From mboxrd@z Thu Jan 1 00:00:00 1970 From: Fabio Estevam Subject: Re: [PATCH 5/8] serial: imx: save and restore context in the suspend path Date: Mon, 10 Aug 2015 00:30:18 -0300 Message-ID: References: <1439144349-10494-1-git-send-email-edubezval@gmail.com> <1439144349-10494-6-git-send-email-edubezval@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Return-path: In-Reply-To: <1439144349-10494-6-git-send-email-edubezval@gmail.com> Sender: linux-pm-owner@vger.kernel.org To: Eduardo Valentin Cc: Greg Kroah-Hartman , Jiri Slaby , Linux PM , "linux-serial@vger.kernel.org" , linux-kernel , Shenwei Wang List-Id: linux-serial@vger.kernel.org Hi Eduardo, On Sun, Aug 9, 2015 at 3:19 PM, Eduardo Valentin wrote: > This change teaches the imx serial driver to save its > context and restore it across suspend and resume path. > To do so, it introduces serial_imx_restore_context() > and serial_imx_save_context() functions. They use > a shadow set of registers to save key registers > and restore them accordingly. These functions can > be reused on other situations, when the device > context is lost. > > Cc: Fabio Stevam s/Stevam/Estevam (applies to all the patches) > Cc: Greg Kroah-Hartman > Cc: Jiri Slaby > Cc: linux-serial@vger.kernel.org > Cc: linux-kernel@vger.kernel.org > Signed-off-by: Eduardo Valentin > --- > drivers/tty/serial/imx.c | 34 ++++++++++++++++++++++++++++++++++ > 1 file changed, 34 insertions(+) > > diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c > index a03855d..494b182 100644 > --- a/drivers/tty/serial/imx.c > +++ b/drivers/tty/serial/imx.c > @@ -217,6 +217,8 @@ struct imx_port { > unsigned int tx_bytes; > unsigned int dma_tx_nents; > wait_queue_head_t dma_wait; > + unsigned int saved_reg[8]; > + bool context_saved; > }; > > struct imx_port_ucrs { > @@ -1803,6 +1805,36 @@ static struct uart_driver imx_reg = { > .cons = IMX_CONSOLE, > }; > > +static void serial_imx_restore_context(struct imx_port *sport) > +{ > + if (!sport->context_saved) > + return; > + > + writel(sport->saved_reg[4], sport->port.membase + UFCR); > + writel(sport->saved_reg[5], sport->port.membase + UBIR); > + writel(sport->saved_reg[6], sport->port.membase + UBMR); > + writel(sport->saved_reg[7], sport->port.membase + IMX21_UTS); > + writel(sport->saved_reg[0], sport->port.membase + UCR1); > + writel(sport->saved_reg[1] | 0x1, sport->port.membase + UCR2); > + writel(sport->saved_reg[2], sport->port.membase + UCR3); > + writel(sport->saved_reg[3], sport->port.membase + UCR4); > + sport->context_saved = false; > +} > + > +static void serial_imx_save_context(struct imx_port *sport) > +{ > + /* Save necessary regs */ > + sport->saved_reg[0] = readl(sport->port.membase + UCR1); > + sport->saved_reg[1] = readl(sport->port.membase + UCR2); > + sport->saved_reg[2] = readl(sport->port.membase + UCR3); > + sport->saved_reg[3] = readl(sport->port.membase + UCR4); > + sport->saved_reg[4] = readl(sport->port.membase + UFCR); > + sport->saved_reg[5] = readl(sport->port.membase + UBIR); > + sport->saved_reg[6] = readl(sport->port.membase + UBMR); > + sport->saved_reg[7] = readl(sport->port.membase + IMX21_UTS); > + sport->context_saved = true; > +} This seems to be similar to what Shenwei posted earlier: http://lists.infradead.org/pipermail/linux-arm-kernel/2015-July/360841.html Regards, Fabio Estevam