Skip to contentSkip to conversation

firs round of SCSI bug fixes for 2.6.29-rc1

PATCH In review James Bottomley 5 participants 13 messages Jan 16, 2009 – Feb 11, 2009

Subsystems (4)
lore
Share & tools
lore

James Bottomley

author maintainer
#0 lore raw

PATCH 1/1 firs round of SCSI bug fixes for 2.6.29-rc1

This is acutally four bug fixes and part of a fusion update which was a bit late processing through the merge window, but which should be perfectly acceptable for -rc1.

The patch is available here:

master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6.git

The short changelog is:

Anirban Chakraborty (1):
qla2xxx: Fix ISP restart bug in multiq code

Hannes Reinecke (1):
Skip deleted devices in __scsi_device_lookup_by_target()

ILLES, Marton (1):
Add SUN Universal Xport to no attach blacklist

Karen Xie (1):
iscsi_tcp: make padbuf non-static

Kashyap, Desai (3):
mpt fusion: Add Firmware debug support
mpt fusion: Add separate msi enable disable for FC,SPI,SAS mpt fusion: Update MPI Headers to version 01.05.19

and the diffstat:

 drivers/message/fusion/lsi/mpi.h           |    7 +-
 drivers/message/fusion/lsi/mpi_cnfg.h      |   47 +++++++++++++++
 drivers/message/fusion/lsi/mpi_fc.h        |    2 
 drivers/message/fusion/lsi/mpi_history.txt |   86 ++++++++++++++++++++--------
 drivers/message/fusion/lsi/mpi_init.h      |    2 
 drivers/message/fusion/lsi/mpi_ioc.h       |   22 ++++++-
 drivers/message/fusion/lsi/mpi_lan.h       |    2 
 drivers/message/fusion/lsi/mpi_log_fc.h    |    2 
 drivers/message/fusion/lsi/mpi_log_sas.h   |   37 ++++++++++--
 drivers/message/fusion/lsi/mpi_raid.h      |   11 +++
 drivers/message/fusion/lsi/mpi_sas.h       |   18 ++++--
 drivers/message/fusion/lsi/mpi_targ.h      |    2 
 drivers/message/fusion/lsi/mpi_tool.h      |    2 
 drivers/message/fusion/lsi/mpi_type.h      |    4 -
 drivers/message/fusion/mptbase.c           |   87 ++++++++++++++++++++++++-----
 drivers/message/fusion/mptbase.h           |    3 +
 drivers/message/fusion/mptscsih.c          |    3 +
 drivers/scsi/libiscsi_tcp.c                |    3 -
 drivers/scsi/qla2xxx/qla_init.c            |   58 ++++++++++++-------
 drivers/scsi/qla2xxx/qla_mid.c             |    1 
 drivers/scsi/qla2xxx/qla_os.c              |    7 +-
 drivers/scsi/scsi.c                        |    5 +
 drivers/scsi/scsi_devinfo.c                |    1 
 include/scsi/libiscsi_tcp.h                |    1 
 24 files changed, 318 insertions(+), 95 deletions(-)

Full diff below.

James

drivers/message/fusion/lsi/mpi.h +5 -2
index 10b6ef7..11c0f46 100644
@@ -6,7 +6,7 @@
  *          Title:  MPI Message independent structures and definitions
  *  Creation Date:  July 27, 2000
  *
- *    mpi.h Version:  01.05.13
+ *    mpi.h Version:  01.05.16
  *
  *  Version History
  *  ---------------
@@ -79,6 +79,9 @@
  *  03-27-06  01.05.11  Bumped MPI_HEADER_VERSION_UNIT.
  *  10-11-06  01.05.12  Bumped MPI_HEADER_VERSION_UNIT.
  *  05-24-07  01.05.13  Bumped MPI_HEADER_VERSION_UNIT.
+ *  08-07-07  01.05.14  Bumped MPI_HEADER_VERSION_UNIT.
+ *  01-15-08  01.05.15  Bumped MPI_HEADER_VERSION_UNIT.
+ *  03-28-08  01.05.16  Bumped MPI_HEADER_VERSION_UNIT.
  *  --------------------------------------------------------------------------
  */
 
@@ -109,7 +112,7 @@
 /* Note: The major versions of 0xe0 through 0xff are reserved */
 
 /* versioning for this MPI header set */
-#define MPI_HEADER_VERSION_UNIT             (0x10)
+#define MPI_HEADER_VERSION_UNIT             (0x13)
 #define MPI_HEADER_VERSION_DEV              (0x00)
 #define MPI_HEADER_VERSION_UNIT_MASK        (0xFF00)
 #define MPI_HEADER_VERSION_UNIT_SHIFT       (8)
drivers/message/fusion/lsi/mpi_cnfg.h +45 -2

