On Tue, Jul 02, 2013 at 04:31:34PM +0100, Dean Jenkins wrote: > Binding SLIP with a PTY/TTY can cause truncated SLIP frames to be > transmitted when the first write to the TTY fails to consume all > the characters of the SLIP frame. > > Asynchronous to the write function is a wakeup event from the TTY > that indicates the TTY can accept more data. The wakeup event > calls tty_wakeup() which calls slip_write_wakeup() when > TTY_DO_WRITE_WAKEUP is set. > > To complete the transmission of a SLIP frame to the TTY, > slip_write_wakeup() must run at least once. Unfortunately, pty_write() > also calls tty_wakeup() and when TTY_DO_WRITE_WAKEUP is set, > slip_write_wakeup() is called. xleft is always zero and > causes slip_write_wakeup() to complete the transmission and > clears the TTY_DO_WRITE_WAKEUP flag. This can cause a truncated > SLIP frame because any remaining characters will not get sent. > > The wakeup event is unable to process the remaining characters > because the TTY_DO_WRITE_WAKEUP flag has been cleared. > > The code modification fixes the transmission segmentation > mechanism by preventing pty_write() from calling > slip_write_wakeup() by clearing TTY_DO_WRITE_WAKEUP before calling > pty_write(). After pty_write() returns, TTY_DO_WRITE_WAKEUP is set > to allow the TTY wakeup event to call slip_write_wakeup() to > attempt to complete the transmission of the SLIP frame. > > This may not be foolproof because a timeout is needed to > break out of the cycle of transmission attempts. Note error > codes from the TTY layer will break out of the cycle of > transmission attempts. > > Signed-off-by: Dean Jenkins <dean_jenk...@mentor.com>
Acked-by: Greg Kroah-Hartman <gre...@linuxfoundation.org> -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
