scsi: lpfc: Driver NVME load fails when CPU cnt > WQ resource cnt

If the cpu count is larger than the number of WQ resources available,
adapter attachment eventually failes due to a WQ_CREATE failure.

Calculate the number of WQs desired (which initializes to cpu count)
after accounting for the number of queues the adapter supports and the
number allocated to SCSI and the control/ELS path, and scale down if
necessary.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
This commit is contained in:
James Smart 2018-05-04 20:37:54 -07:00 committed by Martin K. Petersen
parent 23288b78a1
commit d38f33b304

View file

@ -7790,6 +7790,40 @@ lpfc_sli4_read_config(struct lpfc_hba *phba)
phba->sli4_hba.max_cfg_param.max_wq,
phba->sli4_hba.max_cfg_param.max_rq);
/*
* Calculate NVME queue resources based on how
* many WQ/CQs are available.
*/
if (phba->cfg_enable_fc4_type & LPFC_ENABLE_NVME) {
length = phba->sli4_hba.max_cfg_param.max_wq;
if (phba->sli4_hba.max_cfg_param.max_cq <
phba->sli4_hba.max_cfg_param.max_wq)
length = phba->sli4_hba.max_cfg_param.max_cq;
/*
* Whats left after this can go toward NVME.
* The minus 6 accounts for ELS, NVME LS, MBOX
* fof plus a couple extra. When configured for
* NVMET, FCP io channel WQs are not created.
*/
length -= 6;
if (!phba->nvmet_support)
length -= phba->cfg_fcp_io_channel;
if (phba->cfg_nvme_io_channel > length) {
lpfc_printf_log(
phba, KERN_ERR, LOG_SLI,
"2005 Reducing NVME IO channel to %d: "
"WQ %d CQ %d NVMEIO %d FCPIO %d\n",
length,
phba->sli4_hba.max_cfg_param.max_wq,
phba->sli4_hba.max_cfg_param.max_cq,
phba->cfg_nvme_io_channel,
phba->cfg_fcp_io_channel);
phba->cfg_nvme_io_channel = length;
}
}
}
if (rc)