rpms/mlton/devel mlton-20070826-no-execmem.patch,NONE,1.1

[Adam Goode (agoode)]

Author: agoode

Update of /cvs/extras/rpms/mlton/devel
In directory cvs-int.fedora.redhat.com:/tmp/cvs-serv2429

Added Files:
	mlton-20070826-no-execmem.patch 
Log Message:
* Sat Feb  9 2008 Adam Goode <adam at spicenitz.org> - 20070826-12
- Introduce patch to not call mprotect with PROT_EXEC


mlton-20070826-no-execmem.patch:

--- NEW FILE mlton-20070826-no-execmem.patch ---
>From 544930de3b1c754fa8803169902a63bce7cc02ba Mon Sep 17 00:00:00 2001
From: Adam Goode <adam at spicenitz.org>
Date: Wed, 6 Feb 2008 20:17:51 -0500
Subject: [PATCH] Remove PROT_EXEC from mprotect

It looks like mprotect is used here as part of signal handling.
There doesn't seems to be a reason to have the area of memory
marked as executable. In fact, on Fedora 9, this causes MLton
compiled binaries (including MLton itself) to fail.
---
 runtime/platform/mmap-protect.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/runtime/platform/mmap-protect.c b/runtime/platform/mmap-protect.c
index f0dea49..df42215 100644
--- a/runtime/platform/mmap-protect.c
+++ b/runtime/platform/mmap-protect.c
@@ -7,7 +7,7 @@ void *GC_mmapAnon_safe_protect (void *start, size_t length,
         if (mprotect (low, dead_low, PROT_NONE))
                 diee ("mprotect failed");
         result = (void*)((pointer)low + dead_low);
-        if (mprotect (result, length, PROT_READ | PROT_WRITE | PROT_EXEC))
+        if (mprotect (result, length, PROT_READ | PROT_WRITE))
                 diee ("mprotect failed");
         high = (void*)((pointer)result + length);
         if (mprotect (high, dead_high, PROT_NONE))
-- 
1.5.4