[lttng-dev] [PATCH 0/3] rculfhash: error checking fixes

Eric Wong normalperson at yhbt.net
Mon Jun 23 21:20:29 EDT 2014


Hi Mathieu, I noticed the pthread_create handling is a bit dangerous in
rculfhash since it may kill the process long after the process is up and
running.  I also fixed the calloc failure handling since I was in the
area.

In the future, it would also be nice if the alloc_bucket_table behavior
could fail gracefully on ENOMEM, but that would break ABI...

The following changes since commit d863e15371666dff14e7f3191de5ff91d9b24d7f:

  Cleanup: tests: cast console write return value as void (2014-05-13 09:27:36 -0400)

are available in the git repository at:

  git://80x24.org/urcu rculfhash-errcheck

for you to fetch changes up to d54dd3e6ada2edf4599fc0f26e647a6942a73a39:

  rculfhash: remove duplicated code (2014-06-24 00:39:26 +0000)

----------------------------------------------------------------
Eric Wong (3):
      rculfhash: fall back to single-threaded resize on calloc failure
      rculfhash: handle pthread_create failures
      rculfhash: remove duplicated code

 rculfhash.c | 49 +++++++++++++++++++++++++++++++++----------------
 1 file changed, 33 insertions(+), 16 deletions(-)

P.S. I still hope to resume the wfcqueue-related epoll improvements
     we started on last year in the kernel.



More information about the lttng-dev mailing list