95 lines not on this page · raw message

drivers/message/fusion/lsi/mpi_fc.h +1 -1
index 627acfb..7d663ce 100644
@@ -1,5 +1,5 @@
 /*
- *  Copyright (c) 2000-2004 LSI Corporation.
+ *  Copyright (c) 2000-2008 LSI Corporation.
  *
  *
  *           Name:  mpi_fc.h
drivers/message/fusion/lsi/mpi_history.txt +60 -26

159 lines not on this page · raw message

drivers/message/fusion/lsi/mpi_init.h +1 -1
index a9e3693..4295d06 100644
@@ -1,5 +1,5 @@
 /*
- *  Copyright (c) 2000-2007 LSI Corporation.
+ *  Copyright (c) 2000-2008 LSI Corporation.
  *
  *
  *           Name:  mpi_init.h
drivers/message/fusion/lsi/mpi_ioc.h +20 -2
index 5cbb6bd..8faa4fa 100644
@@ -1,12 +1,12 @@
 /*
- *  Copyright (c) 2000-2007 LSI Corporation.
+ *  Copyright (c) 2000-2008 LSI Corporation.
  *
  *
  *           Name:  mpi_ioc.h
  *          Title:  MPI IOC, Port, Event, FW Download, and FW Upload messages
  *  Creation Date:  August 11, 2000
  *
- *    mpi_ioc.h Version:  01.05.14
+ *    mpi_ioc.h Version:  01.05.16
  *
  *  Version History
  *  ---------------
@@ -113,6 +113,16 @@
  *                      added _MULTI_PORT_DOMAIN.
  *  05-24-07  01.05.14  Added Common Boot Block type to FWDownload Request.
  *                      Added Common Boot Block type to FWUpload Request.
+ *  08-07-07  01.05.15  Added MPI_EVENT_SAS_INIT_RC_REMOVED define.
+ *                      Added MPI_EVENT_IR2_RC_DUAL_PORT_ADDED and
+ *                      MPI_EVENT_IR2_RC_DUAL_PORT_REMOVED for IR2 event data.
+ *                      Added SASAddress field to SAS Initiator Device Table
+ *                      Overflow event data structure.
+ *  03-28-08  01.05.16  Added two new ReasonCode values to SAS Device Status
+ *                      Change Event data to indicate completion of internally
+ *                      generated task management.
+ *                      Added MPI_EVENT_DSCVRY_ERR_DS_SATA_INIT_FAILURE define.
+ *                      Added MPI_EVENT_SAS_INIT_RC_INACCESSIBLE define.
  *  --------------------------------------------------------------------------
  */
 
@@ -612,6 +622,8 @@ typedef struct _EVENT_DATA_SAS_DEVICE_STATUS_CHANGE
 #define MPI_EVENT_SAS_DEV_STAT_RC_CLEAR_TASK_SET_INTERNAL   (0x0B)
 #define MPI_EVENT_SAS_DEV_STAT_RC_QUERY_TASK_INTERNAL       (0x0C)
 #define MPI_EVENT_SAS_DEV_STAT_RC_ASYNC_NOTIFICATION        (0x0D)
+#define MPI_EVENT_SAS_DEV_STAT_RC_CMPL_INTERNAL_DEV_RESET   (0x0E)
+#define MPI_EVENT_SAS_DEV_STAT_RC_CMPL_TASK_ABORT_INTERNAL  (0x0F)
 
 
 /* SCSI Event data for Queue Full event */
@@ -708,6 +720,8 @@ typedef struct _MPI_EVENT_DATA_IR2
 #define MPI_EVENT_IR2_RC_PD_REMOVED                 (0x05)
 #define MPI_EVENT_IR2_RC_FOREIGN_CFG_DETECTED       (0x06)
 #define MPI_EVENT_IR2_RC_REBUILD_MEDIUM_ERROR       (0x07)
+#define MPI_EVENT_IR2_RC_DUAL_PORT_ADDED            (0x08)
+#define MPI_EVENT_IR2_RC_DUAL_PORT_REMOVED          (0x09)
 
 /* defines for logical disk states */
 #define MPI_LD_STATE_OPTIMAL                        (0x00)
@@ -867,6 +881,7 @@ typedef struct _EVENT_DATA_DISCOVERY_ERROR
 #define MPI_EVENT_DSCVRY_ERR_DS_UNSUPPORTED_DEVICE          (0x00000800)
 #define MPI_EVENT_DSCVRY_ERR_DS_MAX_SATA_TARGETS            (0x00001000)
 #define MPI_EVENT_DSCVRY_ERR_DS_MULTI_PORT_DOMAIN           (0x00002000)
+#define MPI_EVENT_DSCVRY_ERR_DS_SATA_INIT_FAILURE           (0x00004000)
 
