Re: [Ipmitool-devel] Error compiling the Intel imb driver with Cywin

[Juan P. B. <jba...@gm...>]

Hi all,
I commented the PAGESIZE definition as Zdenek suggested and, as he
predicted, got errors with libtool (at least no compilation error). All
errors are related to IOCTL_IMB_SEND_MESSAGE definition. If some on the
list have a clue to go over this, I can try again, if not, I will resort to
create a plugin for ipmiutil.

TIA

Juan Pablo

Juan Pablo Baserga
Consultor Independiente/Independent Consultant
Mobile/Linux/High Availability/Internet/Business Intelligence
Cel: +54-911-5318-1127
Phone: +54-11-4799-8237
Skype: jbaserga


2015-02-05 15:25 GMT-03:00 Juan Pablo Baserga <jba...@gm...>:

> Hello Hank,
> thanks very much for your answer. Fortunately my case is not one of size
> but your comment is true. IPMI or not IPMI 3,000 machines deserves a
> special architecture (collectors, etc) to be feasible.
>
> As I told Zdenek, I cannot connect the BMCs to LAN, so I need an inband
> check. I did it for the Linux boxes, but I cannot reproduce this in Windows
>
> Thank you very much for your input.
>
> Best regards,
>
> Juan Pablo
>
> Juan Pablo Baserga
> Consultor Independiente/Independent Consultant
> Mobile/Linux/High Availability/Internet/Business Intelligence
> Cel: +54-911-5318-1127
> Phone: +54-11-4799-8237
> Skype: jbaserga
>
>
> 2015-02-05 7:33 GMT-03:00 Hank Bruning <ha...@jb...>:
>
> I agree with Zdenek,
>> Rewrite the plug in. The reason the reason windows (or other Operating
>> Systems) are not supported is that the identical information is available
>> using IPMI RMCP when no operating system is running.
>> Either way, OS vs no OS, you still are going to run into the problem of
>> lack of CPU threads with Nagios on large scale deployments, Over 3000
>> servers and the monitoring system spends more time thread context switching
>> than collecting IPMI data. Solve that problem and the IMB becomes
>> irrelevant.
>>
>> Hank Bruning
>> JBlade
>>
>> On Wed, Feb 4, 2015 at 1:46 PM, Zdenek Styblik <zde...@gm...>
>> wrote:
>>
>>> On Wed, Feb 4, 2015 at 7:25 PM, Juan Pablo Baserga <jba...@gm...>
>>> wrote:
>>> > Hi Zdenek,
>>> > thank you again very much for your support. IPMIutil is working
>>> (although it
>>> > depends on VB dlls to run) with both the MS and the Intel drivers, but
>>> it
>>> > has no Nagios plugin, that's why I tried to give IPMItool a chance (as
>>> it
>>> > has many Nagios plugins). As I told you I will try a little more (I am
>>> > surprised nobody tried to do this before as it is a very basic need of
>>> > hardware supervision, with the Linux servers it was a piece of cake to
>>> make
>>> > this work!) and let the list know the results. If not I will try to
>>> make a
>>> > VB plugin that works with IPMIutil, guess that in any way it would be
>>> good.
>>> >
>>>
>>> Hello Juan,
>>>
>>> shooting ideas is cheap. To put code where mouth is, that's another
>>> story. :)
>>> Now, bad project management and so on, but given your use case, I'd
>>> almost question whether it wouldn't be faster to rewrite those plugins
>>> to use IPMIutil. I'm not familiar with it, though, so I might be
>>> wrong. But Nagios plugins - should be easy enough. And yes, any work
>>> on Windows port of IPMI Tool will be/is appreciated!
>>> As for VB thing, I believe the main reason for it was the proprietary
>>> driver. And I'd almost bet IMB wasn't even enabled and commented out
>>> via #ifdef. I really don't remember this particular part of the code,
>>> sorry.
>>>
>>> Good luck,
>>> Z.
