Need to disable MSR interrupts in 8250 driver. Request for guidance...

Need to disable MSR interrupts in 8250 driver. Request for guidance...

[Michael Williamson]

Hello,

I am working on platform from the davinci architecture that uses the 8520 UART
driver.  However, there are some configurations that do not have a valid
CTS input pin (it is a multi-purpose pin on a SoC part, and it may be configured
for other functions).  These configurations can cause a pile of "false" 
MSR interrupts.  If, in 8250.c, I set the UART_BUG_NOMSR flag as part of
the up->bugs information, the problem clears up.

The problem is that there is no way to pass bugs information via platform data.  

The link (below) contains a work-around patch that I had submitted to the davinci
group.  It was suggested that I get guidance from the folks on the davinci-serial
list, and that I should consider figuring out a way to set the UART_BUG_NOMSR 
flag from the platform or initialization data.  The patch submission also has a 
more detailed description of the problem for anyone interested.

https://patchwork.kernel.org/patch/442671/

Should I create a new port type, add a new UPF_ flag in the flags field, figure 
out how to pass bugs information via platform data, or continue along the
work-around path?  

Thanks for any insight.

-Mike

Re: Need to disable MSR interrupts in 8250 driver. Request for guidance...

[Kevin Hilman]

Michael Williamson <michael.williamson@criticallink.com> writes:

> I am working on platform from the davinci architecture that uses the 8520 UART
> driver.  However, there are some configurations that do not have a valid
> CTS input pin (it is a multi-purpose pin on a SoC part, and it may be configured
> for other functions).  These configurations can cause a pile of "false" 
> MSR interrupts.  If, in 8250.c, I set the UART_BUG_NOMSR flag as part of
> the up->bugs information, the problem clears up.

[...]

> Should I create a new port type, add a new UPF_ flag in the flags field, figure 
> out how to pass bugs information via platform data, or continue along the
> work-around path?  

I added Greg KH to Cc as he's maintaining the 8250 core now.

IMO, adding UPF_ flag(s) to indicate this bug seems like the right way
to go to me.

Kevin

Re: Need to disable MSR interrupts in 8250 driver. Request for guidance...

[Greg KH]

On Mon, Jan 03, 2011 at 03:56:45PM -0800, Kevin Hilman wrote:
> Michael Williamson <michael.williamson@criticallink.com> writes:
> 
> > I am working on platform from the davinci architecture that uses the 8520 UART
> > driver.  However, there are some configurations that do not have a valid
> > CTS input pin (it is a multi-purpose pin on a SoC part, and it may be configured
> > for other functions).  These configurations can cause a pile of "false" 
> > MSR interrupts.  If, in 8250.c, I set the UART_BUG_NOMSR flag as part of
> > the up->bugs information, the problem clears up.
> 
> [...]
> 
> > Should I create a new port type, add a new UPF_ flag in the flags field, figure 
> > out how to pass bugs information via platform data, or continue along the
> > work-around path?  
> 
> I added Greg KH to Cc as he's maintaining the 8250 core now.
> 
> IMO, adding UPF_ flag(s) to indicate this bug seems like the right way
> to go to me.

Yes, it sounds correct to me as well.

thanks,

greg k-h