Well, after spending some time looking into this and doing some /tmp cleanup and more rebooting everything seems to be working for me again for both l2fwd_simple and this program. However I do see this:
bill@Ubuntu15:~/linaro/review/test/common_plat/performance$ ./odp_pktio_ordered_run.sh using PCAP_IN = ./udp64.pcap _ishm.c:859:_odp_ishm_reserve():No huge pages, fall back to normal pages. check: /proc/sys/vm/nr_hugepages. PKTIO: initialized loop interface. PKTIO: initialized pcap interface. PKTIO: initialized socket mmap, use export ODP_PKTIO_DISABLE_SOCKET_MMAP=1 to disable. PKTIO: initialized socket mmsg,use export ODP_PKTIO_DISABLE_SOCKET_MMSG=1 to disable. ODP system info --------------- ODP API version: 1.12.0 ODP impl name: odp-linux CPU model: Intel(R) Core(TM) i7-4790K CPU CPU freq (hz): 4000000000 Cache line size: 64 CPU count: 6 Running ODP appl: "odp_pktio_ordered" ----------------- IF-count: 2 Using IFs: pcap:in=./udp64.pcap pcap:out=pcapout.pcap Num worker threads: 4 First CPU: 2 CPU mask: 0x3C Pool info --------- pool 0 name packet pool pool type packet pool shm 10 user area shm 11 num 8192 align 64 headroom 66 data size 8126 max data len 48756 max seg len 8126 tailroom 0 block size 8896 uarea size 64 shm size 72876032 base addr 0x7fad4a243000 uarea shm size 524288 uarea base addr 0x7fad80c7b000 Created pktio 1 (pcap:in=./udp64.pcap) Allocating 1 shared output queues, 12 requested Created 1 input and 1 output queues on (pcap:in=./udp64.pcap) pktio handle 1 name pcap:in=./udp64.pcap type pcap state opened mac 02:e9:34:80:73:04 mtu 65536 promisc yes max input queues 1 max output queues 1 Created pktio 2 (pcap:out=pcapout.pcap) Allocating 1 shared output queues, 12 requested Created 1 input and 1 output queues on (pcap:out=pcapout.pcap) pktio handle 2 name pcap:out=pcapout.pcap type pcap state opened mac 02:e9:34:80:73:04 mtu 65536 promisc yes max input queues 1 max output queues 1 Application parameters ---------------------- Input queues: 1 Mode: PKTIN_SCHED_ORDERED Flows: 1 Extra rounds: 15 linux.c:273:odpthread_run_start_routine():helper: ODP worker thread started as linux pthread. (pid=22302) linux.c:273:odpthread_run_start_routine():helper: ODP worker thread started as linux pthread. (pid=22302) linux.c:273:odpthread_run_start_routine():helper: ODP worker thread started as linux pthread. (pid=22302) linux.c:273:odpthread_run_start_routine():helper: ODP worker thread started as linux pthread. (pid=22302) 200 pps, 200 max pps, 0 rx drops, 0 tx drops, 0 invalid seq 0 pps, 200 max pps, 0 rx drops, 0 tx drops, 0 invalid seq TEST RESULT: 200 maximum packets per second. _ishm.c:1556:_odp_ishm_term_global():block 'shm_args' (file ) was never freed (cleaning up...). bill@Ubuntu15:~/linaro/review/test/common_plat/performance$ So it looks like an shm area is not properly freed as part of termination processing. On Tue, Jan 3, 2017 at 9:36 AM, Christophe Milard <christophe.mil...@linaro.org> wrote: > The problem seems to be the name of the shm memory block: > "pktio-pcap:in=./udp64.pcap" > The name of the block is used for creating the filename used to get > the file descriptor. Filenames containing "/" and other strange > characters will be problems. > I guess I can create other "random" filenames, but I thought that > getting the file name as its underlying block would be nice. Another > alternative would be to convert strange char to underscore, but that > would bug if 2 blocks are distinct only by some strange char ,e.g. > "test" and "test/" > Or we could restrict block names to "identifier-like" syntax. I think > this makes more sense (we are programmers). > Anyway, the API should be clear on what is allowed. I guess we don't > want control characters or international characters (like öäå) in > there... do we? > > On 3 January 2017 at 13:44, Bill Fischofer <bill.fischo...@linaro.org> wrote: >> On Tue, Jan 3, 2017 at 12:51 AM, Elo, Matias (Nokia - FI/Espoo) >> <matias....@nokia-bell-labs.com> wrote: >>> >>>> I see the same sort of behavior with l2fwd_simple, so I'm inclined to >>>> agree with you. I've verified that libpcap-dev is installed. What more >>>> should be needed to run these? >>>> >>> >>> Libpcap-dev should be enough to run these tests. Based on the error log I'm >>> guessing shm is causing this. I would try cleaning /tmp, check that there >>> aren’t any other ODP processes running in the background, or reboot the >>> system. >> >> After doing both of these I get some more detail: >> >> bill@Ubuntu15:~/linaro/review/example/l2fwd_simple$ ./l2fwd_simple_run.sh >> using PCAP_IN = ./udp64.pcap >> _ishm.c:859:_odp_ishm_reserve():No huge pages, fall back to normal >> pages. check: /proc/sys/vm/nr_hugepages. >> PKTIO: initialized loop interface. >> PKTIO: initialized socket mmap, use export >> ODP_PKTIO_DISABLE_SOCKET_MMAP=1 to disable. >> PKTIO: initialized socket mmsg,use export >> ODP_PKTIO_DISABLE_SOCKET_MMSG=1 to disable. >> pktio/socket_mmap.c:78:mmap_pkt_socket():socket(SOCK_RAW): Operation >> not permitted >> _ishm.c:451:create_file():open failed for >> /tmp/odp-3150-ishm-pktio-pcap:in=./udp64.pcap: No such file or >> directory. >> _ishm.c:891:_odp_ishm_reserve():_ishm_reserve failed. >> odp_packet_io.c:224:setup_pktio_entry():Unable to init any I/O type. >> Failed to open pcap:in=./udp64.pcap >> ./l2fwd_simple_run.sh: line 15: kill: (3150) - No such process >> Error: status was: 1, expected 0 >> >> +cc Christophe as the _ishm msgs may be a clue. My recollection is >> that the last time I successfully ran pcap stuff we prior to the >> integration of the _ishm code. >> >>> >>> -Matias >>>
