vt: Deadlock workaround

2.6.26 corrected the mutex locking on tty resizing to fix the case where
you could get the tty/vt sizing out of sync. That turns out to have a
deadlock.

The actual fix is really major and I've got it lined up as part of the ops
changes for 2.6.28 so for 2.6.26/2.6.27 it is safer to reintroduce this
ages old minor bug.

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/drivers/char/vt.c b/drivers/char/vt.c
index 82a51f38a5469cd1a56637c6635b9c16f69f5f89..1bc00c9d860d70f917f2d945162aad3755b31749 100644 (file)
--- a/drivers/char/vt.c
+++ b/drivers/char/vt.c
@@ -916,7 +916,6 @@ int vc_resize(struct vc_data *vc, unsigned int cols, unsigned int lines)
                ws.ws_col = vc->vc_cols;
                ws.ws_ypixel = vc->vc_scan_lines;
 
-               mutex_lock(&vc->vc_tty->termios_mutex);
                spin_lock_irq(&vc->vc_tty->ctrl_lock);
                if ((ws.ws_row != cws->ws_row || ws.ws_col != cws->ws_col))
                        pgrp = get_pid(vc->vc_tty->pgrp);
@@ -926,7 +925,6 @@ int vc_resize(struct vc_data *vc, unsigned int cols, unsigned int lines)
                        put_pid(pgrp);
                }
                *cws = ws;
-               mutex_unlock(&vc->vc_tty->termios_mutex);
        }
 
        if (CON_IS_VISIBLE(vc))