On Mon, Jun 05, 2023 at 12:09:56PM -0700, John Hubbard wrote:
On 6/5/23 08:59, Peter Xu wrote: ...
-$(OUTPUT)/uffd-stress: uffd-common.c -$(OUTPUT)/uffd-unit-tests: uffd-common.c +$(OUTPUT)/uffd-stress: uffd-common.c +$(OUTPUT)/uffd-unit-tests: uffd-common.c +$(OUTPUT)/hugepage-mremap: uffd-common.c +$(OUTPUT)/write_to_hugetlbfs: uffd-common.c +$(OUTPUT)/ksm_functional_tests: uffd-common.c
Sorry, John, I still cannot follow..
As I said before uffd-common.[ch] was for uffd stress/unit tests. I confess my fault to not have named it uffd-test-common.[ch] already.
Actually, given that there is nothing *except* test code in this directory, I think your original choice of file names is just right.
I think it's fine to keep uffd_*() helpers in vm_util.[ch] for now, until it grows. Just like if one day we'll have a pagemap.c test we don't necessary need to move pagemap_*() helpers from vm_utils.[ch] into pagemap.[ch]. It just keeps common test helpers.
Can we avoid linking those into other tests in whatever way? Maybe renaming it to uffd-test-common.[ch] may be cleaner?
It sounds like you are suggesting this:
$(OUTPUT)/uffd-stress: uffd-common.c uffd-test-common.c $(OUTPUT)/uffd-unit-tests: uffd-common.c uffd-test-common.c $(OUTPUT)/hugepage-mremap: uffd-test-common.c $(OUTPUT)/write_to_hugetlbfs: uffd-test-common.c $(OUTPUT)/ksm_functional_tests: uffd-test-common.c
...approximately. Do I have that correct? I can arrange it that way if you feel it's a better end result. (And it's better than leaving uffd*() helpers in vm_utils, imho.)
Yes, as long as we don't link (especially) the uffd test specific globals into non-uffd test programs I'll have no issue. Thanks.