We originally used our added BLKBSZSET ioctl to set the blocksize on a device
to 512 bytes during mkfs, this was in case we inherited another blocksize
from a previous filesystem (i.e. 4k from ext2). Otherwise, some writes
during mkfs.xfs time would fail. However, those writes are now gone from
mkfs.xfs, and there doesn't seem to be any other compelling reason to do this.
At this point, our added BLKBSZSET/BLKBSZGET ioctls just make our diff
from Linus' kernel that much bigger.
By turning this off in userspace now, we can start getting the updated
tools to users before we take the ioctls back out of the kernel.
If mkfs.xfs suddenly starts going wrong for you as a result, just set
cmd/xfsprogs/libxfs/Makefile:LCFLAGS += -DHAVE_BLKBSZSET=0
back to
cmd/xfsprogs/libxfs/Makefile:LCFLAGS += -DHAVE_BLKBSZSET=1
and post a bug to the list...!
Date: Mon Jul 16 13:56:07 PDT 2001
Workarea: chuckle.americas.sgi.com:/export/xfs1/eric/linux-2.4-xfs/workarea
The following file(s) were checked into:
bonnie.engr.sgi.com:/isms/slinx/2.4.x-xfs
Modid: 2.4.x-xfs:slinx:98965a
cmd/xfsprogs/libxfs/Makefile - 1.7
- Turn off use of the BLKBSZSET ioctl (which will likely go away soon)
|