31a1b193a2
It is possible to specify io_size which is not a multiple of sector size and in that case real IO will have size of g_io_size_bytes/sector_size. It is integer division, so IO will be less than requested by the user. At the same time performance statistics are reported using io_size specified by the user (which is bigger than real IO size) so we have wrong statistics (e.g. we can see BW higher than NIC line rate). To avoid this confusing situation, add a check that user's io_size can be evenly divided by ns sector size. Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Change-Id: Ie63b5b7b3d99b5c51c10ec09e084fd30d6cb33dd Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/9733 Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com> Community-CI: Mellanox Build Bot Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com> |
||
---|---|---|
.. | ||
.gitignore | ||
Makefile | ||
perf.c | ||
README.md |
Compiling perf on FreeBSD
To use perf test on FreeBSD over NVMe-oF, explicitly link userspace library of HBA. For example, on a setup with Mellanox HBA,
LIBS += -lmlx5