Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Sagi Grimberg
It looks like this. What do you think Sagi? Yes, can you send a proper patch? You can add my: Reviewed-by: Sagi Grimberg

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Sagi Grimberg
It looks like this. What do you think Sagi? Yes, can you send a proper patch? You can add my: Reviewed-by: Sagi Grimberg

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Marta Rybczynska
> On 6/5/2017 12:45 PM, Marta Rybczynska wrote: >> This patch improves the way the RDMA IB signalling is done >> by using atomic operations for the signalling variable. This >> avoids race conditions on sig_count. >> >> The signalling interval changes slightly and is now the >> largest power of

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Marta Rybczynska
> On 6/5/2017 12:45 PM, Marta Rybczynska wrote: >> This patch improves the way the RDMA IB signalling is done >> by using atomic operations for the signalling variable. This >> avoids race conditions on sig_count. >> >> The signalling interval changes slightly and is now the >> largest power of

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Max Gurtovoy
On 6/5/2017 12:45 PM, Marta Rybczynska wrote: This patch improves the way the RDMA IB signalling is done by using atomic operations for the signalling variable. This avoids race conditions on sig_count. The signalling interval changes slightly and is now the largest power of two not larger

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Max Gurtovoy
On 6/5/2017 12:45 PM, Marta Rybczynska wrote: This patch improves the way the RDMA IB signalling is done by using atomic operations for the signalling variable. This avoids race conditions on sig_count. The signalling interval changes slightly and is now the largest power of two not larger

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Marta Rybczynska
t; "keith busch" <keith.bu...@intel.com>, "Doug Ledford" <dledf...@redhat.com>, > "Bart Van Assche" > <bart.vanass...@sandisk.com>, h...@lst.de, "Jason Gunthorpe" > <jguntho...@obsidianresearch.com> > Envoyé: Lundi 5 Jui

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Marta Rybczynska
Van Assche" > , h...@lst.de, "Jason Gunthorpe" > > Envoyé: Lundi 5 Juin 2017 13:39:40 > Objet: Re: [PATCH] nvme-rdma: remove race conditions from IB signalling >>> -static inline int nvme_rdma_queue_sig_limit(struct nvme_rdma_qu

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Marta Rybczynska
>> -static inline int nvme_rdma_queue_sig_limit(struct nvme_rdma_queue *queue) >> +static inline int nvme_rdma_init_sig_count(int queue_size) >> { >> - int sig_limit; >> - >> - /* >> -* We signal completion every queue depth/2 and also handle the >> -* degenerated

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Marta Rybczynska
>> -static inline int nvme_rdma_queue_sig_limit(struct nvme_rdma_queue *queue) >> +static inline int nvme_rdma_init_sig_count(int queue_size) >> { >> - int sig_limit; >> - >> - /* >> -* We signal completion every queue depth/2 and also handle the >> -* degenerated

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Sagi Grimberg
On 05/06/17 12:45, Marta Rybczynska wrote: This patch improves the way the RDMA IB signalling is done by using atomic operations for the signalling variable. This avoids race conditions on sig_count. The signalling interval changes slightly and is now the largest power of two not larger than

Re: [PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Sagi Grimberg
On 05/06/17 12:45, Marta Rybczynska wrote: This patch improves the way the RDMA IB signalling is done by using atomic operations for the signalling variable. This avoids race conditions on sig_count. The signalling interval changes slightly and is now the largest power of two not larger than

[PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Marta Rybczynska
This patch improves the way the RDMA IB signalling is done by using atomic operations for the signalling variable. This avoids race conditions on sig_count. The signalling interval changes slightly and is now the largest power of two not larger than queue depth / 2. ilog() usage idea by Bart Van

[PATCH] nvme-rdma: remove race conditions from IB signalling

2017-06-05 Thread Marta Rybczynska
This patch improves the way the RDMA IB signalling is done by using atomic operations for the signalling variable. This avoids race conditions on sig_count. The signalling interval changes slightly and is now the largest power of two not larger than queue depth / 2. ilog() usage idea by Bart Van