Thomas Luetzel
|
io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/15/2016 3:58 AM
post117111
|
io-pkt leak data section - ppc8260 ethernet driver
Hello,
we do see on our system that io-pkt while running shows in different scenarios a memory leak. The memory in the data
section grows until the system has no resources (RAM) left and crashes!
----------------------------------------------------------------------------------
QNX Version used:
------------------------------------
QNX6.5 SP2 - PPC Platform
NIC Driver uses:
------------------------------------
from BSP 8260 pq2fads (QNX PQ2FADS BSP V1.0.2 (File revision: 1.43))
Start sequence of drivers (extract from a start up script):
------------------------------------
mount -T io-pkt -o channel=1,mac=$MAC1 devn-ppc8260-nxa.so
mount -T io-pkt -o channel=2,mac=$MAC2 devn-ppc8260-nxa.so
mount -T io-pkt -o channel=3,mac=$MAC3 devn-ppc8260-nxa.so
netmanager -r all -f $NETCONFIG
mount -T io-pkt -o bind=ip0,resolve=dns,resolve=autoresolve /usr/lib/lsm-qnet.so
----------------------------------------------------------------------------------
Other embedded sysetm are also connected on the network. If we now perform ls /net/$NODENAME the new node do appear.
By simple performing the "df -h" we can reproduce the leak.
The issue is, that our application does not execute this command - but we can see that ~4.5 hours the data section
(pidin -mem) of io-pkt do increase by 4kBytes.
Any help is welcome, also if more information is needed - please let me know
Kind Regards
Thomas L.
|
|
|
Hugh Brown
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Hugh Brown
11/15/2016 2:39 PM
post117115
|
Re: io-pkt leak data section - ppc8260 ethernet driver
This is most probably a driver issue, but this driver wasn¹t written by
QNX and we don¹t have the source code.
On 2016-11-15, 3:58 AM, "Thomas Luetzel" <community-noreply@qnx.com> wrote:
>Hello,
>
>we do see on our system that io-pkt while running shows in different
>scenarios a memory leak. The memory in the data section grows until the
>system has no resources (RAM) left and crashes!
>
>--------------------------------------------------------------------------
>--------
>QNX Version used:
>------------------------------------
>QNX6.5 SP2 - PPC Platform
>
>NIC Driver uses:
>------------------------------------
>from BSP 8260 pq2fads (QNX PQ2FADS BSP V1.0.2 (File revision: 1.43))
>
>Start sequence of drivers (extract from a start up script):
>------------------------------------
>mount -T io-pkt -o channel=1,mac=$MAC1 devn-ppc8260-nxa.so
>mount -T io-pkt -o channel=2,mac=$MAC2 devn-ppc8260-nxa.so
>mount -T io-pkt -o channel=3,mac=$MAC3 devn-ppc8260-nxa.so
>
>netmanager -r all -f $NETCONFIG
>
>mount -T io-pkt -o bind=ip0,resolve=dns,resolve=autoresolve
>/usr/lib/lsm-qnet.so
>--------------------------------------------------------------------------
>--------
>
>Other embedded sysetm are also connected on the network. If we now
>perform ls /net/$NODENAME the new node do appear.
>
>By simple performing the "df -h" we can reproduce the leak.
>The issue is, that our application does not execute this command - but we
>can see that ~4.5 hours the data section (pidin -mem) of io-pkt do
>increase by 4kBytes.
>
>Any help is welcome, also if more information is needed - please let me
>know
>
>Kind Regards
>Thomas L.
>
>
>
>
>
>
>
>
>
>
>
>
>_______________________________________________
>
>Networking Drivers
>http://community.qnx.com/sf/go/post117111
>To cancel your subscription to this discussion, please e-mail
>drivers-networking-unsubscribe@community.qnx.com
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/16/2016 8:43 AM
post117123
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Hi,
if needed, wanted or allowed from QNX point of view - I can attach the driver code.
Kind Regards
|
|
|
Hugh Brown
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Hugh Brown
11/16/2016 8:50 AM
post117124
|
Re: io-pkt leak data section - ppc8260 ethernet driver
If you don¹t have a support contract with QNX, you will have to speak to
your sales rep to sort something out.
On 2016-11-16, 8:43 AM, "Thomas Luetzel" <community-noreply@qnx.com> wrote:
>Hi,
>
>if needed, wanted or allowed from QNX point of view - I can attach the
>driver code.
