2023-04-14 15:32           ` Wilco Dijkstra
2023-04-14 16:03             ` Wilco Dijkstra
2023-04-14 16:35               ` Cupertino Miranda
2023-04-14 23:10                 ` Wilco Dijkstra
2023-04-14 16:27             ` Cupertino Miranda
  -- strict thread matches above, loose matches on Subject: below --
2023-03-28 15:22 [PATCH v5 0/1] *** " Cupertino Miranda
2023-03-28 15:22 ` [PATCH v5 1/1] " Cupertino Miranda
2023-04-11 19:56   ` Adhemerval Zanella Netto
2023-04-12  8:53     ` Cupertino Miranda
2023-04-12 14:10       ` Adhemerval Zanella Netto
2023-04-13 16:13         ` Cupertino Miranda
2023-04-14 11:41       ` Adhemerval Zanella Netto
2023-04-14 12:27         ` Cupertino Miranda
2023-04-14 13:06           ` Adhemerval Zanella Netto
2023-04-14 14:33             ` Cupertino Miranda

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=PAWPR08MB89827ECD77A70FB1D3B81B6983989@PAWPR08MB8982.eurprd08.prod.outlook.com \
    --to=wilco.dijkstra@arm.com \
    --cc=adhemerval.zanella@linaro.org \
    --cc=cupertino.miranda@oracle.com \
    --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).