diff options
author | Paul Gortmaker <paul.gortmaker@windriver.com> | 2017-01-04 15:08:15 -0500 |
---|---|---|
committer | Alex Williamson <alex.williamson@redhat.com> | 2017-01-04 13:22:38 -0700 |
commit | c6ef7fd40eddad38a8825cbd6bb2ce8bdbba88f5 (patch) | |
tree | d7b398cea5abb297dbf873d7185dfb87f44f9a01 /Documentation | |
parent | e19f32da5ded958238eac1bbe001192acef191a2 (diff) | |
download | linux-c6ef7fd40eddad38a8825cbd6bb2ce8bdbba88f5.tar.bz2 |
vfio-mdev: fix non-standard ioctl return val causing i386 build fail
What appears to be a copy and paste error from the line above gets
the ioctl a ssize_t return value instead of the traditional "int".
The associated sample code used "long" which meant it would compile
for x86-64 but not i386, with the latter failing as follows:
CC [M] samples/vfio-mdev/mtty.o
samples/vfio-mdev/mtty.c:1418:20: error: initialization from incompatible pointer type [-Werror=incompatible-pointer-types]
.ioctl = mtty_ioctl,
^
samples/vfio-mdev/mtty.c:1418:20: note: (near initialization for ‘mdev_fops.ioctl’)
cc1: some warnings being treated as errors
Since in this case, vfio is working with struct file_operations; as such:
long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
...and so here we just standardize on long vs. the normal int that user
space typically sees and documents as per "man ioctl" and similar.
Fixes: 9d1a546c53b4 ("docs: Sample driver to demonstrate how to use Mediated device framework.")
Cc: Kirti Wankhede <kwankhede@nvidia.com>
Cc: Neo Jia <cjia@nvidia.com>
Cc: kvm@vger.kernel.org
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Diffstat (limited to 'Documentation')
0 files changed, 0 insertions, 0 deletions