BTW st SCC-XXdma stuff *is* enabled on my G3 by default.
Beige or blue? On my B&W G3 it's not enabled, I only see the plain
non-DMA RX and TX interrupts.
Beige - but I was burbling (again) - I build all my own kernels... so YMMV
FWIW:
With the same kernel...
On my Lombard - the SCC XXdma does *not* show
On my 9600 - it does.
Config options supplied if you like :-)
Iain.
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
From: Timothy A. Seufert <hidden> Date: 2000-08-01 10:28:00
At 1:22 PM +0100 7/31/00, Iain Sandoe wrote:
On Mon, Jul 31, 2000, I wrote...
quoted
On Mon, Jul 31, 2000, Timothy A. Seufert wrote:
quoted
At 10:11 AM +0100 7/31/00, Iain Sandoe wrote:
quoted
BTW st SCC-XXdma stuff *is* enabled on my G3 by default.
Beige or blue? On my B&W G3 it's not enabled, I only see the plain
non-DMA RX and TX interrupts.
Beige - but I was burbling (again) - I build all my own kernels... so YMMV
FWIW:
With the same kernel...
On my Lombard - the SCC XXdma does *not* show
On my 9600 - it does.
So it looks like it's not getting enabled on New World machines.
After a brief look at the driver init code my impression (quite
possibly a wildly inaccurate one BTW) is that the driver wants DMA
channels and/or IRQs already partially configured. On New World
machines booted via yaboot these may not be configured by OF. On Old
World machines booted via BootX MacOS does the initialization.
Tim Seufert
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
From: Benjamin Herrenschmidt <hidden> Date: 2000-08-02 16:22:29
So it looks like it's not getting enabled on New World machines.
After a brief look at the driver init code my impression (quite
possibly a wildly inaccurate one BTW) is that the driver wants DMA
channels and/or IRQs already partially configured. On New World
machines booted via yaboot these may not be configured by OF. On Old
World machines booted via BootX MacOS does the initialization.
Can you all test with my rsync tree ? I changed the test against the
number of addresses as Takashi pointed out. If it still doesn't work, we
may have to add missing addresses "manually" (I hate OF tree fixup code ;)
ben.
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
Hi,
According to Darwin's SCC driver, there is a chip bug in that the receiver
may lock up when certain conditions are met. The affected machines listed
in the driver source are:
Performa 5400/6400/6500, 20th Anniversary Mac, PowerBook 2400/3400,
StarMax 3000, PowerMac 4400, Power Macintosh G3 (Gossamer).
The workaround is in FixZeroBug() of Darwin source:
IO/Drivers/Port/AppleSCCSerial/AppleSCCSerial.kmodproj/SccChipPrimatives.cpp
Last I checked, this workaround is not in Linux's
drivers/macintosh/macserial.c, and I suspect this bug might account for
occassional failure reports with mac serial.
Anybody care to code it for Linux? I would do it if I had any of the
affected machine, but I don't, so .... anybody? :)
Cheers,
Takashi Oe
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/