don't use zero data for the first file in RAW-OFFLINE

the most likely bugs in HSM involve zero data, due to the
dm_punch_hole() request
(This used to be commit 330ab956ea7e0b344450eee60b1357d854fbef28)

diff --git a/source4/torture/raw/offline.c b/source4/torture/raw/offline.c
index 9c66c3be9c955b2b62cfc51afc99a60d619b90a5..9240c76b6b67133fb8774cd62281737ecc086225 100644 (file)
--- a/source4/torture/raw/offline.c
+++ b/source4/torture/raw/offline.c
@@ -108,7 +108,7 @@ static void loadfile_callback(struct composite_context *ctx)
        }
 
        for (i=0;i<FILE_SIZE;i++) {
-               if (state->loadfile->out.data[i] != state->fnumber % 256) {
+               if (state->loadfile->out.data[i] != 1+(state->fnumber % 255)) {
                        printf("Bad data in file %u\n", state->fnumber);
                        test_failed++;
                        return;
@@ -436,7 +436,7 @@ bool torture_test_offline(struct torture_context *torture)
                char buf[FILE_SIZE];
                NTSTATUS status;
 
-               memset(buf, i % 256, sizeof(buf));
+               memset(buf, 1+(i % 255), sizeof(buf));
 
                fnum = smbcli_open(state[0].tree, fname, O_RDWR|O_CREAT, DENY_NONE);
                if (fnum == -1) {