>
>Kind Regards
>
>
>
>_______________________________________________
>
>Networking Drivers
>http://community.qnx.com/sf/go/post117123
>To cancel your subscription to this discussion, please e-mail
>drivers-networking-unsubscribe@community.qnx.com
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/16/2016 10:44 AM
post117125
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Some more test results,
I just started the OS and killed all Ethernet related things.
I started io-pkt-v4 again and kept a script in the background running:
===================
while [ 0 -lt 1 ]
do
df >> /mnt/flash.log
sleep 1
done
===================
RESULT after 30 minutes:
COMMAND: pidin -l -d 100 io-pkt-v4 mem
pid tid name prio STATE code data stack
208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 308K 4096(516K)*
208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 308K 8192(132K)
libc.so.3 @fe300000 512K 12K
... CUT ...
pid tid name prio STATE code data stack
208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 312K 4096(516K)*
208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 312K 8192(132K)
libc.so.3 @fe300000 512K 12K
... CUT ...
pid tid name prio STATE code data stack
208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 316K 4096(516K)*
208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 316K 8192(132K)
libc.so.3 @fe300000 512K 12K
... CUT ...
pid tid name prio STATE code data stack
208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 320K 4096(516K)*
208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 320K 8192(132K)
libc.so.3 @fe300000 512K 12K
... CUT ...
pid tid name prio STATE code data stack
208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 324K 4096(516K)*
208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 324K 8192(132K)
libc.so.3 @fe300000 512K 12K
... CUT ...
pid tid name prio STATE code data stack
208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 328K 4096(516K)*
208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 328K 8192(132K)
libc.so.3 @fe300000 512K 12K
... CUT ...
pid tid name prio STATE code data stack
208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 332K 4096(516K)*
208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 332K 8192(132K)
libc.so.3 @fe300000 512K 12K
... CUT ...
pid tid name prio STATE code data stack
208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 336K 4096(516K)*
208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 336K 8192(132K)
libc.so.3 @fe300000 512K 12K
What make io-pkt-v4 to allocate memory ????
Thanks in advance for any help
Thomas L.
|
|
|
Hugh Brown
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Hugh Brown
11/16/2016 10:49 AM
post117127
|
Re: io-pkt leak data section - ppc8260 ethernet driver
io-pkt will allocate memory as it is needed, so the data size will carry
on increasing, but should stabilize after a while.
On 2016-11-16, 10:44 AM, "Thomas Luetzel" <community-noreply@qnx.com>
wrote:
>Some more test results,
>
>I just started the OS and killed all Ethernet related things.
>I started io-pkt-v4 again and kept a script in the background running:
>
>===================
>while [ 0 -lt 1 ]
>do
> df >> /mnt/flash.log
> sleep 1
>done
>===================
>RESULT after 30 minutes:
>
>COMMAND: pidin -l -d 100 io-pkt-v4 mem
>
> pid tid name prio STATE code data
>stack
>208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 308K
>4096(516K)*
>208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 308K
>8192(132K)
> libc.so.3 @fe300000 512K 12K
>
>... CUT ...
>
> pid tid name prio STATE code data
>stack
>208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 312K
>4096(516K)*
>208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 312K
>8192(132K)
> libc.so.3 @fe300000 512K 12K
>
>... CUT ...
>
> pid tid name prio STATE code data stack
>208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 316K
>4096(516K)*
>208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 316K
>8192(132K)
> libc.so.3 @fe300000 512K 12K
>
>... CUT ...
>
> pid tid name prio STATE code data stack
>208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 320K
>4096(516K)*
>208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 320K
>8192(132K)
> libc.so.3 @fe300000 512K 12K
>
>... CUT ...
>
> pid tid name prio STATE code data
>stack
>208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 324K
>4096(516K)*
>208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 324K
>8192(132K)
> libc.so.3 @fe300000 512K 12K
>
>... CUT ...
>
> pid tid name prio STATE code data
>stack
>208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 328K
>4096(516K)*
>208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 328K
>8192(132K)
> libc.so.3 @fe300000 512K 12K
>
>... CUT ...
>
> pid tid name prio STATE code data
>stack
>208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 332K
>4096(516K)*
>208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 332K
>8192(132K)
> libc.so.3 @fe300000 512K 12K
>
>... CUT ...
