extreme RTAS printks

linas at austin.ibm.com linas at austin.ibm.com
Thu Feb 12 04:10:39 EST 2004


On Tue, Feb 10, 2004 at 11:25:44AM -0600, Olof Johansson wrote:
>
> I thought the SP only logged checkstops and other severe errors. Or does
> it log ECC parity errors/corrections and other "minor" problems too?

Its reported minor/ridiculous warnings in the past ...

> If you're getting the RTAS messages on the console, the machine is
> likely going to be unstable anyway. Would you prefer it to be quiet and
> not warn at all?

The preference that hollis & the majority express is for an english-language
message.  When I asked about this ages ago, the answer I got back was:

"The logic to decode these messages is complex and is getting more complex
every day with new additions. This logic is too big to fit in the kernel
and should be a userland process/daemon (closed source, at that) instead."

I dunno, I'm not sure I buy the "too complex' story. Sure, there's a
lot of data in the hex dump, but a simple two-sentance english language
summary sure would be nice.  *especially* during boot of a failing machine.
And that amount of code would not be big or complex.

> Also, the data needs to be accessible from a debugger, be it via dmesg
> or in other ways. Hex is much better than nothing at all.

well, a quick extension to add rtas error decode to kdb sure would be cute!
I don't think its hard at all, not the basics.


--linas

** Sent via the linuxppc64-dev mail list. See http://lists.linuxppc.org/





More information about the Linuxppc64-dev mailing list