From: Jerin Jacob <jerin.jacob@caviumnetworks.com>
To: Thomas Monjalon <thomas.monjalon@6wind.com>
Cc: dev@dpdk.org
Subject: Re: [dpdk-dev] DPDK ArmV7 autotests
Date: Tue, 10 Nov 2015 18:15:55 +0530 [thread overview]
Message-ID: <20151110124551.GA17906@localhost.localdomain> (raw)
In-Reply-To: <7285692.g8AiFBK0OF@xps13>
On Tue, Nov 10, 2015 at 11:30:30AM +0100, Thomas Monjalon wrote:
> 2015-11-10 10:22, Hunt, David:
> > On 09/11/2015 17:46, Jan Viktorin wrote:
> > > Here is the log. You an see, that many tests fail just because
> > > of the missing hugetlb support. This is strange as I modified the
> > > autotest_runner.py to inject "--no-huge --no-pci" options when it
> > > detects armv7 architecture.
> >
> > I think publishing these test results to the mailing list confuses
> > matters somewhat. The tests are not tuned for armv7, and there are many
> > false negatives, giving the impression that the results are worse than
> > they actually should be.
> >
> > The tests need to be analysed to see if they respect the --no-huge and
> > --no-pci flags correctly, I suspect some tests are being run even though
> > these flags are enabled.
> >
> > I believe enough of the tests pass to allow the initial version of the
> > patch set to be accepted. We can look at the failing cases later, and
> > improve the test suite as time goes on.
>
> I agree.
> The unit tests are know to require some refactoring.
> I suggest to continue ARM integration while checking how to fix the
Make sense.
> unit tests later. Maybe that Declan has some ideas or plans about the tests.
I have tried autotest on armv8, results looks very promising.
Few more features need for armv8 port like lpm, optimized hash,
removing x86 dependencies from examples directory etc.
But these needs change in the common code
so IMO have existing code as base port and common code
changes we will add step by step otherwise it will be difficult manage.
root@arm64:/export/dpdk-armv8# python app/test/autotest.py build/app/test
build
build/app/test -c f -n 4
Test name Test result Test
Total
================================================================================
Start group_1: Success [00m 02s]
Timer autotest: Success [00m 41s]
Debug autotest: Success [00m 00s]
Errno autotest: Success [00m 00s]
Meter autotest: Success [00m 00s]
Common autotest: Success [00m 01s]
Dump log history: Success [00m 00s]
Dump rings: Success [00m 00s]
Dump mempools: Success [00m 00s][00m 46s]
Start group_2: Success [00m 02s]
Memory autotest: Success [00m 01s]
Read/write lock autotest: Success [00m 00s]
Logs autotest: Success [00m 00s]
CPU flags autotest: Success [00m 00s]
Version autotest: Success [00m 00s]
EAL filesystem autotest: Success [00m 00s]
EAL flags autotest: Fail(failes on x86 too) [00m 00s]
Hash autotest: Success [05m 04s][05m 55s]
Start group_3: Success [00m 02s]
LPM autotest: Fail [Not found] [00m 00s]
IVSHMEM autotest: Fail [Not found](NA for arm64)[00m 00s]
Memcpy autotest: Success [00m 21s]
Memzone autotest: Success [00m 00s]
String autotest: Success [00m 00s]
Alarm autotest: Fail(failes on x86 too) [00m 00s][06m 20s]
Start group_4: Success [00m 02s]
PCI autotest: Success [00m 00s]
Malloc autotest: Success [00m 00s]
Multi-process autotest: Fail(failes on x86 too) [00m 03s]
Mbuf autotest: Success [01m 04s]
Per-lcore autotest: Success [00m 05s]
Ring autotest: Success [00m 02s][07m 37s]
Start group_5: Success [00m 02s]
Spinlock autotest: Success [00m 15s]
Byte order autotest: Success [00m 00s]
TAILQ autotest: Success [00m 00s]
Command-line autotest: Success [00m 00s]
Interrupts autotest: Success [00m 08s][08m 03s]
Start group_6: Success [00m 02s]
Function reentrancy autotest: Success [00m 00s]
Mempool autotest: Success [00m 08s]
Atomics autotest: Success [00m 00s]
Prefetch autotest: Success [00m 00s]
Red autotest: Success [01m 35s][09m 50s]
Start group_7: Success [00m 02s]
PMD ring autotest: Fail(needs NICS provided on cmdline) [00m 00s]
Access list control autotest: Success [00m 02s]
Sched autotest: Success [00m 00s][09m 56s]
Start kni: Success [00m 02s]
KNI autotest: Success [09m 59s]
Start mempool_perf: Success [00m 02s]
Cycles autotest: Success [00m 01s]
Mempool performance autotest: Success [09m 00s][19m 02s]
Start memcpy_perf: Success [00m 02s]
Memcpy performance autotest: Fail [Timeout] [15m 00s][34m 05s]
Start hash_perf: Success [00m 02s]
Hash performance autotest: Success [03m 44s][37m 52s]
Start power: Success [00m 02s]
Power autotest: Success [00m 00s][37m 54s]
Start power_acpi_cpufreq: Success [00m 02s]
Power ACPI cpufreq autotest: Fail(failes on x86 too) [00m 00s][37m 57s]
Start power_kvm_vm: Success [00m 02s]
Power KVM VM autotest: Fail [00m 00s][37m 59s]
Start lpm6: Success [00m 02s]
LPM6 autotest: Fail [Not found] [00m 00s][38m 01s]
Start timer_perf: Success [00m 02s]
Timer performance autotest: Success [00m 14s][38m 18s]
Start ring_perf: Success [00m 02s]
Ring performance autotest: Success [00m 22s][38m 43s]
================================================================================
prev parent reply other threads:[~2015-11-10 12:46 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <5640CB56.4080506@intel.com>
2015-11-09 17:12 ` Jan Viktorin
2015-11-09 17:26 ` Hunt, David
2015-11-09 17:46 ` Jan Viktorin
2015-11-10 10:22 ` Hunt, David
2015-11-10 10:30 ` Thomas Monjalon
2015-11-10 12:45 ` Jerin Jacob [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20151110124551.GA17906@localhost.localdomain \
--to=jerin.jacob@caviumnetworks.com \
--cc=dev@dpdk.org \
--cc=thomas.monjalon@6wind.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).