>
> pid tid name prio STATE code data
>stack
>208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 336K
>4096(516K)*
>208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 336K
>8192(132K)
> libc.so.3 @fe300000 512K 12K
>
>
>What make io-pkt-v4 to allocate memory ????
>
>Thanks in advance for any help
>
>Thomas L.
>
>
>
>
>_______________________________________________
>
>Networking Drivers
>http://community.qnx.com/sf/go/post117125
>To cancel your subscription to this discussion, please e-mail
>drivers-networking-unsubscribe@community.qnx.com
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/16/2016 10:58 AM
post117128
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Understood,
I will keep the system/test running over night and will let you know if the memory stabalized.
If yes - good, then I have to look into another direction, BUT IF NOT?
Again no network driver is mounted as you can see in the pidin log, so for me the original request/answer that there is
a bug in the driver itself can be crossed out - as it is not mounted/loaded/started....
Thanks for the replies
Kind Regards
Thomas
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/18/2016 5:31 AM
post117141
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Hello,
kept on running the test.
io-pkt.-v4 data section of the memory is NOT settling down!
====================
# pidin -l -d 100 -p io-pkt-v4 mem
pid tid name prio STATE code data stack
208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 4184K 4096(516K)*
208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 4184K 8192(132K)
====================
script code:
===========
while [ 0 -lt 1 ]
do
df -h > /dev/null
done
===========
what can cause the system/io-pkt to eat up memory while executing this script. If we find out the reason, then I can
have a look into our application what might cause the same issue.
Thanks in advance
Thomas L.
|
|
|
Hugh Brown
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Hugh Brown
11/18/2016 7:31 AM
post117142
|
Re: io-pkt leak data section - ppc8260 ethernet driver
What version of the O/S are you running, and what is the command line you
are using to start io-pkt? The output from pidin would also be useful.
On 2016-11-18, 5:31 AM, "Thomas Luetzel" <community-noreply@qnx.com> wrote:
>Hello,
>
>kept on running the test.
>io-pkt.-v4 data section of the memory is NOT settling down!
>
>====================
># pidin -l -d 100 -p io-pkt-v4 mem
> pid tid name prio STATE code data
>stack
>208330770 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 572K 4184K
>4096(516K)*
>208330770 2 usr/bin/io-pkt-v4 21r RECEIVE 572K 4184K
>8192(132K)
>====================
>
>script code:
>===========
>while [ 0 -lt 1 ]
>do
> df -h > /dev/null
>done
>===========
>
>what can cause the system/io-pkt to eat up memory while executing this
>script. If we find out the reason, then I can have a look into our
>application what might cause the same issue.
>
>Thanks in advance
>Thomas L.
>
>
>
>
>
>_______________________________________________
>
>Networking Drivers
>http://community.qnx.com/sf/go/post117141
>To cancel your subscription to this discussion, please e-mail
>drivers-networking-unsubscribe@community.qnx.com
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/21/2016 10:21 AM
post117154
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Hi,
we do have two system running
1. QNX 6.5.0 2010/07/09-14:35:59EDT
# use -i io-pkt-v4-hc
NAME=io-pkt-v4-hc
DESCRIPTION=TCP/IP protocol module.
DATE=2010/07/09-13:43:56-EDT
STATE=stable
HOST=mainbuild
USER=builder
VERSION=6.5.0
TAGID=89
2. QNX 6.5.0 2012/06/20-13:49:31EDT
# use -i io-pkt-v4
NAME=io-pkt-v4
DESCRIPTION=TCP/IP protocol module.
DATE=2014/10/15-11:25:52-EDT
STATE=stable
HOST=pspbuildvm
USER=pspbuild
VERSION=650SP1-4004
TAGID=5664
on both systems we start io-pkt-v4 the following way:
io-pkt-v4
mount -T io-pkt -o channel=1,mac=$MAC1 devn-ppc8260-nxa.so
mount -T io-pkt -o channel=2,mac=$MAC2 devn-ppc8260-nxa.so
mount -T io-pkt -o channel=3,mac=$MAC3 devn-ppc8260-nxa.so
mount -T io-pkt -o bind=ip0,resolve=dns,resolve=autoresolve /usr/lib/lsm-qnet.so
Both systems showing the same leak behavior.
