Menu
▾
▴
Re: [Sblim-devel] Alignment Trap in SFCB
|
From: rajib s. <raj...@ya...> - 2007-03-21 15:44:59
|
Hi Victor,=0A=0AI have two points to clarify.=0A=0A1) I need to have a cust= om provider which will be loaded as for the lifetime of sfcb, i.e it should= be like the classprovider or interopprovider that sfcb provides. How am i = to register such a provider? As of the current scenario all the custom prov= iders registered gets invoked while a request comes for that provider, but = i need the provider up and running at the startup.=0A=0A2) sfcb does not fe= tch the method qualifiers (1.2.1) yet. Is there any patch available for thi= s?=0A=0AThanks,=0ARajib. =0A =0A=0A----- Original Message ----=0AFrom: Vikt= or Mihajlovski <MIH...@de...>=0ATo: rajib saha <rajib_saha81@yahoo.c= o.in>=0ACc: ar...@am...; chitrakg <chi...@am...>; Konrad Rzeszutek <k= on...@li...>; sanjoy maity <sa...@am...>; sblim-devel@list= s.sourceforge.net; sbl...@li...=0ASent: Wednes= day, 28 February, 2007 1:15:14 PM=0ASubject: Re: [Sblim-devel] Alignment Tr= ap in SFCB=0A=0A=0AHi Rajib, =0A=0Asorry for not replying earlier. I guess = that you made a good point. I need to do some more experimenting, but it se= ems that all common ABIs allow to use pointer to union and pointer to union= field synonymously. At least it will make sfcb's code more robust, so I'll= most likely follow your suggestion. =0A=0AMit freundlichen Gr=FC=DFen / Ki= nd Regards=0A Viktor Mihajlovski=0A=0ALinux Technology Center=0A=0A Phone= +49-7031-16-2560=0A E-Mail mih...@de...=0A=0AIBM Deutschland Entwi= cklung GmbH=0AVorsitzender des Aufsichtsrats: Johann Weihen=0AGesch=E4ftsf= =FChrung: Herbert Kircher=0ASitz der Gesellschaft: B=F6blingen=0ARegisterge= richt: Amtsgericht Stuttgart, HRB 243294 =0A=0A=0A=0Arajib saha <rajib_sa= ha...@ya...> =0ASent by: sbl...@li... =0A= 02/26/07 05:32 PM =0A To: Viktor Mihajlovski/Germany/= IBM@IBMDE =0A cc: sanjoy maity <sa...@am...>, chitrakg <ch= it...@am...>, sbl...@li..., ar...@am..., Konrad R= zeszutek <ko...@li...> =0A Subject: Re: [Sblim= -devel] Alignment Trap in SFCB=0A=0A=0A=0AHi Viktor, =0A =0AThanks much f= or your reply. Actually we were following the second approach but prefixed = by a typecast to CMPIValue union. The CMSetProperty usage suggested in the = sblim-cmpi-base package is also the same. =0AI will try out the usage you d= escribed and let you know. But don't you think thats going to an overhead f= or the provider writers as they have to use CMPIValue elements at each plac= e they will call CMSetProperty, instead if we can make the changes in the f= unction definition itself we can relax on the provider side. =0A =0A =0AT= hanks, =0ARajib. =0A =0A=0A=0ARajib Saha. =0A=0A=0A----- Original Message = ----=0AFrom: Viktor Mihajlovski <MIH...@de...>=0ATo: rajib saha <raj= ib_...@ya...>=0ACc: ar...@am...; chitrakg <chi...@am...>; Ko= nrad Rzeszutek <ko...@li...>; sanjoy maity <sa...@am...>= ; sbl...@li...=0ASent: Monday, 26 February, 2007 11:15= :58 AM=0ASubject: Re: [Sblim-devel] Alignment Trap in SFCB=0A=0A=0AHi Rajib= , =0A=0Asorry for the delay. I am actually having trouble to understand the= suggested patch. Generally there should be no problem in assigning entire = unions, if they are aligned properly. This should be the case if you're pas= sing in a pointer to a CMPIValue in the provider's setProperty call. In ord= er to illustrate what I mean, see the following samples. Please make sure t= hat your providers are correct in this respect. =0A=0ACorrect Usage: =0A=0A= CMPIValue cmpiValue.uint16 =3D 42; =0A... =0ACMSetProperty(inst,&cmpiValue,= CMPI_uint16); =0A/* this works only on little endian, alignment-insensitive= archs, but even there it can lead to invalid memory access */ =0A=0AWrong = Usage: =0A=0Aunsigned short uint16Value =3D 42; =0A... =0ACMSetProperty(ins= t,&uint16Value,CMPI_uint16); =0A/* this may lead to unaligned access (littl= e endian) or access to invalid union members (big endian) */ =0A=0AMit freu= ndlichen Gr=FC=DFen / Kind Regards=0A Viktor Mihajlovski=0A=0ALinux Technol= ogy Center=0A=0A Phone +49-7031-16-2560=0A E-Mail mih...@de...=0A=0A= IBM Deutschland Entwicklung GmbH=0AVorsitzender des Aufsichtsrats: Johann W= eihen=0AGesch=E4ftsf=FChrung: Herbert Kircher=0ASitz der Gesellschaft: B=F6= blingen=0ARegistergericht: Amtsgericht Stuttgart, HRB 243294 =0A=0A=0Araj= ib saha <raj...@ya...> =0ASent by: sbl...@li...= urceforge.net =0A02/21/07 12:30 AM =0A To: Konrad Rzes= zutek <ko...@li...> =0A cc: sblim-devel@lists.s= ourceforge.net, vi...@am..., sanjoy maity <sa...@am...>, arunv@ami.= com, chitrakg <chi...@am...> =0A Subject: Re: [Sblim-devel= ] Alignment Trap in SFCB=0A=0A=0A=0A=0A=0AThanks for your response. I hav= e attatched the patch with this mail. =0A=0AThanks, =0ARajib Saha. =0A=0A= =0A----- Original Message ---- =0AFrom: Konrad Rzeszutek <ko...@li...e= t.ibm.com> =0ATo: rajib saha <raj...@ya...> =0ACc: sblim-devel@= lists.sourceforge.net; chitrakg <chi...@am...>; sanjoy maity <sanjoym@a= mi.com>; ar...@am...; vi...@am... =0ASent: Tuesday, 20 February, 2007= 9:46:37 AM =0ASubject: Re: [Sblim-devel] Alignment Trap in SFCB =0A=0ACan= you attach the patch? =0A=0AOn Fri, Feb 16, 2007 at 10:12:28PM +0530, raji= b saha wrote: =0A> We found the issues when we tested on the above two plat= forms and we assume that people using sfcb on embedded platforms will face = similar issues when they use SFCB. =0A> Thank You in advance for your help.= =0A=0A=0A-- =0ARegards, =0A=0AKonrad Rzeszutek =0AIBM, (617)-693-1718 =0A= =0A=0A=0AHere=92s a new way to find what you're looking for - Yahoo! Answer= s =0A----------------------------------------------------------------------= ---=0ATake Surveys. Earn Cash. Influence the Future of IT=0AJoin SourceForg= e.net's Techsay panel and you'll get the chance to share your=0Aopinions on= IT & business topics through brief surveys-and earn cash=0Ahttp://www.tech= say.com/default.php?page=3Djoin.php&p=3Dsourceforge&CID=3DDEVDEV =0A_______= ________________________________________=0ASblim-devel mailing list=0ASblim= -d...@li...=0Ahttps://lists.sourceforge.net/lists/listinfo= /sblim-devel =0A=0A=0A#### alignment_path has been removed from this note o= n February 26, 2007 by Viktor Mihajlovski =0A=0A=0A=0A=0A=0AHere=92s a new = way to find what you're looking for - Yahoo! Answers ----------------------= ---------------------------------------------------=0ATake Surveys. Earn Ca= sh. Influence the Future of IT=0AJoin SourceForge.net's Techsay panel and y= ou'll get the chance to share your=0Aopinions on IT & business topics throu= gh brief surveys-and earn cash=0Ahttp://www.techsay.com/default.php?page=3D= join.php&p=3Dsourceforge&CID=3DDEVDEV______________________________________= _________=0ASblim-devel mailing list=0AS...@li...=0A= https://lists.sourceforge.net/lists/listinfo/sblim-devel=0A=0A=0A=09=09=0A_= _________________________________________________________=0AYahoo! India An= swers: Share what you know. Learn something new=0Ahttp://in.answers.yahoo.c= om/ |