vBulletin Search Engine Optimization
| |||||||
| Register | FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
| ||||
| I am using Gentoo with a modular vanilla kernel. I am getting an error "Cannot allocate memory" when I try to load the lockd kernel module: modprobe lockd Error inserting lockd (/lib/modules/2.6.22.9/kernel/fs/lockd/lockd.ko): Cannot allocate memory The top utility reveals: Mem 387268k total 15524k used 371744k free Swap 2706912k total 0k used 2706912k free Any ideas what is wrong? Thanks in advance to anyone who can help. Mark. -- Mark Hobley, 393 Quinton Road West, Quinton, BIRMINGHAM. B32 1QE. |
| |||
| In alt.os.linux.gentoo Mark Hobley <markhobley@hotpop.donottypethisbit.com> wrote: > The lsmod tool reveals that the sunrpc module has loaded successfully, > but lockd has not loaded. Do I need to configure some parameters somehow to load the lockd module? Or do I need a statd module? I don't appear to have such a module. Is there some way that I can get more verbose output relating to this error, or is there some sort of tool that can example the module file for errors, or possibly even load some sort or dummy module into the kernel that can explain why the module is failing? I could possibly do with a tool that does a walkthrough of the kernel modules, systematically loading and unloading them to test for errors. Any suggestions? Mark. -- Mark Hobley, 393 Quinton Road West, Quinton, BIRMINGHAM. B32 1QE. |
| |||
| In alt.os.linux.gentoo Mark Hobley <markhobley@hotpop.donottypethisbit.com> wrote: > I could possibly do with a tool that does a walkthrough of the kernel > modules, systematically loading and unloading them to test for errors. I just tried to insert a large selection of random modules using modprobe. They all load no problem, apart from the lockd module and its dependents nfs and nfsd. Mark. -- Mark Hobley, 393 Quinton Road West, Quinton, BIRMINGHAM. B32 1QE. |
| |||
| In alt.os.linux.gentoo Mark Hobley <markhobley@hotpop.donottypethisbit.com> wrote: > The kernel and modules were built at the same time, and this is a new > installation. There has never been a kernel on the system up to this > point. modinfo lockd produces the following output: filename: /lib/modules/2.6.22.9/kernel/fs/lockd/lockd.ko license: GPL description: NFS file locking service version 0.5. author: Olaf Kirch <okir@monad.swb.de> depends: sunrpc vermagic: 2.6.22.9 preempt mod_unload 586 parm: nsm_use_hostnames:bool I don't know what that last line is, my other (older) systems don't have this. Do I have to pass a parameter or something here? Mark. -- Mark Hobley, 393 Quinton Road West, Quinton, BIRMINGHAM. B32 1QE. |
| |||
| Mark Hobley <markhobley@hotpop.donottypethisbit.com> wrote: > In alt.os.linux.gentoo Mark Hobley > <markhobley@hotpop.donottypethisbit.com> wrote: > >> The lsmod tool reveals that the sunrpc module has loaded >> successfully, but lockd has not loaded. > > Do I need to configure some parameters somehow to load the lockd > module? Not to my knowledge. > Or do I need a statd module? I don't appear to have such a module. statd is implemented as an rpc service, and is part of net-fs/nfs-utils (which truly should be required if using NFS), and starts as part of the nfs start-up script. > Is there some way that I can get more verbose output relating to this > error, or is there some sort of tool that can example the module file > for errors, or possibly even load some sort or dummy module into the > kernel that can explain why the module is failing? I assume you have CONFIG_DEBUG_BUGVERBOSE=y in .config? Other than that, "rmmod lockd; modprobe -v lockd" might give /slightly/ more info. Regards, -- *Art |
| |||
| In alt.os.linux.gentoo Arthur Hagen <art@broomstick.com> wrote: > I assume you have CONFIG_DEBUG_BUGVERBOSE=y in .config? No I didn't because I never thought I would be having a problem like this. My systems have always been pretty reliable until now. I notice that the kernel has been updated, since I downloaded my version two months ago, and I think NFS is being aggressively developed, so maybe there are some updates. I possibly just chose a bad time to go for a build. I'll probably end up trying again at a later date with a newer kernel. I've stuck a note out to the NFS developer mailing list, so we shall see if I get any responses there. > Other than that, "rmmod lockd; modprobe -v lockd" might give /slightly/ more > info. Nahh, it just tells me that the error is occuring after running insmod, which I had already established. Mark. -- Mark Hobley, 393 Quinton Road West, Quinton, BIRMINGHAM. B32 1QE. |
| |||
| Mark Hobley wrote: > In alt.os.linux.gentoo Arthur Hagen <art@broomstick.com> wrote: > >> I assume you have CONFIG_DEBUG_BUGVERBOSE=y in .config? > > No I didn't because I never thought I would be having a problem like > this. My systems have always been pretty reliable until now. > > I notice that the kernel has been updated, since I downloaded my version > two months ago, and I think NFS is being aggressively developed, so > maybe there are some updates. > > I possibly just chose a bad time to go for a build. I'll probably end up > trying again at a later date with a newer kernel. > > I've stuck a note out to the NFS developer mailing list, so we shall see > if I get any responses there. I can say I haven't had any trouble with the later versions of NFS, but I have chosen to build NFS into my kernel instead of using modules. -- //Aho |
| |||
| In alt.os.linux.gentoo J.O. Aho <user@example.net> wrote: > > I can say I haven't had any trouble with the later versions of NFS, but I > have chosen to build NFS into my kernel instead of using modules. Yeah I reckon there is some bug in the lockd which causes it to fail as a module. (On my system everything is a module, if it is not boot critical.) Mark. -- Mark Hobley, 393 Quinton Road West, Quinton, BIRMINGHAM. B32 1QE. |
| |||
| Mark Hobley wrote: > In alt.os.linux.gentoo J.O. Aho <user@example.net> wrote: >> I can say I haven't had any trouble with the later versions of NFS, but I >> have chosen to build NFS into my kernel instead of using modules. > > Yeah I reckon there is some bug in the lockd which causes it to fail as > a module. (On my system everything is a module, if it is not boot critical.) Make it boot root from NFS then it's boot critical and you would build it in Myself I think everything that I would need if something would go wrong with the content of /lib/moudles, to still be able to do most basic stuff and as I have most of my good to use files NFS share, NFS been important for me to have in the kernel. -- //Aho |
| ||||
| J.O. Aho <user@example.net> wrote: > Mark Hobley wrote: >> In alt.os.linux.gentoo J.O. Aho <user@example.net> wrote: >>> I can say I haven't had any trouble with the later versions of NFS, >>> but I have chosen to build NFS into my kernel instead of using >>> modules. >> >> Yeah I reckon there is some bug in the lockd which causes it to fail >> as >> a module. (On my system everything is a module, if it is not boot >> critical.) > > Make it boot root from NFS then it's boot critical and you would > build it in > > Myself I think everything that I would need if something would go > wrong with the content of /lib/moudles, to still be able to do most > basic stuff and as I have most of my good to use files NFS share, NFS > been important for me to have in the kernel. On some prod systems, I put /everything/ in the kernel and disable modules. This both enhances security, slightly decreases memory useage and increases speed. The downside is to have to recompile a kernel if adding/modifying hardware or capabilities. Regards, -- *Art |