Menu
▾
▴
Re: [Openhpi-devel] Plug-In
|
From: Pierre S. <psa...@es...> - 2007-11-20 08:38:12
|
In order to compile, I had to make a few modifications. I=92ll send =
directly
to you a copy of my plugin directory in a separate email. I did not do =
any
test with the OEM plugin section present in openhpi.conf since I don=92t =
have
any of the related hardware. However, when the OEM plugin section is =
missing
I get the following error message when openhpid starts:
=20
(process:14609): GLib-CRITICAL **: file gstrfuncs.c: line 2152 =
(g_strsplit):
assertion `string !=3D NULL' failed
=20
Apart from that, the only comments I have are the following:
=20
- OEM functionality should be verified by the vendors who =
already
have dedicated code in the plugin. I think Andy is taking care of that =
for
Intel, I don=92t know who is going to do that for Force (Thomas ?) or if =
we
should remove that piece of code if it=92s not used/maintained anymore.
=20
- My understanding of this OEM mechanism is that it should be
reserved for things that are not standardized in the various =
specifications
such as ATCA/MicroTCA =85
=20
Pierre
=20
_____ =20
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Audet,
Jean-Michel
Sent: Monday, November 19, 2007 22:47
To: ope...@li...
Subject: Re: [Openhpi-devel] Plug-In
=20
Thanks for the feedback.
=20
I will wait for more feedback and process a new patch base on comments.
=20
=20
JM =20
=20
_____ =20
De : ope...@li...
[mailto:ope...@li...] De la part de =
Cress,
Andrew R
Envoy=E9 : Monday, November 19, 2007 4:38 PM
=C0 : ope...@li...
Objet : Re: [Openhpi-devel] Plug-In
=20
Jean-Michel,
=20
My comments:
- I couldn't compile it until I changed sprintf() calls to snprintf in =
the
core ipmidirect.
- For the openhpi.conf.example, the oem plugin line needs to copied for =
the
smi interface also.
- The 'make install' didn't copy the oem libs to /usr/lib/openhpi/.
I'm still having some issues in discover_resources, not sure about root
cause, so I don't have final results yet.
=20
Andy
=20
_____ =20
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Audet,
Jean-Michel
Sent: Monday, November 19, 2007 2:16 PM
To: ope...@li...
Subject: Re: [Openhpi-devel] Plug-In
Any updates, results, comments on this patch?
=20
Thanks!
Jean-Michel Audet
=20
_____ =20
De : Audet, Jean-Michel=20
Envoy=E9 : Monday, November 05, 2007 3:53 PM
=C0 : 'ope...@li...'
Objet : RE: [Openhpi-devel] Plug-In
=20
OK, I have downloaded a fresh copy.
=20
Moved manually the ipmi_mc_vendor_intel.cpp in oem_plugins/intel
Moved manually the ipmi_mc_vendor_intel.h in oem_plugins/intel
Moved manually the ipmi_mc_vendor_force.cpp in oem_plugins/force
Moved manually the ipmi_mc_vendor_force.h in oem_plugins/force
=20
Applied the patch using the following command: patch =96p0
add_ipmidirect_miniplugins.patch
=20
=20
It then compiles OK. =20
=20
=20
Is there a way to apply the patch using svn. Files move doesn=92t work =
using
the diff/patch mechanism.
=20
Let me know
=20
JMA
=20
_____ =20
De : ope...@li...
[mailto:ope...@li...] De la part de =
Pierre
Sangouard
Envoy=E9 : Friday, November 02, 2007 10:46 AM
=C0 : ope...@li...
