Hi James,
On Tue, Jan 17, 2017 at 05:20:45PM -0800, James Smart wrote:
>
> NVME Initiator: Base modifications
>
> This is part A of parts A..F.
>
> Part A is the bulk of the file list changed by the modifications as
> most mods are small-ish. Changes may be for any aspect below.
>
> *********
>
> This set of patches (6 parts) adds base modifications for NVME initiator
> support to the lpfc driver.
>
> The base modifications consist of:
> - Formal split of SLI3 rings from SLI-4 WQs (sometimes referred to as
> rings as well) as implementation now widely varies between the two.
> - Addition of configuration modes:
> SCSI initiator only; NVME initiator only; NVME target only; and
> SCSI and NVME initiator.
> The configuration mode drives overall adapter configuration,
> offloads enabled, and resource splits.
> NVME support is only available on SLI-4 devices and newer fw.
> - Implements the following based on configuration mode:
> - Exchange resources are split by protocol; Obviously, if only
> 1 mode, then no split occurs. Default is 50/50. module attribute
> allows tuning.
> - Each protocol has it's own set of queues, but share interrupt
> vectors.
> SCSI:
> SLI3 devices have few queues and the original style of queue
> allocation remains.
> SLI4 devices piggy back on an "io-channel" concept that
> eventually needs to merge with scsi-mq/blk-mq support (it is
> underway). For now, the paradigm continues as it existed
> prior. io channel allocates N msix and N WQs (N=4 default)
> and either round robins or uses cpu # modulo N for scheduling.
> A bunch of module parameters allow the configuration to be
> tuned.
> NVME (initiator):
> Allocates an msix per cpu (or whatever pci_alloc_irq_vectors
> gets)
> Allocates a WQ per cpu, and maps the WQs to msix on a WQ #
> modulo msix vector count basis.
> Module parameters exist to cap/control the config if desired.
> - Each protocol has its own buffer and dma pools.
>
> Unfortunately, it was very difficult to break out the above into
> functional patches. A such, they are presented as a 6-patch set to
> keep email size reasonable. All patches in the set must be applied to
> create a functional unit.
>
> Signed-off-by: Dick Kennedy <dick.kenn...@broadcom.com>
> Signed-off-by: James Smart <james.sm...@broadcom.com>
> ---
[...]
> @@ -2704,7 +2710,7 @@ static int lpfcdiag_loop_get_xri(struct lpfc_hba *phba,
> uint16_t rpi,
> * @phba: Pointer to HBA context object
> *
> * This function allocates BSG_MBOX_SIZE (4KB) page size dma buffer and.
> - * returns the pointer to the buffer.
> + * retruns the pointer to the buffer.
This hunk introduces a spelling error.
> **/
> static struct lpfc_dmabuf *
> lpfc_bsg_dma_page_alloc(struct lpfc_hba *phba)
> @@ -2876,7 +2882,7 @@ static int lpfcdiag_loop_post_rxbufs(struct lpfc_hba
> *phba, uint16_t rxxri,
> size_t len)
> {
> struct lpfc_sli *psli = &phba->sli;
> - struct lpfc_sli_ring *pring = &psli->ring[LPFC_ELS_RING];
> + struct lpfc_sli_ring *pring;
> struct lpfc_iocbq *cmdiocbq;
> IOCB_t *cmd = NULL;
> struct list_head head, *curr, *next;
> @@ -2890,6 +2896,11 @@ static int lpfcdiag_loop_post_rxbufs(struct lpfc_hba
> *phba, uint16_t rxxri,
> int iocb_stat;
> int i = 0;
>
> + if (phba->sli_rev == LPFC_SLI_REV4)
> + pring = phba->sli4_hba.els_wq->pring;
> + else
> + pring = &psli->sli3_ring[LPFC_ELS_RING];
> +
> cmdiocbq = lpfc_sli_get_iocbq(phba);
> rxbmp = kmalloc(sizeof(struct lpfc_dmabuf), GFP_KERNEL);
> if (rxbmp != NULL) {
> @@ -5258,7 +5269,7 @@ static int
> lpfc_forced_link_speed(struct bsg_job *job)
> {
> struct Scsi_Host *shost = fc_bsg_to_shost(job);
> - struct lpfc_vport *vport = shost_priv(shost);
> + struct lpfc_vport *vport = (struct lpfc_vport *)shost->hostdata;
Please don't. A) it's unnecessary to cast a void* and b) shost_priv() just
returns shost->hostdata (casted to void*). So all's fine.
