Re: komd issues - was [Re: [openMosix-general] openmosix 2.4.26 in 2.4.31?]
"Matt Dew" <[email protected]> Thu, 21 Sep 2006 13:27:05 -0600
| Newsgroups | gmane.linux.cluster.openmosix.devel |
|---|---|
| Message-ID | <[email protected]> |
On Thu, 2006-09-21 at 18:13 +0000, Florian Delizy wrote: > Matt Dew wrote: > > Setting LD_ASSUME_KERNEL=2.4.1 fixes the system call segfaults. > > WOOHOO!! Now we know where to look to fix the problem. > > > > > Do you mean you need to set the variable LD_ASSUME_KERNEL before calling > insmod ? or something like that ?? or to set the LD_ASSUME_KERNEL > variable before starting (and/or) migrating a process ? > > Unfortunately, I don't really understand LD_ASSUME_KERNEL in great > > detail. instead of running './loop' run 'LD_ASSUME_KERNEL=2.4.1 ./loop' > > > AFAIK, it's something that tells the dynamic linker not to load a shared > object if the version of the kernel is less than what's in this flag ... > > Are we going to have to redo the sys_call_table? Have each entry call a > > new om_sys_call_table that calls the appropriate function? > > > What's the link between LD_ASSUME_KERNEL and the syscall table ? >From what I understand so far: One of LD_ASSUME_KERNEL's functions is to tell glibc how to handle system calls. In linux 2.4, system calls use int 0x80 so we could modify arch/i386/kernel/entry.S to check for migrated processes and then handle the system call differently if need be. As of 2.6, system calls can be called other ways as well such as jumping straight to the syscall handler address. Trapping 0x80 won't work all the time. We have to change the interrupt table, rather than trying to conditionally use our own. http://sourceware.org/ml/gdb/2003-04/msg00153.html helps explain it better than I can. Also, here's my post to kerneltrap.org http://kerneltrap.org/node/7142 Apparently 2.6.18 also changes the way it handles system calls. Matt > > Matt > > > > On Wed, 2006-09-20 at 13:02 +0000, "Matt Dew" wrote: > > > >> Hi Ian, > >> I wish I could narrow it down, but unfortunately I don't know where > >> the bug is. The best I can do is to refer you to Florian's patches or > >> the big patch itself, > >> (against vanilla kernel) > >> http://osource.org/openmosix/patches/om-2.6.17-full-20060831.patch.bz2 > >> > > > > > >> (against git tree) > >> http://osource.org/openmosix/patches/om-2.6.17-git-20060831.patch.bz2 > >> > >> > >> As the bug doesn't happen for every migrated process my current > >> > > thinking > > > >> is that when a process is migrated and kcomd creates a new process the > >> task_struct slab that is allocated isn't correctly initialized and > >> > > only > > > >> by luck is it sometimes set up correctly. But I've searched through > >> > > the > > > >> task structure and mm_struct and vma_struct and haven't found > >> > > anything. > > > >> I could be completely wrong though. > >> > >> Any help would be greatly appreciated. > >> Matt > >> > >> > >> On Wed, 2006-09-20 at 01:40 +0000, "Ian Latter" wrote: > >> > >>> Hello, > >>> > >>> > >>> Is there a patch that you can refer us to for the kcomd > >>> segfault? (as in, the one that contains the kcomd code, > >>> not the one that contains the fix ;-) > >>> > >>> I'm not in a position to test it, but I can run a set of eyes > >>> over it - though I suspect I've coded more segfaults than > >>> I've fixed. Late night code and pointers to pointers don't > >>> mix ... > >>> > >>> > >>> > >>> > >>> > >>> ----- Original Message ----- > >>> > >>>> From: "Matt Dew" <[email protected]> > >>>> To: <moshe-ay74M1d3r6RWk0Htik3J/[email protected]> > >>>> Subject: [Openmosix-devel] komd