[Top] [All Lists]

Re: Alchemy Au1550 Early Console

To: Jon Anders Haugum <>
Subject: Re: Alchemy Au1550 Early Console
From: Mark Schank <>
Date: Sat, 25 Mar 2006 21:24:04 -0600
In-reply-to: <>
Original-recipient: rfc822;
References: <> <>
User-agent: Thunderbird 1.5 (X11/20051201)
Hi Jon,

Thanks for the reply. I do have CONFIG_SERIAL_8250_CONSOLE enabled.
The console does work once the kernel boots and the driver init routines run. However, if the kernel crashes before the driver init routines are run, then I wasn't getting any output. The old au1x00_uart.c driver was able to support an early console, console_init() in start_kernel(), on the Alchemy uart. However, after looking at the new 8250 driver more closely, I see that it can not support the Alchemy uart until after the
driver init routines are run.

I finally wrote a simple early console driver that relied on the boot monitor setting up the uart. This enabled me to figure out why the kernel was crashing during boot.

Thanks again,

Jon Anders Haugum wrote:
On Wed, 22 Mar 2006, Mark Schank wrote:

I am working with a Au1550 based system an am trying to upgrade
from the 2.6.12 kernel to the 2.6.16 kernel . I have noticed that the
au1x00_uart.c driver has been remove and replaced with functionality in
8250.c and 8250_au1x00.c. So far, I have been unable to get the early
console running. Under this new driver model, what is the proper
way to bring up an early console on a Au1550 internal serial port?

Have you enabled the console for 8250-based serial ports?

Another issue might be if you have got some PCI-based serial ports in the system, because they will be assigned before the internal ones. So you might end up having the console on a PCI serial port.

Third: There is a bug in the new driver regarding divisor settings, but the console should still be working since the boot loader will init the uart properly. (I've posted a patch for this bug earlier:


<Prev in Thread] Current Thread [Next in Thread>