posix: more tests & minor fixes

[mslusarz]

This change is 

[codecov[bot]]

Codecov Report

Merging #431 into master will increase coverage by 2.84%. The diff coverage is 80.76%.

@@            Coverage Diff             @@
##           master     #431      +/-   ##
==========================================
+ Coverage   75.95%   78.79%   +2.84%     
==========================================
  Files          70       70              
  Lines        8815     8796      -19     
  Branches     1781     1772       -9     
==========================================
+ Hits         6695     6931     +236     
+ Misses       1628     1424     -204     
+ Partials      492      441      -51

Flag	Coverage Δ
#ltp_tests	45.15% <38.46%> (+0.4%)	:arrow_up:
#sqlite_tests	45.13% <50%> (+0.26%)	:arrow_up:
#tests_antool	15.08% <0%> (+0.03%)	:arrow_up:
#tests_posix_multi_threaded	25.68% <23.07%> (+0.14%)	:arrow_up:
#tests_posix_single_threaded	56.87% <80.76%> (+3.37%)	:arrow_up:
#tests_preload	44.69% <50%> (+0.27%)	:arrow_up:

Impacted Files	Coverage Δ
src/libpmemfile-posix/utils.h	80% <ø> (ø)	:arrow_up:
src/libpmemfile-posix/utils.c	68.51% <0%> (+0.18%)	:arrow_up:
src/libpmemfile-posix/write.c	87.55% <100%> (+3.05%)	:arrow_up:
src/libpmemfile-posix/inode.c	89.07% <100%> (ø)	:arrow_up:
src/libpmemfile-posix/rmdir.c	94.94% <100%> (+2.87%)	:arrow_up:
src/libpmemfile-posix/statfs.c	91.66% <100%> (+16.66%)	:arrow_up:
src/libpmemfile-posix/read.c	94.44% <100%> (+1.97%)	:arrow_up:
src/libpmemfile-posix/chown.c	98.97% <100%> (ø)	:arrow_up:
src/libpmemfile-posix/timestamps.c	93.71% <100%> (+2.4%)	:arrow_up:
src/libpmemfile-posix/rename.c	91.9% <100%> (+0.77%)	:arrow_up:
... and 16 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 7f181b5...7051b18. Read the comment docs.

[krzycz]

Reviewed 16 of 16 files at r1. Review status: all files reviewed at latest revision, all discussions resolved.

---

Comments from Reviewable

[GBuella]

Review status: all files reviewed at latest revision, 1 unresolved discussion, all commit checks successful.

---

src/libpmemfile-posix/utils.c, line 50 at r1 (raw file):

  pmemfile_timespec_t tm;
  if (clock_gettime(CLOCK_REALTIME, &tm)) {
      ERR("!clock_gettime");

Looks like overkill. I'm thinking it is probably pointless to prepare for failures of clock_gettime. According to the docs, CLOCK_REALTIME should be supported everywhere, so the only possible error would be EFAULT. The buffer is on the stack, declared in the line right before the clock_gettime call. The fault injection code for this returns EINVAL, which would suggest here that CLOCK_REALTIME is supported sometimes, and not supported perhaps a few microseconds later -- seems like pointless to prepare for this scenario, I mean, if that happens, then maybe adding 2 + 2 will result in 5, and we don't check for that either.

---

Comments from Reviewable

[mslusarz]

Review status: all files reviewed at latest revision, 1 unresolved discussion.

---

src/libpmemfile-posix/utils.c, line 50 at r1 (raw file):

Previously, GBuella (Gabor Buella) wrote…

Looks like overkill. I'm thinking it is probably pointless to prepare for failures of clock_gettime. According to the docs, `CLOCK_REALTIME` should be supported everywhere, so the only possible error would be EFAULT. The buffer is on the stack, declared in the line right before the clock_gettime call. The fault injection code for this returns EINVAL, which would suggest here that CLOCK_REALTIME is supported sometimes, and not supported perhaps a few microseconds later -- seems like pointless to prepare for this scenario, I mean, if that happens, then maybe adding 2 + 2 will result in 5, and we don't check for that either.

Ok. I handled this failure with an expectation that this function may need to be changed in the future. But I just realized none of those changes would mean that this function could fail...

---

Comments from Reviewable

[GBuella]

---

Reviewed 5 of 16 files at r1, 17 of 20 files at r2. Review status: all files reviewed at latest revision, all discussions resolved.

---

Comments from Reviewable