netperf: update netperf to make it work with gcc 10 and above

[yanan-fu]

patch HewlettPackard/netperf#46
to the existing netperf

id: 1917877

Signed-off-by: Yanan Fu [email protected]

[yanan-fu]

Test result:

1. For cases that use netperf-2.6.0.tar.bz2

22:13:05 INFO | Copy file from /var/lib/avocado/data/avocado-vt/virttest/test-providers.d/downloads/io-github-autotest-qemu/deps/netperf/netperf-2.6.0.tar.bz2 to 10.73.224.76:/var/tmp/, elapsed time: 0.9154536724090576
22:13:05 DEBUG| Compiling netserver from source
22:13:05 DEBUG| Sending command: arch
22:13:05 DEBUG| Sending command: rm -rf /var/tmp/netperf-2.6.0
22:13:05 DEBUG| Sending command: echo $?
22:13:05 DEBUG| Sending command: cd /var/tmp/  && tar jxf netperf-2.6.0.tar.bz2 && cd /var/tmp/netperf-2.6.0  && ./autogen.sh > /dev/null 2>&1 && ./configure --build=x86_64  > /dev/null 2>&1 && make > /dev/null 2>&1
22:13:14 DEBUG| Sending command: echo $?
22:13:14 DEBUG| Using local netperf: /var/tmp/netperf-2.6.0/src/netperf and /var/tmp/netperf-2.6.0/src/netserver
22:13:14 DEBUG| Create remote session
22:13:16 INFO | Start netserver ...
22:13:16 DEBUG| Sending command (safe): ps -C netserver
22:13:16 INFO | Start netserver with cmd: '/var/tmp/netperf-2.6.0/src/netserver > /dev/null'
22:13:16 DEBUG| Sending command (safe): /var/tmp/netperf-2.6.0/src/netserver > /dev/null
22:13:16 DEBUG| Sending command (safe): ps -C netserver
22:13:16 INFO | Netserver start successfully
...

2. For cases that use netperf-2.7.1.tar.bz2

22:16:10 INFO | Copy file from /var/lib/avocado/data/avocado-vt/virttest/test-providers.d/downloads/io-github-autotest-qemu/deps/netperf/netperf-2.7.1.tar.bz2 to 10.73.224.76:/var/tmp/, elapsed time: 0.8781707286834717
22:16:10 DEBUG| Compiling netserver from source
22:16:10 DEBUG| Sending command: arch
22:16:10 DEBUG| Sending command: rm -rf /var/tmp/netperf-2.7.1
22:16:10 DEBUG| Sending command: echo $?
22:16:10 DEBUG| Sending command: cd /var/tmp/  && tar jxf netperf-2.7.1.tar.bz2 && cd /var/tmp/netperf-2.7.1  && ./autogen.sh > /dev/null 2>&1 && ./configure --build=x86_64  > /dev/null 2>&1 && make > /dev/null 2>&1
22:16:19 DEBUG| Sending command: echo $?
22:16:20 DEBUG| Using local netperf: /var/tmp/netperf-2.7.1/src/netperf and /var/tmp/netperf-2.7.1/src/netserver
22:16:20 DEBUG| Create remote session
22:16:21 INFO | Start netserver ...
22:16:21 DEBUG| Sending command (safe): ps -C netserver
22:16:21 DEBUG| Sending command (safe): ps -C netserver
22:16:21 INFO | Netserver start successfully
...

[yanan-fu]

@yanglei-rh @PaulYuuu Could you help review ? Thanks!

[PaulYuuu]

@yfu generic/deps/performance/ has netperf tar archives as well.

# ls generic/deps/performance/
netperf-2.6.0.tar.bz2  netperf-2.7.1.tar.bz2

But in order to maintain one set of netperf archives, do you also plan to make soft links? If no, let @yanglei-rh find out which scripts use performance/netperf* and update to deps/netperf/netperf* in the next patch.

[yanan-fu]

@PaulYuuu It is useless from my checking, so i leave it.
Several cases use performance, but not netperf under performance. It is MoonGen.zip, fio and some test_patch for performance test.

[PaulYuuu]

@yanan-fu , thanks for the explanation. If they are indeed useless, let's delete them to leave disagreement. Waiting for the feedback of feature owners.

[yanan-fu]

@yanan-fu , thanks for the explanation. If they are indeed useless, let's delete them to leave disagreement. Waiting for the feedback of feature owners.

Delete useless pkg is another topic, let's focus on if netperf under generic/deps/performance/ needed be updated or not in this patch.
@quanwenli @yanglei-rh Could you please check it ? Thanks!

[yanglei-rh]

@yanan-fu , thanks for the explanation. If they are indeed useless, let's delete them to leave disagreement. Waiting for the feedback of feature owners.

Delete useless pkg is another topic, let's focus on if netperf under generic/deps/performance/ needed be updated or not in this patch.
@quanwenli @yanglei-rh Could you please check it ? Thanks!

Hi @yanan-fu I double-checked that the netperf package I used comes from deps/netperf/. I will add the test results later.

[yanglei-rh]

LGTM

Test pass
netperf-2.6.0:
(1/2) Host_RHEL.m8.u4.product_av.qcow2.virtio_scsi.up.virtio_net.Guest.RHEL.8.4.0.x86_64.io-github-autotest-qemu.nic_opt.opt_test_netperf.indirect_desc.opt_on.q35: PASS (644.11 s)
(2/2) Host_RHEL.m8.u4.product_av.qcow2.virtio_scsi.up.virtio_net.Guest.RHEL.8.4.0.x86_64.io-github-autotest-qemu.nic_opt.opt_test_netperf.indirect_desc.opt_off.q35: PASS (623.18 s)
netperf-2.7.1:
(1/2) Host_RHEL.m8.u4.product_av.qcow2.virtio_scsi.up.virtio_net.Guest.RHEL.8.4.0.x86_64.io-github-autotest-qemu.netperf_stress_test.TCP_STREAM.guest2guest.q35: PASS (466.26 s)
(2/2) Host_RHEL.m8.u4.product_av.qcow2.virtio_scsi.up.virtio_net.Guest.RHEL.8.4.0.x86_64.io-github-autotest-qemu.netperf_stress_test.TCP_STREAM.host2guest.q35: PASS (565.67 s)

[PaulYuuu]

@yanan-fu In order to continue our testing, let's delete the performance folder, others are well.

[yanan-fu]

@yanan-fu In order to continue our testing, let's delete the performance folder, others are well.

Could you explain the reason ?
performance folder is not a block for our test, to clean up the pkgs of netperf is not my purpose of this patch, i prefer feature owner to do it with a separate patch later, thanks!

[PaulYuuu]

It's also okay for me. The ultimate goal is that we only maintain a set of netperf tools.
Ack.

[yanan-fu]

@quanwenli Looking forward for your result, thanks a lot!

[yanan-fu]

Thanks all~