Hey Radek,

The bug is fixed with this patch. Thanks for the prompt response.

Regards

Praveen

-----Original Message-----
From: Radek Novacek [mailto:rnovacek@redhat.com]
Sent: Tuesday, August 04, 2015 5:37 AM
To: openlmi-devel@lists.fedorahosted.org
Cc: Paladugu, Praveen
Subject: Re: issues with konkretreg

Hi,

thanks for the report. You're right that systems add the KONKRET_REGISTRATION macro into the library. I've added a workaround for this issue [0].

Let me know if it fixed the issue for you.

[0]
https://github.com/rnovacek/konkretcmpi/commit/4934aad4dca4ef0ec493b7871028f94a7bbfc0f0

Radek

On Friday 31 of July 2015 10:25:33 Praveen_Paladugu@Dell.com wrote:
> All,
>
> A provider to publish Service Processor information is at
> https://github.com/praveen-pk/lmi-bmc. This provider is based on
> openlmi framework. I talked about this provider in this forum a few
> months ago. We are trying to package this provider for some distros
> and are running some issues. Hoping someone can provide some guidance on this issue.
>
> After going thru the build steps, the registration file created for
> this provider is as follows:
>
> LMI_BMC.reg:
>
> [LMI_BMC]
> provider: LMI_BMC
> location: cmpiLMIBmc
> type: instance method
> namespace: root/cimv2
> #
> [" CLASS "]
> provider: " PROVIDERNAME "
> location: cmpiLMIBmc
> type: " TYPES;
> namespace: " NAMESPACE "
> #
>
>
> Registering this provider with the above reg file with SFCB CIMOM is
> causing the CIMOM to crash with the following messages:
>
> sfcb[6370]: --- registration statement not recognized:
> sfcb[6370]: 147: [" CLASS "]
> sfcb[6370]: --- Broker terminated because of previous error(s)
>
> The build process creates the reg file " konkretreg libcmpiLMIBmc.so"
>
> The konkretreg tool walks through the binary shared object
> (libcmpiLMIBmc.so) and tries to match lines starting with
> "@(#)KONKRET_REGISTRATION=". The first match is from one of the source
> files, actually, the following macro expanded: KONKRET_REGISTRATION(
> "root/cimv2",
> "LMI_BMC",
> "LMI_BMC",
> "instance method")
>
> The second entry in the reg seems to be from the macro definition itself:
> #define KONKRET_REGISTRATION(NAMESPACE,CLASS,PROVIDERNAME,TYPES)
> static volatile KUNUSED const char __konkret_registration[] =
> "@(#)KONKRET_REGISTRATION=" NAMESPACE ":" CLASS ":" PROVIDERNAME ":"
> TYPES;
>
>
> I am guessing this needs to be fixed in konkretreg tool? Anything else
> I could be missing here?
>
>
>
> Thank you
> Praveen K Paladugu
> Dell Linux Engineering