linux_old1/drivers/scsi/lpfc/lpfc_scsi.c

1288 lines
36 KiB
C
Raw Normal View History

/*******************************************************************
* This file is part of the Emulex Linux Device Driver for *
* Fibre Channel Host Bus Adapters. *
* Copyright (C) 2004-2006 Emulex. All rights reserved. *
* EMULEX and SLI are trademarks of Emulex. *
* www.emulex.com *
* Portions Copyright (C) 2004-2005 Christoph Hellwig *
* *
* This program is free software; you can redistribute it and/or *
* modify it under the terms of version 2 of the GNU General *
* Public License as published by the Free Software Foundation. *
* This program is distributed in the hope that it will be useful. *
* ALL EXPRESS OR IMPLIED CONDITIONS, REPRESENTATIONS AND *
* WARRANTIES, INCLUDING ANY IMPLIED WARRANTY OF MERCHANTABILITY, *
* FITNESS FOR A PARTICULAR PURPOSE, OR NON-INFRINGEMENT, ARE *
* DISCLAIMED, EXCEPT TO THE EXTENT THAT SUCH DISCLAIMERS ARE HELD *
* TO BE LEGALLY INVALID. See the GNU General Public License for *
* more details, a copy of which can be found in the file COPYING *
* included with this package. *
*******************************************************************/
#include <linux/pci.h>
#include <linux/interrupt.h>
#include <linux/delay.h>
#include <scsi/scsi.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_tcq.h>
#include <scsi/scsi_transport_fc.h>
#include "lpfc_version.h"
#include "lpfc_hw.h"
#include "lpfc_sli.h"
#include "lpfc_disc.h"
#include "lpfc_scsi.h"
#include "lpfc.h"
#include "lpfc_logmsg.h"
#include "lpfc_crtn.h"
#define LPFC_RESET_WAIT 2
#define LPFC_ABORT_WAIT 2
/*
* This routine allocates a scsi buffer, which contains all the necessary
* information needed to initiate a SCSI I/O. The non-DMAable buffer region
* contains information to build the IOCB. The DMAable region contains
* memory for the FCP CMND, FCP RSP, and the inital BPL. In addition to
* allocating memeory, the FCP CMND and FCP RSP BDEs are setup in the BPL
* and the BPL BDE is setup in the IOCB.
*/
static struct lpfc_scsi_buf *
lpfc_new_scsi_buf(struct lpfc_hba * phba)
{
struct lpfc_scsi_buf *psb;
struct ulp_bde64 *bpl;
IOCB_t *iocb;
dma_addr_t pdma_phys;
uint16_t iotag;
psb = kmalloc(sizeof(struct lpfc_scsi_buf), GFP_KERNEL);
if (!psb)
return NULL;
memset(psb, 0, sizeof (struct lpfc_scsi_buf));
psb->scsi_hba = phba;
/*
* Get memory from the pci pool to map the virt space to pci bus space
* for an I/O. The DMA buffer includes space for the struct fcp_cmnd,
* struct fcp_rsp and the number of bde's necessary to support the
* sg_tablesize.
*/
psb->data = pci_pool_alloc(phba->lpfc_scsi_dma_buf_pool, GFP_KERNEL,
&psb->dma_handle);
if (!psb->data) {
kfree(psb);
return NULL;
}
/* Initialize virtual ptrs to dma_buf region. */
memset(psb->data, 0, phba->cfg_sg_dma_buf_size);
/* Allocate iotag for psb->cur_iocbq. */
iotag = lpfc_sli_next_iotag(phba, &psb->cur_iocbq);
if (iotag == 0) {
pci_pool_free(phba->lpfc_scsi_dma_buf_pool,
psb->data, psb->dma_handle);
kfree (psb);
return NULL;
}
psb->cur_iocbq.iocb_flag |= LPFC_IO_FCP;
psb->fcp_cmnd = psb->data;
psb->fcp_rsp = psb->data + sizeof(struct fcp_cmnd);
psb->fcp_bpl = psb->data + sizeof(struct fcp_cmnd) +
sizeof(struct fcp_rsp);
/* Initialize local short-hand pointers. */
bpl = psb->fcp_bpl;
pdma_phys = psb->dma_handle;
/*
* The first two bdes are the FCP_CMD and FCP_RSP. The balance are sg
* list bdes. Initialize the first two and leave the rest for
* queuecommand.
*/
bpl->addrHigh = le32_to_cpu(putPaddrHigh(pdma_phys));
bpl->addrLow = le32_to_cpu(putPaddrLow(pdma_phys));
bpl->tus.f.bdeSize = sizeof (struct fcp_cmnd);
bpl->tus.f.bdeFlags = BUFF_USE_CMND;
bpl->tus.w = le32_to_cpu(bpl->tus.w);
bpl++;
/* Setup the physical region for the FCP RSP */
pdma_phys += sizeof (struct fcp_cmnd);
bpl->addrHigh = le32_to_cpu(putPaddrHigh(pdma_phys));
bpl->addrLow = le32_to_cpu(putPaddrLow(pdma_phys));
bpl->tus.f.bdeSize = sizeof (struct fcp_rsp);
bpl->tus.f.bdeFlags = (BUFF_USE_CMND | BUFF_USE_RCV);
bpl->tus.w = le32_to_cpu(bpl->tus.w);
/*
* Since the IOCB for the FCP I/O is built into this lpfc_scsi_buf,
* initialize it with all known data now.
