Incorrect test for default_striping_factor
Originally by [email protected] on 2011-04-19 08:19:25 -0500
When the current ~mpich2-1.3.2p1/src/mpi/romio/test/file_info.c is run and the inquiry for the default_striping_factor returns a value of 2, an error message will always be printed about a striping_factor having an incorrect value. Just plug '2' into the equations to see that this is the case. The mod proposed below corrects this problem and is probably the code that was intended to be run. A couple of comments are also repaired or enhanced as well.
wilber~/MPICH2/functional/runrt/romio_io> diff file_info.1.3.2p1.mod.c file_info.1.3.2p1.orig.c
28c28
< * error, default will be to increment error count for true error conditions
---
> * error, default will be to increment errror cound for true error conditions
202c202
< on Lustre, Intel PFS and IBM PIOFS file systems and are ignored elsewhere.
---
> on Intel PFS and IBM PIOFS file systems and are ignored elsewhere.
253,256c253,255
< if (default_striping_factor - 1 > 0) {
< if (atoi(value) != default_striping_factor-1) {
< errs++;
< if (verbose) fprintf(stderr, "striping_factor is %d; should be %d\n",
---
> if ((default_striping_factor - 1 > 0) && (atoi(value) != default_striping_factor-1)) {
> errs++;
> if (verbose) fprintf(stderr, "striping_factor is %d; should be %d\n",
258d256
< }
260,263c258,260
< else {
< if (atoi(value) != default_striping_factor) {
< errs++;
< if (verbose) fprintf(stderr, "striping_factor IS %d; should be %d\n",
---
> else if (atoi(value) != default_striping_factor) {
> errs++;
> if (verbose) fprintf(stderr, "striping_factor is %d; should be %d\n",
265d261
< }
Originally by [email protected] on 2011-04-19 08:23:16 -0500
Attachment added: file_info.c (11.9 KiB)
mpicc -o file_info file_info.c && mpirun -n 2 ./file_info -fname tmp.txt
No Errors
Looks like it has been fixed?
Oh, this is about src/mpi/romio/test/file_info.c
@roblatham00 Could you take care of this (old) ticket?
I determined that the reported issue no longer exist. The check of default_striping_factor has been removed already.
This commit -- d2cb66671016753efd22f3293ed8984c97b91441 -- fixes it.