[mpich2-dev] Fw: [PATCH 21/21] Add read resized to the test.
Bob Cernohous
bobc at us.ibm.com
Fri Apr 10 17:15:52 CDT 2009
How about adding a read of the resized type to the test? Did I do this
right? It didn't work as written until I copied the ad_read_str.c changes
into ad_bgl, so I think I got this correct.
Signed-off-by: Bob Cernohous <bobc at us.ibm.com>
---
mpich2/test/mpi/io/resized.c | 18 +++++++++++++++++-
1 files changed, 17 insertions(+), 1 deletions(-)
diff --git a/mpich2/test/mpi/io/resized.c b/mpich2/test/mpi/io/resized.c
index 1abab1a..bba22c4 100644
--- a/mpich2/test/mpi/io/resized.c
+++ b/mpich2/test/mpi/io/resized.c
@@ -52,7 +52,7 @@ int main(int argc, char **argv)
MPI_File_delete(filename, MPI_INFO_NULL);
- /* create a resized type comprising an integer with an lb at 0 and
extent = 3*sizeof(int) */
+ /* create a resized type comprising an integer with an lb at
sizeof(int) and extent = 3*sizeof(int) */
lb = sizeof(int);
extent = 3*sizeof(int);
MPI_Type_create_resized(MPI_INT, lb, extent, &newtype);
@@ -82,6 +82,22 @@ int main(int argc, char **argv)
MPI_File_close(&fh);
+ /* read back file view with resized type and verify */
+
+ MPI_File_open(MPI_COMM_WORLD, filename, MPI_MODE_RDONLY,
MPI_INFO_NULL, &fh);
+
+ mpi_errno = MPI_File_set_view(fh, 0, MPI_INT, newtype, "native",
MPI_INFO_NULL);
+ if (mpi_errno != MPI_SUCCESS) errs++;
+
+ for (i=0; i<4; i++) newbuf[i] = 100;
+ MPI_File_read(fh, newbuf, 2, MPI_INT, MPI_STATUS_IGNORE);
+ if ((newbuf[0] != 10) || (newbuf[1] != 20) || (newbuf[2] !=100) ||
(newbuf[3] != 100)) {
+ errs++;
+ fprintf(stderr, "newbuf[0] is %d, should be 10,\n newbuf[1] is %d,
should be 20\n newbuf[2] is %d, should be 100,\n newbuf[3] is %d, should
be 100,\n", newbuf[0], newbuf[1], newbuf[2], newbuf[3]);
+ }
+
+ MPI_File_close(&fh);
+
/* read file back and verify */
MPI_File_open(MPI_COMM_WORLD, filename, MPI_MODE_DELETE_ON_CLOSE |
--
1.5.3.7
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mcs.anl.gov/mailman/private/mpich2-dev/attachments/20090410/59b8003a/attachment.htm>
More information about the mpich2-dev
mailing list