Dear all,
mmap() on VME bridge devices is a feature present in many existing Linux and
UNIX drivers. The proposed solution follows the approach so that mmap()
offsets are coherent with read() and write() offsets, which seems to be the
only manner compatible with the current vme_user API. It also ends up adding
vme_master_mmap call to the VME subsystem API. On one hand this might come
useful for a board driver that will be able to expose one of its windows to the
user that way. On the other hand it introduces dependency on vma_area_struct
into vme.h, which doesn't seem very right.
Also I'm bringing up a bugfix by Martyn Welch that was published on the list,
but for some reason didn't make it to the kernel.
CHANGES IN V2:
* use pr_* instead of printk
* use file_inode() helper
Cheers,
Dmitry
Dmitry Kalinkin (2):
staging: vme: use image mutex for ioctl()
staging: vme: mmap() support for vme_user
Martyn Welch (1):
vme: tsi148: Master windows support USERx and CR/CSR accesses, not
slaves
drivers/staging/vme/devices/vme_user.c | 94 ++++++++++++++++++++++++++++++++--
drivers/vme/bridges/vme_tsi148.c | 11 ++--
drivers/vme/vme.c | 26 ++++++++++
include/linux/vme.h | 1 +
4 files changed, 120 insertions(+), 12 deletions(-)
--
1.9.1
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
