From: Vasanth <vas...@gm...> - 2007-11-24 21:24:08
|
Hi, I am using sshfs to mount some Xilinx tools remotely. I am able to read and write to the mounted file systems. However, when I run some applications, I encounter shared library load errors from the application. On digging deeper, it appears that open() system calls in the application fail with an EPERM error. However, the files do have valid permissions. Placing a local copy of the .so file and adding the local path to LD_LIBRARY_PATH causes the application to proceed further but then eventually fail with a similiar error on some other DLL. Copying all the DLLs locally and running the application works fine to completion. Some other applications work fine when run fully remotely. The error is always on the same .so file. Here is the log from running sshfs with a -d flag (from the machine that exports the file system) <snip> unique: 671, opcode: LOOKUP (1), nodeid: 4, insize: 55 LOOKUP /ISE9.2i/bin/lin64/libXst_Core.so NODEID: 105 unique: 671, error: 0 (Success), outsize: 136 unique: 672, opcode: OPEN (14), nodeid: 105, insize: 48 unique: 672, error: -1 (Operation not permitted), outsize: 16 </snip> I have tried all the -o options supported by sshfs and one of them seem to have any effect, except the -o use_ino flag. When using this flag, the .so load error is not seen anymore, but other DLLs run into a symbol lookup error. I have also tried the -o workaround=all option. Here are some details of my machine configuration: <snip> uname -a Linux domU-12-31-38-00-5D-71 2.6.16.33-xenU #2 SMP Wed Aug 15 17:27:36 SAST 2007 x86_64 x86_64 x86_64 GNU/Linux cat /proc/version Linux version 2.6.16.33-xenU (ro...@do...ernal) (gcc version 4.1.1 20070105 (Red Hat 4.1.1-52)) #2 SMP Wed Aug 15 17:27:36 SAST 2007 </snip> The machines are Fedora Core 6 images on a Xen virtual environment. Here are the FUSE version details: <snip> sshfs -V SSHFS version 1.8 FUSE library version: 2.7.0 fusermount version: 2.7.0 using FUSE kernel interface version 7.8 dmesg | grep fuse fuse init (API version 7.6) </snip> What could be going wrong? Is there anything I can try to workaround the problem? thanks, Vasanth |