Re: [Cluster-devel] [DLM PATCH] DLM: Don't specify WQ_UNBOUND for the ast callback workqueue

2016-10-19 Thread David Teigland
On Wed, Oct 19, 2016 at 11:34:54AM -0400, Bob Peterson wrote: > Hi, > > This patch removes the WQ_UNBOUND flag (which implies WQ_HIGHPRI) > from the DLM's ast work queue, in favor of just WQ_HIGHPRI. > This has been shown to cause a 19 percent performance increase for > simultaneous inode creates

Re: [Cluster-devel] [DLM PATCH] DLM: Remove lock_sock to avoid scheduling while atomic

2016-10-19 Thread Bob Peterson
- Original Message - | Hi, | | Before this patch, functions save_callbacks and restore_callbacks | called function lock_sock and release_sock to prevent other processes | from messing with the struct sock while the callbacks were saved and | restored. However, function add_sock calls write

Re: [Cluster-devel] [DLM PATCH] DLM: Don't save callbacks after accept

2016-10-19 Thread Bob Peterson
- Original Message - | Hi, | | When DLM calls accept() on a socket, the comm code copies the sk | after we've saved its callbacks. Afterward, it calls add_sock which | saves the callbacks a second time. Since the error reporting function | lowcomms_error_report calls the previous callback

[Cluster-devel] [DLM PATCH] DLM: Don't specify WQ_UNBOUND for the ast callback workqueue

2016-10-19 Thread Bob Peterson
Hi, This patch removes the WQ_UNBOUND flag (which implies WQ_HIGHPRI) from the DLM's ast work queue, in favor of just WQ_HIGHPRI. This has been shown to cause a 19 percent performance increase for simultaneous inode creates on GFS2 with fs_mark. Signed-off-by: Bob Peterson --- diff --git a/fs/dl