Re: [PATCH]serial: 8250: Fix detect XScale port wrong

[Wang YanQing]

On Fri, Mar 01, 2013 at 02:17:16PM -0500, Paul Gortmaker wrote:
> On 13-03-01 12:56 AM, Wang YanQing wrote:
> > Some UARTs add enhanced functions with unused bit in
> > 16550 standard, like UART_IER_UUE bit, it cause XScale
> 
> Which xscale platform?  It would be nice to know the specifics.
The "XScale" mean the XScale port detect code in autoconfig_16550a,
I meet the problem in a normal pc motherboard, I should rephrase
the subject use "Fix detect 16550A ports as XScale ports wrong"
> 
> > detect wrong. Now detect UART_IER_UUE and UART_IER_RTOIE
> > to reduce the annoying wrong result which cause UARTs don't
> > work.
> 
> You should ideally identify the original commit which caused
> the regression.  Assuming you have fully understood the problem
> you should be able to do this with "git blame" and not need to
> do the full bisect.
>
No it is not a regression, it is caused just like you say
there are many 8250/16550 variations, and I meet one, that is it.

> 
> > 
> > Serial controller: Device 4348:3253(CH352 PCI based Multi-I/O Controller)
> > is a example. It use UART_IER_UUE as the LOWPOWER function,
> > you can get the datasheet from below urls:
> > 
> > http://wch-ic.com/download/list.asp?id=116
> > CH352DS1.PDF
> > 
> > http://wch-ic.com/download/list.asp?id=117
> > CH352DS2.PDF.
> 
> Rather than quote links that will expire and no longer be
> valid in six months, it would be better if you described
> exactly how and why these ports are different directly in
> your commit long log.

Device 4348:3253 use the UART_IER_UUE
as the LOWPOWER function , so it is readable and writable,
quote out the origin words in datasheet.

"
LOWPOWER:When the bit is 1, close the internal benchmark clock
of serial port to set into low-power status.
"

That is  the convict cause our XScale detect code work wrong.

> What does your change do to Xscale that do not do UART_IER_RTOIE?
Yes, you are right, I don't have experience with Xscale, so I will
fix the problem in another way, set it fix type in 8250_pci code 
maybe is good.

Thanks.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/