Am 31.07.2010, 16:26 Uhr, schrieb Alan Hourihane <alanh@fairlite.co.uk>:
Helmut - if there are fixes on your branch that should be in the trunk
please commit them now.
Roughly the fixes are:
1. fix shutdown (esp. if GEM=ROM) - see
http://sparemint.org/bugtracker/view.php?id=162
2. allow a tty to be opened for non-blocking IO. This is just a few
days old and has not been well tested nor discussed.
3. enable gcc2-compilation.
4. improved load-calculation: use time not spent in kernel to get load
instead of counting running processes every 5 seconds. But this
contradicts to the docs in
http://info2html.sourceforge.net/cgi-bin/info2html-demo/info2html?(libc.info.gz)Processor%2520Resources
"
processors are not already overused. Unix systems calculate something
called the "load average". This is a number indicating how many
processes were running. This number is average over different periods
of times (normally 1, 5, and 15 minutes).
"
Obviously the new method is more exact e.g. the original did not see
any process that starts and terminates between two samples.
Also there's a lot of debug-code because I encountered some
timing-issues on aranym that may affect the load-calculation and other
things.
5. Add some stack-checking code to find the bug that caused
pixel-detection to fail (obsolete IMHO).
6. because sys_b_bioskeys is not reentrant, don't permit it to be
entered again using kbd_lock-semaphore. There's still the possibility
that the process that called sys_b_bioskeys crashes during execution
of sys_b_bioskeys, and kbd_lock is never reset, so one can only reboot.
7. Alans patch to fix the key-repeat on aranym.
If no one that uses my kernel sees no problems with the shutdown (1.)
I will commit that. Any other on demand.
Did you find a solution for the renaming loaded files on FAT?
I guess I have to start another thread for my pending XaAES-changes.