*/
pdma_phys += (sizeof (struct fcp_rsp));
iocb = &psb->cur_iocbq.iocb;
iocb->un.fcpi64.bdl.ulpIoTag32 = 0;
iocb->un.fcpi64.bdl.addrHigh = putPaddrHigh(pdma_phys);
iocb->un.fcpi64.bdl.addrLow = putPaddrLow(pdma_phys);
iocb->un.fcpi64.bdl.bdeSize = (2 * sizeof (struct ulp_bde64));
iocb->un.fcpi64.bdl.bdeFlags = BUFF_TYPE_BDL;
iocb->ulpBdeCount = 1;
iocb->ulpClass = CLASS3;
return psb;
}
static struct lpfc_scsi_buf*
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
lpfc_get_scsi_buf(struct lpfc_hba * phba)
{
struct lpfc_scsi_buf * lpfc_cmd = NULL;
struct list_head *scsi_buf_list = &phba->lpfc_scsi_buf_list;
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
unsigned long iflag = 0;
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
spin_lock_irqsave(&phba->scsi_buf_list_lock, iflag);
list_remove_head(scsi_buf_list, lpfc_cmd, struct lpfc_scsi_buf, list);
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
spin_unlock_irqrestore(&phba->scsi_buf_list_lock, iflag);
return lpfc_cmd;
}
static void
lpfc_release_scsi_buf(struct lpfc_hba * phba, struct lpfc_scsi_buf * psb)
{
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
unsigned long iflag = 0;
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
spin_lock_irqsave(&phba->scsi_buf_list_lock, iflag);
psb->pCmd = NULL;
list_add_tail(&psb->list, &phba->lpfc_scsi_buf_list);
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
spin_unlock_irqrestore(&phba->scsi_buf_list_lock, iflag);
}
static int
lpfc_scsi_prep_dma_buf(struct lpfc_hba * phba, struct lpfc_scsi_buf * lpfc_cmd)
{
struct scsi_cmnd *scsi_cmnd = lpfc_cmd->pCmd;
struct scatterlist *sgel = NULL;
struct fcp_cmnd *fcp_cmnd = lpfc_cmd->fcp_cmnd;
struct ulp_bde64 *bpl = lpfc_cmd->fcp_bpl;
IOCB_t *iocb_cmd = &lpfc_cmd->cur_iocbq.iocb;
dma_addr_t physaddr;
uint32_t i, num_bde = 0;
int datadir = scsi_cmnd->sc_data_direction;
int dma_error;
/*
* There are three possibilities here - use scatter-gather segment, use
* the single mapping, or neither. Start the lpfc command prep by
* bumping the bpl beyond the fcp_cmnd and fcp_rsp regions to the first
* data bde entry.
*/
bpl += 2;
if (scsi_cmnd->use_sg) {
/*
* The driver stores the segment count returned from pci_map_sg
* because this a count of dma-mappings used to map the use_sg
* pages. They are not guaranteed to be the same for those
* architectures that implement an IOMMU.
*/
sgel = (struct scatterlist *)scsi_cmnd->request_buffer;
lpfc_cmd->seg_cnt = dma_map_sg(&phba->pcidev->dev, sgel,
scsi_cmnd->use_sg, datadir);
if (lpfc_cmd->seg_cnt == 0)
return 1;
if (lpfc_cmd->seg_cnt > phba->cfg_sg_seg_cnt) {
printk(KERN_ERR "%s: Too many sg segments from "
"dma_map_sg. Config %d, seg_cnt %d",
__FUNCTION__, phba->cfg_sg_seg_cnt,
lpfc_cmd->seg_cnt);
dma_unmap_sg(&phba->pcidev->dev, sgel,
lpfc_cmd->seg_cnt, datadir);
return 1;
}
/*
* The driver established a maximum scatter-gather segment count
* during probe that limits the number of sg elements in any
* single scsi command. Just run through the seg_cnt and format
* the bde's.
*/
for (i = 0; i < lpfc_cmd->seg_cnt; i++) {
physaddr = sg_dma_address(sgel);
bpl->addrLow = le32_to_cpu(putPaddrLow(physaddr));
bpl->addrHigh = le32_to_cpu(putPaddrHigh(physaddr));
bpl->tus.f.bdeSize = sg_dma_len(sgel);
if (datadir == DMA_TO_DEVICE)
bpl->tus.f.bdeFlags = 0;
else
bpl->tus.f.bdeFlags = BUFF_USE_RCV;
bpl->tus.w = le32_to_cpu(bpl->tus.w);
bpl++;
sgel++;
num_bde++;
}
} else if (scsi_cmnd->request_buffer && scsi_cmnd->request_bufflen) {
physaddr = dma_map_single(&phba->pcidev->dev,
scsi_cmnd->request_buffer,
scsi_cmnd->request_bufflen,
datadir);
dma_error = dma_mapping_error(physaddr);
if (dma_error) {
lpfc_printf_log(phba, KERN_ERR, LOG_FCP,
"%d:0718 Unable to dma_map_single "
"request_buffer: x%x\n",
phba->brd_no, dma_error);
return 1;
}
lpfc_cmd->nonsg_phys = physaddr;
bpl->addrLow = le32_to_cpu(putPaddrLow(physaddr));
bpl->addrHigh = le32_to_cpu(putPaddrHigh(physaddr));
bpl->tus.f.bdeSize = scsi_cmnd->request_bufflen;
if (datadir == DMA_TO_DEVICE)
bpl->tus.f.bdeFlags = 0;
else
bpl->tus.f.bdeFlags = BUFF_USE_RCV;
bpl->tus.w = le32_to_cpu(bpl->tus.w);
num_bde = 1;
bpl++;
}
/*
* Finish initializing those IOCB fields that are dependent on the
* scsi_cmnd request_buffer. Note that the bdeSize is explicitly
* reinitialized since all iocb memory resources are used many times
* for transmit, receive, and continuation bpl's.
*/
iocb_cmd->un.fcpi64.bdl.bdeSize = (2 * sizeof (struct ulp_bde64));
iocb_cmd->un.fcpi64.bdl.bdeSize +=
(num_bde * sizeof (struct ulp_bde64));
iocb_cmd->ulpBdeCount = 1;
iocb_cmd->ulpLe = 1;
fcp_cmnd->fcpDl = be32_to_cpu(scsi_cmnd->request_bufflen);
return 0;
}
static void
lpfc_scsi_unprep_dma_buf(struct lpfc_hba * phba, struct lpfc_scsi_buf * psb)
{
/*
* There are only two special cases to consider. (1) the scsi command
* requested scatter-gather usage or (2) the scsi command allocated
* a request buffer, but did not request use_sg. There is a third
* case, but it does not require resource deallocation.
*/
if ((psb->seg_cnt > 0) && (psb->pCmd->use_sg)) {
dma_unmap_sg(&phba->pcidev->dev, psb->pCmd->request_buffer,
psb->seg_cnt, psb->pCmd->sc_data_direction);
} else {
if ((psb->nonsg_phys) && (psb->pCmd->request_bufflen)) {
dma_unmap_single(&phba->pcidev->dev, psb->nonsg_phys,
psb->pCmd->request_bufflen,
psb->pCmd->sc_data_direction);
}
}
}
static void
lpfc_handle_fcp_err(struct lpfc_scsi_buf *lpfc_cmd)
{
struct scsi_cmnd *cmnd = lpfc_cmd->pCmd;
struct fcp_cmnd *fcpcmd = lpfc_cmd->fcp_cmnd;
struct fcp_rsp *fcprsp = lpfc_cmd->fcp_rsp;
struct lpfc_hba *phba = lpfc_cmd->scsi_hba;
uint32_t fcpi_parm = lpfc_cmd->cur_iocbq.iocb.un.fcpi.fcpi_parm;
uint32_t resp_info = fcprsp->rspStatus2;
uint32_t scsi_status = fcprsp->rspStatus3;
uint32_t host_status = DID_OK;
uint32_t rsplen = 0;
/*
* If this is a task management command, there is no
* scsi packet associated with this lpfc_cmd. The driver
* consumes it.