 /* SAS SMP Error Event data */
 
@@ -902,6 +917,8 @@ typedef struct _EVENT_DATA_SAS_INIT_DEV_STATUS_CHANGE
 
 /* defines for the ReasonCode field of the SAS Initiator Device Status Change event */
 #define MPI_EVENT_SAS_INIT_RC_ADDED                 (0x01)
+#define MPI_EVENT_SAS_INIT_RC_REMOVED               (0x02)
+#define MPI_EVENT_SAS_INIT_RC_INACCESSIBLE          (0x03)
 
 /* SAS Initiator Device Table Overflow Event data */
 
@@ -910,6 +927,7 @@ typedef struct _EVENT_DATA_SAS_INIT_TABLE_OVERFLOW
     U8                      MaxInit;                    /* 00h */
     U8                      CurrentInit;                /* 01h */
     U16                     Reserved1;                  /* 02h */
+    U64                     SASAddress;                 /* 04h */
 } EVENT_DATA_SAS_INIT_TABLE_OVERFLOW,
   MPI_POINTER PTR_EVENT_DATA_SAS_INIT_TABLE_OVERFLOW,
   MpiEventDataSasInitTableOverflow_t,
drivers/message/fusion/lsi/mpi_lan.h +1 -1
index 03253b5..f41fcb6 100644
@@ -1,5 +1,5 @@
 /*
- *  Copyright (c) 2000-2004 LSI Corporation.
+ *  Copyright (c) 2000-2008 LSI Corporation.
  *
  *
  *           Name:  mpi_lan.h
drivers/message/fusion/lsi/mpi_log_fc.h +1 -1
index e4dafce..face6e7 100644
@@ -1,5 +1,5 @@
 /*
- *  Copyright (c) 2000-2001 LSI Corporation. All rights reserved.
+ *  Copyright (c) 2000-2008 LSI Corporation. All rights reserved.
  *
  *  NAME:           fc_log.h
  *  SUMMARY:        MPI IocLogInfo definitions for the SYMFC9xx chips
drivers/message/fusion/lsi/mpi_log_sas.h +31 -6
index af9da03..691620d 100644
@@ -1,6 +1,6 @@
 /***************************************************************************
  *                                                                         *
- *  Copyright 2003 LSI Corporation.  All rights reserved.            *
+ *  Copyright (c) 2000-2008 LSI Corporation.  All rights reserved.         *
  *                                                                         *
  * Description                                                             *
  * ------------                                                            *
@@ -73,6 +73,8 @@
 #define IOP_LOGINFO_CODE_TARGET_MODE_ABORT_EXACT_IO          (0x00070004)
 #define IOP_LOGINFO_CODE_TARGET_MODE_ABORT_EXACT_IO_REQ      (0x00070005)
 
+#define IOP_LOGINFO_CODE_LOG_TIMESTAMP_EVENT                 (0x00080000)
+
 /****************************************************************************/
 /* PL LOGINFO_CODE defines, valid if IOC_LOGINFO_ORIGINATOR = PL            */
 /****************************************************************************/
@@ -92,7 +94,7 @@
 #define PL_LOGINFO_SUB_CODE_OPEN_FAIL_OPEN_TIMEOUT_EXP       (0x0000000C)
 #define PL_LOGINFO_SUB_CODE_OPEN_FAIL_UNUSED_0D              (0x0000000D)
 #define PL_LOGINFO_SUB_CODE_OPEN_FAIL_DVTBLE_ACCSS_FAIL      (0x0000000E)
-#define PL_LOGINFO_SUB CODE_OPEN_FAIL_BAD_DEST               (0x00000011)
+#define PL_LOGINFO_SUB_CODE_OPEN_FAIL_BAD_DEST               (0x00000011)
 #define PL_LOGINFO_SUB_CODE_OPEN_FAIL_RATE_NOT_SUPP          (0x00000012)
 #define PL_LOGINFO_SUB_CODE_OPEN_FAIL_PROT_NOT_SUPP          (0x00000013)
 #define PL_LOGINFO_SUB_CODE_OPEN_FAIL_RESERVED_ABANDON0      (0x00000014)
@@ -159,10 +161,11 @@
 
 #define PL_LOGINFO_SUB_CODE_INVALID_SGL                      (0x00000200)
 #define PL_LOGINFO_SUB_CODE_WRONG_REL_OFF_OR_FRAME_LENGTH    (0x00000300)
-#define PL_LOGINFO_SUB_CODE_FRAME_XFER_ERROR                 (0x00000400) /* Bits 0-3 encode Transport Status Register (offset 0x08) */
-                                                                          /* Bit 0 is Status Bit 0: FrameXferErr */
-                                                                          /* Bit 1 & 2 are Status Bits 16 and 17: FrameXmitErrStatus */
-                                                                          /* Bit 3 is Status Bit 18 WriteDataLengthGTDataLengthErr */
+#define PL_LOGINFO_SUB_CODE_FRAME_XFER_ERROR                 (0x00000400)
+/* Bits 0-3 encode Transport Status Register (offset 0x08) */
+/* Bit 0 is Status Bit 0: FrameXferErr */
+/* Bit 1 & 2 are Status Bits 16 and 17: FrameXmitErrStatus */
+/* Bit 3 is Status Bit 18 WriteDataLenghtGTDataLengthErr */
 
