[U-Boot] [PATCH] dm: ns16550: Add support for reg-offset property

[Alexander Graf <agraf@suse.de>]

On 02/25/2016 02:38 PM, Derald D. Woods wrote:
> On Thu, Feb 25, 2016 at 09:11:24AM +0100, Michal Simek wrote:
>> On 25.2.2016 05:47, Derald D. Woods wrote:
>>> On Wed, Feb 24, 2016 at 12:26:09PM +0100, Michal Simek wrote:
>>>> On 24.2.2016 11:56, Adam Ford wrote:
>>>>> On Tue, Feb 23, 2016 at 12:38 AM, Simon Glass <sjg@chromium.org> wrote:
>>>>>> Hi Michal,
>>>>>>
>>>>>> On 22 February 2016 at 00:40, Michal Simek <michal.simek@xilinx.com> wrote:
>>>>>>> On 19.2.2016 21:55, Simon Glass wrote:
>>>>>>>> Hi Michal,
>>>>>>>>
>>>>>>>> On 16 February 2016 at 08:17, Michal Simek <michal.simek@xilinx.com> wrote:
>>>>>>>>> reg-offset is the part of standard 8250 binding in the kernel.
>>>>>>>>> It is shifting start of address space by reg-offset.
>>>>>>>>> On Xilinx platform this offset is typically 0x1000.
>>>>>>>>>
>>>>>>>>> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
>>>>>>>>> ---
>>>>>>>>>
>>>>>>>>>   drivers/serial/ns16550.c | 6 ++++--
>>>>>>>>>   include/ns16550.h        | 1 +
>>>>>>>>>   2 files changed, 5 insertions(+), 2 deletions(-)
>>>>>>>> Reviewed-by: Simon Glass <sjg@chromium.org>
>>>>>>>>
>>>>>>>> Do you support the debug UART feature on your boards?
>>>>>>> yes. I do support it but there you can put just address plus offset and
>>>>>>> there is no reason to add one more option to Kconfig.
>>>>>>> But let me know if you think that this is incorrect flow.
>>>>> This patch seems to break my OMAP3 board.  Does anyone know if I need
>>>>> to set a certain offset for OMAP3 to make this work (and where is the
>>>>> right place for it) ?
>>>> Are you using DT init? Check your DT description if there is reg-offset
>>>> property. I expect if your board worked before and you remove this
>>>> property it will start to work again.
>>>>
>>> I am seeing the same problem with my BeagleBoard Rev. C4. There is
>>> something common, to more than one board, happening with this commit.
>> You should enable debug console and send the log.
>> Do you have enough space for malloc?
>>
> I will have little time this weekend to go further. Some things will
> need to be un-configured to have enough space. I am around 7 KiB over
> with DEBUG enabled.

I'm not quite sure what exactly is going wrong here - maybe some asm 
code is accessing the fields without proper offset generation?

Either way, the patch below seems to fix the issue for me (on beaglebone):

diff --git a/include/ns16550.h b/include/ns16550.h
index 5eeacd6..1311f4c 100644
--- a/include/ns16550.h
+++ b/include/ns16550.h
@@ -54,9 +54,9 @@
   */
  struct ns16550_platdata {
         unsigned long base;
-       int reg_offset;
         int reg_shift;
         int clock;
+       int reg_offset;
  };

  struct udevice;