Hello jemicheng, As you might know, your version linux-2.6.38.8 and "aufs 2.2-standalone.tree-38-20111024" is too old, and are not maintained now.
While your report is good since the necessary information is provided as aufs README file described, ideally the all kernel messages should be provided instead of extracting the call trace. Anyway I've checked shmem_getpage() in linux-2.6.38.0 and found several places calling __alloc_pages_nodemask(). I don't know which one is the trigger of your problem. Additionally at the top of your call trace, we can see > [52226.790000] [<c00983a8>] (__alloc_pages_nodemask+0x0/0x728) from It looks strage to me. Why 0x0? I am afraid something wrong happened before aufs trunc_xino is activated. Didn't there happen any other error on your system? Last but not at least, I'd strongly suggest you to upgrade your kernel as 2.6.38 is toooo old. If you really think the problem is really related to aufs trunc_xino, then I'd suggest you not to specify trunc_xino because several people reported the old behaviour was not so effective. The most effective solution to reduce the size of XINO files is tmpfs-idr.patch in aufs3-standalone.git. Of cource it doesn't support such old kernel 2.6.38. J. R. Okajima ------------------------------------------------------------------------------