 #define PL_LOGINFO_SUB_CODE_TX_FM_CONNECTED_LOW              (0x00000500)
 #define PL_LOGINFO_SUB_CODE_SATA_NON_NCQ_RW_ERR_BIT_SET      (0x00000600)
@@ -177,6 +180,11 @@
 #define PL_LOGINFO_SUB_CODE_DISCOVERY_REMOTE_SEP_RESET       (0x00000E01)
 #define PL_LOGINFO_SUB_CODE_SECOND_OPEN                      (0x00000F00)
 #define PL_LOGINFO_SUB_CODE_DSCVRY_SATA_INIT_TIMEOUT         (0x00001000)
+#define PL_LOGINFO_SUB_CODE_BREAK_ON_SATA_CONNECTION         (0x00002000)
+/* not currently used in mainline */
+#define PL_LOGINFO_SUB_CODE_BREAK_ON_STUCK_LINK              (0x00003000)
+#define PL_LOGINFO_SUB_CODE_BREAK_ON_STUCK_LINK_AIP          (0x00004000)
+#define PL_LOGINFO_SUB_CODE_BREAK_ON_INCOMPLETE_BREAK_RCVD   (0x00005000)
 
 #define PL_LOGINFO_CODE_ENCL_MGMT_SMP_FRAME_FAILURE          (0x00200000) /* Can't get SMP Frame */
 #define PL_LOGINFO_CODE_ENCL_MGMT_SMP_READ_ERROR             (0x00200010) /* Error occured on SMP Read */
@@ -243,6 +251,8 @@
 #define IR_LOGINFO_VOLUME_ACTIVATE_VOLUME_FAILED               (0x00010014)
 /* Activation failed trying to import the volume */
 #define IR_LOGINFO_VOLUME_ACTIVATING_IMPORT_VOLUME_FAILED      (0x00010015)
+/* Activation failed trying to import the volume */
+#define IR_LOGINFO_VOLUME_ACTIVATING_TOO_MANY_PHYS_DISKS       (0x00010016)
 
 /* Phys Disk failed, too many phys disks */
 #define IR_LOGINFO_PHYSDISK_CREATE_TOO_MANY_DISKS              (0x00010020)
@@ -285,6 +295,21 @@
 /* Compatibility Error : IME size limited to < 2TB */
 #define IR_LOGINFO_COMPAT_ERROR_IME_VOL_NOT_CURRENTLY_SUPPORTED (0x0001003D)
 
+/* Device Firmware Update: DFU can only be started once */
+#define IR_LOGINFO_DEV_FW_UPDATE_ERR_DFU_IN_PROGRESS            (0x00010050)
+/* Device Firmware Update: Volume must be Optimal/Active/non-Quiesced */
+#define IR_LOGINFO_DEV_FW_UPDATE_ERR_DEVICE_IN_INVALID_STATE    (0x00010051)
+/* Device Firmware Update: DFU Timeout cannot be zero */
+#define IR_LOGINFO_DEV_FW_UPDATE_ERR_INVALID_TIMEOUT            (0x00010052)
+/* Device Firmware Update: CREATE TIMER FAILED */
+#define IR_LOGINFO_DEV_FW_UPDATE_ERR_NO_TIMERS                  (0x00010053)
+/* Device Firmware Update: Failed to read SAS_IO_UNIT_PG_1 */
+#define IR_LOGINFO_DEV_FW_UPDATE_ERR_READING_CFG_PAGE           (0x00010054)
+/* Device Firmware Update: Invalid SAS_IO_UNIT_PG_1 value(s) */
+#define IR_LOGINFO_DEV_FW_UPDATE_ERR_PORT_IO_TIMEOUTS_REQUIRED  (0x00010055)
+/* Device Firmware Update: Unable to allocate memory for page */
+#define IR_LOGINFO_DEV_FW_UPDATE_ERR_ALLOC_CFG_PAGE             (0x00010056)
+
 
