On 09/14/2016 02:32 PM, Arnd Bergmann wrote:
> On Wednesday, September 14, 2016 10:50:44 PM CEST zhichang.yuan wrote:
>> And there are probably multiple child devices under LPC, the global arm64_extio_ops only can cover one PIO range. It is fortunate only ipmi driver can not support I/O
>> operation registering, serial driver has serial_in/serial_out to
>> be registered. So, only the PIO range for ipmi device is stored
>> in arm64_extio_ops and the indirect-IO
>> works well for ipmi device.
>
> You should not do that in the serial driver, please just use the
> normal 8250 driver that works fine once you handle the entire
> port range.
Just for the record, Arnd has the right idea. There is only one type of
UART permitted by SBSA (PL011). We carved out an exception for a design
that was already in flight and allowed it to be 16550. That other design
was then corrected in future generations to be PL011 as we required it
to be. Then there's the Hip06. I've given feedback elsewhere about the
need for there to be (at most) two types of UART in the wild. This "LPC"
stuff needs cleaning up (feedback given elsewhere already on that), but
we won't be adding a third serial driver into the mix in order to make
it work. There will be standard ARM servers. There will not be the
kinda-sorta-standard. Thanks.
Jon.
On 02/10/2016 23:03, Jon Masters wrote:
> On 09/14/2016 02:32 PM, Arnd Bergmann wrote:
>> On Wednesday, September 14, 2016 10:50:44 PM CEST zhichang.yuan wrote:
>
>>> And there are probably multiple child devices under LPC, the global arm64_extio_ops only can cover one PIO range. It is fortunate only ipmi driver can not support I/O
>>> operation registering, serial driver has serial_in/serial_out to
>>> be registered. So, only the PIO range for ipmi device is stored
>>> in arm64_extio_ops and the indirect-IO
>>> works well for ipmi device.
>>
>> You should not do that in the serial driver, please just use the
>> normal 8250 driver that works fine once you handle the entire
>> port range.
>
> Just for the record, Arnd has the right idea. There is only one type of
> UART permitted by SBSA (PL011). We carved out an exception for a design
> that was already in flight and allowed it to be 16550. That other design
> was then corrected in future generations to be PL011 as we required it
> to be. Then there's the Hip06. I've given feedback elsewhere about the
> need for there to be (at most) two types of UART in the wild. This "LPC"
> stuff needs cleaning up (feedback given elsewhere already on that), but
> we won't be adding a third serial driver into the mix in order to make
> it work. There will be standard ARM servers. There will not be the
> kinda-sorta-standard. Thanks.
>
Right, so I think Zhichang can make the necessary generic changes to
8250 OF driver to support IO port as well as MMIO-based.
However an LPC-based earlycon driver is still required.
A note on hip07-based D05 (for those unaware): this does not use
LPC-based uart. It uses PL011. The hardware guys have managed some
trickery where they loopback the serial line around the BMC/CPLD. But we
still need it for hip06 D03 and any other boards which want to use LPC
bus for uart.
A question on SBSA: does it propose how to provide serial via BMC for SOL?
> Jon.
>
>
> .
>
On Tue, 2016-10-04 at 13:02 +0100, John Garry wrote:
> Right, so I think Zhichang can make the necessary generic changes to
> 8250 OF driver to support IO port as well as MMIO-based.
>
> However an LPC-based earlycon driver is still required.
>
> A note on hip07-based D05 (for those unaware): this does not use
> LPC-based uart. It uses PL011. The hardware guys have managed some
> trickery where they loopback the serial line around the BMC/CPLD. But we
> still need it for hip06 D03 and any other boards which want to use LPC
> bus for uart.
>
> A question on SBSA: does it propose how to provide serial via BMC for SOL?
Probably another reason to keep 8250 as a legal option ... The (very
popular) Aspeed BMCs tend to do this via a 8250-looking virtual UART on
LPC.
Cheers,
Ben,
On 06/10/2016 01:18, Benjamin Herrenschmidt wrote:
> On Tue, 2016-10-04 at 13:02 +0100, John Garry wrote:
>> Right, so I think Zhichang can make the necessary generic changes to
>> 8250 OF driver to support IO port as well as MMIO-based.
>>
>> However an LPC-based earlycon driver is still required.
>>
>> A note on hip07-based D05 (for those unaware): this does not use
>> LPC-based uart. It uses PL011. The hardware guys have managed some
>> trickery where they loopback the serial line around the BMC/CPLD. But we
>> still need it for hip06 D03 and any other boards which want to use LPC
>> bus for uart.
>>
>> A question on SBSA: does it propose how to provide serial via BMC for SOL?
>
> Probably another reason to keep 8250 as a legal option ... The (very
> popular) Aspeed BMCs tend to do this via a 8250-looking virtual UART on
> LPC.
>
> Cheers,
> Ben,
I think we're talking about the same thing for our LPC-based UART.
John
>
>
> .
>