issues - was > >>>> > >> [Re: [openMosix-general] openmosix > >> > >>> 2.4.26 in 2.4.31?] > >>> > >>>> Date: Tue, 19 Sep 2006 17:11:49 -0600 > >>>> > >>>> Moshe, > >>>> kcomd has one show stopper. There is a bug that causes migrated > >>>> processes to segfault whenever a system call is issued. Since > >>>> > > even > > > >>>> exit() is a system call it's a major problem. This bug doesn't > >>>> > >> happen > >> > >>>> on every migrated process so it's been difficult to pin down. > >>>> > >>>> Sadly, I've spent a large amount of time on this bug to no avail. > >>>> > > I > > > >>>> need some assistance in fixing it. > >>>> > >>>> Other issues are to finish the spinlocks for smp safety and change > >>>> > >> the > >> > >>>> migration code to speed up migrations. Neither of these two do I > >>>> consider major issues. > >>>> > >>>> > >>>> Florian has been very helpful in breaking up the big patch into > >>>> > >> smaller > >> > >>>> ones. Thanks again Florian. > >>>> > >>>> Matt > >>>> > >>>> > >>>> On Mon, 2006-09-18 at 09:16 +0000, moshe bar wrote: > >>>> > >>>>> We need to have a strong effort to test your tools by as many > >>>>> > >> people > >> > >>>>> as possible and then release them as one package by Kris. > >>>>> > >>>>> Matt what's missing on the kcomd side? > >>>>> > >>>>> Moshe > >>>>> > >>>>> > >>>>> On Sep 18, 2006, at 4:24 AM, Carl Radford ((DC0854-IT)) wrote: > >>>>> > >>>>> > >>>>>> Hi Moshe, > >>>>>> > >>>>>> The 2.6 tools are not the problem (my version is already > >>>>>> > > usable, > > > >> > >> > >>>>>> awaiting user feedback). > >>>>>> Matt Dew has been putting some serious effort into kcomd, and > >>>>>> > >> needs > >> > >>>>>> the help of others to put some > >>>>>> effort in. The only person that appears to be onboard with > >>>>>> > > this > > > >> so > >> > >>>>>> far is Florian. > >>>>>> > >>>>>> -----Original Message----- > >>>>>> From: moshe bar [mailto:moshe-ay74M1d3r6RWk0Htik3J/[email protected]] > >>>>>> Sent: 17 September 2006 16:25 > >>>>>> To: [email protected] > >>>>>> Cc: openmosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org; Irfan Habib > >>>>>> Subject: Re: [openMosix-general] openmosix 2.4.26 in 2.4.31? > >>>>>> > >>>>>> > >>>>>> Martin > >>>>>> > >>>>>> Yes, we are glad to have oMFS and DFSA behind us. Using > >>>>>> > >> something > >> > >>>>>> like Red Hat's GFS is so much more better. > >>>>>> > >>>>>> David (Orcero), where are we with the user space tools? > >>>>>> > >>>>>> Moshe > >>>>>> > >>>>>> > >>>>>> On Sep 17, 2006, at 8:51 AM, Deadpan110 wrote: > >>>>>> > >>>>>> > >>>>>>> I read somewhere in the oM development user lists that > >>>>>>> > > someone > > > >> was > >> > >>>>>>> working on a 2.4.32 kernel patch (without checking, I wouldnt > >>>>>>> > >> know > >> > >>>>>>> how > >>>>>>> stable it is or even if it is finished). > >>>>>>> > >>>>>>> openMosix could definately do with a patch for 2.4.32 while > >>>>>>> > > we > > > >> have > >> > >>>>>>> the long wait for the 2.6.* versions with userspace tools. > >>>>>>> > >>>>>>> Out of curiosity, I tried the Mosix 2.4.32 kernel patch and > >>>>>>> > >> found > >> > >>>> it > >>>> > >>>>>>> fairly unstable along with the old problems of MFS and DFSA > >>>>>>> (expecially if you use pivot_root). > >>>>>>> > >>>>>>> It didn't take me long to decide to switch back to openMosix > >>>>>>> > >> 2.4.26 > >> > >>>>>>> instead! > >>>>>>> > >>>>>>> Martin > >>>>>>> > >>>>>>> On 15/09/06, James Wagner <jwagner-I/[email protected]> wrote: > >>>>>>> > >>>>>>>> I've tried to do so, but after 3 hours of beating my head > >>>>>>>> > > into > > > >> the > >> > >>>>>>>> wall > >>>>>>>> trying to determine the differences and exactly where to > >>>>>>>> > >> manually > >> > >>>>>>>> apply > >>>>>>>> the changes, gave up. It's far easier to simply download > >>>>>>>> > > the > > > >> full > >> > >>>>>>>> (already-patched) 2.4.26 kernel and configure lilo to > >>>>>>>> > >> dual-boot. > >> > >>>>>>>> I only > >>>>>>>> needed a cluster's power for a single project though, so > >>>>>>>> > > this > > > >> may > >> > >>>>>>>> not be > >>>>>>>> acceptable for your situation. > >>>>>>>> > >>>>>>>> I take it that you're another slackware fan? :) > >>>>>>>> > >>>>>>>> Jim > >>>>>>>> > >>>>>>>> Irfan Habib wrote: > >>>>>>>> > >>>>>>>> > >>>>>>>>> Hi, > >>>>>>>>> > >>>>>>>>> Is it possible for me to patch the 2.4.31 kernel with > >>>>>>>>> the available OM 2.4.26 patch? > >>>>>>>>> > >>>>>>>>> regards > >>>>>>>>> Irfan > >>>>>>>>> > >>>>>>>>> __________________________________________________ > >>>>>>>>> Do You Yahoo!? > >>>>>>>>> Tired of spam? Yahoo! Mail has the best spam protection > >>>>>>>>> > >> around > >> > >>>>>>>>> http://mail.yahoo.com > >>>>>>>>> > >>>>>>>>> > >>>>>>>>> > > ------------------------------------------------------------------- > > > >>>>>>>>> - > >>>>>>>>> ----- > >>>>>>>>> Using Tomcat but need to do more? Need to support web > >>>>>>>>> > >> services, > >> > >>>>>>>>> security? > >>>>>>>>> Get stuff done quickly with pre-integrated technology to > >>>>>>>>> > > make > > > >>>>>>>>> your job easier > >>>>>>>>> Download IBM WebSphere Application Server v.1.0.1 based on > >>>>>>>>> > >> Apache > >> > >>>>>>>>> Geronimo > >>>>>>>>> http://sel.as-us.falkag.net/sel? > >>>>>>>>> cmd=lnk&kid=120709&bid=263057&dat=121642 > >>>>>>>>> _______________________________________________ > >>>>>>>>> openMosix-general mailing list > >>>>>>>>> openMosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org > >>>>>>>>> > >>>>>>>>> > >> https://lists.sourceforge.net/lists/listinfo/openmosix-general > >> > >>>>>>>>> > >>>>>>>>> > >>>>>>>> > > -------------------------------------------------------------------- > > > >>>>>>>> - > >>>>>>>> ---- > >>>>>>>> Using Tomcat but need to do more? Need to support web > >>>>>>>> > >> services, > >> > >>>>>>>> security? > >>>>>>>> Get stuff done quickly with pre-integrated technology to > >>>>>>>> > > make > > > >> your > >> > >>>>>>>> job easier > >>>>>>>> Download IBM WebSphere Application Server v.1.0.1 based on > >>>>>>>> > >> Apache > >> > >>>>>>>> Geronimo > >>>>>>>> http://sel.as-us.falkag.net/sel? > >>>>>>>> cmd=lnk&kid=120709&bid=263057&dat=121642 > >>>>>>>> _______________________________________________ > >>>>>>>> openMosix-general mailing list > >>>>>>>> openMosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org > >>>>>>>> > >>>>>>>> > > https://lists.sourceforge.net/lists/listinfo/openmosix-general > > > >>>>>>> > >> --------------------------------------------------------------------- > >> > >>>>>>> - > >>>>>>> --- > >>>>>>> Using Tomcat but need to do more? Need to support web > >>>>>>> > > services, > > > >>>>>>> security? > >>>>>>> Get stuff done quickly with pre-integrated technology to make > >>>>>>> > >> your > >> > >>>>>>> job easier > >>>>>>> Download IBM WebSphere Application Server v.1.0.1 based on > >>>>>>> > >> Apache > >> > >>>>>>> Geronimo > >>>>>>> http://sel.as-us.falkag.net/sel? > >>>>>>> cmd=lnk&kid=120709&bid=263057&dat=121642 > >>>>>>> _______________________________________________ > >>>>>>> openMosix-general mailing list > >>>>>>> openMosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org > >>>>>>> > >>>>>>> > > https://lists.sourceforge.net/lists/listinfo/openmosix-general > > > >>>>>> > >>>>>> > >> ---------------------------------------------------------------------- > >> > > > > > >>>>>> --- > >>>>>> Using Tomcat but need to do more? Need to support web > >>>>>> > > services, > > > >>>>>> security? > >>>>>> Get stuff done quickly with pre-integrated technology to make > >>>>>> > >> your > >> > >>>>>> job easier > >>>>>> Download IBM WebSphere Application Server v.1.0.1 based on > >>>>>> > >> Apache > >> > >>>>>> Geronimo > >>>>>> http://sel.as-us.falkag.net/sel? > >>>>>> cmd=lnk&kid=120709&bid=263057&dat=121642 > >>>>>> _______________________________________________ > >>>>>> openMosix-general mailing list > >>>>>> openMosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org > >>>>>> https://lists.sourceforge.net/lists/listinfo/openmosix-general > >>>>>> > >>>>> > >>>>> > > ------------------------------------------------------------------------- > > > >>>>> Using Tomcat but need to do more? Need to support web services, > >>>>> > >>>> security? > >>>> > >>>>> Get stuff done quickly with pre-integrated technology to make > >>>>> > > your > > > >> job > >> > >>>> easier > >>>> > >>>>> Download IBM WebSphere Application Server v.1.0.1 based on > >>>>> > > Apache > > > >>>> Geronimo > >>>> > > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > > > >>>>> _______________________________________________ > >>>>> openMosix-devel mailing list > >>>>> openMosix-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org > >>>>> https://lists.sourceforge.net/lists/listinfo/openmosix-devel > >>>>> > >>>> > >>>> > > ------------------------------------------------------------------------- > > > >>>> Take Surveys. Earn Cash. Influence the Future of IT > >>>> Join SourceForge.net's Techsay panel and you'll get the chance to > >>>> > >> share your > >> > >>>> opinions on IT & business topics through brief surveys -- and earn > >>>> > >> cash > >> > > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > > > >>>> _______________________________________________ > >>>> openMosix-devel mailing list > >>>> openMosix-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org > >>>> https://lists.sourceforge.net/lists/listinfo/openmosix-devel > >>>> > >>>> > >>> -- > >>> Ian Latter > >>> Late night coder .. > >>> http://midnightcode.org/ > >>> > >>> > >> > >> > > ------------------------------------------------------------------------- > > > >> Take Surveys. Earn Cash. Influence the Future of IT > >> Join SourceForge.net's Techsay panel and you'll get the chance to > >> > > share your > > > >> opinions on IT & business topics through brief surveys -- and earn > >> > > cash > > > > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > > > >> _______________________________________________ > >> openMosix-devel mailing list > >> openMosix-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org > >> https://lists.sourceforge.net/lists/listinfo/openmosix-devel > >> > > > > > > ------------------------------------------------------------------------- > > Take Surveys. Earn Cash. Influence the Future of IT > > Join SourceForge.net's Techsay panel and you'll get the chance to share your > > opinions on IT & business topics through brief surveys -- and earn cash > > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > > _______________________________________________ > > openMosix-devel mailing list > > openMosix-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org > > https://lists.sourceforge.net/lists/listinfo/openmosix-devel > > > > > ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys -- and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV