On Thu, 23 Sep 2010, Henry C Chang wrote: > Hi Sage, > > There are two patches on our github for the ceph-client-standalone. > Would you also help to take a look at them? > > 1. http://github.com/tcloud/ceph-client-standalone/commit/121810874701393756dc21e4415e9ccf0dfb3b06 > > This is another mistake about list_* function series. Great, fixed this to ceph-client.git. That's the third one of these now! :) > 2. http://github.com/tcloud/ceph-client-standalone/commit/2c9c3cf2a9234b6b7a2b7e7ee12441d5a82bfb69 > > We are using kernel 2.6.32. I am not sure whether all kernel<2.6.35 > have the same behavior. > We hit this bug when re-exporting ceph via NFS. The NFS client can > mount ceph successfully, > but when copying in a file, NULL pointer exception occurs. > > We found that when ceph_dir_fsync(struct file *file, struct dentry > *dentry, int datasync) is called, > the "file" pointer passed in is NULL. Thus, we use "dentry" to get the > inode ptr instead of using "file". Yeah, your fix looks good. It should apply to all versions prior to the API change (.35?). Applied this to ceph-client-standalone.git. Compile tested on .27 to .35. > BTW, could you backport recent patches in ceph-client to > ceph-client-standalone? I've updated the master and master-backport branches. The unstable branch includes the code refactor into multiple modules (libceph, ceph, and rbd), which will require some reorganization and changes to my update scripts. Thanks! sage -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
