Re: [dpdk-dev] Memory leak in rte_pci_scan

From: David Marchand <david.marchand@redhat.com>
To: Owen Hilyard <ohilyard@iol.unh.edu>
Cc: dev <dev@dpdk.org>, dpdklab <dpdklab@iol.unh.edu>,
	Gaetan Rivet <grive@u256.net>,
	 Jerin Jacob Kollanukkaran <jerinj@marvell.com>,
	Aaron Conole <aconole@redhat.com>
Subject: Re: [dpdk-dev] Memory leak in rte_pci_scan
Date: Wed, 16 Jun 2021 11:37:29 +0200	[thread overview]
Message-ID: <CAJFAV8xB3m-aSbLqZHnsk5ka=_c==39bOokfShkNTEZt6R981Q@mail.gmail.com> (raw)
In-Reply-To: <CAHx6DYBNx40S_hMSYtSQYrqLzjdHKPAkHVEJwmVqW_AibyN=0w@mail.gmail.com>

On Tue, Jun 15, 2021 at 5:16 PM Owen Hilyard <ohilyard@iol.unh.edu> wrote:
>
> The issue may have been the interactive docker session I was running it in. The last few tests (150-157) were all taking until the timeout the lab uses for unit tests (2 hours since the timeout was multiplied by 10). I had to leave for the day so I restarted it in a non-interactive container and it ran in 2 hours. If we were to just run the fast-tests suite, then it would have taken 42 minutes to run. This is mostly due to timeouts in eal_flags_c_opt_autotest, eal_flags_hpet_autotest, eal_flags_misc_autotest and multiprocess_autotest, each taking 600 seconds. Finding out what caused these to stall would bring the runtime down to 3 minutes. All of the failures should be ASAN-related.

- The perf-tests testsuite is heavy and is probably not suited for per
patchset runs.
It may be worth running every once in a while but at the project/main
branches level.

- For the fast-tests testsuite, the default timeout should be 10s, not 600s.
See timeout_seconds_fast,
https://git.dpdk.org/dpdk/tree/app/test/meson.build#n446
Odd that a 600s timeout has been applied to fast-tests in your run.
How do you invoke meson?

- There is an interesting trace that shows issues with starting
secondary processes in this environment.
EAL: Detected 16 lcore(s)
EAL: Detected 2 NUMA nodes
EAL: Detected static linkage of DPDK
EAL: Multi-process socket
/var/run/dpdk/eal_flags_c_opt_autotest/mp_socket_264_50a7b93b648fa
EAL: Selected IOVA mode 'PA'
EAL: VFIO support initialized
EAL: Cannot get a virtual area at requested address: 0x7f1a8f400000
(got 0x7f19dc1fe000)
EAL: Cannot reserve 17179869184 bytes at [0x7f1a8f400000] - please use
'--base-virtaddr' option
EAL: Cannot preallocate VA space for hugepage memory
EAL: FATAL: Cannot init memory
EAL: Cannot init memory
EAL: Cannot destroy local memory map
EAL: Could not release memory subsystem data

It seems like there are multiple dpdk processes running in // in this
environment.
Any idea of what is happening on your system at the moment you tried
to run this test?

-- 
David Marchand