On 22/03/2023 18:04, Luis Chamberlain wrote:

On Tue, Mar 21, 2023 at 08:23:50PM -0700, Luis Chamberlain wrote:

quoted

Example uses:

sudo ./stress-ng --module 1 --module-name xfs
sudo ./stress-ng --module 1 --module-name xfs --module-sharedfd

The use case with --module 8192 was causing some errors from
stress-ng having unexpected bail out messages before ramp up.

quoted

diff --git a/stress-module.c b/stress-module.c
new file mode 100644
index 00000000..cee581bd
--- /dev/null
+++ b/stress-module.c
+			//snprintf(module_path, strlen(module_path), "%s/%s/%s",
+			snprintf(module_path, PATH_MAX*2, "%s/%s/%s",
+				 dirname_default_prefix,
+				 u.release, module);
+			ret = 0;

I forgot to remove this stray comment.

No worries, I can fix that up when I apply the patch.

quoted

+	/*
+	 * We're not stressing the modules.dep --> module path lookup,
+	 * just the finit_module() calls and so only do the lookup once.
+	 */
+	if (args->instance != 0) {
+		if (!module_path_found)
+			return EXIT_SUCCESS;
+	}

So here was the reason for the complaints, Although changing this to
return just EXIT_NO_RESOURCE cures the warning, I don't think the
non instance 0 workers are doing anything then. Is that right Colin?

I'll have a look at that when I test this out later tonight.

   Luis