Hi Phil, On 04.06.2016 09:35, Phil Sutter wrote: > [Cc'ing other interested parties, therefore full-quoting.] > On Fri, Jun 03, 2016 at 10:22:26PM +0200, Helge Deller wrote: >> I'm testing 4.7.0-rc1-64bit on a parisc/hppa machine and get >> those message with CONFIG_TEST_RHASHTABLE=y. >> >> Is this expected/normal ? > > No, this shouldn't happen. Can you possibly bisect the issue? I did some more testing. >> [ 11.004021] Running rhashtable test nelem=8, max_size=0, shrinking=0 >> [ 11.060013] Test 00: >> [ 11.181391] Adding 50000 keys >> [ 11.299533] Info: encountered resize >> [ 11.344499] Info: encountered resize >> [ 11.392971] Info: encountered resize >> [ 11.447632] Traversal complete: counted=50628, nelems=50000, entries=50000, table-jumps=3 >> [ 11.556024] Test failed: Total count mismatch ^^^ I still see this randomly on my boxes. Still need to check further. >> [ 11.647150] Traversal complete: counted=50000, nelems=50000, entries=50000, table-jumps=0 >> [ 11.760022] Deleting 50000 keys >> [ 11.822941] Duration of test: 596000000 ns >> [ 11.876057] Test 01: >> [ 11.919907] Adding 50000 keys >> [ 12.024027] Info: encountered resize >> [ 12.024767] Info: encountered resize >> [ 12.072981] Info: encountered resize >> [ 12.169265] Traversal complete: counted=53763, nelems=50000, entries=50000, table-jumps=3 >> [ 12.172021] Test failed: Total count mismatch ^^^ >> [ 12.377004] Traversal complete: counted=50000, nelems=50000, entries=50000, table-jumps=0 >> [ 12.380022] Deleting 50000 keys >> [ 12.552900] Duration of test: 596000000 ns >> [ 12.556069] Test 02: >> [ 12.650017] Adding 50000 keys >> [ 12.746381] Info: encountered resize >> [ 12.792486] Info: encountered resize >> [ 12.840977] Info: encountered resize >> [ 12.894604] Traversal complete: counted=52831, nelems=50000, entries=50000, table-jumps=3 >> [ 13.004025] Test failed: Total count mismatch ^^^ >> [ 13.094046] Traversal complete: counted=50000, nelems=50000, entries=50000, table-jumps=0 >> [ 13.204022] Deleting 50000 keys >> [ 13.269555] Duration of test: 576000000 ns >> [ 13.324062] Test 03: >> [ 13.366586] Adding 50000 keys >> [ 13.466624] Info: encountered resize >> [ 13.468255] Info: encountered resize >> [ 13.516485] Info: encountered resize >> [ 13.564977] Info: encountered resize >> [ 13.662159] Traversal complete: counted=52381, nelems=50000, entries=50000, table-jumps=4 >> [ 13.664022] Test failed: Total count mismatch ^^^ >> [ 13.870009] Traversal complete: counted=50000, nelems=50000, entries=50000, table-jumps=0 >> [ 13.876015] Deleting 50000 keys >> [ 14.045912] Duration of test: 640000000 ns >> [ 14.048065] Average test time: 602000000 >> [ 14.104017] Testing concurrent rhashtable access from 10 threads >> [ 14.537134] thread[6]: rhashtable_insert_fast failed >> [ 14.540722] thread[5]: rhashtable_insert_fast failed >> [ 14.540747] thread[1]: rhashtable_insert_fast failed >> [ 14.540757] thread[3]: rhashtable_insert_fast failed >> [ 14.540767] thread[0]: rhashtable_insert_fast failed >> [ 14.540862] thread[7]: rhashtable_insert_fast failed >> [ 14.540873] thread[9]: rhashtable_insert_fast failed >> [ 14.540874] thread[2]: rhashtable_insert_fast failed >> [ 14.540874] thread[4]: rhashtable_insert_fast failed >> [ 14.676061] Test failed: thread 0 returned: -12 >> [ 14.676102] Test failed: thread 1 returned: -12 >> [ 14.676120] thread[8]: rhashtable_insert_fast failed >> [ 14.676153] Test failed: thread 2 returned: -12 >> [ 14.676176] Test failed: thread 3 returned: -12 >> [ 14.676199] Test failed: thread 4 returned: -12 >> [ 14.676224] Test failed: thread 5 returned: -12 >> [ 15.516892] Test failed: thread 6 returned: -12 >> [ 15.644271] Test failed: thread 7 returned: -12 >> [ 15.708319] Test failed: thread 8 returned: -12 >> [ 15.772074] Test failed: thread 9 returned: -12 >> [ 15.832013] Started 10 threads, 10 failed Those -12 error messages seem to indicate, that there is memory pressure on my box (ENOMEM), right? That might be true for parisc right now as reported by Mikulas at http://www.spinics.net/lists/linux-parisc/msg07066.html and fixed by this patch: http://marc.info/?l=linux-mm&m=146468933302010&w=2 which I applied then. Additionally I enabled the enomem_retry kernel option. With those changes I get: [ 71.752666] Testing concurrent rhashtable access from 10 threads [ 72.548724] 8812 insertions retried after -ENOMEM [ 72.611804] 8799 insertions retried after -ENOMEM [ 72.674785] 8906 insertions retried after -ENOMEM [ 72.737836] 8895 insertions retried after -ENOMEM [ 72.800804] 9337 insertions retried after -ENOMEM [ 72.863843] 11122 insertions retried after -ENOMEM [ 72.927993] 11163 insertions retried after -ENOMEM [ 72.992219] 11121 insertions retried after -ENOMEM [ 73.056331] 11011 insertions retried after -ENOMEM [ 73.120492] 11121 insertions retried after -ENOMEM [ 73.621555] thread[4]: 8799 insertions retried due to memory pressure [ 73.709060] thread[5]: 8906 insertions retried due to memory pressure [ 73.804561] thread[3]: 8895 insertions retried due to memory pressure [ 73.894631] thread[1]: 9337 insertions retried due to memory pressure [ 73.988483] thread[0]: 11122 insertions retried due to memory pressure [ 74.087575] thread[7]: 11163 insertions retried due to memory pressure [ 74.180283] thread[6]: 11121 insertions retried due to memory pressure [ 74.276768] thread[2]: 11121 insertions retried due to memory pressure [ 74.367538] thread[9]: 11011 insertions retried due to memory pressure [ 74.472908] thread[8]: 8812 insertions retried due to memory pressure [ 79.281516] Started 10 threads, 0 failed I'll try to investigate further on this. Helge -- To unsubscribe from this list: send the line "unsubscribe linux-parisc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