=======================================
# pidin -pio-pkt-v4
pid tid name prio STATE Blocked
393246 1 usr/bin/io-pkt-v4 21r SIGWAITINFO
393246 2 usr/bin/io-pkt-v4 10r RECEIVE 1
393246 3 usr/bin/io-pkt-v4 21r RECEIVE 24
393246 4 usr/bin/io-pkt-v4 10r RECEIVE 28
393246 5 usr/bin/io-pkt-v4 10r RECEIVE 32
393246 6 usr/bin/io-pkt-v4 21r RECEIVE 39
393246 7 usr/bin/io-pkt-v4 10r CONDVAR (0xfe3cd6a4)
393246 8 usr/bin/io-pkt-v4 20r RECEIVE 44
393246 9 usr/bin/io-pkt-v4 10r RECEIVE 36
393246 10 usr/bin/io-pkt-v4 10r CONDVAR (0xfe3cc96c)
========================================
# pidin -p io-pkt-v4 mem
pid tid name prio STATE code data stack
393246 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 660K 1172K 8192(516K)*
393246 2 usr/bin/io-pkt-v4 10r RECEIVE 660K 1172K 8192(132K)
393246 3 usr/bin/io-pkt-v4 21r RECEIVE 660K 1172K 4096(132K)
393246 4 usr/bin/io-pkt-v4 10r RECEIVE 660K 1172K 4096(132K)
393246 5 usr/bin/io-pkt-v4 10r RECEIVE 660K 1172K 4096(132K)
393246 6 usr/bin/io-pkt-v4 21r RECEIVE 660K 1172K 4096(132K)
393246 7 usr/bin/io-pkt-v4 10r CONDVAR 660K 1172K 4096(132K)
393246 8 usr/bin/io-pkt-v4 20r RECEIVE 660K 1172K 4096(132K)
393246 9 usr/bin/io-pkt-v4 10r RECEIVE 660K 1172K 4096(132K)
393246 10 usr/bin/io-pkt-v4 10r CONDVAR 660K 1172K 4096(132K)
libc.so.3 @fe300000 512K 12K
devnp-shim.so @fe386000 36K 4096
evn-ppc8260-nxa.so @fe390000 48K 4096
lsm-qnet.so @fe39d000 188K 44K
libsocket.so.3 @fe3d7000 184K 24K
/dev/mem @80100000 (30000000) 128K
/dev/mem @80120000 (30000000) 128K
/dev/mem @80140000 (30000000) 128K
========================================
Thanks
Thomas
|
|
|
Hugh Brown
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Hugh Brown
11/21/2016 12:11 PM
post117155
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Unfortunately there is nothing we can do about this without getting access
to your BSP as well as the source to the io-pkt driver. We cannot give you
the latest io-pkt, as it will require the network driver to be re-compiled
with the latest libc and libsocket. Your best bet is to contact your sales
rep and make a formal request to have this work done.
On 2016-11-21, 10:21 AM, "Thomas Luetzel" <community-noreply@qnx.com>
wrote:
>Hi,
>
>we do have two system running
>1. QNX 6.5.0 2010/07/09-14:35:59EDT
># use -i io-pkt-v4-hc
>NAME=io-pkt-v4-hc
>DESCRIPTION=TCP/IP protocol module.
>DATE=2010/07/09-13:43:56-EDT
>STATE=stable
>HOST=mainbuild
>USER=builder
>VERSION=6.5.0
>TAGID=89
>
>2. QNX 6.5.0 2012/06/20-13:49:31EDT
># use -i io-pkt-v4
>NAME=io-pkt-v4
>DESCRIPTION=TCP/IP protocol module.
>DATE=2014/10/15-11:25:52-EDT
>STATE=stable
>HOST=pspbuildvm
>USER=pspbuild
>VERSION=650SP1-4004
>TAGID=5664
>
>on both systems we start io-pkt-v4 the following way:
>
>io-pkt-v4
>mount -T io-pkt -o channel=1,mac=$MAC1 devn-ppc8260-nxa.so
>mount -T io-pkt -o channel=2,mac=$MAC2 devn-ppc8260-nxa.so
>mount -T io-pkt -o channel=3,mac=$MAC3 devn-ppc8260-nxa.so
>mount -T io-pkt -o bind=ip0,resolve=dns,resolve=autoresolve
>/usr/lib/lsm-qnet.so
>
>Both systems showing the same leak behavior.
