Learn how to create and mount a DAX-capable file system.

1. Format the device by specifying the stripe unit size (su) as 2 MB and stripe width size (sw) as 1.

   # mkfs.xfs -f -d su=2m,sw=1 /dev/pmem0
   ...
   ...
   data     = bsize=4096 blocks=777240064, imaxpct=5
            = sunit=512 swidth=512 blks
   naming   =version 2 bsize=4096 ascii-ci=0 ftype=1
   log      =internal log bsize=4096 blocks=379511, version=2
            = sectsz=4096 sunit=1 blks, lazy-count=1
   realtime =none

   In this example, the data section of the output has bsize=4096. So, the data block size for this file system is 4096 bytes. The sunit=512 and the swidth=512 blks. This implies that the stripe unit is 512*4096 bytes = 2 MB and the stripe width is 512*4096 bytes = 2 MB. A single stripe of this file system is composed of a single stripe unit (512 blocks / 512 blocks per unit).

   A minimum contiguous allocation of 2 MB or contiguous allocations in exact multiples of 2 MB is enforced for data to enable the address space to be represented with HugePages mappings and to bypass the page cache.

2. Mount the device using the dax option.

   # mount -o dax /dev/pmem0 /mnt/pmem0

3. Verify if the file system is mounted with the dax option.

   # mount|grep dax
   /dev/pmem0 on /mnt/pmem0 type xfs
   (rw,relatime,attr2,dax,inode64,sunit=4096,swidth=4096,noquota)
   

   In this example, the sunit and swidth from the mount are in 512-byte blocks.

4. Format the /dev/pmem1 device and mount and verify the dax option in the mount point.

   # mkfs.xfs -f -d su=2m,sw=1 /dev/pmem1
   # mount -o dax /dev/pmem0 /mnt/pmem1
   # mount|grep dax
   /dev/pmem0 on /mnt/pmem0 type xfs
   (rw,relatime,attr2,dax,inode64,sunit=4096,swidth=4096,noquota)
   /dev/pmem1 on /mnt/pmem1 type xfs
   (rw,relatime,attr2,dax,inode64,sunit=4096,swidth=4096,noquota)