From: Dominique L. <Dom...@TM...> - 2008-10-22 07:59:54
|
On Tue, 2008-10-21 at 09:51 -0700, Aaron Rolett wrote: > Dominique, > The vsock module warning messages have always been there (its > been on our list of things to cleanup for awhile). They appear because > the vsock build has no knowledge of the vmci module build even though > vsock depends on vmci symbols. I believe that the kernel module loader > changed sometime before 2.6.27 to prevent modules with unknown symbols > from loading if the kernel was built with CONFIG_MODVERSIONS. This is > something I was going to look into internally but haven't had a chance > to test out yet. > The reporters have seen the following from dmesg: > home/user/vmxnet3h/hosted08/bora# insmod build/obj-x64/ws/modules/ > local/2.6.27-4-generic/vsock/vsock.ko > insmod: error inserting 'build/obj-x64/ws/modules/local/2.6.27-4- > generic/vsock/vsock.ko': -1 Unknown symbol in module > > A quick test to see if this is the problem would be to: > 1. Build the vmci module > 2. Copy Module.symvers from vmci => vsock > 3. Build vsock > 4. Try loading the module. > Thank you very much for your help. I tried this. Built all the modules, before compiling vsock, I copied the Modules.symvers from vmci-only to vsock-only, and as suggested by you, vsock compiled correctly like this. As suggested by you, I can now load all the modules (vmblock, vmmon, vmnet, vmci and vsock). So a little bit better... just yet: /etc/init.d/vmware start fails on loading all of the modules. It seems not to be happy with this setup yet. When loading the modules manually, I don't have a /dev/vmnet0 does not get created (was probably normally done by the vmware init script.. will have to look into this). > All of that being said, ws shouldn't be segfaulting on startup (it > should be able to load and run fine without the vsock module ... you > just will not be able to use vmci sockets). > Indeed, this should not happen... I'm performing my tests on openSUSE 11.1beta3; openSUSE 11.1 is scheduled to be released before x-mas... and I hope it's in everybody's interest to snatch this out before the release... After all that's what for beta releases are. > A couple of questions: > 1. Which process specifically is dying? vmware / vmware-vmx something > else? When not having the modules loaded / compiled and I start vmware, I get this on the console: dle3ams@3120-2914:~/Desktop/RPMs/var/cache/vmware> vmware Logging to /tmp/vmware-dle3ams/setup-11031.log modinfo: could not find module vmmon modinfo: could not find module vmnet modinfo: could not find module vmblock modinfo: could not find module vmci modinfo: could not find module vsock modinfo: could not find module vmmon modinfo: could not find module vmnet modinfo: could not find module vmblock modinfo: could not find module vmci modinfo: could not find module vsock /usr/lib/vmware/bin/launcher.sh: line 231: 11031 Segmentation fault "$binary" "$@" The content of the mentioned log file is: dle3ams@3120-2914:~/Desktop/RPMs/var/cache/vmware> cat /tmp/vmware-dle3ams/setup-11031.log Oct 22 09:50:06.868: app| Log for VMware Workstation pid=11031 version=6.5.0 build=build-118166 option=Release Oct 22 09:50:06.868: app| Host codepage=UTF-8 encoding=UTF-8 Oct 22 09:50:06.868: app| Logging to /tmp/vmware-dle3ams/setup-11031.log on dmesg I get the following output: vmware-modconfi[10656]: segfault at 0 ip 00007f54bd915347 sp 00007fffccbde380 error 4 in libc-2.8.90.so[7f54bd896000+152000] A coredump is attached to this mail) > 2. Can you get us a core dump from the segfault? Of course.. attached > 3. Can you upload the relevant logs. On my machine they live in /tmp/ > vmware-${USER} and <VM Directory>/vmware.log Attached (the one from tmp)... the one from the VMwares itself is not very interesting, as vm gui does not even come up. > 4. Any other relevant debugging information you can think of. > Well, as said, I'm performing my tests on openSUSE 11.1, beta3, x86_64 setup. Not being a complete noob in the linux area I hope I can be of further assistance to you in finding out what goes wrong and testing possible solutions. In very best case, the problem is not caused by vmware ;) > Aaron Dominique |