 /****************************************************************************/
 /* Defines for convenience                                                  */
drivers/message/fusion/lsi/mpi_raid.h +9 -2
index 2856108..add60cc 100644
@@ -1,12 +1,12 @@
 /*
- *  Copyright (c) 2001-2007 LSI Corporation.
+ *  Copyright (c) 2001-2008 LSI Corporation.
  *
  *
  *           Name:  mpi_raid.h
  *          Title:  MPI RAID message and structures
  *  Creation Date:  February 27, 2001
  *
- *    mpi_raid.h Version:  01.05.03
+ *    mpi_raid.h Version:  01.05.05
  *
  *  Version History
  *  ---------------
@@ -34,6 +34,9 @@
  *                      _SET_RESYNC_RATE and _SET_DATA_SCRUB_RATE.
  *  02-28-07  01.05.03  Added new RAID Action, Device FW Update Mode, and
  *                      associated defines.
+ *  08-07-07  01.05.04  Added Disable Full Rebuild bit to the ActionDataWord
+ *                      for the RAID Action MPI_RAID_ACTION_DISABLE_VOLUME.
+ *  01-15-08  01.05.05  Added define for MPI_RAID_ACTION_SET_VOLUME_NAME.
  *  --------------------------------------------------------------------------
  */
 
@@ -93,6 +96,7 @@ typedef struct _MSG_RAID_ACTION
 #define MPI_RAID_ACTION_SET_RESYNC_RATE             (0x13)
 #define MPI_RAID_ACTION_SET_DATA_SCRUB_RATE         (0x14)
 #define MPI_RAID_ACTION_DEVICE_FW_UPDATE_MODE       (0x15)
+#define MPI_RAID_ACTION_SET_VOLUME_NAME             (0x16)
 
 /* ActionDataWord defines for use with MPI_RAID_ACTION_CREATE_VOLUME action */
 #define MPI_RAID_ACTION_ADATA_DO_NOT_SYNC           (0x00000001)
@@ -105,6 +109,9 @@ typedef struct _MSG_RAID_ACTION
 #define MPI_RAID_ACTION_ADATA_KEEP_LBA0             (0x00000000)
 #define MPI_RAID_ACTION_ADATA_ZERO_LBA0             (0x00000002)
 
+/* ActionDataWord defines for use with MPI_RAID_ACTION_DISABLE_VOLUME action */
+#define MPI_RAID_ACTION_ADATA_DISABLE_FULL_REBUILD  (0x00000001)
+
 /* ActionDataWord defines for use with MPI_RAID_ACTION_ACTIVATE_VOLUME action */
 #define MPI_RAID_ACTION_ADATA_INACTIVATE_ALL        (0x00000001)
 
drivers/message/fusion/lsi/mpi_sas.h +13 -5
index 33fca83..ab41003 100644
@@ -1,12 +1,12 @@
 /*
- *  Copyright (c) 2004-2006 LSI Corporation.
+ *  Copyright (c) 2004-2008 LSI Corporation.
  *
  *
  *           Name:  mpi_sas.h
  *          Title:  MPI Serial Attached SCSI structures and definitions
  *  Creation Date:  August 19, 2004
  *
- *    mpi_sas.h Version:  01.05.04
+ *    mpi_sas.h Version:  01.05.05
  *
  *  Version History
  *  ---------------
@@ -23,6 +23,10 @@
  *                      reply.
  *  10-11-06  01.05.04  Fixed the name of a define for Operation field of SAS IO
  *                      Unit Control request.
+ *  01-15-08  01.05.05  Added support for MPI_SAS_OP_SET_IOC_PARAMETER,
+ *                      including adding IOCParameter and IOCParameter value
+ *                      fields to SAS IO Unit Control Request.
+ *                      Added MPI_SAS_DEVICE_INFO_PRODUCT_SPECIFIC define.
  *  --------------------------------------------------------------------------
  */
 
@@ -60,6 +64,8 @@
  * Values for the SAS DeviceInfo field used in SAS Device Status Change Event
  * data and SAS IO Unit Configuration pages.
  */
+#define MPI_SAS_DEVICE_INFO_PRODUCT_SPECIFIC    (0xF0000000)
+
 #define MPI_SAS_DEVICE_INFO_SEP                 (0x00004000)
 #define MPI_SAS_DEVICE_INFO_ATAPI_DEVICE        (0x00002000)
 #define MPI_SAS_DEVICE_INFO_LSI_DEVICE          (0x00001000)
@@ -216,7 +222,7 @@ typedef struct _MSG_SAS_IOUNIT_CONTROL_REQUEST
     U8                      ChainOffset;        /* 02h */
     U8                      Function;           /* 03h */
     U16                     DevHandle;          /* 04h */
-    U8                      Reserved3;          /* 06h */
+    U8                      IOCParameter;       /* 06h */
     U8                      MsgFlags;           /* 07h */
     U32                     MsgContext;         /* 08h */
     U8                      TargetID;           /* 0Ch */
@@ -225,7 +231,7 @@ typedef struct _MSG_SAS_IOUNIT_CONTROL_REQUEST
     U8                      PrimFlags;          /* 0Fh */
     U32                     Primitive;          /* 10h */
     U64                     SASAddress;         /* 14h */
-    U32                     Reserved4;          /* 1Ch */
+    U32                     IOCParameterValue;  /* 1Ch */
 } MSG_SAS_IOUNIT_CONTROL_REQUEST, MPI_POINTER PTR_MSG_SAS_IOUNIT_CONTROL_REQUEST,
   SasIoUnitControlRequest_t, MPI_POINTER pSasIoUnitControlRequest_t;
 
