[parisc-linux] touching several drivers
Sat, 9 Feb 2002 20:58:42 +0000
Hey, I've jut produced ftp.parisc-linux.org/patches/random-cleanups.diff
It fixes a number of things that have been bugging me on the C360 for
arch/parisc/kernel/drivers.c | 59 ++++++++++++++++++++----------------------
drivers/gsc/dino.c | 26 ++++++++----------
drivers/gsc/gsc.c | 8 -----
drivers/gsc/serial.c | 18 ++++--------
drivers/parport/parport_gsc.c | 16 -----------
drivers/scsi/lasi700.c | 8 -----
drivers/sound/harmony.c | 4 --
so if anyone wants to object, now's a good time to do it. If anyone has
the urge, testing it on other machines would be worthwhile. I've been
careful to consider other machines, but thinking is no substitute for
actually trying the code, and I touched some interesting parts of
* Stop checking hversion_rev in match_device. All drivers specify
* Don't print a warning if a driver declines to claim a device. That's
* Revert get_node_path to its previous form. I found another bug in
doing the inventory and adding another hack would have been silly.
* Neaten up walk_native_bus a little, I have moral objections to the
* Fix a bug in fixup_child_irqs. If the device had no children, it
would assign irqs to siblings instead, which led to the bogus serial
port being detected on Cujo.
* A few more Dino -> Cujo changes. Print a little more information when
we get a stuck interrupt.
* Make the GSC serial driver return silently for Cujo's serial port (if
it's unassigned). Improve the error message and make it not dependent
on any CONFIG options.
* Clean up a comment and remove the A180's serial port from the
* Call request_mem_region for each successful device/driver claim.
* Make gsc_common_irqsetup a little more sensible about what regions
* Delete request/release regions from lasi700.c, parport_gsc.c, harmony.c
Revolutions do not require corporate support.