[...]
> diff --git a/drivers/scsi/lpfc/lpfc_disc.h b/drivers/scsi/lpfc/lpfc_disc.h
> index 361f5b3..1d07a5f 100644
> --- a/drivers/scsi/lpfc/lpfc_disc.h
> +++ b/drivers/scsi/lpfc/lpfc_disc.h
> @@ -133,6 +133,7 @@ struct lpfc_node_rrq {
> /* Defines for nlp_flag (uint32) */
> #define NLP_IGNR_REG_CMPL 0x00000001 /* Rcvd rscn before we cmpl reg login
> */
> #define NLP_REG_LOGIN_SEND 0x00000002 /* sent reglogin to adapter */
> +#define NLP_SUPPRESS_RSP 0x00000010 /* Remote NPort supports
> suppress rsp */
> #define NLP_PLOGI_SND 0x00000020 /* sent PLOGI request for this
> entry */
> #define NLP_PRLI_SND 0x00000040 /* sent PRLI request for this
> entry */
> #define NLP_ADISC_SND 0x00000080 /* sent ADISC request for this
> entry */
> diff --git a/drivers/scsi/lpfc/lpfc_els.c b/drivers/scsi/lpfc/lpfc_els.c
> index 27f0cbb..f61e1c2 100644
> --- a/drivers/scsi/lpfc/lpfc_els.c
> +++ b/drivers/scsi/lpfc/lpfc_els.c
> @@ -1323,7 +1323,10 @@ lpfc_els_abort_flogi(struct lpfc_hba *phba)
> "0201 Abort outstanding I/O on NPort x%x\n",
> Fabric_DID);
>
> - pring = &phba->sli.ring[LPFC_ELS_RING];
> + if (phba->sli_rev != LPFC_SLI_REV4)
> + pring = &phba->sli.sli3_ring[LPFC_ELS_RING];
> + else
> + pring = phba->sli4_hba.els_wq->pring;
>
You're doing this quite often, can you make a small helper for it?
static inline struct lpfc_sli_ring *lpfc_sli_ring_from_hba(struct lpfc_hba
*phba)
{
if (phba->sli_rev != LPFC_SLI_REV4)
pring = &phba->sli.sli3_ring[LPFC_ELS_RING];
else
pring = phba->sli4_hba.els_wq->pring;
}
[...]
> @@ -1751,7 +1757,7 @@ lpfc_sli4_new_fcf_random_select(struct lpfc_hba *phba,
> uint32_t fcf_cnt)
> uint32_t rand_num;
>
> /* Get 16-bit uniform random number */
> - rand_num = 0xFFFF & prandom_u32();
> + rand_num = (0xFFFF & prandom_u32());
Unnecessary introduction of parenthesis.
[...]
> @@ -4452,28 +4471,51 @@ lpfc_no_rpi(struct lpfc_hba *phba, struct
> lpfc_nodelist *ndlp)
> psli = &phba->sli;
> if (ndlp->nlp_flag & NLP_RPI_REGISTERED) {
> /* Now process each ring */
> - for (i = 0; i < psli->num_rings; i++) {
> - pring = &psli->ring[i];
> -
> - spin_lock_irq(&phba->hbalock);
> - list_for_each_entry_safe(iocb, next_iocb, &pring->txq,
> - list) {
> - /*
> - * Check to see if iocb matches the nport we are
> - * looking for
> - */
> + if (phba->sli_rev != LPFC_SLI_REV4) {
> + for (i = 0; i < psli->num_rings; i++) {
> + pring = &psli->sli3_ring[i];
> + spin_lock_irq(&phba->hbalock);
> + list_for_each_entry_safe(iocb, next_iocb,
> + &pring->txq, list) {
> + /*
> + * Check to see if iocb matches the
> + * nport we are looking for
> + */
> + if ((lpfc_check_sli_ndlp(phba, pring,
> + iocb, ndlp))) {
> + /* It matches, so deque and
> + * call compl with an error
> + */
> + list_move_tail(&iocb->list,
> + &completions);
> + }
> + }
> + spin_unlock_irq(&phba->hbalock);
> + }
> + goto out;
> + }
Can you please factor the for loop into an own function so that we get
if (phba->sli_rev != LPFC_SLI_REV4)
lpfc_handle_sli_rev4();
The level of indentation is just too high. Btw checkpatch.pl tells you about
it with '--type DEEP_INDENTATION'.
