Changes to UART ns8250
Zbigniew Bodek
zbb at freebsd.org
Wed Oct 9 21:33:08 UTC 2013
Hello Ganbold.
Thank you for testing the patch and pointing those issue out.
My detection log from Armada XP:
uart0: <16550 or compatible> mem 0xd0012000-0xd001201f irq 41 on simplebus0
uart0: console (115200,n,8,1)
uart1: <16550 or compatible> mem 0xd0012100-0xd001211f irq 42 on simplebus0
uart2: <16550 or compatible> mem 0xd0012200-0xd001221f irq 43 on simplebus0
uart3: <16550 or compatible> mem 0xd0012300-0xd001231f irq 44 on simplebus0
Is there a possibility to download a datasheet for RK30xx so that I could
verify what is required for it's UART?
The patch is causing that we only wait until UART is not busy anymore. I
can't find why would that cause problems
if your UART requires busy detection anyway.
Best regards
Zbigniew Bodek
2013/10/9 Ganbold Tsagaankhuu <ganbold at gmail.com>
>
>
>
> On Tue, Oct 8, 2013 at 9:58 AM, Ganbold Tsagaankhuu <ganbold at gmail.com>wrote:
>
>> Zbigniew,
>>
>>
>> On Tue, Oct 8, 2013 at 3:54 AM, Zbigniew Bodek <zbb at freebsd.org> wrote:
>>
>>> Hello.
>>>
>>> I would like to present a patch for ns8250 serial that I would like to
>>> commit in the near future (if there are no objections).
>>>
>>> The patch is fixing newest DesignWare UART with busy detection.
>>> During frequency divisors configuration when UART is busy transferring or
>>> receiving data, line control register manipulation will not take effect.
>>> Therefore, we will not set divisor latch access bit and we will corrupt
>>> LCR
>>> instead of configuring divisors.
>>> It is necessary to wait until UART finishes all transfers to proceed with
>>> the configuration.
>>>
>>> This was detected on Armada XP as UART fails on this issue 100/100
>>> attempts.
>>> The patch was tested by kevlo@ and me and it works on our Armada XP -
>>> based
>>> systems.
>>>
>>> Please send your comment or remarks if there are any.
>>>
>>
>> I'm trying your patch on r254983.
>> Tried on 2 boards (Cubieboard2 (Allwinner A20 SoC - dual Cortex A7) and
>> Radxa Rock (Rockchip RK3188 - Quad Cortex A9)). Both seem to have some sort
>> of DesignWare uart.
>>
>> 1. It works fine on Cubieboard2. Uart dmesg is like:
>>
>> uart0: <16750 or compatible> mem 0x1c28000-0x1c283ff irq 33 on simplebus0
>>
>> uart0: console (115200,n,8,1)
>>
>> 2. No any printing on screen in case of Radxa Rock. Without your patch
>> uart dmesg is like:
>>
>> uart0: <16650 or compatible> mem 0x20064000-0x200643ff irq 68 on
>> simplebus0
>> uart0: console (115200,n,8,1)
>>
>> In case of RK3188 SoC, it seems booting FreeBSD kernel seems very
>> fragile, not sure yet what is causing the problem.
>> Even with stock ns8250 some version later than r254983 didn't show/print
>> anything on serial console few days ago.
>> Only thing so far I know is this r254983 (with some patch) works in my
>> case on RK3188 SoC based board.
>>
>
>
> Zbigniew,
>
> Just tried again your patch on RK30xx board. I was able to see boot
> messages on screen.
> This uart detected as:
> ...
> uart0: <16650 or compatible> mem 0x20064000-0x200643ff irq 68 on
> simplebus0
> uart0: console (115200,n,8,1)
>
> uart0: fast interrupt
> ...
> Can you show me your uart detection log?
> It seems this DW uart of RK30xx is different than DW uart of A10/A20.
> Boot simply stops printing "start_init: trying /sbin/init".
>
> thanks,
>
> Ganbold
>
>
>
>>
>> thanks,
>>
>> Ganbold
>>
>>
>>
>>
>>>
>>> Best regards
>>> Zbigniew Bodek
>>>
>>> _______________________________________________
>>> freebsd-current at freebsd.org mailing list
>>> http://lists.freebsd.org/mailman/listinfo/freebsd-current
>>> To unsubscribe, send any mail to "
>>> freebsd-current-unsubscribe at freebsd.org"
>>>
>>
>>
>
More information about the freebsd-current
mailing list