From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755797AbbLQNPQ (ORCPT ); Thu, 17 Dec 2015 08:15:16 -0500 Received: from mail-yk0-f182.google.com ([209.85.160.182]:35980 "EHLO mail-yk0-f182.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755225AbbLQNPN (ORCPT ); Thu, 17 Dec 2015 08:15:13 -0500 MIME-Version: 1.0 In-Reply-To: <56700A45.3070109@arm.com> References: <1449048790-25859-1-git-send-email-vladimir.murzin@arm.com> <1449048790-25859-5-git-send-email-vladimir.murzin@arm.com> <56700A45.3070109@arm.com> Date: Thu, 17 Dec 2015 15:15:12 +0200 Message-ID: Subject: Re: [PATCH v1 04/10] serial: mps2-uart: add MPS2 UART driver From: Andy Shevchenko To: Vladimir Murzin Cc: Arnd Bergmann , Russell King , Greg Kroah-Hartman , Daniel Lezcano , Thomas Gleixner , =?UTF-8?Q?Uwe_Kleine=2DK=C3=B6nig?= , =?UTF-8?Q?Andreas_F=C3=A4rber?= , Maxime Coquelin , Mark Rutland , Pawel Moll , "ijc+devicetree" , Kumar Gala , Jiri Slaby , Rob Herring , devicetree , "linux-serial@vger.kernel.org" , "linux-api@vger.kernel.org" , linux-arm Mailing List , "linux-kernel@vger.kernel.org" Content-Type: text/plain; charset=UTF-8 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, Dec 15, 2015 at 2:40 PM, Vladimir Murzin wrote: > On 12/12/15 23:39, Andy Shevchenko wrote: >> On Wed, Dec 2, 2015 at 11:33 AM, Vladimir Murzin >> wrote: >>> This driver adds support to the UART controller found on ARM MPS2 >>> platform. >> >> Just few comments (have neither time not big desire to do full review). >> > > Still better than nothing ;) I'm mostly agree on points you had, so I've > just left some I'm doubt about... > >>> + >>> +static void mps2_uart_enable_ms(struct uart_port *port) >>> +{ >>> +} >>> + >>> +static void mps2_uart_break_ctl(struct uart_port *port, int ctl) >>> +{ >>> +} >> >> Are those required to be present? If not, remove them until you have >> alive code there. > > A quick grep shows that core calls mps2_uart_break_ctl() > unconditionally, but, yes, it checks for presence of > mps2_uart_enable_ms() before jumping there, so it is safe to remove latter. OK. >>> +static irqreturn_t mps2_uart_oerrirq(int irq, void *data) >>> +{ >>> + irqreturn_t handled = IRQ_NONE; >>> + struct uart_port *port = data; >>> + u8 irqflag = mps2_uart_read8(port, UARTn_INT); >>> + >>> + spin_lock(&port->lock); >>> + >>> + if (irqflag & UARTn_INT_RX_OVERRUN) { >>> + struct tty_port *tport = &port->state->port; >>> + >>> + mps2_uart_write8(port, UARTn_INT_RX_OVERRUN, UARTn_INT); >>> + tty_insert_flip_char(tport, 0, TTY_OVERRUN); >>> + port->icount.overrun++; >>> + handled = IRQ_HANDLED; >>> + } >>> + >>> + /* XXX: this shouldn't happen? */ >> >> If shouldn't why it's there? Otherwise better to explain which >> conditions may lead to this. >> > > In practice I've never seen that happened and I think it never *should* > happen since we check if there is room in TX buffer. However, I could be > wrong here, so it is why that statement has question mark. So, worth to have a proper comment then. >>> +static int __init mps2_uart_init(void) >>> +{ >>> + int ret; >>> + >>> + ret = uart_register_driver(&mps2_uart_driver); >>> + if (ret) >>> + return ret; >>> + >>> + ret = platform_driver_register(&mps2_serial_driver); >>> + if (ret) >>> + uart_unregister_driver(&mps2_uart_driver); >>> + >>> + pr_info("MPS2 UART driver initialized\n"); >>> + >>> + return ret; >>> +} >>> +module_init(mps2_uart_init); >>> + >>> +static void __exit mps2_uart_exit(void) >>> +{ >>> + platform_driver_unregister(&mps2_serial_driver); >>> + uart_unregister_driver(&mps2_uart_driver); >>> +} >>> +module_exit(mps2_uart_exit); >> >> module_platform_driver(); >> And