>
>=======================================
>
># pidin -pio-pkt-v4
> pid tid name prio STATE Blocked
> 393246 1 usr/bin/io-pkt-v4 21r SIGWAITINFO
> 393246 2 usr/bin/io-pkt-v4 10r RECEIVE 1
> 393246 3 usr/bin/io-pkt-v4 21r RECEIVE 24
> 393246 4 usr/bin/io-pkt-v4 10r RECEIVE 28
> 393246 5 usr/bin/io-pkt-v4 10r RECEIVE 32
> 393246 6 usr/bin/io-pkt-v4 21r RECEIVE 39
> 393246 7 usr/bin/io-pkt-v4 10r CONDVAR (0xfe3cd6a4)
> 393246 8 usr/bin/io-pkt-v4 20r RECEIVE 44
> 393246 9 usr/bin/io-pkt-v4 10r RECEIVE 36
> 393246 10 usr/bin/io-pkt-v4 10r CONDVAR (0xfe3cc96c)
>========================================
># pidin -p io-pkt-v4 mem
> pid tid name prio STATE code data
>stack
> 393246 1 usr/bin/io-pkt-v4 21r SIGWAITINFO 660K 1172K
>8192(516K)*
> 393246 2 usr/bin/io-pkt-v4 10r RECEIVE 660K 1172K
>8192(132K)
> 393246 3 usr/bin/io-pkt-v4 21r RECEIVE 660K 1172K
>4096(132K)
> 393246 4 usr/bin/io-pkt-v4 10r RECEIVE 660K 1172K
>4096(132K)
> 393246 5 usr/bin/io-pkt-v4 10r RECEIVE 660K 1172K
>4096(132K)
> 393246 6 usr/bin/io-pkt-v4 21r RECEIVE 660K 1172K
>4096(132K)
> 393246 7 usr/bin/io-pkt-v4 10r CONDVAR 660K 1172K
>4096(132K)
> 393246 8 usr/bin/io-pkt-v4 20r RECEIVE 660K 1172K
>4096(132K)
> 393246 9 usr/bin/io-pkt-v4 10r RECEIVE 660K 1172K
>4096(132K)
> 393246 10 usr/bin/io-pkt-v4 10r CONDVAR 660K 1172K
>4096(132K)
> libc.so.3 @fe300000 512K 12K
> devnp-shim.so @fe386000 36K 4096
> evn-ppc8260-nxa.so @fe390000 48K 4096
> lsm-qnet.so @fe39d000 188K 44K
> libsocket.so.3 @fe3d7000 184K 24K
> /dev/mem @80100000 (30000000) 128K
> /dev/mem @80120000 (30000000) 128K
> /dev/mem @80140000 (30000000) 128K
>
>========================================
>
>Thanks
>
>Thomas
>
>
>
>
>_______________________________________________
>
>Networking Drivers
>http://community.qnx.com/sf/go/post117154
>To...
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/22/2016 9:29 AM
post117158
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Hi all,
I just proofed that this issue is also present on an X86 machine!
If downloaded the vmware image from (http://www.qnx.com/download/feature.html?programid=23665) and executed the same
scripte
======================
while [ 0 - lt 1 ]
do
df >> /dev/null
done
======================
While monitoring via pidin the io-pkt module (pidin -l -d 100 -p io-pkt-v4-hc mem) I can see the memory is increasing,
increasing, increasing, increasing, ...
See attached screenshot.
I think it is now prooven it has nothing todo with my BSP or modified network driver. It is a general issue!
Need help to identify which is causing this memory leak - relation of DF command and io-pkt?
Then I would know what we are doing WRONG in our application which shows the same behavior.
Thank in advance
If needed - how can I file this defect directly to the QNX development team?
Kind Regards
|
|
|
Hugh Brown
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Hugh Brown
11/22/2016 9:38 AM
post117159
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Well then there must be something strange going on on your systems, as I
have a 6.5.0 SP1 x86 machine and have no problem with the data segment
increasing.
On 2016-11-22, 9:29 AM, "Thomas Luetzel" <community-noreply@qnx.com> wrote:
>Hi all,
>
>I just proofed that this issue is also present on an X86 machine!
