Bug #14963: Since r64107, on Solaris, spec test of rb_io_wait_readable hangs up

• Status: Closed
• Assignee: -
• Target version: -
• Backport: 2.3: UNKNOWN, 2.4: UNKNOWN, 2.5: UNKNOWN
• Tags: thread, solaris

[ngoto (Naohisa Goto)]

Since r64107 (removal of timer thread), on Solaris (both on Solaris 10 and Solaris 11, both on x86 and sparc),
make test-rubyspec never ends in the test of rb_io_wait_readable in spec/ruby/optional/capi/io_spec.rb.

C-API IO function rb_io_wait_writable
- returns false if there is no error condition (FAILED - 1)
- raises an IOError if the IO is closed

C-API IO function rb_thread_fd_writable
- waits til an fd is ready for writing

C-API IO function rb_io_wait_readable
- returns false if there is no error condition

On rubyci.org, it stopped by 600s timeout.
(confirmed with r64171 or r64169)

Solaris10 i386: https://rubyci.org/logs/rubyci.s3.amazonaws.com/unstable10x/ruby-trunk/log/20180803T121806Z.fail.html.gz
Solaris11 i386: https://rubyci.org/logs/rubyci.s3.amazonaws.com/unstable11x/ruby-trunk/log/20180803T122406Z.fail.html.gz
Solaris10 sparc: https://rubyci.org/logs/rubyci.s3.amazonaws.com/unstable10s/ruby-trunk/log/20180803T071912Z.fail.html.gz
Solaris11 sparc: https://rubyci.org/logs/rubyci.s3.amazonaws.com/unstable11s/ruby-trunk/log/20180803T112505Z.fail.html.gz

I cannot determine whether the changes of removing timer thread contain some bug on Solaris or the test code is essentially wrong.

[hsbt (Hiroshi SHIBATA)]

• Tags set to solaris, thread

[jeremyevans0 (Jeremy Evans)]

• Status changed from Open to Closed

r64107 (708bfd21156828526fe72de2cedecfaca6647dc1) was reverted at [ruby-core:88306] Revert "process.c: ensure t..." href="/projects/ruby-master/repository/git/revisions/194a6a2c68e9c8a3536b24db18ceac87535a6051">194a6a2c68e9c8a3536b24db18ceac87535a6051, and the timer thread still exists, so I think this can be closed.