public inbox for fortran@gcc.gnu.org
 help / color / mirror / Atom feed
From: Mikael Morin <mikael@gcc.gnu.org>
To: gcc-patches@gcc.gnu.org, fortran@gcc.gnu.org
Subject: [PATCH] fortran: Remove reference count update [PR108957]
Date: Fri, 15 Sep 2023 09:18:47 +0200	[thread overview]
Message-ID: <20230915071847.135585-1-mikael@gcc.gnu.org> (raw)

Hello,

Harald reminded me recently that there was a working patch attached to the PR.
I added a documentation comment with the hope that it may help avoid
making the same mistake in the future.
Regression tested on x86_64-pc-linux-gnu.
OK for master?

-- >8 --

Remove one reference count incrementation following the assignment of a
symbol pointer to a local variable.  Most symbol pointers are "weak" pointer
and don't need any reference count update when they are assigned, and it is
especially the case of local variables.

This fixes a memory leak with the testcase from the PR (not included).

	PR fortran/108957

gcc/fortran/ChangeLog:

	* gfortran.h (gfc_symbol): Add comment documenting reference counting.
	* parse.cc (parse_interface): Remove reference count incrementation.
---
 gcc/fortran/gfortran.h | 20 ++++++++++++++++++++
 gcc/fortran/parse.cc   |  3 ---
 2 files changed, 20 insertions(+), 3 deletions(-)

diff --git a/gcc/fortran/gfortran.h b/gcc/fortran/gfortran.h
index f4a1c106cea..6caf7765ac6 100644
--- a/gcc/fortran/gfortran.h
+++ b/gcc/fortran/gfortran.h
@@ -1944,7 +1944,27 @@ typedef struct gfc_symbol
      according to the Fortran standard.  */
   unsigned pass_as_value:1;
 
+  /* Reference counter, used for memory management.
+
+     Some symbols may be present in more than one namespace, for example
+     function and subroutine symbols are present both in the outer namespace and
+     the procedure body namespace.  Freeing symbols with the namespaces they are
+     in would result in double free for those symbols.  This field counts
+     references and is used to delay the memory release until the last reference
+     to the symbol is removed.
+
+     Not every symbol pointer is accounted for reference counting.  Fields
+     gfc_symtree::n::sym are, and gfc_finalizer::proc_sym as well.  But most of
+     them (dummy arguments, generic list elements, etc) are "weak" pointers;
+     the reference count isn't updated when they are assigned, and they are
+     ignored when the surrounding structure memory is released.  This is not a
+     problem because there is always a namespace as surrounding context and
+     symbols have a name they can be referred with in that context, so the
+     namespace keeps the symbol from being freed, keeping the pointer valid.
+     When the namespace ceases to exist, and the symbols with it, the other
+     structures referencing symbols cease to exist as well.  */
   int refs;
+
   struct gfc_namespace *ns;	/* namespace containing this symbol */
 
   tree backend_decl;
diff --git a/gcc/fortran/parse.cc b/gcc/fortran/parse.cc
index 8f09ddf753c..58386805ffe 100644
--- a/gcc/fortran/parse.cc
+++ b/gcc/fortran/parse.cc
@@ -4064,9 +4064,6 @@ loop:
   accept_statement (st);
   prog_unit = gfc_new_block;
   prog_unit->formal_ns = gfc_current_ns;
-  if (prog_unit == prog_unit->formal_ns->proc_name
-      && prog_unit->ns != prog_unit->formal_ns)
-    prog_unit->refs++;
 
 decl:
   /* Read data declaration statements.  */
-- 
2.40.1


             reply	other threads:[~2023-09-15  7:18 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-09-15  7:18 Mikael Morin [this message]
2023-09-15  9:16 ` Paul Richard Thomas

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=20230915071847.135585-1-mikael@gcc.gnu.org \
    --to=mikael@gcc.gnu.org \
    --cc=fortran@gcc.gnu.org \
    --cc=gcc-patches@gcc.gnu.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).