Objet : Re: [Openhpi-devel] Plug-In
=20
Well I tried again but first encountered the following error:
=20
make[5]: Entering directory
`/root/public/svnhpi/newhead/plugins/ipmidirect/oem_plugins/kontron'
make[6]: Entering directory
`/root/public/svnhpi/newhead/plugins/ipmidirect/oem_plugins/kontron'
if /bin/sh ../../../../libtool --mode=3Dcompile --tag=3DCXX g++ =
-DHAVE_CONFIG_H
-I. -I. -I../../../.. -I../../../../include -I../../../../utils
-I../../../../utils -I../../../../plugins/ipmidirect
-I../../../../plugins/ipmidirect/oem_plugins/kontron -O2 -pthread
-I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -Wall
-Wpointer-arith -Wformat=3D2 -Wformat-security
-Wformat-nonliteral -Wcast-qual -Wcast-align
-Wshadow -Wwrite-strings -Wredundant-decls
-Wno-strict-aliasing -Wno-unused-value -Werror -D_GNU_SOURCE
-D_REENTRANT -fexceptions -MT ipmi_plugin_kontron.lo -MD -MP -MF
".deps/ipmi_plugin_kontron.Tpo" -c -o ipmi_plugin_kontron.lo
ipmi_plugin_kontron.cpp; \
then mv -f ".deps/ipmi_plugin_kontron.Tpo" =
".deps/ipmi_plugin_kontron.Plo";
else rm -f ".deps/ipmi_plugin_kontron.Tpo"; exit 1; fi
mkdir .libs
g++ -DHAVE_CONFIG_H -I. -I. -I../../../.. -I../../../../include
-I../../../../utils -I../../../../utils -I../../../../plugins/ipmidirect
-I../../../../plugins/ipmidirect/oem_plugins/kontron -O2 -pthread
-I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -Wall =
-Wpointer-arith
-Wformat=3D2 -Wformat-security -Wformat-nonliteral -Wcast-qual =
-Wcast-align
-Wshadow -Wwrite-strings -Wredundant-decls -Wno-strict-aliasing
-Wno-unused-value -Werror -D_GNU_SOURCE -D_REENTRANT -fexceptions -MT
ipmi_plugin_kontron.lo -MD -MP -MF .deps/ipmi_plugin_kontron.Tpo -c
ipmi_plugin_kontron.cpp -fPIC -DPIC -o .libs/ipmi_plugin_kontron.o
g++ -DHAVE_CONFIG_H -I. -I. -I../../../.. -I../../../../include
-I../../../../utils -I../../../../utils -I../../../../plugins/ipmidirect
-I../../../../plugins/ipmidirect/oem_plugins/kontron -O2 -pthread
-I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -Wall =
-Wpointer-arith
-Wformat=3D2 -Wformat-security -Wformat-nonliteral -Wcast-qual =
-Wcast-align
-Wshadow -Wwrite-strings -Wredundant-decls -Wno-strict-aliasing
-Wno-unused-value -Werror -D_GNU_SOURCE -D_REENTRANT -fexceptions -MT
ipmi_plugin_kontron.lo -MD -MP -MF .deps/ipmi_plugin_kontron.Tpo -c
ipmi_plugin_kontron.cpp -o ipmi_plugin_kontron.o >/dev/null 2>&1
make[6]: *** No rule to make target `ipmi_mc_vendor_kontron.c', needed =
by
`ipmi_mc_vendor_kontron.lo'. Stop.