*/
if (fcpcmd->fcpCntl2) {
scsi_status = 0;
goto out;
}
lpfc_printf_log(phba, KERN_WARNING, LOG_FCP,
"%d:0730 FCP command failed: RSP "
"Data: x%x x%x x%x x%x x%x x%x\n",
phba->brd_no, resp_info, scsi_status,
be32_to_cpu(fcprsp->rspResId),
be32_to_cpu(fcprsp->rspSnsLen),
be32_to_cpu(fcprsp->rspRspLen),
fcprsp->rspInfo3);
if (resp_info & RSP_LEN_VALID) {
rsplen = be32_to_cpu(fcprsp->rspRspLen);
if ((rsplen != 0 && rsplen != 4 && rsplen != 8) ||
(fcprsp->rspInfo3 != RSP_NO_FAILURE)) {
host_status = DID_ERROR;
goto out;
}
}
if ((resp_info & SNS_LEN_VALID) && fcprsp->rspSnsLen) {
uint32_t snslen = be32_to_cpu(fcprsp->rspSnsLen);
if (snslen > SCSI_SENSE_BUFFERSIZE)
snslen = SCSI_SENSE_BUFFERSIZE;
memcpy(cmnd->sense_buffer, &fcprsp->rspInfo0 + rsplen, snslen);
}
cmnd->resid = 0;
if (resp_info & RESID_UNDER) {
cmnd->resid = be32_to_cpu(fcprsp->rspResId);
lpfc_printf_log(phba, KERN_INFO, LOG_FCP,
"%d:0716 FCP Read Underrun, expected %d, "
"residual %d Data: x%x x%x x%x\n", phba->brd_no,
be32_to_cpu(fcpcmd->fcpDl), cmnd->resid,
fcpi_parm, cmnd->cmnd[0], cmnd->underflow);
/*
* The cmnd->underflow is the minimum number of bytes that must
* be transfered for this command. Provided a sense condition
* is not present, make sure the actual amount transferred is at
* least the underflow value or fail.
*/
if (!(resp_info & SNS_LEN_VALID) &&
(scsi_status == SAM_STAT_GOOD) &&
(cmnd->request_bufflen - cmnd->resid) < cmnd->underflow) {
lpfc_printf_log(phba, KERN_INFO, LOG_FCP,
"%d:0717 FCP command x%x residual "
"underrun converted to error "
"Data: x%x x%x x%x\n", phba->brd_no,
cmnd->cmnd[0], cmnd->request_bufflen,
cmnd->resid, cmnd->underflow);
host_status = DID_ERROR;
}
} else if (resp_info & RESID_OVER) {
lpfc_printf_log(phba, KERN_WARNING, LOG_FCP,
"%d:0720 FCP command x%x residual "
"overrun error. Data: x%x x%x \n",
phba->brd_no, cmnd->cmnd[0],
cmnd->request_bufflen, cmnd->resid);
host_status = DID_ERROR;
/*
* Check SLI validation that all the transfer was actually done
* (fcpi_parm should be zero). Apply check only to reads.
*/
} else if ((scsi_status == SAM_STAT_GOOD) && fcpi_parm &&
(cmnd->sc_data_direction == DMA_FROM_DEVICE)) {
lpfc_printf_log(phba, KERN_WARNING, LOG_FCP,
"%d:0734 FCP Read Check Error Data: "
"x%x x%x x%x x%x\n", phba->brd_no,
be32_to_cpu(fcpcmd->fcpDl),
be32_to_cpu(fcprsp->rspResId),
fcpi_parm, cmnd->cmnd[0]);
host_status = DID_ERROR;
cmnd->resid = cmnd->request_bufflen;
}
out:
cmnd->result = ScsiResult(host_status, scsi_status);
}
static void
lpfc_scsi_cmd_iocb_cmpl(struct lpfc_hba *phba, struct lpfc_iocbq *pIocbIn,
struct lpfc_iocbq *pIocbOut)
{
struct lpfc_scsi_buf *lpfc_cmd =
(struct lpfc_scsi_buf *) pIocbIn->context1;
struct lpfc_rport_data *rdata = lpfc_cmd->rdata;
struct lpfc_nodelist *pnode = rdata->pnode;
struct scsi_cmnd *cmd = lpfc_cmd->pCmd;
int result;
struct scsi_device *sdev, *tmp_sdev;
int depth = 0;
lpfc_cmd->result = pIocbOut->iocb.un.ulpWord[4];
lpfc_cmd->status = pIocbOut->iocb.ulpStatus;
if (lpfc_cmd->status) {
if (lpfc_cmd->status == IOSTAT_LOCAL_REJECT &&
(lpfc_cmd->result & IOERR_DRVR_MASK))
lpfc_cmd->status = IOSTAT_DRIVER_REJECT;
else if (lpfc_cmd->status >= IOSTAT_CNT)
lpfc_cmd->status = IOSTAT_DEFAULT;
lpfc_printf_log(phba, KERN_WARNING, LOG_FCP,
"%d:0729 FCP cmd x%x failed <%d/%d> status: "
"x%x result: x%x Data: x%x x%x\n",
phba->brd_no, cmd->cmnd[0], cmd->device->id,
cmd->device->lun, lpfc_cmd->status,
lpfc_cmd->result, pIocbOut->iocb.ulpContext,
lpfc_cmd->cur_iocbq.iocb.ulpIoTag);
switch (lpfc_cmd->status) {
case IOSTAT_FCP_RSP_ERROR:
/* Call FCP RSP handler to determine result */
lpfc_handle_fcp_err(lpfc_cmd);
break;
case IOSTAT_NPORT_BSY:
case IOSTAT_FABRIC_BSY:
cmd->result = ScsiResult(DID_BUS_BUSY, 0);
break;
default:
cmd->result = ScsiResult(DID_ERROR, 0);
break;
}
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
if ((pnode == NULL )
|| (pnode->nlp_state != NLP_STE_MAPPED_NODE))
cmd->result = ScsiResult(DID_BUS_BUSY, SAM_STAT_BUSY);
} else {
cmd->result = ScsiResult(DID_OK, 0);
}
if (cmd->result || lpfc_cmd->fcp_rsp->rspSnsLen) {
uint32_t *lp = (uint32_t *)cmd->sense_buffer;
lpfc_printf_log(phba, KERN_INFO, LOG_FCP,
"%d:0710 Iodone <%d/%d> cmd %p, error x%x "
"SNS x%x x%x Data: x%x x%x\n",
phba->brd_no, cmd->device->id,
cmd->device->lun, cmd, cmd->result,
*lp, *(lp + 3), cmd->retries, cmd->resid);
}
result = cmd->result;
sdev = cmd->device;
cmd->scsi_done(cmd);
if (phba->cfg_poll & ENABLE_FCP_RING_POLLING) {
lpfc_scsi_unprep_dma_buf(phba, lpfc_cmd);
lpfc_release_scsi_buf(phba, lpfc_cmd);
return;
}
if (!result && pnode != NULL &&
((jiffies - pnode->last_ramp_up_time) >
LPFC_Q_RAMP_UP_INTERVAL * HZ) &&
((jiffies - pnode->last_q_full_time) >
LPFC_Q_RAMP_UP_INTERVAL * HZ) &&
(phba->cfg_lun_queue_depth > sdev->queue_depth)) {
shost_for_each_device(tmp_sdev, sdev->host) {
if (phba->cfg_lun_queue_depth > tmp_sdev->queue_depth) {
if (tmp_sdev->id != sdev->id)
continue;
if (tmp_sdev->ordered_tags)
scsi_adjust_queue_depth(tmp_sdev,
MSG_ORDERED_TAG,
tmp_sdev->queue_depth+1);
else
scsi_adjust_queue_depth(tmp_sdev,
MSG_SIMPLE_TAG,
tmp_sdev->queue_depth+1);
pnode->last_ramp_up_time = jiffies;
}
}
}
/*
* Check for queue full. If the lun is reporting queue full, then
* back off the lun queue depth to prevent target overloads.
