diff options
author | David Woodhouse <David.Woodhouse@intel.com> | 2008-12-03 22:12:38 -0800 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2008-12-03 22:12:38 -0800 |
commit | 8865c418caf4e9dd2c24bdfae3a5a4106e143e60 (patch) | |
tree | 4d7fac2c74af56c6aaf83b324b349b76c5f949d9 /drivers/net/acenic.c | |
parent | dcd39c90290297f6e6ed8a04bb20da7ac2b043c5 (diff) | |
download | linux-8865c418caf4e9dd2c24bdfae3a5a4106e143e60.tar.gz linux-8865c418caf4e9dd2c24bdfae3a5a4106e143e60.tar.bz2 linux-8865c418caf4e9dd2c24bdfae3a5a4106e143e60.zip |
atm: 32-bit ioctl compatibility
We lack compat ioctl support through most of the ATM code. This patch
deals with most of it, and I can now at least use BR2684 and PPPoATM
with 32-bit userspace.
I haven't added a .compat_ioctl method to struct atm_ioctl, because
AFAICT none of the current users need any conversion -- so we can just
call the ->ioctl() method in every case. I looked at br2684, clip, lec,
mpc, pppoatm and atmtcp.
In svc_compat_ioctl() the only mangling which is needed is to change
COMPAT_ATM_ADDPARTY to ATM_ADDPARTY. Although it's defined as
_IOW('a', ATMIOC_SPECIAL+4,struct atm_iobuf)
it doesn't actually _take_ a struct atm_iobuf as an argument -- it takes
a struct sockaddr_atmsvc, which _is_ the same between 32-bit and 64-bit
code, so doesn't need conversion.
Almost all of vcc_ioctl() would have been identical, so I converted that
into a core do_vcc_ioctl() function with an 'int compat' argument.
I've done the same with atm_dev_ioctl(), where there _are_ a few
differences, but still it's relatively contained and there would
otherwise have been a lot of duplication.
I haven't done any of the actual device-specific ioctls, although I've
added a compat_ioctl method to struct atmdev_ops.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'drivers/net/acenic.c')
0 files changed, 0 insertions, 0 deletions