[BK PATCH] devfs cleanups for 2.5.29 - take 2

Greg KH (greg@kroah.com)
Wed, 31 Jul 2002 11:33:59 -0700

Messages sorted by: [ date ][ thread ][ subject ][ author ]
Next message: Greg KH: "Re: [BK PATCH] devfs cleanups for 2.5.29 - take 2"
Previous message: Ray Lee: "[PATCH] Guarantee APM power status change notifications"

Hi,

First off, I've added one more small patch to this series, that removes
all of the devfs_should* functions that I created, with a simple
devfs_only() call that is much easier to understand what is happening.

Secondly, as to Richard's assertion that this patch will break things,
here are a list of the block and then character drivers that "might"
break if "devfs=only" is enabled, and this patch is applied:

Block drivers whose usage in "devfs=only" mode will be affected:
cpqarray.c - actually this driver looks broken for devfs anyway
stram.c
fd177d.c
mfmhd.c
amiflop.c
cciss.c
pf.c
swim_iop.c
z2ram.c
ataraid.c
i20_block.c
ftl.c
ntflcore.c
jsflash.c

Char drivers whose usage in "devfs=only" mode will be affected:
ds1302.c - arch/cris driver
eeprom.c - arch/cris driver
gpio.c - arch/cris driver
i2c.c - arch/cris driver
sync_serial.c - arch/cris driver
cpuid.c
msr.c
drm_stub.c
raw.c
pcilynx.c
lincfg.c
ds.c
dpt_i2o.c
megaraid.c
phonedev.c
psdev.c - fs/intermesso driver

As you can see, this list is _much_ smaller than the list of drivers
that have had to have the "devfs_*" function change to them. Please do
not penalize all of the kernel because these few drivers have not been
"fixed". If "devfs=only" mode is important enough to users of these
drivers, they will be fixed.

Thirdly, I got a machine up and running with devfs, and this patch does
not effect it at all. To be fair, I could not boot _any_ recent 2.5
kernel without this patch in "devfs=only" mode, but that must be because
the Gentoo people are not crazy enough to set up their distro to run in
this devfs mode. And these people are pretty dedicated to devfs, so I
can't test "devfs=only" mode myself with this patch. But from looking
at the above list of drivers, I don't see anything preventing this from
happening with my (and most everyone else's) hardware.

So Linus, please take these changesets. A large number of driver
authors have told me they would like this patch into the kernel, and it
cleans up the driver API to be much saner (as well as making devfs less
intrusive to all drivers).

Please pull from:
bk://linuxusb.bkbits.net/devfs-2.5

thanks,

greg k-h

Overall diffstat for these changesets:

ChangeSet@1.546, 2002-07-30 21:46:34-07:00, greg@kroah.com
Remove the devfs_should* functions I added, and replace them with one devfs_only() call

This now explains what is really going on much better than before.

fs/block_dev.c | 4 +--
fs/devfs/base.c | 50 +++-------------------------------------
fs/devices.c | 4 +--
include/linux/devfs_fs_kernel.h | 19 +--------------
4 files changed, 10 insertions(+), 67 deletions(-)
------

ChangeSet@1.545, 2002-07-30 15:40:49-07:00, greg@kroah.com
Removed devfs_register_blkdev and devfs_unregister_blkdev.

Use register_blkdev and unregister_blkdev as before, and everything will work just fine.

ChangeSet@1.544, 2002-07-30 14:38:11-07:00, greg@kroah.com
Removed devfs_register_chrdev and devfs_unregister_chrdev.

Use register_chrdev and unregister_chrdev as before, and everything will work.

Next message: Greg KH: "Re: [BK PATCH] devfs cleanups for 2.5.29 - take 2"
Previous message: Ray Lee: "[PATCH] Guarantee APM power status change notifications"