>
>If downloaded the vmware image from
>(http://www.qnx.com/download/feature.html?programid=23665) and executed
>the same scripte
>
>======================
>while [ 0 - lt 1 ]
>do
> df >> /dev/null
>done
>======================
>
>While monitoring via pidin the io-pkt module (pidin -l -d 100 -p
>io-pkt-v4-hc mem) I can see the memory is increasing, increasing,
>increasing, increasing, ...
>
>
>See attached screenshot.
>
>I think it is now prooven it has nothing todo with my BSP or modified
>network driver. It is a general issue!
>Need help to identify which is causing this memory leak - relation of DF
>command and io-pkt?
>
>Then I would know what we are doing WRONG in our application which shows
>the same behavior.
>
>Thank in advance
>
>If needed - how can I file this defect directly to the QNX development
>team?
>
>Kind Regards
>
>
>
>
>_______________________________________________
>
>Networking Drivers
>http://community.qnx.com/sf/go/post117158
>To cancel your subscription to this discussion, please e-mail
>drivers-networking-unsubscribe@community.qnx.com
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/22/2016 9:42 AM
post117160
|
Re: io-pkt leak data section - ppc8260 ethernet driver
... ???? it the vmware X86 provided by QNX, downloaded - started. Nothing else done?
I don't really know what I do wrong here?
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/22/2016 10:01 AM
post117161
|
Re: io-pkt leak data section - ppc8260 ethernet driver
here some screen recordings what I'm doing.
No magic!
Thanks Hugh for all your help so far
|
|
|
Hugh Brown
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Hugh Brown
11/22/2016 10:11 AM
post117162
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Does the x86 data segment on the x86 machine stabilize after a while?
On 2016-11-22, 10:01 AM, "Thomas Luetzel" <community-noreply@qnx.com>
wrote:
>here some screen recordings what I'm doing.
>No magic!
>
>Thanks Hugh for all your help so far
>
>
>
>_______________________________________________
>
>Networking Drivers
>http://community.qnx.com/sf/go/post117161
>To cancel your subscription to this discussion, please e-mail
>drivers-networking-unsubscribe@community.qnx.com
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/22/2016 10:13 AM
post117163
|
Re: io-pkt leak data section - ppc8260 ethernet driver
No. I just kept it running and it eating up already 5MB
I'm just about to test the same on QNX6.6.0 and let you know
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/22/2016 10:18 AM
post117164
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Just used the QNX SDP 6.6.0 runtime ISO for VMware [or virtual machine] to install a virtual machine.
If I do the same test in here
io-pkt-v4-hc is stable! 504K and stable!
But as we have a PPC architecture on or target I can't switch I have to stick with 6.5
> No. I just kept it running and it eating up already 5MB
>
> I'm just about to test the same on QNX6.6.0 and let you know
|
|
|
Dennis Kellly
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Dennis Kellly
11/22/2016 10:35 AM
post117165
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Your issue does not reproduce for me on 6.5 VMware image. Are you sure the number of processes stays constant? Copies
of sh are being left in memory?
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/22/2016 10:44 AM
post117166
|
Re: io-pkt leak data section - ppc8260 ethernet driver
I use the vmware QNX 6.5.0 2012/06/20-13:50:50EDT x86pc x86
Number of processes is constant ~40 (39-40)
Number of threads (pidin info) constant
O.k with every call of "df" it does get a new process ID but the calling shell stays the same.
Of all processes only io-pkt-v4-hc do increase
Kind Regards
|
|
|
Dennis Kellly
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Dennis Kellly
11/22/2016 10:45 AM
post117167
|
Re: io-pkt leak data section - ppc8260 ethernet driver
What is your "df" output? I can't see any relationship of df to network UNLESS there is a sharepoint mounted...
|
|
|
Thomas Luetzel
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Thomas Luetzel
11/22/2016 10:47 AM
post117168
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
|
|
Dennis Kellly
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Dennis Kellly
11/22/2016 11:01 AM
post117170
|
Re: io-pkt leak data section - ppc8260 ethernet driver
My io-pkt-v4-hc is exactly the same. Has to be something with the network - but I don't see how it is related to the "
df" call you are making.
What is you replace "df >>/dev/null" with "ls >>/dev/null"... does it have to be "df" to see the leak?
|
|
|
Mario Charest
|
RE: io-pkt leak data section - ppc8260 ethernet driver
|
Mario Charest
11/22/2016 11:05 AM
post117171
|
RE: io-pkt leak data section - ppc8260 ethernet driver
A capture with tracelogger should answer all these questions? Assuming the VM is using the instrumented kernel ?
