[PATCH] Fortran: fix minor front-end memleaks
Dear all,
the attached patch fixes two minor front-end memleaks I saw when working
on recent PRs (pr117774 is one of them) and running f951 under valgrind.
Regtested on x86_64-pc-linux-gnu. OK for mainline?
Thanks,
Harald
From d5236446302a01eb11c017862a0813fafd0b95a1 Mon Sep 17 00:00:00 2001
From: Harald Anlauf
Date: Tue, 26 Nov 2024 20:37:35 +0100
Subject: [PATCH] Fortran: fix minor front-end memleaks
gcc/fortran/ChangeLog:
* expr.cc (find_inquiry_ref): Fix memleak introduced by scanning
the reference chain to find and simplify inquiry references.
* symbol.cc (gfc_copy_formal_args_intr): Free formal namespace
when not needed to avoid a front-end memleak.
---
gcc/fortran/expr.cc | 6 +-
gcc/fortran/symbol.cc | 2 ++
2 files changed, 7 insertions(+), 1 deletion(-)
diff --git a/gcc/fortran/expr.cc b/gcc/fortran/expr.cc
index 6b40e8e0aa5..a997bdae726 100644
--- a/gcc/fortran/expr.cc
+++ b/gcc/fortran/expr.cc
@@ -1833,6 +1833,7 @@ find_inquiry_ref (gfc_expr *p, gfc_expr **newp)
{
gfc_ref *ref;
gfc_ref *inquiry = NULL;
+ gfc_ref *inquiry_head;
gfc_expr *tmp;
tmp = gfc_copy_expr (p);
@@ -1858,6 +1859,7 @@ find_inquiry_ref (gfc_expr *p, gfc_expr **newp)
return false;
}
+ inquiry_head = inquiry;
gfc_resolve_expr (tmp);
/* Leave these to the backend since the type and kind is not confirmed until
@@ -1930,7 +1932,7 @@ find_inquiry_ref (gfc_expr *p, gfc_expr **newp)
mpc_imagref (tmp->value.complex), GFC_RND_MODE);
break;
}
- // TODO: Fix leaking expr tmp, when simplify is done twice.
+
if (inquiry->next)
gfc_replace_expr (tmp, *newp);
}
@@ -1944,10 +1946,12 @@ find_inquiry_ref (gfc_expr *p, gfc_expr **newp)
}
gfc_free_expr (tmp);
+ gfc_free_ref_list (inquiry_head);
return true;
cleanup:
gfc_free_expr (tmp);
+ gfc_free_ref_list (inquiry_head);
return false;
}
diff --git a/gcc/fortran/symbol.cc b/gcc/fortran/symbol.cc
index e803cdd93c9..f13cb1883ea 100644
--- a/gcc/fortran/symbol.cc
+++ b/gcc/fortran/symbol.cc
@@ -4910,6 +4910,8 @@ gfc_copy_formal_args_intr (gfc_symbol *dest, gfc_intrinsic_sym *src,
if (dest->formal != NULL)
/* The current ns should be that for the dest proc. */
dest->formal_ns = gfc_current_ns;
+ else
+gfc_free_namespace (gfc_current_ns);
/* Restore the current namespace to what it was on entry. */
gfc_current_ns = parent_ns;
}
--
2.35.3
Re: [PATCH] Fortran: fix minor front-end memleaks
Hi Harald, Looks good to me. Thanks Paul On Tue, 26 Nov 2024 at 19:51, Harald Anlauf wrote: > Dear all, > > the attached patch fixes two minor front-end memleaks I saw when working > on recent PRs (pr117774 is one of them) and running f951 under valgrind. > > Regtested on x86_64-pc-linux-gnu. OK for mainline? > > Thanks, > Harald > >
Re: [PATCH] Fortran: passing inquiry ref of complex array to assumed rank dummy [PR117774]
Am 25.11.24 um 23:33 schrieb Steve Kargl: On Mon, Nov 25, 2024 at 10:05:49PM +, Harald Anlauf wrote: Dear all, the attached patch fixes an ICE when passing an inquiry reference of a complex array to an assumed-rank dummy argument by terminating the search for the array reference before we hit the inquiry reference. (The array reference is needed for a possible descriptor update w.r.t. bounds.) Regtested on x86_64-pc-linux-gnu. OK for mainline? Looks good to me. Thanks for the review, Steve. Pushed as r15-5701-geff7e72815ada5 . What about backporting to 14-branch? If you have the time and energy, I see no reason not to fix it on 14. I've put it on my list for backports. Thanks for expanding the testcase to include array sections.
Re: [PATCH] Fortran: fix minor front-end memleaks
Am 26.11.24 um 21:09 schrieb Paul Richard Thomas: Hi Harald, Looks good to me. Thanks Paul Thanks, Paul. Pushed as r15-5701-geff7e72815ada5 . On Tue, 26 Nov 2024 at 19:51, Harald Anlauf wrote: Dear all, the attached patch fixes two minor front-end memleaks I saw when working on recent PRs (pr117774 is one of them) and running f951 under valgrind. Regtested on x86_64-pc-linux-gnu. OK for mainline? Thanks, Harald
Re: [PATCH] Fortran: fix minor front-end memleaks
Am 26.11.24 um 23:12 schrieb Harald Anlauf: Am 26.11.24 um 21:09 schrieb Paul Richard Thomas: Hi Harald, Looks good to me. Thanks Paul Thanks, Paul. Pushed as r15-5701-geff7e72815ada5 . This should have read r15-5708-g94f98f67f7271f, the testcase in the other one was used in the testing for the present one. Duh! On Tue, 26 Nov 2024 at 19:51, Harald Anlauf wrote: Dear all, the attached patch fixes two minor front-end memleaks I saw when working on recent PRs (pr117774 is one of them) and running f951 under valgrind. Regtested on x86_64-pc-linux-gnu. OK for mainline? Thanks, Harald
