Closed edwardhartnett closed 1 year ago
BTW, I noticed in the developer CI run that a number of your new ufbstp tests were failing due to a SIGSEGV, rather than failing within the bort routine like we want:
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Also, at least two new tests are failing with memory leaks:
1: ==6968==ERROR: LeakSanitizer: detected memory leaks
1:
1: Direct leak of 32 byte(s) in 1 object(s) allocated from:
1: #0 0x7fb0dc424867 in __interceptor_malloc ../../../../src/libsanitizer/asan/asan_malloc_linux.cpp:145
1: #1 0x7fb0dc0b7ef8 (/lib/x86_64-linux-gnu/libgfortran.so.5+0x22ef8)
1:
1: Indirect leak of 512 byte(s) in 1 object(s) allocated from:
1: #0 0x7fb0dc424867 in __interceptor_malloc ../../../../src/libsanitizer/asan/asan_malloc_linux.cpp:145
1: #1 0x7fb0dc0b7ef8 (/lib/x86_64-linux-gnu/libgfortran.so.5+0x22ef8)
1:
1: SUMMARY: AddressSanitizer: 544 byte(s) leaked in 2 allocation(s).
Of course it can be really hard to notice these things with test_bort, since the roles are reversed and a fail now gets interpreted as a pass. So I think we really need to get in the habit of digging through the developer run output to check for these.
OK, let's separate the bort testing from the documentation changes, in order to divide and conquer this PR. ;-)
I have put the tests (after fixing the ufbstp() seg faults) in new PR #422 .
Once those are merged I will circle around again to this PR. We will have a documentation meeting today to discuss some of the issues with the documentation changes in this PR.
The testing work in the PR has been submitted in #422
We will discuss some documentation issues at today's meeting, and then I will resumbit some of the documentation work as future PRs...
Part of #381 Part of #397
Testing more borts().