@@ -241,6 +247,8 @@ typedef struct _MSG_SAS_IOUNIT_CONTROL_REQUEST
 #define MPI_SAS_OP_TRANSMIT_PORT_SELECT_SIGNAL  (0x0C)
 #define MPI_SAS_OP_TRANSMIT_REMOVE_DEVICE       (0x0D)  /* obsolete name */
 #define MPI_SAS_OP_REMOVE_DEVICE                (0x0D)
+#define MPI_SAS_OP_SET_IOC_PARAMETER            (0x0E)
+#define MPI_SAS_OP_PRODUCT_SPECIFIC_MIN         (0x80)
 
 /* values for the PrimFlags field */
 #define MPI_SAS_PRIMFLAGS_SINGLE                (0x08)
@@ -256,7 +264,7 @@ typedef struct _MSG_SAS_IOUNIT_CONTROL_REPLY
     U8                      MsgLength;          /* 02h */
     U8                      Function;           /* 03h */
     U16                     DevHandle;          /* 04h */
-    U8                      Reserved3;          /* 06h */
+    U8                      IOCParameter;       /* 06h */
     U8                      MsgFlags;           /* 07h */
     U32                     MsgContext;         /* 08h */
     U16                     Reserved4;          /* 0Ch */
drivers/message/fusion/lsi/mpi_targ.h +1 -1
index ff8c37d..c3dea7f 100644
@@ -1,5 +1,5 @@
 /*
- *  Copyright (c) 2000-2004 LSI Corporation.
+ *  Copyright (c) 2000-2008 LSI Corporation.
  *
  *
  *           Name:  mpi_targ.h
drivers/message/fusion/lsi/mpi_tool.h +1 -1
index 8834ae6..53cd715 100644
@@ -1,5 +1,5 @@
 /*
- *  Copyright (c) 2001-2005 LSI Corporation.
+ *  Copyright (c) 2001-2008 LSI Corporation.
  *
  *
  *           Name:  mpi_tool.h
drivers/message/fusion/lsi/mpi_type.h +2 -2
index 08dad9c..888b26d 100644
@@ -1,12 +1,12 @@
 /*
- *  Copyright (c) 2000-2004 LSI Corporation.
+ *  Copyright (c) 2000-2008 LSI Corporation.
  *
  *
  *           Name:  mpi_type.h
  *          Title:  MPI Basic type definitions
  *  Creation Date:  June 6, 2000
  *
- *    mpi_type.h Version:  01.05.01
+ *    mpi_type.h Version:  01.05.02
  *
  *  Version History
  *  ---------------
drivers/message/fusion/mptbase.c +74 -13

118 lines not on this page · raw message

drivers/message/fusion/mptbase.h +3 -0
index dff048c..b3e981d 100644
@@ -922,11 +922,14 @@ extern void	 mpt_free_fw_memory(MPT_ADAPTER *ioc);
 extern int	 mpt_findImVolumes(MPT_ADAPTER *ioc);
 extern int	 mptbase_sas_persist_operation(MPT_ADAPTER *ioc, u8 persist_opcode);
 extern int	 mpt_raid_phys_disk_pg0(MPT_ADAPTER *ioc, u8 phys_disk_num, pRaidPhysDiskPage0_t phys_disk);
+extern void     mpt_halt_firmware(MPT_ADAPTER *ioc);
+
 
 /*
  *  Public data decl's...
  */
 extern struct list_head	  ioc_list;
+extern int mpt_fwfault_debug;
 
 /*=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=*/
 #endif		/* } __KERNEL__ */
drivers/message/fusion/mptscsih.c +3 -0
index ee09041..e62c6bc 100644
@@ -1846,6 +1846,9 @@ mptscsih_abort(struct scsi_cmnd * SCpnt)
 	if (hd->timeouts < -1)
 		hd->timeouts++;
 
+	if (mpt_fwfault_debug)
+		mpt_halt_firmware(ioc);
+
 	/* Most important!  Set TaskMsgContext to SCpnt's MsgContext!
 	 * (the IO to be ABORT'd)
 	 *
drivers/scsi/libiscsi_tcp.c +1 -2
index a745f91..e7705d3 100644
@@ -177,7 +177,6 @@ int iscsi_tcp_segment_done(struct iscsi_tcp_conn *tcp_conn,
 			   struct iscsi_segment *segment, int recv,
 			   unsigned copied)
 {
-	static unsigned char padbuf[ISCSI_PAD_LEN];
 	struct scatterlist sg;
 	unsigned int pad;
 