*/
if (result == SAM_STAT_TASK_SET_FULL && pnode != NULL) {
pnode->last_q_full_time = jiffies;
shost_for_each_device(tmp_sdev, sdev->host) {
if (tmp_sdev->id != sdev->id)
continue;
depth = scsi_track_queue_full(tmp_sdev,
tmp_sdev->queue_depth - 1);
}
/*
* The queue depth cannot be lowered any more.
* Modify the returned error code to store
* the final depth value set by
* scsi_track_queue_full.
*/
if (depth == -1)
depth = sdev->host->cmd_per_lun;
if (depth) {
lpfc_printf_log(phba, KERN_WARNING, LOG_FCP,
"%d:0711 detected queue full - lun queue depth "
" adjusted to %d.\n", phba->brd_no, depth);
}
}
lpfc_scsi_unprep_dma_buf(phba, lpfc_cmd);
lpfc_release_scsi_buf(phba, lpfc_cmd);
}
static void
lpfc_scsi_prep_cmnd(struct lpfc_hba * phba, struct lpfc_scsi_buf * lpfc_cmd,
struct lpfc_nodelist *pnode)
{
struct scsi_cmnd *scsi_cmnd = lpfc_cmd->pCmd;
struct fcp_cmnd *fcp_cmnd = lpfc_cmd->fcp_cmnd;
IOCB_t *iocb_cmd = &lpfc_cmd->cur_iocbq.iocb;
struct lpfc_iocbq *piocbq = &(lpfc_cmd->cur_iocbq);
int datadir = scsi_cmnd->sc_data_direction;
lpfc_cmd->fcp_rsp->rspSnsLen = 0;
/* clear task management bits */
lpfc_cmd->fcp_cmnd->fcpCntl2 = 0;
int_to_scsilun(lpfc_cmd->pCmd->device->lun,
&lpfc_cmd->fcp_cmnd->fcp_lun);
memcpy(&fcp_cmnd->fcpCdb[0], scsi_cmnd->cmnd, 16);
if (scsi_cmnd->device->tagged_supported) {
switch (scsi_cmnd->tag) {
case HEAD_OF_QUEUE_TAG:
fcp_cmnd->fcpCntl1 = HEAD_OF_Q;
break;
case ORDERED_QUEUE_TAG:
fcp_cmnd->fcpCntl1 = ORDERED_Q;
break;
default:
fcp_cmnd->fcpCntl1 = SIMPLE_Q;
break;
}
} else
fcp_cmnd->fcpCntl1 = 0;
/*
* There are three possibilities here - use scatter-gather segment, use
* the single mapping, or neither. Start the lpfc command prep by
* bumping the bpl beyond the fcp_cmnd and fcp_rsp regions to the first
* data bde entry.
*/
if (scsi_cmnd->use_sg) {
if (datadir == DMA_TO_DEVICE) {
iocb_cmd->ulpCommand = CMD_FCP_IWRITE64_CR;
iocb_cmd->un.fcpi.fcpi_parm = 0;
iocb_cmd->ulpPU = 0;
fcp_cmnd->fcpCntl3 = WRITE_DATA;
phba->fc4OutputRequests++;
} else {
iocb_cmd->ulpCommand = CMD_FCP_IREAD64_CR;
iocb_cmd->ulpPU = PARM_READ_CHECK;
iocb_cmd->un.fcpi.fcpi_parm =
scsi_cmnd->request_bufflen;
fcp_cmnd->fcpCntl3 = READ_DATA;
phba->fc4InputRequests++;
}
} else if (scsi_cmnd->request_buffer && scsi_cmnd->request_bufflen) {
if (datadir == DMA_TO_DEVICE) {
iocb_cmd->ulpCommand = CMD_FCP_IWRITE64_CR;
iocb_cmd->un.fcpi.fcpi_parm = 0;
iocb_cmd->ulpPU = 0;
fcp_cmnd->fcpCntl3 = WRITE_DATA;
phba->fc4OutputRequests++;
} else {
iocb_cmd->ulpCommand = CMD_FCP_IREAD64_CR;
iocb_cmd->ulpPU = PARM_READ_CHECK;
iocb_cmd->un.fcpi.fcpi_parm =
scsi_cmnd->request_bufflen;
fcp_cmnd->fcpCntl3 = READ_DATA;
phba->fc4InputRequests++;
}
} else {
iocb_cmd->ulpCommand = CMD_FCP_ICMND64_CR;
iocb_cmd->un.fcpi.fcpi_parm = 0;
iocb_cmd->ulpPU = 0;
fcp_cmnd->fcpCntl3 = 0;
phba->fc4ControlRequests++;
}
/*
* Finish initializing those IOCB fields that are independent
* of the scsi_cmnd request_buffer
*/
piocbq->iocb.ulpContext = pnode->nlp_rpi;
if (pnode->nlp_fcp_info & NLP_FCP_2_DEVICE)
piocbq->iocb.ulpFCP2Rcvy = 1;
piocbq->iocb.ulpClass = (pnode->nlp_fcp_info & 0x0f);
piocbq->context1 = lpfc_cmd;
piocbq->iocb_cmpl = lpfc_scsi_cmd_iocb_cmpl;
piocbq->iocb.ulpTimeout = lpfc_cmd->timeout;
}
static int
lpfc_scsi_prep_task_mgmt_cmd(struct lpfc_hba *phba,
struct lpfc_scsi_buf *lpfc_cmd,
unsigned int lun,
uint8_t task_mgmt_cmd)
{
struct lpfc_sli *psli;
struct lpfc_iocbq *piocbq;
IOCB_t *piocb;
struct fcp_cmnd *fcp_cmnd;
struct lpfc_rport_data *rdata = lpfc_cmd->rdata;
struct lpfc_nodelist *ndlp = rdata->pnode;
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
if ((ndlp == NULL) || (ndlp->nlp_state != NLP_STE_MAPPED_NODE)) {
return 0;
}
psli = &phba->sli;
piocbq = &(lpfc_cmd->cur_iocbq);
piocb = &piocbq->iocb;
fcp_cmnd = lpfc_cmd->fcp_cmnd;
int_to_scsilun(lun, &lpfc_cmd->fcp_cmnd->fcp_lun);
fcp_cmnd->fcpCntl2 = task_mgmt_cmd;
piocb->ulpCommand = CMD_FCP_ICMND64_CR;
piocb->ulpContext = ndlp->nlp_rpi;
if (ndlp->nlp_fcp_info & NLP_FCP_2_DEVICE) {
piocb->ulpFCP2Rcvy = 1;
}
piocb->ulpClass = (ndlp->nlp_fcp_info & 0x0f);
/* ulpTimeout is only one byte */
if (lpfc_cmd->timeout > 0xff) {
/*
* Do not timeout the command at the firmware level.
