Commit e62cca19 authored by kashyap.desai@lsi.com's avatar kashyap.desai@lsi.com Committed by James Bottomley
Browse files

[SCSI] mptfusion: Better handling of DEAD IOC PCI-E Link down error condition 



Find Non-Operation IOC and remove it from OS: Detecting
dead(non-functional) ioc will be done reading doorbell register value
from fault reset thread, which has been called from work thread
context after each specific interval. If doorbell value is 0xFFFFFFFF,
it will be considered as IOC is non-operational and marked as dead
ioc.

Once Dead IOC has been detected, it will be removed at pci layer using
"pci_remove_bus_device" API.

Signed-off-by: default avatarKashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: default avatarJames Bottomley <JBottomley@Parallels.com>
parent 3850b14e

drivers/message/fusion/mptbase.c

+62 −1
Original line number Diff line number Diff line
@@ -63,6 +63,8 @@ 
#ifdef CONFIG_MTRR

#include <asm/mtrr.h>

#endif

#include <linux/kthread.h>

#include <scsi/scsi_host.h>



#include "mptbase.h"

#include "lsi/mpi_log_fc.h"
@@ -323,6 +325,32 @@ mpt_is_discovery_complete(MPT_ADAPTER *ioc) 
	return rc;

}





/**

 *  mpt_remove_dead_ioc_func - kthread context to remove dead ioc

 * @arg: input argument, used to derive ioc

 *

 * Return 0 if controller is removed from pci subsystem.

 * Return -1 for other case.

 */

static int mpt_remove_dead_ioc_func(void *arg)

{

	MPT_ADAPTER *ioc = (MPT_ADAPTER *)arg;

	struct pci_dev *pdev;



	if ((ioc == NULL))

		return -1;



	pdev = ioc->pcidev;

	if ((pdev == NULL))

		return -1;



	pci_remove_bus_device(pdev);

	return 0;

}







/**

 *	mpt_fault_reset_work - work performed on workq after ioc fault

 *	@work: input argument, used to derive ioc
@@ -336,12 +364,45 @@ mpt_fault_reset_work(struct work_struct *work) 
	u32		 ioc_raw_state;

	int		 rc;

	unsigned long	 flags;

	MPT_SCSI_HOST	*hd;

	struct task_struct *p;



	if (ioc->ioc_reset_in_progress || !ioc->active)

		goto out;





	ioc_raw_state = mpt_GetIocState(ioc, 0);

	if ((ioc_raw_state & MPI_IOC_STATE_MASK) == MPI_IOC_STATE_FAULT) {

	if ((ioc_raw_state & MPI_IOC_STATE_MASK) == MPI_IOC_STATE_MASK) {

		printk(MYIOC_s_INFO_FMT "%s: IOC is non-operational !!!!\n",

		    ioc->name, __func__);



		/*

		 * Call mptscsih_flush_pending_cmds callback so that we

		 * flush all pending commands back to OS.

		 * This call is required to aovid deadlock at block layer.

		 * Dead IOC will fail to do diag reset,and this call is safe

		 * since dead ioc will never return any command back from HW.

		 */

		hd = shost_priv(ioc->sh);

		ioc->schedule_dead_ioc_flush_running_cmds(hd);



		/*Remove the Dead Host */

		p = kthread_run(mpt_remove_dead_ioc_func, ioc,

				"mpt_dead_ioc_%d", ioc->id);

		if (IS_ERR(p))	{

			printk(MYIOC_s_ERR_FMT

				"%s: Running mpt_dead_ioc thread failed !\n",

				ioc->name, __func__);

		} else {

			printk(MYIOC_s_WARN_FMT

				"%s: Running mpt_dead_ioc thread success !\n",

				ioc->name, __func__);

		}

		return; /* don't rearm timer */

	}



	if ((ioc_raw_state & MPI_IOC_STATE_MASK)

			== MPI_IOC_STATE_FAULT) {

		printk(MYIOC_s_WARN_FMT "IOC is in FAULT state (%04xh)!!!\n",

		       ioc->name, ioc_raw_state & MPI_DOORBELL_DATA_MASK);

		printk(MYIOC_s_WARN_FMT "Issuing HardReset from %s!!\n",

drivers/message/fusion/mptbase.h

+38 −28
Original line number Diff line number Diff line
@@ -554,10 +554,47 @@ struct mptfc_rport_info 
	u8		flags;

};



/*=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=*/



/*

 * MPT_SCSI_HOST defines - Used by the IOCTL and the SCSI drivers

 * Private to the driver.

 */



#define MPT_HOST_BUS_UNKNOWN		(0xFF)

#define MPT_HOST_TOO_MANY_TM		(0x05)

#define MPT_HOST_NVRAM_INVALID		(0xFFFFFFFF)

#define MPT_HOST_NO_CHAIN		(0xFFFFFFFF)

#define MPT_NVRAM_MASK_TIMEOUT		(0x000000FF)

#define MPT_NVRAM_SYNC_MASK		(0x0000FF00)

#define MPT_NVRAM_SYNC_SHIFT		(8)

#define MPT_NVRAM_DISCONNECT_ENABLE	(0x00010000)

#define MPT_NVRAM_ID_SCAN_ENABLE	(0x00020000)

#define MPT_NVRAM_LUN_SCAN_ENABLE	(0x00040000)

#define MPT_NVRAM_TAG_QUEUE_ENABLE	(0x00080000)

#define MPT_NVRAM_WIDE_DISABLE		(0x00100000)

