On 5/9/20 6:13 PM, Paul Eggert wrote:
> perhaps it'd be better to disable -Wreturn-local-addr instead, at least for
> this
> compilation unit.
I tried doing that, but with GCC 10.1.0 one cannot disable the warning with a
#pragma! So the GCC bug is pretty bad.
I installed the attached to try to limit the damage to programs compiled with
GCC 10.1.0 (I hope they fix it before 10.2.0 comes out).
>From 45e72160bebba64532ffc8dddf34a0882bcaa110 Mon Sep 17 00:00:00 2001
From: Paul Eggert <egg...@cs.ucla.edu>
Date: Sun, 10 May 2020 17:07:47 -0700
Subject: [PATCH] careadlinkat: limit GCC workaround
* lib/careadlinkat.c (careadlinkat): Limit workaround to GCC
10.1.0 and later, since the workaround is pretty bad and the GCC
bug should get fixed.
---
ChangeLog | 7 +++++++
lib/careadlinkat.c | 9 +++++++--
2 files changed, 14 insertions(+), 2 deletions(-)
diff --git a/ChangeLog b/ChangeLog
index 6ef88249c..7468718c3 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,10 @@
+2020-05-10 Paul Eggert <egg...@cs.ucla.edu>
+
+ careadlinkat: limit GCC workaround
+ * lib/careadlinkat.c (careadlinkat): Limit workaround to GCC
+ 10.1.0 and later, since the workaround is pretty bad and the GCC
+ bug should get fixed.
+
2020-05-10 Bruno Haible <br...@clisp.org>
havelib: Enhance documentation.
diff --git a/lib/careadlinkat.c b/lib/careadlinkat.c
index e1f8305e9..fbed634f9 100644
--- a/lib/careadlinkat.c
+++ b/lib/careadlinkat.c
@@ -71,9 +71,14 @@ careadlinkat (int fd, char const *filename,
size_t buf_size_max =
SSIZE_MAX < SIZE_MAX ? (size_t) SSIZE_MAX + 1 : SIZE_MAX;
-#if defined GCC_LINT || defined lint
+#if (defined GCC_LINT || defined lint) && _GL_GNUC_PREREQ (10, 1)
/* Pacify preadlinkat without creating a pointer to the stack
- that gcc -Wreturn-local-addr would cry wolf about. */
+ that a broken gcc -Wreturn-local-addr would cry wolf about. See:
+ https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95044
+ This workaround differs substantially from the mainline code, but
+ no other way to pacify GCC 10.1.0 is known; even an explicit
+ #pragma does not pacify GCC. When the GCC bug is fixed this
+ workaround should be limited to the broken GCC versions. */
static char initial_buf[1];
enum { initial_buf_size = 0 }; /* 0 so that initial_buf never changes. */
#else
--
2.17.1
