[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <47C40D0F.9080708@enea.com>
Date: Tue, 26 Feb 2008 13:58:55 +0100
From: Arvid Brodin <arvid.brodin@...a.com>
To: "linux-os (Dick Johnson)" <linux-os@...logic.com>
CC: linux-kernel@...r.kernel.org
Subject: Re: Print long messages to console from kernel module
On 2008-02-25 23:27, linux-os (Dick Johnson) wrote:
> On Mon, 25 Feb 2008, Arvid Brodin wrote:
>
>> I need to write messages > 1023 characters long to the console from a module*. printk() is limited to 1023 characters, and splitting the message over several printk()'s results in a line break and "Month hh:mm:ss host kernel:" being inserted in my text.
>>
>> I tried including <linux/console.h> and using the console_drivers declared there, but get
>> "WARNING: "console_drivers" [<path>/log.ko] undefined!" when compiling and
>> "insmod: error inserting 'log.ko': -1 Unknown symbol in module" when insmodding.
>>
>> I guess this is because non EXPORT_SYMBOL'd symbols are only accessible to statically linked code, and not to modules? I see in printk.c that console_drivers is set up there, and I haven't been able to find any other interface to console_drivers.
>>
>> In short: is there any way to print messages to the console from a kernel module, except printk()? Is opening /dev/tty and writing to it the way to go?
>>
>>
>> * I'm writing an in-memory logger to be included in a module. The log can be several megabytes. The idea is to use SysRq to print the contents of the log to console after a kernel panic or otherwise when writing to disk might not work.
>>
>
> Write the data to a kernel buffer. Impliment read() or ioctl() and
> poll(). Have a user-mode task sleep in poll, waiting for data to
> become available. That user-mode task can do anything it wants,
> unrestricted, with the data including writing it to files or any
> tty it wants to open.
Thank you for your answer. However, I don't see how a user-mode task will help me print my log after a kernel panic, through SysRq? Please clarify.
What we want is essentially a replacement for printk(), where the messages are instead logged in a big ring buffer, and can be printed with Alt-SysRq-l when need be. And the problem is the actual printing of the buffer to the console, since printk() inserts its timestamp after every linebreak or 1023 characters, whichever comes first.
--
Arvid Brodin
Enea LCC
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists