[Cluster-devel] [PATCH] socket: move compat timeout handling into sock.c

2019-01-08 Thread Arnd Bergmann
This is a cleanup to prepare for the addition of 64-bit time_t in O_SNDTIMEO/O_RCVTIMEO. The existing compat handler seems unnecessarily complex and error-prone, moving it all into the main setsockopt()/getsockopt() implementation requires half as much code and is easier to extend. 32-bit user

Re: [Cluster-devel] [PATCH 2/3] socket: Rename SO_RCVTIMEO/ SO_SNDTIMEO with _OLD suffixes

2019-01-08 Thread Deepa Dinamani
On Tue, Jan 8, 2019 at 12:04 PM Arnd Bergmann wrote: > > On Tue, Jan 8, 2019 at 6:24 AM Deepa Dinamani wrote: > > > > SO_RCVTIMEO and SO_SNDTIMEO socket options use struct timeval > > as the time format. struct timeval is not y2038 safe. > > The subsequent patches in the series add support for

Re: [Cluster-devel] [PATCH] socket: move compat timeout handling into sock.c

2019-01-08 Thread Deepa Dinamani
On Tue, Jan 8, 2019 at 12:10 PM Arnd Bergmann wrote: > > This is a cleanup to prepare for the addition of 64-bit time_t > in O_SNDTIMEO/O_RCVTIMEO. The existing compat handler seems > unnecessarily complex and error-prone, moving it all into the > main setsockopt()/getsockopt() implementation

Re: [Cluster-devel] [PATCH 2/3] socket: Rename SO_RCVTIMEO/ SO_SNDTIMEO with _OLD suffixes

2019-01-08 Thread Arnd Bergmann
On Tue, Jan 8, 2019 at 6:24 AM Deepa Dinamani wrote: > > SO_RCVTIMEO and SO_SNDTIMEO socket options use struct timeval > as the time format. struct timeval is not y2038 safe. > The subsequent patches in the series add support for new socket > timeout options with _NEW suffix that are y2038 safe.

[Cluster-devel] kernel BUG at fs/gfs2/inode.h:64

2019-01-08 Thread Mark Syms
Hi, We've seen this in testing with 4.19. Full trace is at the bottom. Looking at the code though it looks like it will assert if the value of change is equal to the number of blocks currently allocated to the inode. Is this expected or should the assert be using >= instead of > ? Thanks,

[Cluster-devel] [PATCH 3/3] dlm: allow binding to all network interfaces

2019-01-08 Thread David Windsor
Currently, in the kernel, DLM only is able to bind its listen socket to a single network interface. To support more robust network configurations, DLM should be able to bind to all network interfaces. This patch adds a configfs node to enable/disable binding to all network interfaces. When 1 is

[Cluster-devel] [PATCH 1/3] dlm: check if workqueues are NULL before destroying

2019-01-08 Thread David Windsor
If a network failure occurs before any DLM traffic can be generated, the send and receive workqueues can be NULL when work_stop() is called. Check to see if these workqueues are NULL before calling destroy_workqueue(). Signed-off-by: David Windsor --- fs/dlm/lowcomms.c | 6 -- 1 file

[Cluster-devel] [PATCH 2/3] dlm: add TCP multihoming/failover support

2019-01-08 Thread David Windsor
Add the ability to specify multiple source addresses for DLM nodes so that multihomed configurations can use multiple addresses and still be recognized by the receiving node. While each node is capable of being configured for multiple IPs, DLM requires each node have only one active address at a