>>>
>>> > Best regards,
>>> >
>>> > Juan Pablo
>>> >
>>> > Juan Pablo Baserga
>>> > Consultor Independiente/Independent Consultant
>>> > Mobile/Linux/High Availability/Internet/Business Intelligence
>>> > Cel: +54-911-5318-1127
>>> > Phone: +54-11-4799-8237
>>> > Skype: jbaserga
>>> >
>>> >
>>> > 2015-02-04 14:22 GMT-03:00 Zdenek Styblik <zde...@gm...>:
>>> >
>>> >> On Wed, Feb 4, 2015 at 5:54 PM, Zdenek Styblik <
>>> zde...@gm...>
>>> >> wrote:
>>> >> > On Wed, Feb 4, 2015 at 4:24 PM, Juan Pablo Baserga <
>>> jba...@gm...>
>>> >> > wrote:
>>> >> >> Zdenek,
>>> >> >> thanks again on this, The problem with the Oracle/Sun approach is
>>> that
>>> >> >> they
>>> >> >> are using their own propietary IPMI driver, and don't support the
>>> >> >> Intel/Microsoft ones. My goal is to get IPMI inband (I don't have
>>> the
>>> >> >> possibility to link the LAN interface of the IPMI board) so I can
>>> have
>>> >> >> visibility of sensors for Nagios management.
>>> >> >>
>>> >> >> I'll give the patch a try and let the list know!
>>> >> >>
>>> >> >> Best regards
>>> >> >>
>>> >> >> Juan Pablo
>>> >> >>
>>> >> >
>>> >> > Sorry Juan, I've sent that reply from mobile phone and it got messed
>>> >> > up. The patch I've posted isn't necessary(more like how it should be
>>> >> > patched), just removing that line should be enough in your
>>> particular
>>> >> > case. This definition can be found in
>>> /usr/include/asm-generic/param.h
>>> >> > on Linux/Unix platforms and it's written in the same way(ifndef).
>>> I've
>>> >> > also seen somewhere getpagesize(), or something, can be called. But
>>> I
>>> >> > believe this is like low-level stuff which shouldn't be done in IPMI
>>> >> > tool and definition from param.h should be used.
>>> >> > As for Oracle/Sun, true. I know VB is used for compilation; ipmievd
>>> >> > code is commented out, because it wouldn't work on Windows; and that
>>> >> > proprietary driver is used. I have no idea how to compile IPMI tool
>>> on
>>> >> > Windows in pretty much any way, I'm sorry.
>>> >> > Couple people from Intel used to hang around here, but I'm not sure
>>> if
>>> >> > they still do nor really sure how much they'd be interested in or
>>> >> > willing to help.
>>> >> > Also, I'm sure any patches and how-to_s related to Windows are
>>> welcome
>>> >> > and appreciated. As I've said earlier, I thought IMB driver is up
>>> for
>>> >> > removal(I was really close to such proposition). If you're good at
>>> C,
>>> >> > perhaps it could make sense to reverse engineer it from the code and
>>> >> > rewrite from scratch. Or may be ask Microsoft/Intel for help? There
>>> >> > surely must be some kind of documentation/API + may be IPMI v2.0
>>> >> > specification, p. 69, IPMB Interface? Somebody more profound in such
>>> >> > things could/should answer that.
>>> >> > If you have more questions or anything, ask. I'll try to pitch
>>> >> > ideas/help as much as I can.
>>> >> >
>>> >>
>>> >> Actually, one question and that's whether IMB really does what you
>>> >> need and ever worked on Windows. There must be some reason why Sun
>>> >> came up with their own driver. As far as I know, the only working
>>> >> Windows version was from Oracle/Sun. However, that doesn't mean others
>>> >> have tried, failed and given up(= may be nobody really tried).
>>> >> Also, there are other open-source IPMI client tools,
>>> >> http://ipmiutil.sourceforge.net/docs/ipmisw-compare.htm, and they
>>> have
>>> >> Windows driver. Well, it seems only ipmiutil does. May be it would be
>>> >> possible(and better/faster?) to port it.
>>> >>
>>> >> Z.
>>> >>
>>> >> > Z.
>>> >> >
>>> >> > PS: mailing list on CC again. As I've said, my reply got messed up.
>>> >> >
>>> >> >> Juan Pablo Baserga
>>> >> >> Consultor Independiente/Independent Consultant
>>> >> >> Mobile/Linux/High Availability/Internet/Business Intelligence
>>> >> >> Cel: +54-911-5318-1127
>>> >> >> Phone: +54-11-4799-8237
>>> >> >> Skype: jbaserga
>>> >> >>
>>> >> >>
>>> >> >> 2015-02-04 12:10 GMT-03:00 Zdenek Styblik <
>>> zde...@gm...>:
>>> >> >>
>>> >> >>> On Wed, Feb 4, 2015 at 1:36 PM, Juan Pablo Baserga
>>> >> >>> <jba...@gm...>
>>> >> >>> wrote:
>>> >> >>> > Hi Zdenek,
>>> >> >>> > thank you very much for your prompt answer. I thought of that,
>>> but
>>> >> >>> > I'd
>>> >> >>> > rather ask before breaking something... I'll will try and let
>>> the
>>> >> >>> > list
>>> >> >>> > know.