make[6]: Leaving directory
`/root/public/svnhpi/newhead/plugins/ipmidirect/oem_plugins/kontron'
make[5]: *** [all-recursive] Error 1
make[5]: Leaving directory
`/root/public/svnhpi/newhead/plugins/ipmidirect/oem_plugins/kontron'
make[4]: *** [all-recursive] Error 1
make[4]: Leaving directory
`/root/public/svnhpi/newhead/plugins/ipmidirect/oem_plugins'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory =
`/root/public/svnhpi/newhead/plugins/ipmidirect'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/root/public/svnhpi/newhead/plugins'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/root/public/svnhpi/newhead'
make: *** [all] Error 2
=20
=20
I managed to fix it by modifying
plugins/ipmidirect/oem_plugins/kontron/Makefile.am file but then I got =
the
following error:
=20
make[5]: Entering directory
`/root/public/svnhpi/newhead/plugins/ipmidirect/oem_plugins/force'
make[6]: Entering directory
`/root/public/svnhpi/newhead/plugins/ipmidirect/oem_plugins/force'
if /bin/sh ../../../../libtool --mode=3Dcompile --tag=3DCXX g++ =
-DHAVE_CONFIG_H
-I. -I. -I../../../.. -I../../../../include -I../../../../utils
-I../../../../utils -I../../../../plugins/ipmidirect -O2 -pthread
-I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -Wall
-Wpointer-arith -Wformat=3D2 -Wformat-security
-Wformat-nonliteral -Wcast-qual -Wcast-align
-Wshadow -Wwrite-strings -Wredundant-decls
-Wno-strict-aliasing -Wno-unused-value -Werror =
-D_GNU_SOURCE
-D_REENTRANT -fexceptions -MT ipmi_plugin_force.lo -MD -MP -MF
".deps/ipmi_plugin_force.Tpo" -c -o ipmi_plugin_force.lo
ipmi_plugin_force.cpp; \
then mv -f ".deps/ipmi_plugin_force.Tpo" ".deps/ipmi_plugin_force.Plo"; =
else
rm -f ".deps/ipmi_plugin_force.Tpo"; exit 1; fi
mkdir .libs
g++ -DHAVE_CONFIG_H -I. -I. -I../../../.. -I../../../../include
-I../../../../utils -I../../../../utils -I../../../../plugins/ipmidirect =
-O2
-pthread -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -Wall
-Wpointer-arith -Wformat=3D2 -Wformat-security -Wformat-nonliteral =
-Wcast-qual
-Wcast-align -Wshadow -Wwrite-strings -Wredundant-decls =
-Wno-strict-aliasing
-Wno-unused-value -Werror -D_GNU_SOURCE -D_REENTRANT -fexceptions -MT
ipmi_plugin_force.lo -MD -MP -MF .deps/ipmi_plugin_force.Tpo -c
ipmi_plugin_force.cpp -fPIC -DPIC -o .libs/ipmi_plugin_force.o
ipmi_plugin_force.cpp: In function `SaErrorT
PluginRegisterBoards(cIpmiMcVendorFactory*)':
ipmi_plugin_force.cpp:139: error: `cIpmiMcVendorForceShMc' has not been
declared
ipmi_plugin_force.cpp:140: error: `cIpmiMcVendorForceShMc' has not been
declared
make[6]: *** [ipmi_plugin_force.lo] Error 1
make[6]: Leaving directory
`/root/public/svnhpi/newhead/plugins/ipmidirect/oem_plugins/force'
make[5]: *** [all-recursive] Error 1
make[5]: Leaving directory
`/root/public/svnhpi/newhead/plugins/ipmidirect/oem_plugins/force'
make[4]: *** [all-recursive] Error 1
make[4]: Leaving directory
`/root/public/svnhpi/newhead/plugins/ipmidirect/oem_plugins'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory =
`/root/public/svnhpi/newhead/plugins/ipmidirect'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/root/public/svnhpi/newhead/plugins'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/root/public/svnhpi/newhead'
make: *** [all] Error 2
=20
=20
And I gave up. Could you please try the patch against a freshly checked =
out
trunk and resend it?
=20
Thanks,
Pierre
=20
_____ =20
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Audet,
Jean-Michel
Sent: Friday, November 02, 2007 14:31
To: ope...@li...
Subject: Re: [Openhpi-devel] Plug-In
=20
Youps,=20
=20
I forgot both ipmi_plugin.cpp and ipmi_plugin.h
=20
Attached is the new patch.
=20
Thanks for the feedback.
=20
JM
=20
_____ =20
De : ope...@li...
[mailto:ope...@li...] De la part de =
Pierre
Sangouard
Envoy=E9 : Friday, November 02, 2007 6:34 AM
=C0 : ope...@li...
Objet : Re: [Openhpi-devel] Plug-In
=20
It seems that ipmi_plugin.h is missing from the patch =96 see error =
below.
