From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jan Stancek Date: Mon, 8 Oct 2018 10:25:19 -0400 (EDT) Subject: [LTP] [PATCH v4] syscalls/membarrier: Add membarrier() initial test In-Reply-To: <75dc7e79-ddc2-f5ee-4fb5-dfcb7385e4df@linaro.org> References: <243767154.55279006.1538383679362.JavaMail.zimbra@redhat.com> <20181008132914.8246-1-rafael.tinoco@linaro.org> <1320883984.56890565.1539007652586.JavaMail.zimbra@redhat.com> <75dc7e79-ddc2-f5ee-4fb5-dfcb7385e4df@linaro.org> Message-ID: <357830099.56894195.1539008719156.JavaMail.zimbra@redhat.com> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: ltp@lists.linux.it ----- Original Message ----- > On 10/8/18 11:07 AM, Jan Stancek wrote: > > > > > > ----- Original Message ----- > >> Fixes: #265 > >> > >> Initial test for membarrier() syscall. It tests all existing membarrier > >> "commands" (or features), including the need (or not) for previous > >> registration for the call to work. > >> > >> Some features did not exist in older kernels and that is covered by > >> skipping some calls, flagging test as skipped & okay, and forcing > >> others, making sure that return codes and errno are set right in those > >> cases. > >> > >> Tests are done in a child process due to inexistent kernel interface to > >> "unregister" the process from being affected by membarrier() call. > >> > >> Signed-off-by: Rafael David Tinoco > > > > Looks good to me. > > > > We probably can drop linux/membarrier.h configure check, right? > > (since v4 doesn't include it and unsupported kernel should > > hit ENOSYS and TCONF) > > Ooops, true. No need to check HAVE_LINUX_MEMBARRIER_H if we are > declaring all CMDs and will be given ENOSYS. Want me to send a v5 ? Let's give a couple days to other potential reviewers. If nothing else comes up, I can drop it before push. Thanks, Jan