>>> >> >>> > I am not very familiar with the code, is there a definition
>>> (i.e.
>>> >> >>> > ifdef
>>> >> >>> > cygwin) that I can use so I can try to make a patch for future
>>> use?
>>> >> >>> >
>>> >> >>>
>>> >> >>> Juan,
>>> >> >>>
>>> >> >>> you're welcome. To be honest, I don't know where to start.
>>> >> >>> There used to be #ifdef for cygwin very, very long time ago, if I
>>> >> >>> remember
>>> >> >>> correctly. But they are gone and I don't believe there should be
>>> any
>>> >> >>> ifdefs,
>>> >> >>> although I might be wrong about this one.
>>> >> >>> I don't think there is anyone whom has tried to compile IPMI tool
>>> on
>>> >> >>> Windows. Probably not ever ever, but in very long time. In other
>>> >> >>> words,
>>> >> >>> you're probably in uncharted waters here. Also, I thought IMB
>>> driver
>>> >> >>> is dead
>>> >> >>> by now and nobody is using it for real.
>>> >> >>> There is a precompiled version of IPMI tool from Oracle as a part
>>> of
>>> >> >>> Hardware Management Pack. They're not using cygwin and have their
>>> own
>>> >> >>> Windows BMC interface, that I'm sure of ;) And I'm almost sure
>>> there
>>> >> >>> are
>>> >> >>> more problems ahead, eg. Compilation of ipmievd on Windows.
>>> >> >>>
>>> >> >>> As for better work-around/patch in this case, it should be
>>> something
>>> >> >>> like:
>>> >> >>>
>>> >> >>> diff --git a/src/plugins/imb/imbapi.c b/src/plugins/imb/imbapi.c
>>> >> >>> index 111df62..507fa01 100644
>>> >> >>> --- a/src/plugins/imb/imbapi.c
>>> >> >>> +++ b/src/plugins/imb/imbapi.c
>>> >> >>> @@ -104,7 +104,9 @@ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY
>>> OF
>>> >> >>> SUCH
>>> >> >>> DAMAGE.
>>> >> >>> #define IMB_DEVICE "/dev/instru/mismic"
>>> >> >>> #else
>>> >> >>> #define IMB_DEVICE "/dev/imb"
>>> >> >>> -#define PAGESIZE EXEC_PAGESIZE
>>> >> >>> +# ifndef PAGESIZE
>>> >> >>> +# define PAGESIZE EXEC_PAGESIZE
>>> >> >>> +# endif
>>> >> >>> #endif
>>> >> >>>
>>> >> >>> /*Just to make the DEB
>>> >> >>>
>>> >> >>> > Best regards,
>>> >> >>> >
>>> >> >>> > Juan Pablo
>>> >> >>> >
>>> >> >>> > Juan Pablo Baserga
>>> >> >>> > Consultor Independiente/Independent Consultant
>>> >> >>> > Mobile/Linux/High Availability/Internet/Business Intelligence
>>> >> >>> > Cel: +54-911-5318-1127
>>> >> >>> > Phone: +54-11-4799-8237
>>> >> >>> > Skype: jbaserga
>>> >> >>> >
>>> >> >>> >
>>> >> >>> > 2015-02-04 8:29 GMT-03:00 Zdenek Styblik <
>>> zde...@gm...>:
>>> >> >>> >
>>> >> >>> >> On Tue, Feb 3, 2015 at 11:27 PM, Juan Pablo Baserga
>>> >> >>> >> <jba...@gm...>
>>> >> >>> >> wrote:
>>> >> >>> >> > Hi,
>>> >> >>> >> > I am trying to compile ipmitool-1.8.15 under Cygwin. If I
>>> compile
>>> >> >>> >> > it
>>> >> >>> >> > using
>>> >> >>> >> > only the lan driver (./configure --enable-intf-lanplus)
>>> >> >>> >> > everything
>>> >> >>> >> > goes
>>> >> >>> >> > smooth; but if I try to compile the Intel driver (so I can do
>>> >> >>> >> > inband
>>> >> >>> >> > management in Windows) it gives an error.
