Skip to contentSkip to conversation

network/perf_lan: remove the test

PATCH New Alexey Kodanev 2 participants 3 messages Jun 10, 2016 – Jun 23, 2016

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 network/perf_lan: remove the test

There are several reasons to remove it:

* Test does what 'ping -p' command can do (can be added to tcp_cmds/ping);
* It needs cleanup/fixing and IPv6 version is disabled right now;
* ck_packet() is wrongly setting data buffer pointer, assuming that the sender added timestamp but it didn't. Actually timestamp was removed in the packet sending code but not in the receiving one (patch ef77253);
* Also test always passes, even if packet data is not correct, because ck_packet() always returns 0, so it doesn't really check packets.

---
 runtest/net.tcp_cmds                            |    1 -
 runtest/stress.part3                            |    2 -
 testcases/network/.gitignore                    |    2 -
 testcases/network/tcp_cmds/perf_lan/Makefile    |   45 ---
 testcases/network/tcp_cmds/perf_lan/perf_lan    |  122 -------
 testcases/network/tcp_cmds/perf_lan/pingpong.c  |  396 -----------------------
 testcases/network/tcp_cmds/perf_lan/pingpong6.c |  291 -----------------
 7 files changed, 0 insertions(+), 859 deletions(-)
 delete mode 100644 testcases/network/tcp_cmds/perf_lan/Makefile
 delete mode 100755 testcases/network/tcp_cmds/perf_lan/perf_lan
 delete mode 100644 testcases/network/tcp_cmds/perf_lan/pingpong.c
 delete mode 100644 testcases/network/tcp_cmds/perf_lan/pingpong6.c
runtest/net.tcp_cmds +0 -1
index 466718b..de58b35 100644
@@ -8,7 +8,6 @@ finger finger01
 ftp export TCbin=$LTPROOT/testcases/network/tcp_cmds/ftp; ftp01
 host host01
 netstat netstat01
-perf_lan export TCbin=$LTPROOT/testcases/network/tcp_cmds/perf_lan; perf_lan
 ping export LTPROOT; ping01
 rcp export TCbin=$LTPROOT/testcases/network/tcp_cmds/rcp; rcp01
 rdist export TCbin=$LTPROOT/testcases/network/tcp_cmds/rdist; rdist01
runtest/stress.part3 +0 -2
index c5da547..350a2c3 100644
@@ -913,5 +913,3 @@ hangup01 hangup01
 mc_cmds mc_cmds
 mc_opts export TCsrc=$LTPROOT/testcases/network/multicast/mc_opts; mc_opts
 rpc01 rpc01
-perf_lan perf_lan
-
testcases/network/.gitignore +0 -2
index ca4e0f3..63d6361 100644
@@ -40,8 +40,6 @@
 /tcp_cmds/echo/createfile
 /tcp_cmds/echo/echoes
 /tcp_cmds/echo/echoes6
-/tcp_cmds/perf_lan/pingpong
-/tcp_cmds/perf_lan/pingpong6
 /tcp_cmds/sendfile/testsf_c
 /tcp_cmds/sendfile/testsf_c6
 /tcp_cmds/sendfile/testsf_s
testcases/network/tcp_cmds/perf_lan/Makefile deleted +0 -45

45 lines not on this page · raw message

testcases/network/tcp_cmds/perf_lan/perf_lan deleted +0 -122

122 lines not on this page · raw message

testcases/network/tcp_cmds/perf_lan/pingpong.c deleted +0 -396

396 lines not on this page · raw message

testcases/network/tcp_cmds/perf_lan/pingpong6.c deleted +0 -291

291 lines not on this page · raw message

Signature · 1 line
1.7.1
#1 lore raw

PATCH 1/1 network/perf_lan: remove the test

Hi!

Quoted 8 lines

* Test does what 'ping -p' command can do (can be added to tcp_cmds/ping);
* It needs cleanup/fixing and IPv6 version is disabled right now;
* ck_packet() is wrongly setting data buffer pointer, assuming that the sender added timestamp but it didn't. Actually timestamp was removed in the packet sending code but not in the receiving one (patch ef77253);
* Also test always passes, even if packet data is not correct, because ck_packet() always returns 0, so it doesn't really check packets.

The test really seems to be recreating in C what could be done with ping albeit it seems to send count packets as fast as possible, as far as I can tell it does someting equivalent to:

"ping -c $count -f -s $packetsize -p 00010203..."

Also looking closer the $count is wired to 1, so the -f does not have any effect.

I guess that the ping01 covers more or less the same and adding more functionality to it is much easier than fixing the mess called perf_lan.

In short, the removal is acked.

Signature · 2 lines
Cyril Hrubis
[email protected]
#2 lore raw

PATCH 1/1 network/perf_lan: remove the test

Hi,

Quoted 13 lines · Cyril Hrubis

Hi!

Quoted 8 lines

* Test does what 'ping -p' command can do (can be added to tcp_cmds/ping);
* It needs cleanup/fixing and IPv6 version is disabled right now;
* ck_packet() is wrongly setting data buffer pointer, assuming that the sender added timestamp but it didn't. Actually timestamp was removed in the packet sending code but not in the receiving one (patch ef77253);
* Also test always passes, even if packet data is not correct, because ck_packet() always returns 0, so it doesn't really check packets.

The test really seems to be recreating in C what could be done with ping albeit it seems to send count packets as fast as possible, as far as I can tell it does someting equivalent to:

Quoted 7 lines

"ping -c $count -f -s $packetsize -p 00010203..."

Also looking closer the $count is wired to 1, so the -f does not have any effect.

I guess that the ping01 covers more or less the same and adding more functionality to it is much easier than fixing the mess called perf_lan.

Right. I've pushed the patch.

Thanks,
Alexey

Quoted 2 lines

In short, the removal is acked.