* The driver will provide the timeout mechanism.
*/
piocb->ulpTimeout = 0;
} else {
piocb->ulpTimeout = lpfc_cmd->timeout;
}
return (1);
}
static int
lpfc_scsi_tgt_reset(struct lpfc_scsi_buf * lpfc_cmd, struct lpfc_hba * phba,
unsigned tgt_id, unsigned int lun,
struct lpfc_rport_data *rdata)
{
struct lpfc_iocbq *iocbq;
struct lpfc_iocbq *iocbqrsp;
int ret;
lpfc_cmd->rdata = rdata;
ret = lpfc_scsi_prep_task_mgmt_cmd(phba, lpfc_cmd, lun,
FCP_TARGET_RESET);
if (!ret)
return FAILED;
lpfc_cmd->scsi_hba = phba;
iocbq = &lpfc_cmd->cur_iocbq;
iocbqrsp = lpfc_sli_get_iocbq(phba);
if (!iocbqrsp)
return FAILED;
/* Issue Target Reset to TGT <num> */
lpfc_printf_log(phba, KERN_INFO, LOG_FCP,
"%d:0702 Issue Target Reset to TGT %d "
"Data: x%x x%x\n",
phba->brd_no, tgt_id, rdata->pnode->nlp_rpi,
rdata->pnode->nlp_flag);
ret = lpfc_sli_issue_iocb_wait(phba,
&phba->sli.ring[phba->sli.fcp_ring],
iocbq, iocbqrsp, lpfc_cmd->timeout);
if (ret != IOCB_SUCCESS) {
lpfc_cmd->status = IOSTAT_DRIVER_REJECT;
ret = FAILED;
} else {
ret = SUCCESS;
lpfc_cmd->result = iocbqrsp->iocb.un.ulpWord[4];
lpfc_cmd->status = iocbqrsp->iocb.ulpStatus;
if (lpfc_cmd->status == IOSTAT_LOCAL_REJECT &&
(lpfc_cmd->result & IOERR_DRVR_MASK))
lpfc_cmd->status = IOSTAT_DRIVER_REJECT;
}
lpfc_sli_release_iocbq(phba, iocbqrsp);
return ret;
}
const char *
lpfc_info(struct Scsi_Host *host)
{
struct lpfc_hba *phba = (struct lpfc_hba *) host->hostdata;
int len;
static char lpfcinfobuf[384];
memset(lpfcinfobuf,0,384);
if (phba && phba->pcidev){
strncpy(lpfcinfobuf, phba->ModelDesc, 256);
len = strlen(lpfcinfobuf);
snprintf(lpfcinfobuf + len,
384-len,
" on PCI bus %02x device %02x irq %d",
phba->pcidev->bus->number,
phba->pcidev->devfn,
phba->pcidev->irq);
len = strlen(lpfcinfobuf);
if (phba->Port[0]) {
snprintf(lpfcinfobuf + len,
384-len,
" port %s",
phba->Port);
}
}
return lpfcinfobuf;
}
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
static __inline__ void lpfc_poll_rearm_timer(struct lpfc_hba * phba)
{
unsigned long poll_tmo_expires =
(jiffies + msecs_to_jiffies(phba->cfg_poll_tmo));
if (phba->sli.ring[LPFC_FCP_RING].txcmplq_cnt)
mod_timer(&phba->fcp_poll_timer,
poll_tmo_expires);
}
void lpfc_poll_start_timer(struct lpfc_hba * phba)
{
lpfc_poll_rearm_timer(phba);
}
void lpfc_poll_timeout(unsigned long ptr)
{
struct lpfc_hba *phba = (struct lpfc_hba *)ptr;
unsigned long iflag;
spin_lock_irqsave(phba->host->host_lock, iflag);
if (phba->cfg_poll & ENABLE_FCP_RING_POLLING) {
lpfc_sli_poll_fcp_ring (phba);
if (phba->cfg_poll & DISABLE_FCP_RING_INT)
lpfc_poll_rearm_timer(phba);
}
spin_unlock_irqrestore(phba->host->host_lock, iflag);
}
static int
lpfc_queuecommand(struct scsi_cmnd *cmnd, void (*done) (struct scsi_cmnd *))
{
struct lpfc_hba *phba =
(struct lpfc_hba *) cmnd->device->host->hostdata;
struct lpfc_sli *psli = &phba->sli;
struct lpfc_rport_data *rdata = cmnd->device->hostdata;
struct lpfc_nodelist *ndlp = rdata->pnode;
struct lpfc_scsi_buf *lpfc_cmd;
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
struct fc_rport *rport = starget_to_rport(scsi_target(cmnd->device));
int err;
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
err = fc_remote_port_chkready(rport);
if (err) {
cmnd->result = err;
goto out_fail_command;
}
/*
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
* Catch race where our node has transitioned, but the
* transport is still transitioning.
