[PATCH] IP22: Disable EARLY PRINTK, because it breaks serial console

[PATCH] IP22: Disable EARLY PRINTK, because it breaks serial console

[Thomas Bogendoerfer]

Disable EARLY PRINTK, because it breaks serial console

Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
---
 arch/mips/Kconfig |    1 -
 1 files changed, 0 insertions(+), 1 deletions(-)

diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
index 3b807b4..1f0502d 100644
--- a/arch/mips/Kconfig
+++ b/arch/mips/Kconfig
@@ -334,7 +334,6 @@ config SGI_IP22
 	select SWAP_IO_SPACE
 	select SYS_HAS_CPU_R4X00
 	select SYS_HAS_CPU_R5000
-	select SYS_HAS_EARLY_PRINTK
 	select SYS_SUPPORTS_32BIT_KERNEL
 	select SYS_SUPPORTS_64BIT_KERNEL
 	select SYS_SUPPORTS_BIG_ENDIAN
-- 
1.4.4.4

-- 
Crap can work. Given enough thrust pigs will fly, but it's not necessary a
good idea.                                                [ RFC1925, 2.3 ]

Re: [PATCH] IP22: Disable EARLY PRINTK, because it breaks serial console

[Martin Michlmayr]

* Thomas Bogendoerfer <tsbogend@alpha.franken.de> [2007-09-11 12:44]:
> Disable EARLY PRINTK, because it breaks serial console

Ralf, at the moment IP22 output stops after "Serial: IP22 Zilog driver
(1 chips).".  Can you put this patch in until there's a real fix?

Tested-by: Martin Michlmayr <tbm@cyrius.com>

> Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
> ---
>  arch/mips/Kconfig |    1 -
>  1 files changed, 0 insertions(+), 1 deletions(-)
> 
> diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
> index 3b807b4..1f0502d 100644
> --- a/arch/mips/Kconfig
> +++ b/arch/mips/Kconfig
> @@ -334,7 +334,6 @@ config SGI_IP22
>  	select SWAP_IO_SPACE
>  	select SYS_HAS_CPU_R4X00
>  	select SYS_HAS_CPU_R5000
> -	select SYS_HAS_EARLY_PRINTK
>  	select SYS_SUPPORTS_32BIT_KERNEL
>  	select SYS_SUPPORTS_64BIT_KERNEL
>  	select SYS_SUPPORTS_BIG_ENDIAN
> -- 
> 1.4.4.4
> 
> -- 
> Crap can work. Given enough thrust pigs will fly, but it's not necessary a
> good idea.                                                [ RFC1925, 2.3 ]

-- 
Martin Michlmayr
http://www.cyrius.com/

Re: [PATCH] IP22: Disable EARLY PRINTK, because it breaks serial console

[Maciej W. Rozycki]

On Tue, 30 Oct 2007, Martin Michlmayr wrote:

> * Thomas Bogendoerfer <tsbogend@alpha.franken.de> [2007-09-11 12:44]:
> > Disable EARLY PRINTK, because it breaks serial console
> 
> Ralf, at the moment IP22 output stops after "Serial: IP22 Zilog driver
> (1 chips).".  Can you put this patch in until there's a real fix?

 Is it by any chance the same problem that I noticed with the DECstation 
and reported in the thread starting at: 
"http://marc.info/?l=linux-kernel&m=119030963931879&w=2"?  If so, there is 
a fix for the DECstation provided somewhere down the discussion which you 
may consider porting to IP22.  I think the change to the serial core by 
RMK mentioned there has already been applied upstream.

 Ideally, of course, all the SCC drivers should get merged eventually, but 
due to subtle (and sometimes broken, as it is the case with the 
DECstation) differences in wiring for various systems it may never really 
happen, sigh...

  Maciej

Re: [PATCH] IP22: Disable EARLY PRINTK, because it breaks serial console

[Thomas Bogendoerfer]

On Wed, Nov 07, 2007 at 04:55:10PM +0000, Maciej W. Rozycki wrote:
> On Tue, 30 Oct 2007, Martin Michlmayr wrote:
> 
> > * Thomas Bogendoerfer <tsbogend@alpha.franken.de> [2007-09-11 12:44]:
> > > Disable EARLY PRINTK, because it breaks serial console
> > 
> > Ralf, at the moment IP22 output stops after "Serial: IP22 Zilog driver
> > (1 chips).".  Can you put this patch in until there's a real fix?
> 
>  Is it by any chance the same problem that I noticed with the DECstation 
> and reported in the thread starting at: 

it's the same problem

> "http://marc.info/?l=linux-kernel&m=119030963931879&w=2"?  If so, there is 
> a fix for the DECstation provided somewhere down the discussion which you 
> may consider porting to IP22.  I think the change to the serial core by 
> RMK mentioned there has already been applied upstream.

I ported your fix, but it didn't work for me. Maybe because the
IP22 zilog driver is still a little bit different than the DEC one.

>  Ideally, of course, all the SCC drivers should get merged eventually, but 
> due to subtle (and sometimes broken, as it is the case with the 
> DECstation) differences in wiring for various systems it may never really 
> happen, sigh...

having a more generic zilog driver with platform backends is quite high on
my todo list. But I won't make promisses when this happens...

Thomas.

-- 
Crap can work. Given enough thrust pigs will fly, but it's not necessary a
good idea.                                                [ RFC1925, 2.3 ]

Re: [PATCH] IP22: Disable EARLY PRINTK, because it breaks serial console

[Ralf Baechle]

On Wed, Nov 07, 2007 at 10:54:23PM +0100, Thomas Bogendoerfer wrote:

> >  Ideally, of course, all the SCC drivers should get merged eventually, but 
> > due to subtle (and sometimes broken, as it is the case with the 
> > DECstation) differences in wiring for various systems it may never really 
> > happen, sigh...
> 
> having a more generic zilog driver with platform backends is quite high on
> my todo list. But I won't make promisses when this happens...

Right now there are alot of artificial differences between the sun and IP22
8530 drivers by just different prefixes.  A good start would already be
to unify all these needless differences which is largely a mechanical job.
Once that is that is it will become easier to spot all the relevant
functional differences - there aren't that many but by now they're well
hidden.

  Ralf

Re: [PATCH] IP22: Disable EARLY PRINTK, because it breaks serial console

[Maciej W. Rozycki]

On Wed, 7 Nov 2007, Thomas Bogendoerfer wrote:

> having a more generic zilog driver with platform backends is quite high on
> my todo list. But I won't make promisses when this happens...

 Synchronous and DMA operation is the tough part, at least for the 
DECstation.  OTOH, such a framework actually exists already as 
drivers/net/wan/z85230.[ch], but fitting it into the serial subsystem is a 
lot of work as the serial core currently has no notion of synchronous 
modes at all.  Ultimately you'd like to be able to switch between 
asynchronous and synchronous line disciplines on a port by port basis 
(possibly including halves of the same chip).

  Maciej