| Message ID | E1TKTlL-0002ni-MB@rmk-PC.arm.linux.org.uk (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Sat, 06 Oct 2012 13:45:47 +0100 Russell King <rmk+kernel@arm.linux.org.uk> wrote: > This allows drivers (such as OMAP serial) to allocate and free their > transmit buffers in a sane manner, rather than working around the > buffer allocation provided by serial_core. I think this just illustrates how broken the serial_core layering is in that drivers can't treat it as a library but get constrained by it. Fine for now and actually probably a useful hook to begin removing the circ buffers for the kernel generic kfifo and other work that needs doing eventually. Alan -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Sat, Oct 06, 2012 at 04:49:56PM +0100, Alan Cox wrote: > On Sat, 06 Oct 2012 13:45:47 +0100 > Russell King <rmk+kernel@arm.linux.org.uk> wrote: > > > This allows drivers (such as OMAP serial) to allocate and free their > > transmit buffers in a sane manner, rather than working around the > > buffer allocation provided by serial_core. > > I think this just illustrates how broken the serial_core layering is in > that drivers can't treat it as a library but get constrained by it. > > Fine for now and actually probably a useful hook to begin removing the > circ buffers for the kernel generic kfifo and other work that needs doing > eventually. Yes, I was thinking about turning the 'default' buffer allocation into a library call which drivers would set, but I felt that's just asking for a big patch and problems with new drivers appearing. We could transition it by providing a default buffer allocation call, converting all the drivers, and then making that call mandatory. That probably would then give us a better path to convert to kfifo. The view that serial_core should be a library is one that Ted mentioned when I created serial_core originally - I disagreed at the time, and I still believe it would've ended up being much more complex - and hairy - to go that route at that time. Maybe as things have moved forward in the tty layer that's changed, but I've been out of tty stuff for too long to properly comment. -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/tty/serial/serial_core.c b/drivers/tty/serial/serial_core.c index 4dd609e..53274ae 100644 --- a/drivers/tty/serial/serial_core.c +++ b/drivers/tty/serial/serial_core.c @@ -147,12 +147,18 @@ static int uart_port_startup(struct tty_struct *tty, struct uart_state *state, * buffer. */ if (!state->xmit.buf) { - /* This is protected by the per port mutex */ - page = get_zeroed_page(GFP_KERNEL); - if (!page) - return -ENOMEM; + if (uport->ops->alloc_xmit) { + retval = uport->ops->alloc_xmit(uport, &state->xmit); + if (retval) + return retval; + } else { + /* This is protected by the per port mutex */ + page = get_zeroed_page(GFP_KERNEL); + if (!page) + return -ENOMEM; - state->xmit.buf = (unsigned char *) page; + state->xmit.buf = (unsigned char *) page; + } uart_circ_clear(&state->xmit); } @@ -257,7 +263,10 @@ static void uart_shutdown(struct tty_struct *tty, struct uart_state *state) * Free the transmit buffer page. */ if (state->xmit.buf) { - free_page((unsigned long)state->xmit.buf); + if (uport->ops->free_xmit) + uport->ops->free_xmit(uport, &state->xmit); + else + free_page((unsigned long)state->xmit.buf); state->xmit.buf = NULL; } } diff --git a/include/linux/serial_core.h b/include/linux/serial_core.h index b17d4c9..ebe9af1 100644 --- a/include/linux/serial_core.h +++ b/include/linux/serial_core.h @@ -248,6 +248,8 @@ struct uart_ops { void (*break_ctl)(struct uart_port *, int ctl); int (*startup)(struct uart_port *); void (*shutdown)(struct uart_port *); + int (*alloc_xmit)(struct uart_port *, struct circ_buf *); + void (*free_xmit)(struct uart_port *, struct circ_buf *); void (*flush_buffer)(struct uart_port *); void (*set_termios)(struct uart_port *, struct ktermios *new, struct ktermios *old);
This allows drivers (such as OMAP serial) to allocate and free their transmit buffers in a sane manner, rather than working around the buffer allocation provided by serial_core. Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> --- drivers/tty/serial/serial_core.c | 23 ++++++++++++++++------- include/linux/serial_core.h | 2 ++ 2 files changed, 18 insertions(+), 7 deletions(-)