>>> >> >>> >> >
>>> >> >>> >>
>>> >> >>> >> Hello Juan,
>>> >> >>> >>
>>> >> >>> >> you can try to remove (re)definition of PAGESIZE in imbapi.c at
>>> >> >>> >> line
>>> >> >>> >> 107, but it's a wild guess to be honest.
>>> >> >>> >> I'm almost sure there are more surprises ahead.
>>> >> >>> >>
>>> >> >>> >> Good luck to you,
>>> >> >>> >> Z.
>>> >> >>> >>
>>> >> >>> >> > I do:
>>> >> >>> >> >
>>> >> >>> >> > ./configure --enable-intf-lanplus --enable-intf-imb
>>> >> >>> >> >
>>> >> >>> >> > configure runs ok
>>> >> >>> >> >
>>> >> >>> >> > make
>>> >> >>> >> >
>>> >> >>> >> > ends with the following errors:
>>> >> >>> >> >
>>> >> >>> >> > make[4]: Entering directory
>>> >> >>> >> > '/cygdrive/c/ipmitool-1.8.15/src/plugins/imb'
>>> >> >>> >> > /bin/sh ../../../libtool --silent --tag=CC --mode=compile gcc
>>> >> >>> >> > -DHAVE_CONFIG_H
>>> >> >>> >> > -I. -I../../.. -I../../../include -g -O2 -fno-strict-aliasing
>>> >> >>> >> > -Wreturn-type
>>> >> >>> >> > -MT imbapi.lo -MD -MP -MF .deps/imbapi.Tpo -c -o imbapi.lo
>>> >> >>> >> > imbapi.c
>>> >> >>> >> > imbapi.c:107:0: warning: "PAGESIZE" redefined
>>> >> >>> >> > #define PAGESIZE EXEC_PAGESIZE
>>> >> >>> >> > ^
>>> >> >>> >> > In file included from
>>> >> >>> >> > /usr/lib/gcc/i686-pc-cygwin/4.9.2/include-fixed/limits.h:1
>>> >> >>> >> > 68:0,
>>> >> >>> >> > from
>>> >> >>> >> > /usr/lib/gcc/i686-pc-cygwin/4.9.2/include-fixed/syslimits.
>>> >> >>> >> > h:7,
>>> >> >>> >> > from
>>> >> >>> >> > /usr/lib/gcc/i686-pc-cygwin/4.9.2/include-fixed/limits.h:3
>>> >> >>> >> > 4,
>>> >> >>> >> > from /usr/include/sys/param.h:15,
>>> >> >>> >> > from imbapi.c:89:
>>> >> >>> >> > /usr/include/limits.h:211:0: note: this is the location of
>>> the
>>> >> >>> >> > previous
>>> >> >>> >> > definiti
>>> >> >>> >> > on
>>> >> >>> >> > #define PAGESIZE 65536
>>> >> >>> >> > ^
>>> >> >>> >> > imbapi.c: In function 'MapPhysicalMemory':
>>> >> >>> >> > imbapi.c:107:18: error: 'EXEC_PAGESIZE' undeclared (first
>>> use in
>>> >> >>> >> > this
>>> >> >>> >> > function)
>>> >> >>> >> > #define PAGESIZE EXEC_PAGESIZE
>>> >> >>> >> > ^
>>> >> >>> >> > imbapi.c:1999:30: note: in expansion of macro 'PAGESIZE'
>>> >> >>> >> > diff = (int)startpAddress % PAGESIZE;
>>> >> >>> >> > ^
>>> >> >>> >> > imbapi.c:107:18: note: each undeclared identifier is reported
>>> >> >>> >> > only
>>> >> >>> >> > once
>>> >> >>> >> > for
>>> >> >>> >> > each
>>> >> >>> >> > function it appears in
>>> >> >>> >> > #define PAGESIZE EXEC_PAGESIZE
>>> >> >>> >> > ^
>>> >> >>> >> > imbapi.c:1999:30: note: in expansion of macro 'PAGESIZE'
>>> >> >>> >> > diff = (int)startpAddress % PAGESIZE;
>>> >> >>> >> > ^
>>> >> >>> >> > imbapi.c: In function 'UnmapPhysicalMemory':
>>> >> >>> >> > imbapi.c:107:18: error: 'EXEC_PAGESIZE' undeclared (first
>>> use in
>>> >> >>> >> > this
>>> >> >>> >> > function)
>>> >> >>> >> > #define PAGESIZE EXEC_PAGESIZE
>>> >> >>> >> > ^
>>> >> >>> >> > imbapi.c:2045:44: note: in expansion of macro 'PAGESIZE'
>>> >> >>> >> > diff = ((unsigned int) virtualAddress) % PAGESIZE;
>>> >> >>> >> > ^
>>> >> >>> >> > Makefile:417: recipe for target 'imbapi.lo' failed
