Hi Tomasz,
On Mon, Sep 22, 2014 at 4:49 AM, Tomasz Figa <tomasz.figa@xxxxxxxxx> wrote:
> On 22.09.2014 01:10, Alim Akhtar wrote:
>
> [snip]
>
>>>> As you said there is no support for ioremap on ARM, so this is not
>>>> tested on ARM.
>>>
>>> Don't forget that this driver is primarily targeted for ARM platforms
>>> (versus just one ARM64-based Exynos7), so either this feature should be
>>> clearly added as ARM64-specific (and compiled in conditionally) or made
>>> work for all supported platforms.
>>>
>> Well, this will work on every platform which uses
>> "samsung,exynos4210-uart" as a UART controller.
>> Exynos7 also use same, there is nothing special about ARM64 bit here.
>> please see[1].
>> For "s3c24xx-uart", this has to be implemented separably as that is a
>> bit different.
>
> For this feature, they differ only in locations and lengths of
> FIFO-related bit fields. We had this already implemented for all
> hardware variants and so my recommendation to reuse that code.
>
>>>>>
>>>>>> smh Use ARM semihosting calls for early console.
>>>>>>
>>>>>> earlyprintk= [X86,SH,BLACKFIN,ARM,M68k]
>
> [snip]
>
>> Tomasz/Marek,
>> Do you think something like below make sense here?
>>
>> +static void exynos4210_serial_console_putc(struct uart_port *port, int ch)
>> +{
>> + while (!(readl(port->membase + S3C2410_UFCON) & S3C2410_UFCON_FIFOMODE))
>> + ;
>
> Waiting in a loop for a software-writable FIFO mode enable bit doesn't
> look reasonable to me. Probably a typo?
>
>> +
>> + wr_regb(port, S3C2410_UTXH, ch);
>> +
>> + while ((readl(port->membase + S3C2410_UFSTAT) & S5PV210_UFSTAT_TXFULL))
>> + ;
>
> I wonder if you need to wait for the character to be sent. If you ensure
> before writing a character that there is no previous one in the buffer
> or there is space in FIFO then I believe you should be fine.
>
Well, below is what I was about to post:
#define S3C24XX_SERIAL_CONSOLE &s3c24xx_serial_console
+static void s3c24xx_serial_console_putc(struct uart_port *port, int ch)
+{
+ unsigned int ufcon = rd_regl(port, S3C2410_UFCON);
+
+ if (ufcon & S3C2410_UFCON_FIFOMODE) {
+ while((readl(port->membase + S3C2410_UFSTAT) &
+ S5PV210_UFSTAT_TXFULL))
+ ;
+ wr_regb(port, S3C2410_UTXH, ch);
+ }
+}
+
+static void samsung_early_write(struct console *con, const char *s, unsigned n)
+{
+ struct earlycon_device *dev = con->data;
+
+ uart_console_write(&dev->port, s, n, s3c24xx_serial_console_putc);
+}
+
+static int __init samsung_early_console_setup(struct earlycon_device *device,
+ const char *opt)
+{
+ if (!device->port.membase)
+ return -ENODEV;
+ device->con->write = samsung_early_write;
+ return 0;
+}
+EARLYCON_DECLARE(samsung, samsung_early_console_setup);
+OF_EARLYCON_DECLARE(samsung, "samsung,exynos4210-uart",
samsung_early_console_setup);
but just saw patches from Marek few min back which is similar except
for the handle non-fifo mode, which is good anyway. :-)
So lets drops this patch and discussion here.
Thanks!!!
>> +}
>>
>> and call exynos4210_serial_console_putc() in samsung_early_write()?
>> Anyway functions names need to be changes accordingly.
>
> Yes, this is exactly what we had implemented in the patch I mentioned,
> except that the putc function was generic for all hardware variants.
>
> Best regards,
> Tomasz
--
Regards,
Alim
--
To unsubscribe from this list: send the line "unsubscribe linux-samsung-soc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