=20
make[6]: Entering directory
`/root/public/svnhpi/head/plugins/ipmidirect/oem_plugins/kontron'
if /bin/sh ../../../../libtool --mode=3Dcompile --tag=3DCXX g++ =
-DHAVE_CONFIG_H
-I. -I. -I../../../.. -I../../../../include -I../../../../utils
-I../../../../utils -I../../../../plugins/ipmidirect
-I../../../../plugins/ipmidirect/oem_plugins/kontron -O2 -pthread
-I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -Wall
-Wpointer-arith -Wformat=3D2 -Wformat-security
-Wformat-nonliteral -Wcast-qual -Wcast-align
-Wshadow
-Wwrite-strings -Wredundant-decls -Wno-strict-aliasing
-Wno-unused-value -Werror -D_GNU_SOURCE -D_REENTRANT =
-fexceptions
-MT
ipmi_plugin_kontron.lo -MD -MP -MF ".deps/ipmi_plugin_kontron.Tpo" -c -o
ipmi_plugin_kontron.lo ipmi_plugin_kontron.cpp; \
then mv -f ".deps/ipmi_plugin_kontron.Tpo" =
".deps/ipmi_plugin_kontron.Plo";
else rm -f ".deps/ipmi_plugin_kontron.Tpo"; exit 1; fi
mkdir .libs
g++ -DHAVE_CONFIG_H -I. -I. -I../../../.. -I../../../../include
-I../../../../utils -I../../../../utils -I../../../../plugins/ipmidirect
-I../../../../plugins/ipmidirect/oem_plugins/kontron -O2 -pthread
-I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -Wall =
-Wpointer-arith
-Wformat=3D2 -Wformat-security -Wformat-nonliteral -Wcast-qual =
-Wcast-align
-Wshadow -Wwrite-strings -Wredundant-decls -Wno-strict-aliasing
-Wno-unused-value -Werror -D_GNU_SOURCE -D_REENTRANT -fexceptions -MT
ipmi_plugin_kontron.lo -MD -MP -MF .deps/ipmi_plugin_kontron.Tpo -c
ipmi_plugin_kontron.cpp -fPIC -DPIC -o .libs/ipmi_plugin_kontron.o
In file included from ../../../../plugins/ipmidirect/ipmi_mc.h:57,
from ipmi_plugin_kontron.cpp:28:
../../../../plugins/ipmidirect/ipmi_mc_vendor.h:43:25: ipmi_plugin.h: No
such
file or directory
ipmi_plugin_kontron.cpp: In function `int PluginOpen(cIpmiLog*)':
ipmi_plugin_kontron.cpp:63: error: `IPMI_OEM_PLUGIN_ABI_VERSION' =
undeclared
(first use this function)
ipmi_plugin_kontron.cpp:63: error: (Each undeclared identifier is =
reported
only once for each function it appears in.)
make[6]: *** [ipmi_plugin_kontron.lo] Error 1
make[6]: Leaving directory
`/root/public/svnhpi/head/plugins/ipmidirect/oem_plugins/kontron'
make[5]: *** [all-recursive] Error 1
make[5]: Leaving directory
`/root/public/svnhpi/head/plugins/ipmidirect/oem_plugins/kontron'
make[4]: *** [all-recursive] Error 1
make[4]: Leaving directory
`/root/public/svnhpi/head/plugins/ipmidirect/oem_plugins'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory `/root/public/svnhpi/head/plugins/ipmidirect'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/root/public/svnhpi/head/plugins'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/root/public/svnhpi/head'
make: *** [all] Error 2
=20
=20
_____ =20
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Audet,
Jean-Michel
Sent: Thursday, November 01, 2007 21:00
To: ope...@li...
Subject: Re: [Openhpi-devel] Plug-In
=20
Here=92s the patch=85=20
=20
Let me know if I made it wrong=85 or right!
=20
Again, your comments will be greatly appreciated.