-----Message d'origine-----
De : Dennis Kellly [mailto:community-noreply@qnx.com]
Envoyé : 22 novembre 2016 11:01
À : drivers-networking <drivers-networking@community.qnx.com>
Objet : Re: io-pkt leak data section - ppc8260 ethernet driver
My io-pkt-v4-hc is exactly the same. Has to be something with the network - but I don't see how it is related to the "
df" call you are making.
What is you replace "df >>/dev/null" with "ls >>/dev/null"... does it have to be "df" to see the leak?
_______________________________________________
Networking Drivers
http://community.qnx.com/sf/go/post117170
To cancel your subscription to this discussion, please e-mail drivers-networking-unsubscribe@community.qnx.com
|
|
|
Hugh Brown
|
Re: io-pkt leak data section - ppc8260 ethernet driver
|
Hugh Brown
11/22/2016 11:36 AM
post117172
|
Re: io-pkt leak data section - ppc8260 ethernet driver
Here is the output from my x86 PC running just io-pkt-v4-hc. I also ran df
-h several times as well, with no difference to the output.
(root) /usr/qnx650/target/qnx6/x86/sbin-> pidin -pio-pkt-v4-hc mem
pid tid name prio STATE code data
stack
127537170 1 ./io-pkt-v4-hc 21r SIGWAITINFO 948K 344K
4096(516K)*
127537170 2 ./io-pkt-v4-hc 21r RECEIVE 948K 344K
8192(132K)
libc.so.3 @b0300000 488K 16K
(root) /usr/qnx650/target/qnx6/x86/sbin-> pidin -pio-pkt-v4-hc mem
pid tid name prio STATE code data
stack
127537170 1 ./io-pkt-v4-hc 21r SIGWAITINFO 948K 344K
4096(516K)*
127537170 2 ./io-pkt-v4-hc 10r RECEIVE 948K 344K
8192(132K)
libc.so.3 @b0300000 488K 16K
(root) /usr/qnx650/target/qnx6/x86/sbin-> pidin -pio-pkt-v4-hc mem
pid tid name prio STATE code data
stack
127537170 1 ./io-pkt-v4-hc 21r SIGWAITINFO 948K 344K
4096(516K)*
127537170 2 ./io-pkt-v4-hc 21r RECEIVE 948K 344K
8192(132K)
libc.so.3 @b0300000 488K 16K
(root) /usr/qnx650/target/qnx6/x86/sbin-> pidin -pio-pkt-v4-hc mem
pid tid name prio STATE code data
stack
127537170 1 ./io-pkt-v4-hc 21r SIGWAITINFO 948K 344K
4096(516K)*
127537170 2 ./io-pkt-v4-hc 21r RECEIVE 948K 344K
8192(132K)
libc.so.3 @b0300000 488K 16K
(root) /usr/qnx650/target/qnx6/x86/sbin-> pidin -pio-pkt-v4-hc mem
pid tid name prio STATE code data
stack
127537170 1 ./io-pkt-v4-hc 21r SIGWAITINFO 948K 344K
4096(516K)*
127537170 2 ./io-pkt-v4-hc 21r RECEIVE 948K 344K
8192(132K)
libc.so.3 @b0300000 488K 16K
(root) /usr/qnx650/target/qnx6/x86/sbin-> pidin -pio-pkt-v4-hc mem
pid tid name prio STATE code data
stack
127537170 1 ./io-pkt-v4-hc 21r SIGWAITINFO 948K 344K
4096(516K)*
127537170 2 ./io-pkt-v4-hc 21r RECEIVE 948K 344K
8192(132K)
libc.so.3 @b0300000 488K 16K
NAME=io-pkt-v4-hc
DESCRIPTION=TCP/IP protocol module.
DATE=2012/06/20-13:36:58-EDT
STATE=stable
HOST=gusbuild4
USER=builder
VERSION=6.5.0
TAGID=650SP1-166
On 2016-11-22, 10:47 AM, "Thomas Luetzel" <community-noreply@qnx.com>
wrote:
>see attached screen shot
>
>
>
>_______________________________________________
>
>Networking Drivers
>http://community.qnx.com/sf/go/post117168
>To cancel your subscription to this discussion, please e-mail
>drivers-networking-unsubscribe@community.qnx.com
|
|
|
|