| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Thu, 3 Sep 2009 13:46:43 +0800
From: "HU TAO-TGHK48" <taohu@...orola.com>
To: "Pandita, Vikram" <vikram.pandita@...com>,
"Govindraj" <govindraj.ti@...il.com>
Cc: "vimal singh" <vimal.newwork@...il.com>,
<linux-omap@...r.kernel.org>,
"LKML" <linux-kernel@...r.kernel.org>,
<linux-serial@...r.kernel.org>,
"Ye Yuan.Bo-A22116" <yuan-bo.ye@...orola.com>,
"Chen Xiaolong-A21785" <A21785@...orola.com>
Subject: RE: [RFC][PATCH]: Adding support for omap-serail driver
I don't see much added value to use omap_uart_idle_timer() in serial.c since it will be called anyway after timeout.
Is it more simple to do it in omap_uart_prepare_idle()?
Hence omap_uart_prepare_idle() can be move into driver/serial/omap_serial.c eventually.
>
> +extern int are_driveromap_uarts_active(int *);
> +
> static void omap_uart_idle_timer(unsigned long data)
> {
> struct omap_uart_state *uart = (struct omap_uart_state *)data;
> + int dummy;
> +
> + if (are_driveromap_uarts_active(&dummy)){
> + /* Keep UART on NoIdle when DMA channel is
> enabled in omap
> + * serial: do not allow UART to goto Smart-idle
> till its done
> + * dma'ing
> + */
> + omap_uart_block_sleep(uart);
> + return;
> + }
>
> omap_uart_allow_sleep(uart);
> }
>
> -----Original Message-----
> From: Pandita, Vikram [mailto:vikram.pandita@...com]
> Sent: Tuesday, September 01, 2009 10:58 PM
> To: Govindraj; HU TAO-TGHK48
> Cc: vimal singh; linux-omap@...r.kernel.org; LKML;
> linux-serial@...r.kernel.org; Ye Yuan.Bo-A22116; Chen Xiaolong-A21785
> Subject: RE: [RFC][PATCH]: Adding support for omap-serail driver
>
> govindraj
>
> >-----Original Message-----
> >From: linux-omap-owner@...r.kernel.org
> >[mailto:linux-omap-owner@...r.kernel.org] On Behalf Of Govindraj
> >Sent: Tuesday, September 01, 2009 2:14 AM
> >To: HU TAO-TGHK48
> >Cc: vimal singh; linux-omap@...r.kernel.org; LKML;
> >linux-serial@...r.kernel.org; Ye Yuan.Bo-A22116; Chen Xiaolong-A21785
> >Subject: Re: [RFC][PATCH]: Adding support for omap-serail driver
> >
> >On Mon, Aug 31, 2009 at 5:20 PM, HU
> TAO-TGHK48<taohu@...orola.com> wrote:
> >>
> >> 1. Shall we cleanup PM related stuff in
> arch/arm/mach-omap2/serial.c
> >> as well?
> >> Originally serail.c register UART IRQ to decide if
> UART idle for
> >> a while and is able to enter low power mode (e.g. retention).
> >> To work with original 8250 driver, it is probably the only way
> >> since 8250 is not aware of OMAP PM.
> >>
> >> However it would be more reasonable to merge PM stuff to
> >> omap-serial.c. since the new driver is already OMAP specific
> >>
> >> 2. There is an issue for DMA with current implementation
> in serial.c
> >> When Rx DMA is active NO Rx IRQ will be generated.
> >> So serial.c will easily set uart->can_sleep with "1"
> even there is
> >> Rx DMA ongoing
> >> + if ((iir & 0x4) && up->use_dma) {
> >> + up->ier &= ~UART_IER_RDI;
> >> + serial_out(up, UART_IER, up->ier
> >>
> >> In my view, the best way is to do the idle detection in
> >> omap_serial.c.
> >
> >Yes I understand that we cannot adapt 8250 PM model for omap-serial
> >driver in DMA mode I am currently working on that adaption with dma
> >mode and will be posting a separate patch for changes on serial.c.
> >
> >Wouldn't it be cleaner to inherit and adapt the Serial-PM framework
> >from serial.c rather than redefining the PM changes in the driver.
> >
> >As Serial-PM framework already has support for interrupt mode and we
> >have to adapt the same for DMA mode.
> >
> >Also defining PM changes in omap-serial would need changes
> in PM framework.
> >As PM framework calls functions from serail.c file if we are
> defining
> >PM framework in our driver then we may need to redefine the
> functions
> >as in serial.c or modify the PM-framework for uart-activity check.
> >I feel adapting the existing serial-PM framework for DMA
> mode would be
> >better rather than doing these changes.
>
> You can take reference of how to integrate omap-serial driver
> PM with mach-omap2/serial.c as follows, for reference ---
>
>
> ---
> From 69112426bd6009cb11e104b9aafcc65429d662f0 Mon Sep 17 00:00:00 2001
> From: Vikram Pandita <vikram.pandita@...com>
> Date: Fri, 21 Aug 2009 11:15:58 -0500
> Subject: [RFC PATCH] OMAP: Serial: Keep uart in no idle mode
> when DMA ongoing
>
> Keep UART in NoIdle mode when DMA is going on.
>
> Only once UART transfers are complete, switch to smart idle and
> allow OsIdle to kick in
>
> Signed-off-by: Vikram Pandita <vikram.pandita@...com>
> ---
> arch/arm/mach-omap2/serial.c | 12 ++++++++++++
> drivers/serial/omap-serial.c | 2 +-
> 2 files changed, 13 insertions(+), 1 deletions(-)
>
> diff --git a/arch/arm/mach-omap2/serial.c
> b/arch/arm/mach-omap2/serial.c
> index ff9beb7..a6ee6ab 100755
> --- a/arch/arm/mach-omap2/serial.c
> +++ b/arch/arm/mach-omap2/serial.c
> @@ -359,9 +359,21 @@ static void omap_uart_allow_sleep(struct
> omap_uart_state *uart)
> del_timer(&uart->timer);
> }
>
> +extern int are_driveromap_uarts_active(int *);
> +
> static void omap_uart_idle_timer(unsigned long data)
> {
> struct omap_uart_state *uart = (struct omap_uart_state *)data;
> + int dummy;
> +
> + if (are_driveromap_uarts_active(&dummy)){
> + /* Keep UART on NoIdle when DMA channel is
> enabled in omap
> + * serial: do not allow UART to goto Smart-idle
> till its done
> + * dma'ing
> + */
> + omap_uart_block_sleep(uart);
> + return;
> + }
>
> omap_uart_allow_sleep(uart);
> }
> diff --git a/drivers/serial/omap-serial.c
> b/drivers/serial/omap-serial.c
> index 938f29f..f105e24 100644
> --- a/drivers/serial/omap-serial.c
> +++ b/drivers/serial/omap-serial.c
> @@ -1641,6 +1641,7 @@ int omap24xx_uart_cts_wakeup(int
> uart_no, int state)
> return 0;
> }
> EXPORT_SYMBOL(omap24xx_uart_cts_wakeup);
> +#endif
> /**
> * are_driver8250_uarts_active() - Check if any ports managed by this
> * driver are currently busy. This should be called with interrupts
> @@ -1709,7 +1710,6 @@ int are_driveromap_uarts_active(int
> *driver8250_managed)
> }
> EXPORT_SYMBOL(are_driveromap_uarts_active);
>
> -#endif
>
> static void serial_omap_display_reg(struct uart_port *port)
> {
> --
> 1.6.3.3.334.g916e1
>
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists