Re: [Nfs-ganesha-devel] Ganesha 2.3.2 crash in avltree_insert

2016-07-11 Thread Matt Benjamin
The root cause of all previous stack traces of this type were caused by use-after free of an object containing node. Matt - Original Message - > From: "Krishna Harathi" > To: "nfs-ganesha-devel" > Sent: Monday, July 11,

[Nfs-ganesha-devel] Ganesha 2.3.2 crash in avltree_insert

2016-07-11 Thread Krishna Harathi
We got this crash, full stack trace is shown. Any help in resolving and fixing this is appreciated. struct avltree_node *avltree_insert(...) { > for (;;) { > if (parent->left == node) <- CRASH Here > dec_balance(parent); >

Re: [Nfs-ganesha-devel] EBADF error in fsal_rdwr() while writing VFS layer for VxFS

2016-07-11 Thread Frank Filz
This is one of the circumstances which is leading me to push for all FSALs to move to support_ex for 2.4. With support_ex, the FSAL is fully given the responsibility to manage file descriptors the way it needs to, dispensing with the shotgun content_lock from the old cache_inode days with the

[Nfs-ganesha-devel] EBADF error in fsal_rdwr() while writing VFS layer for VxFS

2016-07-11 Thread Vaibhav Gupta
Hi All, I am writing VFS layer for VxFS for Ganesha 2.4. While writing I am getting EIO error on Ganesha NFS V4 Server. Problem: In function fsal_rdwr(), if the file is already open then it does not open the file which races with the another thread