Hi Shijith,

> -----Original Message-----
> From: Shijith Thotton <sthot...@marvell.com>
> Sent: Tuesday, February 15, 2022 9:39 PM
> To: Gujjar, Abhinandan S <abhinandan.guj...@intel.com>; dev@dpdk.org;
> Jerin Jacob Kollanukkaran <jer...@marvell.com>
> Subject: RE: [PATCH v4] app/eventdev: add crypto producer mode
> 
> Hi Abhinandan,
> 
> >>
> >> In crypto producer mode, producer core enqueues cryptodev with
> >> software generated crypto ops and worker core dequeues crypto
> >> completion events from the eventdev. Event crypto metadata used for
> >> above processing is pre- populated in each crypto session.
> >>
> >> Parameter --prod_type_cryptodev can be used to enable crypto producer
> >> mode. Parameter --crypto_adptr_mode can be set to select the crypto
> >> adapter mode, 0 for OP_NEW and 1 for OP_FORWARD.
> >>
> >> This mode can be used to measure the performance of crypto adapter.
> >>
> >> Example:
> >>   ./dpdk-test-eventdev -l 0-2 -w <EVENTDEV> -w <CRYPTODEV> -- \
> >>   --prod_type_cryptodev --crypto_adptr_mode 1 --test=perf_atq \
> >>   --stlist=a --wlcores 1 --plcores 2
> >>
> >> Signed-off-by: Shijith Thotton <sthot...@marvell.com>
> >> ---
> >
> ><SNIP>
> >Reposting your comments from previous patch:
> >I tried adding support for software adapter implementation, but is
> >getting a crash in sw_event PMD after some packets. I have posted the
> >respective changes
> >here: https://urldefense.proofpoint.com/v2/url?u=https-
> >3A__patchwork.dpdk.org_project_dpdk_patch_0677cbafa5145f1b9f64dd0
> 07594e
> >033f2d9ab8a.1644337310.git.sthotton-
> >40marvell.com_&d=DwIFAg&c=nKjWec2b6R0mOyPaz7xtfQ&r=G9w4KsPaQ
> LACBf
> >GCL35PtiRH996yqJDxAZwrWegU2qQ&m=sMNIOpMhfpihE2tsEnuoFNOZNb
> V5fctS
> >mXD33o-
> >7RC1A1o2s6on3bh2npzmlCR4e&s=xitJsdeoR3OqFEkVzPdJMVrU3xbkF3Qs0
> x2k2Pgi
> >jeE&e=
> >Please take it forward.
> >
> >Command used to test is:
> >dpdk-test-eventdev -l 0-8 -s 0xf0 --vdev=event_sw0  --vdev="crypto_null" --
> \
> >      --prod_type_cryptodev --crypto_adptr_mode 1 --test=perf_queue --
> stlist=a \
> >      --wlcores 1 --plcores 2
> >
> >I tried to get the above command working with your patches. As you
> >mentioned, I see a crash! If I change the --test to any other queue, I
> >see a totally irrelevant
> 
> I have mentioned the same in my previous message. The crash is in event_sw
> PMD.
> 
> Thread 10 "lcore-worker-7" received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0x7ffff3767400 (LWP 1140492)] 0x00005555569f2c8a in
> sw_event_schedule ()
> (gdb) bt
> #0  0x00005555569f2c8a in sw_event_schedule ()
> #1  0x0000555556a05e3d in sw_sched_service_func ()
> #2  0x000055555632ada3 in service_runner_func ()
> #3  0x0000555555749b4b in eal_thread_loop.cold ()
> #4  0x00007ffff7a0f609 in start_thread (arg=<optimized out>) at
> pthread_create.c:477
> #5  0x00007ffff7936293 in clone () at
> ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
> 
> So requested to take it forward as I'm not well versed in SW PMD
> implementation.
I can't take it forward. Please check with PMD maintainer.
> 
> >error message:
> >error: pipeline_opt_check() Invalid producer type 'Event crypto
> >adapter' valid producer 'Ethdev Rx Adapter'
> >error: main() invalid command line argument
> >
> 
> Looks like, the test used is pipeline (from the log). Only --prod_type_ethdev
> is supported by the pipeline test. Similar error can be found for all other
> modes.
> 
> Refer doc/guides/tools/testeventdev.rst for the list of supported options for
> each test.
> 
> >I think, running a test with minimal configuration like sw_event, sw
> >adapter & null cryptodev without any vendor specific HW is very
> >important to acknowledge a patch. Later, this code will be a reference
> >for adding performance test for any specific HW.
> >Right now, I can't think of any other options to see a working test.
> >If you know any other ways to test this code, please let me.
> >
> 
> Please let me know if any change is needed in the application side. I will 
> take
> a look.
Right now, the app is not working as expected.
Please debug and fix the issue.
> 
> Thanks,
> Shijith

Reply via email to