@@ -233,7 +232,7 @@ int iscsi_tcp_segment_done(struct iscsi_tcp_conn *tcp_conn,
 			debug_tcp("consume %d pad bytes\n", pad);
 			segment->total_size += pad;
 			segment->size = pad;
-			segment->data = padbuf;
+			segment->data = segment->padbuf;
 			return 0;
 		}
 	}
drivers/scsi/qla2xxx/qla_init.c +36 -22

100 lines not on this page · raw message

drivers/scsi/qla2xxx/qla_mid.c +1 -0
index 8863231..f53179c 100644
@@ -629,6 +629,7 @@ qla25xx_create_req_que(struct qla_hw_data *ha, uint16_t options,
 	req->ring_index = 0;
 	req->cnt = req->length;
 	req->id = que_id;
+	req->max_q_depth = ha->req_q_map[0]->max_q_depth;
 	mutex_unlock(&ha->vport_lock);
 
 	ret = qla25xx_init_req_que(base_vha, req, options);
drivers/scsi/qla2xxx/qla_os.c +3 -4
index 4a71f52..cf32653 100644
@@ -1158,8 +1158,8 @@ qla2x00_abort_all_cmds(scsi_qla_host_t *vha, int res)
 	struct req_que *req;
 
 	spin_lock_irqsave(&ha->hardware_lock, flags);
-	for (que = 0; que < QLA_MAX_HOST_QUES; que++) {
-		req = ha->req_q_map[vha->req_ques[que]];
+	for (que = 0; que < ha->max_queues; que++) {
+		req = ha->req_q_map[que];
 		if (!req)
 			continue;
 		for (cnt = 1; cnt < MAX_OUTSTANDING_COMMANDS; cnt++) {
@@ -1193,7 +1193,7 @@ qla2xxx_slave_configure(struct scsi_device *sdev)
 	scsi_qla_host_t *vha = shost_priv(sdev->host);
 	struct qla_hw_data *ha = vha->hw;
 	struct fc_rport *rport = starget_to_rport(sdev->sdev_target);
-	struct req_que *req = ha->req_q_map[0];
+	struct req_que *req = ha->req_q_map[vha->req_ques[0]];
 
 	if (sdev->tagged_supported)
 		scsi_activate_tcq(sdev, req->max_q_depth);
@@ -1998,7 +1998,6 @@ qla2x00_probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
 	return 0;
 
 probe_failed:
-	qla2x00_free_que(ha, req, rsp);
 	qla2x00_free_device(base_vha);
 
 	scsi_host_put(base_vha->host);
drivers/scsi/scsi.c +4 -1
index 42e72a2..cbcd3f6 100644
@@ -1095,7 +1095,8 @@ EXPORT_SYMBOL(__starget_for_each_device);
  * Description: Looks up the scsi_device with the specified @lun for a given
  * @starget.  The returned scsi_device does not have an additional
  * reference.  You must hold the host's host_lock over this call and
- * any access to the returned scsi_device.
+ * any access to the returned scsi_device. A scsi_device in state
+ * SDEV_DEL is skipped.
  *
  * Note:  The only reason why drivers should use this is because
  * they need to access the device list in irq context.  Otherwise you
@@ -1107,6 +1108,8 @@ struct scsi_device *__scsi_device_lookup_by_target(struct scsi_target *starget,
 	struct scsi_device *sdev;
 
 	list_for_each_entry(sdev, &starget->devices, same_target_siblings) {
+		if (sdev->sdev_state == SDEV_DEL)
+			continue;
 		if (sdev->lun ==lun)
 			return sdev;
 	}
drivers/scsi/scsi_devinfo.c +1 -0
index 4969e4e..099b545 100644
@@ -224,6 +224,7 @@ static struct {
 	{"SGI", "TP9100", "*", BLIST_REPORTLUN2},
 	{"SGI", "Universal Xport", "*", BLIST_NO_ULD_ATTACH},
 	{"IBM", "Universal Xport", "*", BLIST_NO_ULD_ATTACH},
+	{"SUN", "Universal Xport", "*", BLIST_NO_ULD_ATTACH},
 	{"SMSC", "USB 2 HS-CF", NULL, BLIST_SPARSELUN | BLIST_INQUIRY_36},
 	{"SONY", "CD-ROM CDU-8001", NULL, BLIST_BORKEN},
 	{"SONY", "TSL", NULL, BLIST_FORCELUN},		/* DDS3 & DDS4 autoloaders */
include/scsi/libiscsi_tcp.h +1 -0
index 83e32f6..9e3182e 100644
@@ -39,6 +39,7 @@ struct iscsi_segment {
 	unsigned int		total_copied;
 
 	struct hash_desc	*hash;
+	unsigned char		padbuf[ISCSI_PAD_LEN];
 	unsigned char		recv_digest[ISCSI_DIGEST_SIZE];
 	unsigned char		digest[ISCSI_DIGEST_SIZE];
 	unsigned int		digest_len;
#1 lore raw
Quoted 7 lines · On Fri, 16 Jan 2009 09:56:11 -0500 James Bottomley…

This is acutally four bug fixes and part of a fusion update which was a bit late processing through the merge window, but which should be perfectly acceptable for -rc1.

