From mboxrd@z Thu Jan 1 00:00:00 1970 From: Cyril Hrubis Date: Tue, 1 Aug 2017 17:04:32 +0200 Subject: [LTP] [PATCH v3 00/15] Convert LTP mem/testcase to new API In-Reply-To: <20170718082312.3946-1-liwang@redhat.com> References: <20170718082312.3946-1-liwang@redhat.com> Message-ID: <20170801150432.GA3925@rei> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: ltp@lists.linux.it Hi! I've pushed this patches, after careful testing and with many changes, thanks. Most of the changes were more or less cosmetic, such as removing -1 handling after SAFE_FORK(), make use of more SAFE_MACROS() etc. But there were some problems as well. It introduced a compilation failure on 32bit for min_free_kbytes.c since it removed the total_mem variable that is used only on 32bit. The ksm06.c cleanup had to be changed, since in a case that merge_across_nodes is missing it attempted to run cleanup (previously there was NULL cleanup pointer for TCONF!) and failed the test. There is probably more unvanted cleanups possibly run for the ksm test, we should have a look at these later. I've also reverted the change to the thp01 in the runtest file since I belive that the value was reduced only for testing and slipped to the patchset by accident. I've changed the patch that adds possibility to disable timeouts, mainly so that the API is consistent, which means that the tst_set_timeout() takes -1 as parameter as well + wrote documentation for the change. Also unsolved is a problem with TCONF tests. At the moment newlib test fails if we pass any parameters to it, since the option parsing runs with the tst_test structure without the option stucture. This fails, for instance, ksm test if they were compiled out since we pass parameters to them in the runtest file. I will have to think about this a bit more and figure out a good solution. -- Cyril Hrubis chrubis@suse.cz