move uart_*register calls to probe/remove. >> > > With this move we'll get uart_*register for every device probed, no? Don't see a problem, maybe someone else could share an authoritive opinion. Some of the drivers do that, though most do in __init stage. So, see above. -- With Best Regards, Andy Shevchenko From mboxrd@z Thu Jan 1 00:00:00 1970 From: andy.shevchenko@gmail.com (Andy Shevchenko) Date: Thu, 17 Dec 2015 15:15:12 +0200 Subject: [PATCH v1 04/10] serial: mps2-uart: add MPS2 UART driver In-Reply-To: <56700A45.3070109@arm.com> References: <1449048790-25859-1-git-send-email-vladimir.murzin@arm.com> <1449048790-25859-5-git-send-email-vladimir.murzin@arm.com> <56700A45.3070109@arm.com> Message-ID: To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Tue, Dec 15, 2015 at 2:40 PM, Vladimir Murzin wrote: > On 12/12/15 23:39, Andy Shevchenko wrote: >> On Wed, Dec 2, 2015 at 11:33 AM, Vladimir Murzin >> wrote: >>> This driver adds support to the UART controller found on ARM MPS2 >>> platform. >> >> Just few comments (have neither time not big desire to do full review). >> > > Still better than nothing ;) I'm mostly agree on points you had, so I've > just left some I'm doubt about... > >>> + >>> +static void mps2_uart_enable_ms(struct uart_port *port) >>> +{ >>> +} >>> + >>> +static void mps2_uart_break_ctl(struct uart_port *port, int ctl) >>> +{ >>> +} >> >> Are those required to be present? If not, remove them until you have >> alive code there. > > A quick grep shows that core calls mps2_uart_break_ctl() > unconditionally, but, yes, it checks for presence of > mps2_uart_enable_ms() before jumping there, so it is safe to remove latter. OK. >>> +static irqreturn_t mps2_uart_oerrirq(int irq, void *data) >>> +{ >>> + irqreturn_t handled = IRQ_NONE; >>> + struct uart_port *port = data; >>> + u8 irqflag = mps2_uart_read8(port, UARTn_INT); >>> + >>> + spin_lock(&port->lock); >>> + >>> + if (irqflag & UARTn_INT_RX_OVERRUN) { >>> + struct tty_port *tport = &port->state->port; >>> + >>> + mps2_uart_write8(port, UARTn_INT_RX_OVERRUN, UARTn_INT); >>> + tty_insert_flip_char(tport, 0, TTY_OVERRUN); >>> + port->icount.overrun++; >>> + handled = IRQ_HANDLED; >>> + } >>> + >>> + /* XXX: this shouldn't happen? */ >> >> If shouldn't why it's there? Otherwise better to explain which >> conditions may lead to this. >> > > In practice I've never seen that happened and I think it never *should* > happen since we check if there is room in TX buffer. However, I could be > wrong here, so it is why that statement has question mark. So, worth to have a proper comment then. >>> +static int __init mps2_uart_init(void) >>> +{ >>> + int ret; >>> + >>> + ret = uart_register_driver(&mps2_uart_driver); >>> + if (ret) >>> + return ret; >>> + >>> + ret = platform_driver_register(&mps2_serial_driver); >>> + if (ret) >>> + uart_unregister_driver(&mps2_uart_driver); >>> + >>> + pr_info("MPS2 UART driver initialized\n"); >>> + >>> + return ret; >>> +} >>> +module_init(mps2_uart_init); >>> + >>> +static void __exit mps2_uart_exit(void) >>> +{ >>> + platform_driver_unregister(&mps2_serial_driver); >>> + uart_unregister_driver(&mps2_uart_driver); >>> +} >>> +module_exit(mps2_uart_exit); >> >> module_platform_driver(); >> And move uart_*register calls to probe/remove. >> > > With this move we'll get uart_*register for every device probed, no? Don't see a problem, maybe someone else could share an authoritive opinion. Some of the drivers do that, though most do in __init stage. So, see above. -- With Best Regards, Andy Shevchenko