#define MPT_NVRAM_BOOT_CHOICE		(0x00200000)



typedef enum {

	FC,

	SPI,

	SAS

} BUS_TYPE;



typedef struct _MPT_SCSI_HOST {

	struct _MPT_ADAPTER		 *ioc;

	ushort			  sel_timeout[MPT_MAX_FC_DEVICES];

	char			  *info_kbuf;

	long			  last_queue_full;

	u16			  spi_pending;

	struct list_head	  target_reset_list;

} MPT_SCSI_HOST;



typedef void (*MPT_ADD_SGE)(void *pAddr, u32 flagslength, dma_addr_t dma_addr);

typedef void (*MPT_ADD_CHAIN)(void *pAddr, u8 next, u16 length,

		dma_addr_t dma_addr);

typedef void (*MPT_SCHEDULE_TARGET_RESET)(void *ioc);

typedef void (*MPT_FLUSH_RUNNING_CMDS)(MPT_SCSI_HOST *hd);



/*

 *  Adapter Structure - pci_dev specific. Maximum: MPT_MAX_ADAPTERS
@@ -717,6 +754,7 @@ typedef struct _MPT_ADAPTER 
	u8			 taskmgmt_quiesce_io;

	u8			 ioc_reset_in_progress;

	MPT_SCHEDULE_TARGET_RESET schedule_target_reset;

	MPT_FLUSH_RUNNING_CMDS schedule_dead_ioc_flush_running_cmds;

	struct work_struct	 sas_persist_task;



	struct work_struct	 fc_setup_reset_work;
@@ -830,19 +868,6 @@ typedef struct _MPT_LOCAL_REPLY { 
	u32	pad;

} MPT_LOCAL_REPLY;



#define MPT_HOST_BUS_UNKNOWN		(0xFF)

#define MPT_HOST_TOO_MANY_TM		(0x05)

#define MPT_HOST_NVRAM_INVALID		(0xFFFFFFFF)

#define MPT_HOST_NO_CHAIN		(0xFFFFFFFF)

#define MPT_NVRAM_MASK_TIMEOUT		(0x000000FF)

#define MPT_NVRAM_SYNC_MASK		(0x0000FF00)

#define MPT_NVRAM_SYNC_SHIFT		(8)

#define MPT_NVRAM_DISCONNECT_ENABLE	(0x00010000)

#define MPT_NVRAM_ID_SCAN_ENABLE	(0x00020000)

#define MPT_NVRAM_LUN_SCAN_ENABLE	(0x00040000)

#define MPT_NVRAM_TAG_QUEUE_ENABLE	(0x00080000)

#define MPT_NVRAM_WIDE_DISABLE		(0x00100000)

#define MPT_NVRAM_BOOT_CHOICE		(0x00200000)



/* The TM_STATE variable is used to provide strict single threading of TM

 * requests as well as communicate TM error conditions.
@@ -851,21 +876,6 @@ typedef struct _MPT_LOCAL_REPLY { 
#define	TM_STATE_IN_PROGRESS   (1)

#define	TM_STATE_ERROR	       (2)



typedef enum {

	FC,

	SPI,

	SAS

} BUS_TYPE;



typedef struct _MPT_SCSI_HOST {

	MPT_ADAPTER		 *ioc;

	ushort			  sel_timeout[MPT_MAX_FC_DEVICES];

	char 			  *info_kbuf;

	long			  last_queue_full;

	u16			  spi_pending;

	struct list_head	  target_reset_list;

} MPT_SCSI_HOST;



/*=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=*/

/*

 *	More Dynamic Multi-Pathing stuff...

drivers/message/fusion/mptsas.c

+2 −0
Original line number Diff line number Diff line
@@ -5147,6 +5147,8 @@ mptsas_probe(struct pci_dev *pdev, const struct pci_device_id *id) 
	ioc->TaskCtx = mptsasTaskCtx;

	ioc->InternalCtx = mptsasInternalCtx;

	ioc->schedule_target_reset = &mptsas_schedule_target_reset;

	ioc->schedule_dead_ioc_flush_running_cmds =

				&mptscsih_flush_running_cmds;

	/*  Added sanity check on readiness of the MPT adapter.

	 */

	if (ioc->last_state != MPI_IOC_STATE_OPERATIONAL) {

drivers/message/fusion/mptscsih.c

+2 −1
Original line number Diff line number Diff line
@@ -1024,7 +1024,7 @@ out: 
 *

 *	Must be called while new I/Os are being queued.

 */

static void

void

mptscsih_flush_running_cmds(MPT_SCSI_HOST *hd)

{

	MPT_ADAPTER *ioc = hd->ioc;
@@ -1055,6 +1055,7 @@ mptscsih_flush_running_cmds(MPT_SCSI_HOST *hd) 
		sc->scsi_done(sc);

	}

}

EXPORT_SYMBOL(mptscsih_flush_running_cmds);



/*

 *	mptscsih_search_running_cmds - Delete any commands associated

drivers/message/fusion/mptscsih.h

+1 −0
Original line number Diff line number Diff line
@@ -135,3 +135,4 @@ extern int mptscsih_is_phys_disk(MPT_ADAPTER *ioc, u8 channel, u8 id); 
extern struct device_attribute *mptscsih_host_attrs[];

extern struct scsi_cmnd	*mptscsih_get_scsi_lookup(MPT_ADAPTER *ioc, int i);

extern void mptscsih_taskmgmt_response_code(MPT_ADAPTER *ioc, u8 response_code);

extern void mptscsih_flush_running_cmds(MPT_SCSI_HOST *hd);

CRA Git | Maintained and supported by SUSTech CRA and CCSE