package/trinity/0004-get_syscall_entry-remove-SYSCALL_OFFSET.patch

From 1d9af9d07ae111c253c92112fb50000adac47a0c Mon Sep 17 00:00:00 2001
From: Marcin Nowakowski <marcin.nowakowski@imgtec.com>
Date: Thu, 29 Jun 2017 12:24:10 +0200
Subject: [PATCH] get_syscall_entry: remove SYSCALL_OFFSET

0-based syscall numbers are used throughout the application and the
SYSCALL_OFFSET is only added when invoking the syscall. It is therefore
wrong to substract the offset in get_syscall_entry, as it leads to
dereferencing array entries at negative offsets and segfaults.

Tested on MIPS32/LE.
This change effectively reverts cc5108ff ('make get_syscall_entry take
SYSCALL_OFFSET into account')

Signed-off-by: Marcin Nowakowski <marcin.nowakowski@imgtec.com>
[Upstream commit: https://github.com/kernelslacker/trinity/commit/1d9af9d07ae111c253c92112fb50000adac47a0c]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
---
 tables.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/tables.c b/tables.c
index e7dab85d..8090146a 100644
--- a/tables.c
+++ b/tables.c
@@ -673,15 +673,14 @@ int munge_tables(void)
  */
 struct syscallentry * get_syscall_entry(unsigned int callno, bool do32)
 {
-	unsigned int offset = callno - SYSCALL_OFFSET;
 	if (biarch == FALSE)
-		return syscalls[offset].entry;
+		return syscalls[callno].entry;
 
 	/* biarch case */
 	if (do32 == TRUE)
-		return syscalls_32bit[offset].entry;
+		return syscalls_32bit[callno].entry;
 	else
-		return syscalls_64bit[offset].entry;
+		return syscalls_64bit[callno].entry;
 }
 
 /*