[PATCH] PPC4xx UART0 (8250) problem

[PATCH] PPC4xx UART0 (8250) problem

[Mikhail Zolotaryov]

On PPC4xx embedded we often have UART0 on IRQ0 but 8250 UART driver uses
IRQ0 as "no irq" hack.
It's suggested that platforms with UART on IRQ0 will redefine
is_real_interrupt(irq) macro but 8250 code anyway overrides that.
As the result, serial line on UART0 works on polling-way and we have
character loss on high baud rates.

In patch I've added check if UART0 is on IRQ0, if so we define
is_real_interrupt macro to be 1 (true).
Also, check if is_real_interrupt macro is already defined added to 8250
UART driver.

--- linux-2.6.20.5.orig/include/asm-ppc/ibm4xx.h        2007-03-23
21:52:51.000000000 +0200
+++ linux-2.6.20.5/include/asm-ppc/ibm4xx.h     2007-04-06
17:37:22.000000000 +0300
@@ -54,6 +54,12 @@
 #include <platforms/4xx/xilinx_ml403.h>
 #endif
 
+#ifdef UART0_INT
+#if (UART0_INT == 0)
+#define is_real_interrupt(irq)  (1)
+#endif
+#endif
+
 #ifndef __ASSEMBLY__
 
 #ifdef CONFIG_40x
--- linux-2.6.20.5.orig/drivers/serial/8250.c   2007-03-23
21:52:51.000000000 +0200
+++ linux-2.6.20.5/drivers/serial/8250.c        2007-04-06
16:44:32.000000000 +0300
@@ -77,7 +77,10 @@
  * machine types want others as well - they're free
  * to redefine this in their header file.
  */
+
+#ifndef is_real_interrupt(irq)
 #define is_real_interrupt(irq) ((irq) != 0)
+#endif
 
 #ifdef CONFIG_SERIAL_8250_DETECT_IRQ
 #define CONFIG_SERIAL_DETECT_IRQ 1

[PATCH] PPC4xx UART0 (8250) problem

[Mikhail Zolotaryov]

On PPC4xx embedded we often have UART0 on IRQ0 but 8250 UART driver
uses IRQ0 as "no irq" hack. It's suggested that platforms with UART
on IRQ0 will redefine is_real_interrupt(irq) macro but 8250 code
anyway overrides that. As the result, serial line on UART0 works
on polling-way and we have character loss on high baud rates.

In patch I added check if UART0 is on IRQ0, if so we define
is_real_interrupt macro to be 1 (true). Also, check if
is_real_interrupt macro is already defined added to 8250 UART
driver.

-------------------------------------------------------------------

--- linux.orig/include/asm-ppc/ibm4xx.h        2007-03-23 21:52:51.000000000 +0200
+++ linux/include/asm-ppc/ibm4xx.h     2007-04-06 17:37:22.000000000 +0300
@@ -54,6 +54,12 @@
 #include <platforms/4xx/xilinx_ml403.h>
 #endif

+#ifdef UART0_INT
+#if (UART0_INT == 0)
+#define is_real_interrupt(irq)  (1)
+#endif
+#endif
+
 #ifndef __ASSEMBLY__

 #ifdef CONFIG_40x
--- linux.orig/drivers/serial/8250.c   2007-03-23 21:52:51.000000000 +0200
+++ linux/drivers/serial/8250.c        2007-04-06 16:44:32.000000000 +0300
@@ -77,7 +77,9 @@
  * machine types want others as well - they're free
  * to redefine this in their header file.
  */
+#ifndef is_real_interrupt(irq)
 #define is_real_interrupt(irq)	((irq) != 0)
+#endif
 
 #ifdef CONFIG_SERIAL_8250_DETECT_IRQ
 #define CONFIG_SERIAL_DETECT_IRQ 1

Re: [PATCH] PPC4xx UART0 (8250) problem

[Benjamin Herrenschmidt]

On Sat, 2007-04-07 at 07:21 +0300, Mikhail Zolotaryov wrote:
> On PPC4xx embedded we often have UART0 on IRQ0 but 8250 UART driver
> uses IRQ0 as "no irq" hack. It's suggested that platforms with UART
> on IRQ0 will redefine is_real_interrupt(irq) macro but 8250 code
> anyway overrides that. As the result, serial line on UART0 works
> on polling-way and we have character loss on high baud rates.
> 
> In patch I added check if UART0 is on IRQ0, if so we define
> is_real_interrupt macro to be 1 (true). Also, check if
> is_real_interrupt macro is already defined added to 8250 UART
> driver.

This is an old problem. The proper fix is already implemented for
arch/powerpc and consist of having virtual irq numbers (which helps for
many other things anyway).

Support for 4xx platforms in arch/powerpc is starting to get in, pop on
linuxppc-dev@ozlabs.org where the patches are being posted and you are
welcome to give a hand porting more platforms over :-)

Ben.

Re: [PATCH] PPC4xx UART0 (8250) problem

[Mikhail Zolotaryov]

Benjamin Herrenschmidt wrote: 
> This is an old problem. The proper fix is already implemented for
> arch/powerpc and consist of having virtual irq numbers (which helps for
> many other things anyway).
> 
> Support for 4xx platforms in arch/powerpc is starting to get in, pop on
> linuxppc-dev@ozlabs.org where the patches are being posted and you are
> welcome to give a hand porting more platforms over :-)

First, thanks for response.

Ok, I see IRQ0 mapping to virtual number is good solution but I know the
problem from the times of 2.4 kernel. It was very surprising while moving to
2.6 kernel - serial driver was changed, but we have the same problem there.

Maybe it's good idea to create unofficial ppc linux branch repository
(or it already exists ?) where we can put patches that work (as above) but
not in pretty good way linux community expects.