public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Pierre-Marie de Rodat <derodat@adacore.com>
To: gcc-patches@gcc.gnu.org
Cc: Olivier Hainque <hainque@adacore.com>
Subject: [Ada] Fix indirect calls to imported subprograms within generic
Date: Tue, 05 Dec 2017 12:23:00 -0000	[thread overview]
Message-ID: <20171205122352.GA15912@adacore.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 1437 bytes --]

Our trampoline avoidance scheme based on subprogram descriptors
should always be disconnected for subprograms with foreign convention.
For access to subprogram subtypes, the Set_Convention front-end
procedure is expected to take care of this, but it currently only
does so for subtypes not within a generic instance.

This causes wrong code to be generated on many targets for the
testcase below, where the indirect call from Trigger is performed
through a (non existent) descriptor, resulting in SEGV in most
circumstances.

/* cfoo.c */
void c_foo ()
{
 printf ("hello there from C");
}

-- blob.ads
generic
package Blob is
  procedure Initialize;
  procedure Trigger;
end;

-- blob.adb
package body Blob is

  type Service_Ptr is access procedure;
  pragma Convention (C, Service_Ptr);

  Hook : Service_Ptr;

  procedure C_Foo;
  pragma Import (C, C_Foo, "c_foo");

  procedure Initialize is
  begin
     Hook := C_Foo'Access;
  end;

  procedure Trigger is
  begin
     Hook.all;
  end;
end;

The change proposed here just arranges for the bit clearing
to happen systematically in Set_Convention, which fixes the case
at hand and just confirms the bit value in case the entity was
processed previously.

Tested on x86_64-pc-linux-gnu, committed on trunk

2017-12-05  Olivier Hainque  <hainque@adacore.com>

	* sem_util.adb (Set_Convention): Always clear Can_Use_Internal_Rep
	on access to subprogram types with foreign convention.


[-- Attachment #2: difs --]
[-- Type: text/plain, Size: 840 bytes --]

Index: sem_util.adb
===================================================================
--- sem_util.adb	(revision 255412)
+++ sem_util.adb	(working copy)
@@ -23090,17 +23090,7 @@
         and then Is_Access_Subprogram_Type (Base_Type (E))
         and then Has_Foreign_Convention (E)
       then
-
-         --  A pragma Convention in an instance may apply to the subtype
-         --  created for a formal, in which case we have already verified
-         --  that conventions of actual and formal match and there is nothing
-         --  to flag on the subtype.
-
-         if In_Instance then
-            null;
-         else
-            Set_Can_Use_Internal_Rep (E, False);
-         end if;
+         Set_Can_Use_Internal_Rep (E, False);
       end if;
 
       --  If E is an object, including a component, and the type of E is an

                 reply	other threads:[~2017-12-05 12:23 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

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=20171205122352.GA15912@adacore.com \
    --to=derodat@adacore.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=hainque@adacore.com \
    /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).