I tried yesterday compiling 2.4.0-test7 from the linuxcare.au
linux-pmac-devel tree (I also tried the linux-pmac-benh tree with
similar results). Comments:
The kernel only recognizes 512MB of memory. I was hoping 2.4 would
let me move beyond the 680 (or so) MB I can successfully access with
the 2.2.17 linux-pmac-stable kernel.
I get a series of PCI address conflicts when probing the PCI bus
which result in my ethernet card not being able to be turned on. I am
not using the built-in GMAC, but an older Apple 10/100 PCI card. This
does not happen with the linux-pmac-stable kernel.
The USB mouse and keyboard are completely messed up (using the new
input layer). The mouse is frozen, and the keyboard returns
(thoroughly) scrambled keycodes. I noticed in a previous comment
something about changing keycodes... is the setup completely
different in 2.4?
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
The kernel only recognizes 512MB of memory. I was hoping 2.4 would
let me move beyond the 680 (or so) MB I can successfully access with
the 2.2.17 linux-pmac-stable kernel.
Bitkeeper 2.4 has Paul's new high memory stuff (it's an option you need
to enable in the config). I don't know how well tested it is however.
I get a series of PCI address conflicts when probing the PCI bus
which result in my ethernet card not being able to be turned on. I am
not using the built-in GMAC, but an older Apple 10/100 PCI card. This
does not happen with the linux-pmac-stable kernel.
With recent kernels, GMAC works nicely. Several people (including me) are
working on solving those PCI problems.
The USB mouse and keyboard are completely messed up (using the new
input layer). The mouse is frozen, and the keyboard returns
(thoroughly) scrambled keycodes. I noticed in a previous comment
something about changing keycodes... is the setup completely
different in 2.4?
Add "keyboard_sends_linux_keycodes=0" to your kernel command line to
force the input layer to use the old keycodes. Read Franz Sirl comments
on how to setup the new input layer, the mouse driver is no longer a
busmouse, but now provides PS2 datas to applications. You should find
links on my page (ppclinux.apple.com/~benh).
Ben.
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
quoted
I get a series of PCI address conflicts when probing the PCI bus
which result in my ethernet card not being able to be turned on. I am
not using the built-in GMAC, but an older Apple 10/100 PCI card. This
does not happen with the linux-pmac-stable kernel.
With recent kernels, GMAC works nicely. Several people (including me) are
working on solving those PCI problems.
Oops, I didn't notice you were using a dual-G4. I've not yet merged the
GMAC changes for this machine in 2.4, but if you rsync my current 2.2
tree, you should have a working GMAC driver (and no PCI conflicts BTW).
I'm waiting for some older patches to be accepted by Linus before hacking
more on the bk tree.
Ben.
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/