> + spin_lock_irq(&phba->hbalock);
> + list_for_each_entry(qp, &phba->sli4_hba.lpfc_wq_list, wq_list) {
> + pring = qp->pring;
> + if (!pring)
> + continue;
> + spin_lock_irq(&pring->ring_lock);
> + list_for_each_entry_safe(iocb, next_iocb,
> + &pring->txq, list) {
> if ((lpfc_check_sli_ndlp(phba, pring, iocb,
> ndlp))) {
> - /* It matches, so deque and call compl
> - with an error */
> + /* It matches, so deque and
> + * call compl with an error
> + */
> list_move_tail(&iocb->list,
> &completions);
> }
> }
> - spin_unlock_irq(&phba->hbalock);
> + spin_unlock_irq(&pring->ring_lock);
> }
like above, is it possible to do
spin_lock_irq(...);
handler();
spin_unlock_irq();
This makes it way easier to follow.
[...]
> @@ -5207,7 +5248,10 @@ lpfc_free_tx(struct lpfc_hba *phba, struct
> lpfc_nodelist *ndlp)
> struct lpfc_sli_ring *pring;
>
> psli = &phba->sli;
> - pring = &psli->ring[LPFC_ELS_RING];
> + if (phba->sli_rev != LPFC_SLI_REV4)
> + pring = &phba->sli.sli3_ring[LPFC_ELS_RING];
> + else
> + pring = phba->sli4_hba.els_wq->pring;
>
you really might want to write a coccinelle rule for this pattern...
[...]
> diff --git a/drivers/scsi/lpfc/lpfc_scsi.c b/drivers/scsi/lpfc/lpfc_scsi.c
> index 19d349f..c60dfc9 100644
> --- a/drivers/scsi/lpfc/lpfc_scsi.c
> +++ b/drivers/scsi/lpfc/lpfc_scsi.c
> @@ -242,6 +242,7 @@ lpfc_update_stats(struct lpfc_hba *phba, struct
> lpfc_scsi_buf *lpfc_cmd)
> spin_unlock_irqrestore(shost->host_lock, flags);
> }
>
> +
Stray newine.
[...]
> @@ -605,7 +610,7 @@ lpfc_sli4_fcp_xri_aborted(struct lpfc_hba *phba,
> }
>
> /**
> - * lpfc_sli4_post_scsi_sgl_list - Post blocks of scsi buffer sgls from a list
> + * lpfc_sli4_post_scsi_sgl_list - Psot blocks of scsi buffer sgls from a list
^^^ o.O
[...]
> @@ -3894,7 +3896,7 @@ int lpfc_sli4_scmd_to_wqidx_distr(struct lpfc_hba *phba,
> }
> }
> chann = atomic_add_return(1, &phba->fcp_qidx);
> - chann = (chann % phba->cfg_fcp_io_channel);
> + chann = (chann % phba->cfg_fcp_max_hw_queue);
No need for the parenthesis.
[...]
> @@ -4959,11 +4968,11 @@ lpfc_send_taskmgmt(struct lpfc_vport *vport, struct
> scsi_cmnd *cmnd,
> int status;
>
> rdata = lpfc_rport_data_from_scsi_device(cmnd->device);
> - if (!rdata || !rdata->pnode || !NLP_CHK_NODE_ACT(rdata->pnode))
> - return FAILED;
OK, I don't get this hunk. lpfc_rport_data_from_scsi_device() cannot return
NULL anymore?
I at least expected something like:
rdata = lpfc_rport_data_from_scsi_device(cmnd->device);
if (!rdata || !rdata->pnode)
return FAILED;
pnode = rdata->pnode;
if (!NLP_CHK_NODE_ACT(pnode)
return FAILED;
> pnode = rdata->pnode;
> + if (!pnode || !NLP_CHK_NODE_ACT(pnode))
> + return FAILED;
>
> - lpfc_cmd = lpfc_get_scsi_buf(phba, pnode);
> + lpfc_cmd = lpfc_get_scsi_buf(phba, rdata->pnode);
> if (lpfc_cmd == NULL)
> return FAILED;
> lpfc_cmd->timeout = phba->cfg_task_mgmt_tmo;
[...]
--
Johannes Thumshirn Storage
jthumsh...@suse.de +49 911 74053 689
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: Felix Imendörffer, Jane Smithard, Graham Norton
HRB 21284 (AG Nürnberg)
Key fingerprint = EC38 9CAB C2C4 F25D 8600 D0D0 0393 969D 2D76 0850
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