*/
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
if (!ndlp) {
cmnd->result = ScsiResult(DID_BUS_BUSY, 0);
goto out_fail_command;
}
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
lpfc_cmd = lpfc_get_scsi_buf (phba);
if (lpfc_cmd == NULL) {
lpfc_printf_log(phba, KERN_INFO, LOG_FCP,
"%d:0707 driver's buffer pool is empty, "
"IO busied\n", phba->brd_no);
goto out_host_busy;
}
/*
* Store the midlayer's command structure for the completion phase
* and complete the command initialization.
*/
lpfc_cmd->pCmd = cmnd;
lpfc_cmd->rdata = rdata;
lpfc_cmd->timeout = 0;
cmnd->host_scribble = (unsigned char *)lpfc_cmd;
cmnd->scsi_done = done;
err = lpfc_scsi_prep_dma_buf(phba, lpfc_cmd);
if (err)
goto out_host_busy_free_buf;
lpfc_scsi_prep_cmnd(phba, lpfc_cmd, ndlp);
err = lpfc_sli_issue_iocb(phba, &phba->sli.ring[psli->fcp_ring],
&lpfc_cmd->cur_iocbq, SLI_IOCB_RET_IOCB);
if (err)
goto out_host_busy_free_buf;
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
if (phba->cfg_poll & ENABLE_FCP_RING_POLLING) {
lpfc_sli_poll_fcp_ring(phba);
if (phba->cfg_poll & DISABLE_FCP_RING_INT)
lpfc_poll_rearm_timer(phba);
}
return 0;
out_host_busy_free_buf:
lpfc_scsi_unprep_dma_buf(phba, lpfc_cmd);
lpfc_release_scsi_buf(phba, lpfc_cmd);
out_host_busy:
return SCSI_MLQUEUE_HOST_BUSY;
out_fail_command:
done(cmnd);
return 0;
}
static void
lpfc_block_error_handler(struct scsi_cmnd *cmnd)
{
struct Scsi_Host *shost = cmnd->device->host;
struct fc_rport *rport = starget_to_rport(scsi_target(cmnd->device));
spin_lock_irq(shost->host_lock);
while (rport->port_state == FC_PORTSTATE_BLOCKED) {
spin_unlock_irq(shost->host_lock);
msleep(1000);
spin_lock_irq(shost->host_lock);
}
spin_unlock_irq(shost->host_lock);
return;
}
static int
lpfc_abort_handler(struct scsi_cmnd *cmnd)
{
struct Scsi_Host *shost = cmnd->device->host;
struct lpfc_hba *phba = (struct lpfc_hba *)shost->hostdata;
struct lpfc_sli_ring *pring = &phba->sli.ring[phba->sli.fcp_ring];
struct lpfc_iocbq *iocb;
struct lpfc_iocbq *abtsiocb;
struct lpfc_scsi_buf *lpfc_cmd;
IOCB_t *cmd, *icmd;
unsigned int loop_count = 0;
int ret = SUCCESS;
lpfc_block_error_handler(cmnd);
spin_lock_irq(shost->host_lock);
lpfc_cmd = (struct lpfc_scsi_buf *)cmnd->host_scribble;
BUG_ON(!lpfc_cmd);
/*
* If pCmd field of the corresponding lpfc_scsi_buf structure
* points to a different SCSI command, then the driver has
* already completed this command, but the midlayer did not
* see the completion before the eh fired. Just return
* SUCCESS.
*/
iocb = &lpfc_cmd->cur_iocbq;
if (lpfc_cmd->pCmd != cmnd)
goto out;
BUG_ON(iocb->context1 != lpfc_cmd);
abtsiocb = lpfc_sli_get_iocbq(phba);
if (abtsiocb == NULL) {
ret = FAILED;
goto out;
}
/*
* The scsi command can not be in txq and it is in flight because the
* pCmd is still pointig at the SCSI command we have to abort. There
* is no need to search the txcmplq. Just send an abort to the FW.
*/
cmd = &iocb->iocb;
icmd = &abtsiocb->iocb;
icmd->un.acxri.abortType = ABORT_TYPE_ABTS;
icmd->un.acxri.abortContextTag = cmd->ulpContext;
icmd->un.acxri.abortIoTag = cmd->ulpIoTag;
icmd->ulpLe = 1;
icmd->ulpClass = cmd->ulpClass;
if (phba->hba_state >= LPFC_LINK_UP)
icmd->ulpCommand = CMD_ABORT_XRI_CN;
else
icmd->ulpCommand = CMD_CLOSE_XRI_CN;
abtsiocb->iocb_cmpl = lpfc_sli_abort_fcp_cmpl;
if (lpfc_sli_issue_iocb(phba, pring, abtsiocb, 0) == IOCB_ERROR) {
lpfc_sli_release_iocbq(phba, abtsiocb);
ret = FAILED;
goto out;
}
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
if (phba->cfg_poll & DISABLE_FCP_RING_INT)
lpfc_sli_poll_fcp_ring (phba);
/* Wait for abort to complete */
while (lpfc_cmd->pCmd == cmnd)
{
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
if (phba->cfg_poll & DISABLE_FCP_RING_INT)
lpfc_sli_poll_fcp_ring (phba);
spin_unlock_irq(phba->host->host_lock);
schedule_timeout_uninterruptible(LPFC_ABORT_WAIT*HZ);
spin_lock_irq(phba->host->host_lock);
if (++loop_count
> (2 * phba->cfg_devloss_tmo)/LPFC_ABORT_WAIT)
break;
}
if (lpfc_cmd->pCmd == cmnd) {
ret = FAILED;
lpfc_printf_log(phba, KERN_ERR, LOG_FCP,
"%d:0748 abort handler timed out waiting for "
"abort to complete: ret %#x, ID %d, LUN %d, "
"snum %#lx\n",
phba->brd_no, ret, cmnd->device->id,
cmnd->device->lun, cmnd->serial_number);
}
out:
lpfc_printf_log(phba, KERN_WARNING, LOG_FCP,
"%d:0749 SCSI Layer I/O Abort Request "
"Status x%x ID %d LUN %d snum %#lx\n",
phba->brd_no, ret, cmnd->device->id,
cmnd->device->lun, cmnd->serial_number);
spin_unlock_irq(shost->host_lock);
return ret;
}
static int
lpfc_reset_lun_handler(struct scsi_cmnd *cmnd)
{
struct Scsi_Host *shost = cmnd->device->host;
struct lpfc_hba *phba = (struct lpfc_hba *)shost->hostdata;
struct lpfc_scsi_buf *lpfc_cmd;
struct lpfc_iocbq *iocbq, *iocbqrsp;
struct lpfc_rport_data *rdata = cmnd->device->hostdata;
struct lpfc_nodelist *pnode = rdata->pnode;
uint32_t cmd_result = 0, cmd_status = 0;
int ret = FAILED;
int cnt, loopcnt;
lpfc_block_error_handler(cmnd);
spin_lock_irq(shost->host_lock);
/*
* If target is not in a MAPPED state, delay the reset until
* target is rediscovered or devloss timeout expires.
*/
while ( 1 ) {
if (!pnode)
break;
if (pnode->nlp_state != NLP_STE_MAPPED_NODE) {
spin_unlock_irq(phba->host->host_lock);
schedule_timeout_uninterruptible(msecs_to_jiffies(500));
spin_lock_irq(phba->host->host_lock);
}
if ((pnode) && (pnode->nlp_state == NLP_STE_MAPPED_NODE))
break;
}
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
lpfc_cmd = lpfc_get_scsi_buf (phba);
if (lpfc_cmd == NULL)
goto out;
lpfc_cmd->timeout = 60;
lpfc_cmd->scsi_hba = phba;
lpfc_cmd->rdata = rdata;
ret = lpfc_scsi_prep_task_mgmt_cmd(phba, lpfc_cmd, cmnd->device->lun,
FCP_LUN_RESET);
if (!ret)
goto out_free_scsi_buf;
iocbq = &lpfc_cmd->cur_iocbq;
/* get a buffer for this IOCB command response */
iocbqrsp = lpfc_sli_get_iocbq(phba);
if (iocbqrsp == NULL)
goto out_free_scsi_buf;
lpfc_printf_log(phba, KERN_INFO, LOG_FCP,
"%d:0703 Issue LUN Reset to TGT %d LUN %d "
"Data: x%x x%x\n", phba->brd_no, cmnd->device->id,
cmnd->device->lun, pnode->nlp_rpi, pnode->nlp_flag);
ret = lpfc_sli_issue_iocb_wait(phba,
&phba->sli.ring[phba->sli.fcp_ring],
iocbq, iocbqrsp, lpfc_cmd->timeout);
if (ret == IOCB_SUCCESS)
ret = SUCCESS;
cmd_result = iocbqrsp->iocb.un.ulpWord[4];
cmd_status = iocbqrsp->iocb.ulpStatus;
lpfc_sli_release_iocbq(phba, iocbqrsp);
/*
* All outstanding txcmplq I/Os should have been aborted by the device.
* Unfortunately, some targets do not abide by this forcing the driver
* to double check.
*/
cnt = lpfc_sli_sum_iocb(phba, &phba->sli.ring[phba->sli.fcp_ring],
cmnd->device->id, cmnd->device->lun,
LPFC_CTX_LUN);
if (cnt)
lpfc_sli_abort_iocb(phba,
&phba->sli.ring[phba->sli.fcp_ring],
cmnd->device->id, cmnd->device->lun,
0, LPFC_CTX_LUN);
loopcnt = 0;
while(cnt) {
spin_unlock_irq(phba->host->host_lock);
schedule_timeout_uninterruptible(LPFC_RESET_WAIT*HZ);
spin_lock_irq(phba->host->host_lock);
if (++loopcnt
> (2 * phba->cfg_devloss_tmo)/LPFC_RESET_WAIT)
break;
cnt = lpfc_sli_sum_iocb(phba,
&phba->sli.ring[phba->sli.fcp_ring],
cmnd->device->id, cmnd->device->lun,
LPFC_CTX_LUN);
}
if (cnt) {
lpfc_printf_log(phba, KERN_ERR, LOG_FCP,
"%d:0719 LUN Reset I/O flush failure: cnt x%x\n",
phba->brd_no, cnt);
ret = FAILED;
}
out_free_scsi_buf:
lpfc_release_scsi_buf(phba, lpfc_cmd);
lpfc_printf_log(phba, KERN_ERR, LOG_FCP,
"%d:0713 SCSI layer issued LUN reset (%d, %d) "
"Data: x%x x%x x%x\n",
phba->brd_no, cmnd->device->id,cmnd->device->lun,
ret, cmd_status, cmd_result);
out:
spin_unlock_irq(shost->host_lock);
return ret;
}
static int
lpfc_reset_bus_handler(struct scsi_cmnd *cmnd)
{
struct Scsi_Host *shost = cmnd->device->host;
struct lpfc_hba *phba = (struct lpfc_hba *)shost->hostdata;
struct lpfc_nodelist *ndlp = NULL;
int match;
int ret = FAILED, i, err_count = 0;
int cnt, loopcnt;
struct lpfc_scsi_buf * lpfc_cmd;
lpfc_block_error_handler(cmnd);
spin_lock_irq(shost->host_lock);
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
lpfc_cmd = lpfc_get_scsi_buf(phba);
if (lpfc_cmd == NULL)
goto out;
/* The lpfc_cmd storage is reused. Set all loop invariants. */
lpfc_cmd->timeout = 60;
lpfc_cmd->scsi_hba = phba;
/*
* Since the driver manages a single bus device, reset all
* targets known to the driver. Should any target reset
* fail, this routine returns failure to the midlayer.
*/
for (i = 0; i < LPFC_MAX_TARGET; i++) {
/* Search the mapped list for this target ID */
match = 0;
list_for_each_entry(ndlp, &phba->fc_nlpmap_list, nlp_listp) {
if ((i == ndlp->nlp_sid) && ndlp->rport) {
match = 1;
break;
}
}
if (!match)
continue;
ret = lpfc_scsi_tgt_reset(lpfc_cmd, phba, i, cmnd->device->lun,
ndlp->rport->dd_data);
if (ret != SUCCESS) {
lpfc_printf_log(phba, KERN_ERR, LOG_FCP,
"%d:0700 Bus Reset on target %d failed\n",
phba->brd_no, i);
err_count++;
}
}
if (err_count == 0)
ret = SUCCESS;
lpfc_release_scsi_buf(phba, lpfc_cmd);
/*
* All outstanding txcmplq I/Os should have been aborted by
* the targets. Unfortunately, some targets do not abide by
* this forcing the driver to double check.
