Fix several raidz_test issues

[amotin]

Trying to investigate CI raidz_001_neg failures on FreeBSD 16 I ended up with few patches.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Performance enhancement (non-breaking change which improves efficiency)
• Code cleanup (non-breaking change which makes code smaller or more readable)
• Quality assurance (non-breaking change which makes the code more robust against bugs)
• Breaking change (fix or feature that would cause existing functionality to change)
• Library ABI change (libzfs, libzfs_core, libnvpair, libuutil and libzfsbootenv)
• Documentation (a change to man pages or other documentation)

Checklist:

• My code follows the OpenZFS code style requirements.
• I have updated the documentation accordingly.
• I have read the contributing document.
• I have added tests to cover my changes.
• I have run the ZFS Test Suite with this change applied.
• All commit messages are properly formatted and contain Signed-off-by.

[robn]

Yeah, this is all pretty sensible. What a funny little program.

[robn]

Yeah, you're right about using mprotect() on memory obtained from libc (technically, wherever umem_alloc() got it from, which is kinda worse maybe?). POSIX explicitly says calling it on a pointer you didn't get from mmap() is unspecified, so.

I don't care for this PR, since we were already doing and your change is an improvement, but if you like we could just use an anonymous mapping:

diff --git a/cmd/raidz_test/raidz_test.c b/cmd/raidz_test/raidz_test.c
index 4839e909e4..eb1d6a562a 100644
--- a/cmd/raidz_test/raidz_test.c
+++ b/cmd/raidz_test/raidz_test.c
@@ -820,7 +820,8 @@ main(int argc, char **argv)
 	kernel_init(SPA_MODE_READ);
 
 	/* setup random data because rand() is not reentrant */
-	rand_data = (int *)umem_alloc(SPA_MAXBLOCKSIZE, UMEM_NOFAIL);
+	rand_data = (int *)mmap(NULL, SPA_MAXBLOCKSIZE, PROT_READ|PROT_WRITE,
+	    MAP_ANONYMOUS|MAP_PRIVATE, -1, 0);
 	srand((unsigned)time(NULL) * getpid());
 	for (i = 0; i < SPA_MAXBLOCKSIZE / sizeof (int); i++)
 		rand_data[i] = rand();
@@ -835,7 +836,7 @@ main(int argc, char **argv)
 		err = run_test(NULL);
 	}
 
-	umem_free(rand_data, SPA_MAXBLOCKSIZE);
+	munmap(rand_data, SPA_MAXBLOCKSIZE);
 	kernel_fini();
 
 	return (err);

[behlendorf]

Yup, assuming using an anonymous mapping works as hoped this does seems preferable. Still I don't think we need to hold up this PR on that. We can always follow up with this.