Re: Kernel Debugging Hardware

David Woodhouse (Dave@imladris.ml.org)
Mon, 12 May 1997 16:47:25 +0100


> The actual writing/reading could best be done from userspace, for example

Anything that a userspace daemon manages to chuck out will have been put on
disc by syslogd anyway. What we want to record is the final dying scream of
the kernel as it panics, and it's not going to pass control back to a user
process after that it does that.
We'd need to add kernel support for the device, if we can't just use the
CONFIG_SERIAL_CONSOLE option. Otherwise we're no better off than we were
before, because we'll miss out on what the kernel says on the way down.

---- ------------------------------------------------ ----
David Woodhouse, WWW: http://www.imladris.ml.org
Robinson College, Email: Dave@imladris.ml.org
Cambridge, Tel: +44 (0) 976 658355
CB3 9AN, (n)talk: dwmw2@imladris.ml.org
England. PGP KEY: finger pgp@imladris.ml.org
---- ------------------------------------------------ ----