*/
cnt = lpfc_sli_sum_iocb(phba, &phba->sli.ring[phba->sli.fcp_ring],
0, 0, LPFC_CTX_HOST);
if (cnt)
lpfc_sli_abort_iocb(phba, &phba->sli.ring[phba->sli.fcp_ring],
0, 0, 0, LPFC_CTX_HOST);
loopcnt = 0;
while(cnt) {
spin_unlock_irq(phba->host->host_lock);
schedule_timeout_uninterruptible(LPFC_RESET_WAIT*HZ);
spin_lock_irq(phba->host->host_lock);
if (++loopcnt
> (2 * phba->cfg_devloss_tmo)/LPFC_RESET_WAIT)
break;
cnt = lpfc_sli_sum_iocb(phba,
&phba->sli.ring[phba->sli.fcp_ring],
0, 0, LPFC_CTX_HOST);
}
if (cnt) {
lpfc_printf_log(phba, KERN_ERR, LOG_FCP,
"%d:0715 Bus Reset I/O flush failure: cnt x%x left x%x\n",
phba->brd_no, cnt, i);
ret = FAILED;
}
lpfc_printf_log(phba,
KERN_ERR,
LOG_FCP,
"%d:0714 SCSI layer issued Bus Reset Data: x%x\n",
phba->brd_no, ret);
out:
spin_unlock_irq(shost->host_lock);
return ret;
}
static int
lpfc_slave_alloc(struct scsi_device *sdev)
{
struct lpfc_hba *phba = (struct lpfc_hba *)sdev->host->hostdata;
struct lpfc_scsi_buf *scsi_buf = NULL;
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
struct fc_rport *rport = starget_to_rport(scsi_target(sdev));
uint32_t total = 0, i;
uint32_t num_to_alloc = 0;
unsigned long flags;
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
if (!rport || fc_remote_port_chkready(rport))
return -ENXIO;
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
sdev->hostdata = rport->dd_data;
/*
* Populate the cmds_per_lun count scsi_bufs into this host's globally
* available list of scsi buffers. Don't allocate more than the
* HBA limit conveyed to the midlayer via the host structure. The
* formula accounts for the lun_queue_depth + error handlers + 1
* extra. This list of scsi bufs exists for the lifetime of the driver.
*/
total = phba->total_scsi_bufs;
num_to_alloc = phba->cfg_lun_queue_depth + 2;
if (total >= phba->cfg_hba_queue_depth) {
lpfc_printf_log(phba, KERN_WARNING, LOG_FCP,
"%d:0704 At limitation of %d preallocated "
"command buffers\n", phba->brd_no, total);
return 0;
} else if (total + num_to_alloc > phba->cfg_hba_queue_depth) {
lpfc_printf_log(phba, KERN_WARNING, LOG_FCP,
"%d:0705 Allocation request of %d command "
"buffers will exceed max of %d. Reducing "
"allocation request to %d.\n", phba->brd_no,
num_to_alloc, phba->cfg_hba_queue_depth,
(phba->cfg_hba_queue_depth - total));
num_to_alloc = phba->cfg_hba_queue_depth - total;
}
for (i = 0; i < num_to_alloc; i++) {
scsi_buf = lpfc_new_scsi_buf(phba);
if (!scsi_buf) {
lpfc_printf_log(phba, KERN_ERR, LOG_FCP,
"%d:0706 Failed to allocate command "
"buffer\n", phba->brd_no);
break;
}
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
spin_lock_irqsave(&phba->scsi_buf_list_lock, flags);
phba->total_scsi_bufs++;
list_add_tail(&scsi_buf->list, &phba->lpfc_scsi_buf_list);
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
spin_unlock_irqrestore(&phba->scsi_buf_list_lock, flags);
}
return 0;
}
static int
lpfc_slave_configure(struct scsi_device *sdev)
{
struct lpfc_hba *phba = (struct lpfc_hba *) sdev->host->hostdata;
struct fc_rport *rport = starget_to_rport(sdev->sdev_target);
if (sdev->tagged_supported)
scsi_activate_tcq(sdev, phba->cfg_lun_queue_depth);
else
scsi_deactivate_tcq(sdev, phba->cfg_lun_queue_depth);
/*
* Initialize the fc transport attributes for the target
* containing this scsi device. Also note that the driver's
* target pointer is stored in the starget_data for the
* driver's sysfs entry point functions.
*/
rport->dev_loss_tmo = phba->cfg_devloss_tmo;
[SCSI] lpfc 8.1.1 : Add polled-mode support - Add functionality to run in polled mode only. Includes run time attribute to enable mode. - Enable runtime writable hba settings for coallescing and delay parameters Customers have requested a mode in the driver to run strictly polled. This is generally to support an environment where the server is extremely loaded and is looking to reclaim some cpu cycles from adapter interrupt handling. This patch adds a new "poll" attribute, and the following behavior: if value is 0 (default): The driver uses the normal method for i/o completion. It uses the firmware feature of interrupt coalesing. The firmware allows a minimum number of i/o completions before an interrupt, or a maximum time delay between interrupts. By default, the driver sets these to no delay (disabled) or 1 i/o - meaning coalescing is disabled. Attributes were provided to change the coalescing values, but it was a module-load time only and global across all adapters. This patch allows them to be writable on a per-adapter basis. if value is 1 : Interrupts are left enabled, expecting that the user has tuned the interrupt coalescing values. When this setting is enabled, the driver will attempt to service completed i/o whenever new i/o is submitted to the adapter. If the coalescing values are large, and the i/o generation rate steady, an interrupt will be avoided by servicing completed i/o prior to the coalescing thresholds kicking in. However, if the i/o completion load is high enough or i/o generation slow, the coalescion values will ensure that completed i/o is serviced in a timely fashion. if value is 3 : Turns off FCP i/o interrupts altogether. The coalescing values now have no effect. A new attribute "poll_tmo" (default 10ms) exists to set the polling interval for i/o completion. When this setting is enabled, the driver will attempt to service completed i/o and restart the interval timer whenever new i/o is submitted. This behavior allows for servicing of completed i/o sooner than the interval timer, but ensures that if no i/o is being issued, then the interval timer will kick in to service the outstanding i/o. Signed-off-by: James Smart <James.Smart@emulex.com> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-30 05:32:13 +08:00
if (phba->cfg_poll & ENABLE_FCP_RING_POLLING) {
lpfc_sli_poll_fcp_ring(phba);
if (phba->cfg_poll & DISABLE_FCP_RING_INT)
lpfc_poll_rearm_timer(phba);
}
return 0;
}
static void
lpfc_slave_destroy(struct scsi_device *sdev)
{
sdev->hostdata = NULL;
return;
}
struct scsi_host_template lpfc_template = {
.module = THIS_MODULE,
.name = LPFC_DRIVER_NAME,
.info = lpfc_info,
.queuecommand = lpfc_queuecommand,
.eh_abort_handler = lpfc_abort_handler,
.eh_device_reset_handler= lpfc_reset_lun_handler,
.eh_bus_reset_handler = lpfc_reset_bus_handler,
.slave_alloc = lpfc_slave_alloc,
.slave_configure = lpfc_slave_configure,
.slave_destroy = lpfc_slave_destroy,
.this_id = -1,
.sg_tablesize = LPFC_SG_SEG_CNT,
.cmd_per_lun = LPFC_CMD_PER_LUN,
.use_clustering = ENABLE_CLUSTERING,
.shost_attrs = lpfc_host_attrs,
.max_sectors = 0xFFFF,
};