mn10300: get rid of set_fs(USER_DS) in sigframe setup
authorAl Viro <viro@ftp.linux.org.uk>
Sun, 26 Sep 2010 18:29:02 +0000 (19:29 +0100)
committerLinus Torvalds <torvalds@linux-foundation.org>
Mon, 27 Sep 2010 19:19:53 +0000 (12:19 -0700)
It really has no business being there; short of a serious kernel bug
we should already have USER_DS at that point.  It shouldn't have been
done on x86 either...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
arch/mn10300/kernel/signal.c

index 5a2c0040b166d2587ccc01f4a45a7c21ffd14056..ae36480b450f6ff554a35cd0298a558a00f42ba7 100644 (file)
@@ -333,8 +333,6 @@ static int setup_frame(int sig, struct k_sigaction *ka, sigset_t *set,
        regs->d0 = sig;
        regs->d1 = (unsigned long) &frame->sc;
 
-       set_fs(USER_DS);
-
        /* the tracer may want to single-step inside the handler */
        if (test_thread_flag(TIF_SINGLESTEP))
                ptrace_notify(SIGTRAP);
@@ -416,8 +414,6 @@ static int setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
        regs->d0 = sig;
        regs->d1 = (long) &frame->info;
 
-       set_fs(USER_DS);
-
        /* the tracer may want to single-step inside the handler */
        if (test_thread_flag(TIF_SINGLESTEP))
                ptrace_notify(SIGTRAP);