This makes us behave better out of the box, for the cases where
people use bigger block sizes. Without this change and if the
user set buffer_compress_percentage=50 and bs=128k, then we'd
have 64k of zeroes at the end. This is probably not what the
user expects.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
followed by the remaining zeroed. With this set to some chunk size smaller
than the block size, fio can alternate random and zeroed data throughout the
I/O buffer. This is particularly useful when bigger block sizes are used
- for a job.
+ for a job. Defaults to 512.
.. option:: buffer_pattern=str
followed by the remaining zeroed. With this set to some chunk size smaller
than the block size, fio can alternate random and zeroed data throughout the
I/O buffer. This is particularly useful when bigger block sizes are used
-for a job.
+for a job. Defaults to 512.
.TP
.BI buffer_pattern \fR=\fPstr
If set, fio will fill the I/O buffers with this pattern or with the contents
.parent = "buffer_compress_percentage",
.hide = 1,
.help = "Size of compressible region in buffer",
+ .def = "512",
.interval = 256,
.category = FIO_OPT_C_IO,
.group = FIO_OPT_G_IO_BUF,