From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jan Stancek Date: Wed, 15 May 2019 05:03:55 -0400 (EDT) Subject: [LTP] LTP: mm: overcommit_memory01, 03...06 failed In-Reply-To: References: Message-ID: <543317293.22835729.1557911035979.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 ----- > ltp-mm-tests failed on Linux mainline kernel 5.1.0, > * overcommit_memory01 overcommit_memory > * overcommit_memory03 overcommit_memory -R 30 > * overcommit_memory04 overcommit_memory -R 80 > * overcommit_memory05 overcommit_memory -R 100 > * overcommit_memory06 overcommit_memory -R 200 > > mem.c:814: INFO: set overcommit_memory to 0 > overcommit_memory.c:185: INFO: malloc 8094844 kB successfully > overcommit_memory.c:204: PASS: alloc passed as expected > overcommit_memory.c:189: INFO: malloc 32379376 kB failed > overcommit_memory.c:210: PASS: alloc failed as expected > overcommit_memory.c:185: INFO: malloc 16360216 kB successfully > overcommit_memory.c:212: FAIL: alloc passed, expected to fail > > Failed test log, > https://lkft.validation.linaro.org/scheduler/job/726417#L22852 > > LTP version 20190115 > > Test case link, > https://github.com/linux-test-project/ltp/blob/master/testcases/kernel/mem/tunable/overcommit_memory.c#L212 > > First bad commit: > git branch master > git commit e0654264c4806dc436b291294a0fbf9be7571ab6 > git describe v5.1-10706-ge0654264c480 > git repo https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git > > Last good commit: > git branch master > git commit 7e9890a3500d95c01511a4c45b7e7192dfa47ae2 > git describe v5.1-10326-g7e9890a3500d > git repo https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git Heuristic changed a bit in: commit 8c7829b04c523cdc732cb77f59f03320e09f3386 Author: Johannes Weiner Date: Mon May 13 17:21:50 2019 -0700 mm: fix false-positive OVERCOMMIT_GUESS failures LTP tries to allocate "mem_total + swap_total": alloc_and_check(sum_total, EXPECT_FAIL); which now presumably falls short to trigger failure. > > Best regards > Naresh Kamboju >