=20
(To make is work, don=92t forget to look at the differences in =
openhpi.conf)
=20
Jean-Michel Audet
Kontron
=20
=20
=20
_____ =20
De : ope...@li...
[mailto:ope...@li...] De la part de =
Renier
Morales
Envoy=E9 : Wednesday, October 31, 2007 5:04 PM
=C0 : ope...@li...
Objet : Re: [Openhpi-devel] Plug-In
=20
ope...@li... wrote on 10/31/2007 03:45:45 =
PM:
> What is the version of the SVN that you are using? I have 1.4.2=20
> (r22196) from CollabNet.=20
> =20
> I get the following output when trying to get trunk:=20
> =20
> svn: PROPFIND request failed on '/svnroot/openhpi/openhpi/trunk'
> svn: PROPFIND of '/svnroot/openhpi/openhpi/trunk': Could not resolve
> hostname `openhpi.svn.sourceforge.net': Temporary failure in name
resolution (
> https://openhpi.svn.sourceforge.net)=20
I'm using subversion 1.4.0 (r21228), so probably more of a network =
issue.=20
Can you get to https://openhpi.svn.sourceforge.net on your browser at =
least?
Pinging openhpi.svn.sourceforge.net gives me 66.35.250.145.=20
--Renier
> I=92m used to CVS but SVN is new for me.=20
> =20
> Jean-Michel Audet=20
> =20
>=20
> De : ope...@li... [mailto:openhpi-
> dev...@li...] De la part de Renier Morales
> Envoy=E9 : Wednesday, October 31, 2007 3:37 PM
> =C0 : ope...@li...
> Objet : Re: [Openhpi-devel] Plug-In=20
> =20
>=20
> ope...@li... wrote on 10/31/2007 =
01:52:05
PM:
>=20
> > =20
> > Hi,=20
> > I just completed the implementation of the IPMI OEM mini
> > plug-ins. It=92s a mechanism to add mini plug-ins to the ipmidirect =
> > plug-in. It is using the class cIpmiMcVendorFactory and the=20
> > mechanism already in place in ipmi_mc_vendor.cpp. I have recover=20
> > 100% of the implementation of Force computer and Intel that was=20
> > registering board to the mc vendor factory. They are now (if OK=20
> > with you), moved out from the ipmidirect plug-in and compile as a=20
> stand-alone=20
> > OEM IPMI mini plug-in.=20
>=20
> That's great! I will wait to see what Pierre thinks.=20
>=20
> > In order load the plug-in, the openhpi.conf shall be modified as =
follow:
> > =20
> > In openhpi.conf=20
> > ## Section for ipmidirect plugin using RMCP:
> > handler libipmidirect {
> > entity_root =3D "{SYSTEM_CHASSIS,7}"
> > =85=20
> > ipmi_plugins =3D "libipmidirect_kontron,libipmidirect_intel,
> > libipmidirect_force"
> > }=20
> > =20
> > =20
> > I would like to know how you want to have the new code. I have:=20
> > =20
> > Made changes on the following files:=20
> > - openhpi.conf=20
> > - configure.ac=20
> > - openhpi/plugins/ipmidirect =DF changes in many files in=20
> > this directory. =20
> > =20
> > Added the following files:=20
> > - openhpi/plugins/ipmidirect/ipmi_plugin.cpp=20
> > - openhpi/plugins/ipmidirect/ipmi_plugin.h=20
> > =20
> > Created the following tree=20
> > - openhpi/plugins/ ipmidirect/oem_plugins/*** (Where *** is
> > the manufacturer name, I have created the folder for kontron,=20
> intel and force)=20
> > =20
> > Moved:=20
> > - board factory registration and ipmi_mc_vendor_intel.* and
> > ipmi_mc_vendor_force.* into the oem_plugins vendor folders.=20
>=20
> This is how I'd like to see the changes. Check out the trunk version
> of OpenHPI like this:=20
> svn co =
https://openhpi.svn.sourceforge.net/svnroot/openhpi/openhpi/trunk
> openhpi-trunk=20
>=20
> Make/patch the changes you mention there (adding new files and dirs=20
> with 'svn add') and send the output of "svn diff" to the list.=20
>=20
> > =20
> > I have created the plug-ins for Force and Intel but I do not have=20
> > made any tests. However, I have made no changes to the=20