>>> >> >>> >> > make[4]: *** [imbapi.lo] Error 1
>>> >> >>> >> > make[4]: Leaving directory
>>> >> >>> >> > '/cygdrive/c/ipmitool-1.8.15/src/plugins/imb'
>>> >> >>> >> > Makefile:481: recipe for target 'all-recursive' failed
>>> >> >>> >> > make[3]: *** [all-recursive] Error 1
>>> >> >>> >> > make[3]: Leaving directory
>>> >> >>> >> > '/cygdrive/c/ipmitool-1.8.15/src/plugins'
>>> >> >>> >> > Makefile:580: recipe for target 'all-recursive' failed
>>> >> >>> >> > make[2]: *** [all-recursive] Error 1
>>> >> >>> >> > make[2]: Leaving directory '/cygdrive/c/ipmitool-1.8.15/src'
>>> >> >>> >> > Makefile:454: recipe for target 'all-recursive' failed
>>> >> >>> >> > make[1]: *** [all-recursive] Error 1
>>> >> >>> >> > make[1]: Leaving directory '/cygdrive/c/ipmitool-1.8.15'
>>> >> >>> >> > Makefile:385: recipe for target 'all' failed
>>> >> >>> >> > make: *** [all] Error 2
>>> >> >>> >> >
>>> >> >>> >> > Do I need to use any other switches in configure or lack a
>>> Cywin
>>> >> >>> >> > package?
>>> >> >>> >> >
>>> >> >>> >> > Thanks in advance for your help!
>>> >> >>> >> >
>>> >> >>> >> > Best regards,
>>> >> >>> >> >
>>> >> >>> >> > Juan Pablo Baserga
>>> >> >>> >> > Consultor Independiente/Independent Consultant
>>> >> >>> >> > Mobile/Linux/High Availability/Internet/Business Intelligence
>>> >> >>> >> > Cel: +54-911-5318-1127
>>> >> >>> >> > Phone: +54-11-4799-8237
>>> >> >>> >> > Skype: jbaserga
>>> >> >>> >> >
>>> >> >>> >> >
>>> >> >>> >> >
>>> >> >>> >> >
>>> >> >>> >> >
>>> >> >>> >> >
>>> ------------------------------------------------------------------------------
>>> >> >>> >> > Dive into the World of Parallel Programming. The Go Parallel
>>> >> >>> >> > Website,
>>> >> >>> >> > sponsored by Intel and developed in partnership with Slashdot
>>> >> >>> >> > Media,
>>> >> >>> >> > is
>>> >> >>> >> > your
>>> >> >>> >> > hub for all things parallel software development, from weekly
>>> >> >>> >> > thought
>>> >> >>> >> > leadership blogs to news, videos, case studies, tutorials and
>>> >> >>> >> > more.
>>> >> >>> >> > Take
>>> >> >>> >> > a
>>> >> >>> >> > look and join the conversation now.
>>> >> >>> >> > http://goparallel.sourceforge.net/
>>> >> >>> >> > _______________________________________________
>>> >> >>> >> > Ipmitool-devel mailing list
>>> >> >>> >> > Ipm...@li...
>>> >> >>> >> > https://lists.sourceforge.net/lists/listinfo/ipmitool-devel
>>> >> >>> >> >
>>> >> >>> >
>>> >> >>> >
>>> >> >>
>>> >> >>
>>> >
>>> >
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> Dive into the World of Parallel Programming. The Go Parallel Website,
>>> sponsored by Intel and developed in partnership with Slashdot Media, is
>>> your
>>> hub for all things parallel software development, from weekly thought
>>> leadership blogs to news, videos, case studies, tutorials and more. Take
>>> a
>>> look and join the conversation now. http://goparallel.sourceforge.net/
>>> _______________________________________________
>>> Ipmitool-devel mailing list
>>> Ipm...@li...
>>> https://lists.sourceforge.net/lists/listinfo/ipmitool-devel
>>>
>>
>>
>