open_posix_testsuite: RLIMIT_MEMLOCK is not specified in POSIX #96
Conversation
The RLIMIT_MEMLOCK resource is not defined in POSIX.
| munmap(pa, len); | ||
| printf("Test FAILED: Did not get EAGAIN as expected\n"); | ||
| return PTS_FAIL; | ||
| } |
There was a problem hiding this comment.
It's true that RLIMIT_MEMLOCK is not in POSIX albeit it's commonly available (at least on Linux and BSDs). What about keeping the test but keeping the real test inside #ifdef RLIMIT_MEMLOCK ifdefs and returning PTS_UNTESTED if RLIMIT_MEMLOCK wasn't defined?
There was a problem hiding this comment.
Yea, so it's testing a valid POSIX case, just using a Linux/BSD interface to do so. This POSIX behavior is marked as optional (ML) so I suppose ifdef'ing this and marking unresolved on other platforms is a fine compromise. I don't see any obvious POSIX-compliant method for inducing this behavior. Going to think on it for a little.
| ML [EAGAIN] The mapping could not be locked in memory, if required by mlockall ( ), due to | ||
| a lack of resources. | ||
| </assertion> | ||
|
|
||
| <assertion id="19" tag="ref:XSH6:25310:25342"> | ||
| The mmap( ) function shall fail if: | ||
| [EBADF] The fildes argument is not a valid open file descriptor. |
There was a problem hiding this comment.
There is nothing wrong about this assertion (it does not even mention the rlimit), hence there is no need to remove it.
The RLIMIT_MEMLOCK resource is not defined in POSIX.
This is a fix for #95.