> > ipmi_mc_vendor_xxx files.=20
> > =20
> > Kontron will certainly provide fixes and maybe other=20
> > implementations. Can I join the developer list? (You may want to=20
> > wait after this).=20
>=20
> You can. Let's start with you posting the patch so that everyone can
> review. Then I'll give you devel access to the repository so that=20
> you can check the changes in.=20
>=20
> > =20
> > Your comments will be greatly appreciated.=20
>=20
> This sounds like a very good patch. Thanks again.=20
>=20
> --Renier=20
>=20
> > =20
> > Let me know!=20
> > Jean-Michel Audet=20
> > Kontron Canada=20
> >=20
> > De : ope...@li... [mailto:openhpi-
> > dev...@li...] De la part de Renier Morales
> > Envoy=E9 : Monday, October 22, 2007 10:43 AM
> > =C0 : ope...@li...
> > Objet : Re: [Openhpi-devel] Plug-In=20
> > =20
> >=20
> > ope...@li... wrote on 10/19/2007 =
03:13:28
PM:
> >=20
> > > Hi all,=20
> > > I have a general question regarding the plug-ins. Is=20
> > > there a way to add the OEM specific stuff without modifying the=20
> > > OpenHPI direct plug-in? >From what I can see in the ATCA and uTCA=20
> > > world, it looks like the companies are currently modifying the=20
> > > ipmidirect plug-in in order to provide OEM functionalities. =20
> > > However, an ATCA chassis can be populate with many vendors blades=20
> > > equipped with different IPMI solutions. If each company has their =
> > > own flavor of the ipmidirect plug-in, it is then impossible to =
have=20
> > > a fully integrated solution.=20
> > > =20
> > > Do I understand it correctly?=20
> > > =20
> > > Is there any other solution?=20
> > > =20
> > > =20
> > > One of the answers will probably be that the company just has to=20
> > > commit the changes into the OpenHPI CVS=85 but I am not sure that =
you=20
> > > want to pollute the plug-ins with full of OEM stuff=20
> (maintenance=85 etc.). =20
> > > =20
> >=20
> > You should take a look at the ipmi plugin in OpenHPI which uses=20
> > OpenIPMI. Its purpose is to provide an integrated solution for all=20
> > ipmi-based hardware. There are some problems with it as it is not=20
> > under active development anymore and in need of testing and code=20
> > contributions to get it into shape.=20
> >=20
> > To add OEM specific stuff, the ipmi plugin could have its own set of
> > mini-plugins, each for a specific type of ipmi hardware that=20
> > contains any OEM magic stuff. Vendors would only need to provide the
> > mini-plugin.=20
> >=20
> > If you'd like to work on something like this, it would be a real=20
> > improvement in OpenHPI's ipmi support.=20
> >=20
> > --Renier=20
> >
-------------------------------------------------------------------------=
> > This SF.net email is sponsored by: Splunk Inc.
> > Still grepping through log files to find problems? Stop.
> > Now Search log events and configuration files using AJAX and a =
browser.
> > Download your FREE copy of Splunk now >> http://get.splunk.com/
> > _______________________________________________
> > Openhpi-devel mailing list
> > Ope...@li...
> > https://lists.sourceforge.net/lists/listinfo/openhpi-devel
> =
-------------------------------------------------------------------------=
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems? Stop.
> Now Search log events and configuration files using AJAX and a =
browser.
> Download your FREE copy of Splunk now >> http://get.splunk.com/
> _______________________________________________
> Openhpi-devel mailing list
> Ope...@li...
> https://lists.sourceforge.net/lists/listinfo/openhpi-devel
|