...

Full diff below.

I like full diffs.

Quoted 17 lines
drivers/message/fusion/mptbase.c +8 -3
@@ -79,9 +79,22 @@ MODULE_VERSION(my_VERSION);
 /*
  *  cmd line parameters
  */
-static int mpt_msi_enable = -1;
-module_param(mpt_msi_enable, int, 0);
-MODULE_PARM_DESC(mpt_msi_enable, " MSI Support Enable (default=0)");
+
+static int mpt_msi_enable_spi;
+module_param(mpt_msi_enable_spi, int, 0);
+MODULE_PARM_DESC(mpt_msi_enable_spi, " Enable MSI Support for SPI \
+		controllers (default=0)");
+
+static int mpt_msi_enable_fc;
+module_param(mpt_msi_enable_fc, int, 0);

drivers/message/fusion/mptbase.c MODULE_VERSION(my_VERSION); · line 98

Quoted 10 lines
drivers/message/fusion/mptbase.c +8 -0
+MODULE_PARM_DESC(mpt_msi_enable_fc, " Enable MSI Support for FC \
+		controllers (default=0)");
+
+static int mpt_msi_enable_sas;
+module_param(mpt_msi_enable_sas, int, 1);
+MODULE_PARM_DESC(mpt_msi_enable_sas, " Enable MSI Support for SAS \
+		controllers (default=1)");
+
 
...

Those MODULE_PARM_DESC strings are going to come out funny-looking, with extra tabs and stuff.

drivers/message/fusion/mptbase.c static int mpt_debug_level; · line 115

Quoted 9 lines
drivers/message/fusion/mptbase.c +9 -0
+MODULE_PARM_DESC(mpt_debug_level, " debug level - refer to mptdebug.h \
+	- (default=0)");
+
+int mpt_fwfault_debug;
+EXPORT_SYMBOL(mpt_fwfault_debug);
+module_param_call(mpt_fwfault_debug, param_set_int, param_get_int,
+	  &mpt_fwfault_debug, 0600);
+MODULE_PARM_DESC(mpt_fwfault_debug, "Enable detection of Firmware fault"
+	" and halt Firmware on fault - (default=0)");

That one got it right.

Quoted 58 lines

#ifdef MFCNT
static int mfcounter = 0;

drivers/message/fusion/mptbase.c +34 -9
@@ -1751,16 +1774,25 @@ mpt_attach(struct pci_dev *pdev, const struct pci_device_id *id)
 		ioc->bus_type = SAS;
 	}
 
-	if (mpt_msi_enable == -1) {
-		/* Enable on SAS, disable on FC and SPI */
-		if (ioc->bus_type == SAS)
-			ioc->msi_enable = 1;
-		else
-			ioc->msi_enable = 0;
-	} else
-		/* follow flag: 0 - disable; 1 - enable */
-		ioc->msi_enable = mpt_msi_enable;
 
+	switch (ioc->bus_type) {
+
+	case SAS:
+		ioc->msi_enable = mpt_msi_enable_sas;
+		break;
+
+	case SPI:
+		ioc->msi_enable = mpt_msi_enable_spi;
+		break;
+
+	case FC:
+		ioc->msi_enable = mpt_msi_enable_fc;
+		break;
+
+	default:
+		ioc->msi_enable = 0;
+		break;
+	}
 	if (ioc->errata_flag_1064)
 		pci_disable_io_access(pdev);
 
@@ -6313,6 +6345,33 @@ mpt_print_ioc_summary(MPT_ADAPTER *ioc, char *buffer, int *size, int len, int sh
 	*size = y;
 }
 
+
+/**
+ *	mpt_halt_firmware - Halts the firmware if it is operational and panic
+ *	the kernel
+ *	@ioc: Pointer to MPT_ADAPTER structure
+ *
+ **/
+void
+mpt_halt_firmware(MPT_ADAPTER *ioc)
+{
+	u32	 ioc_raw_state;
+
+	ioc_raw_state = mpt_GetIocState(ioc, 0);
+
+	if ((ioc_raw_state & MPI_IOC_STATE_MASK) == MPI_IOC_STATE_FAULT) {
+		printk(MYIOC_s_ERR_FMT "IOC is in FAULT state (%04xh)!!!\n",

drivers/message/fusion/mptbase.c mpt_print_ioc_summary(MPT_ADAPTER *ioc, char *buffer, int *size, int len, int sh · line 6373

Quoted 10 lines
drivers/message/fusion/mptbase.c +10 -0
+			ioc->name, ioc_raw_state & MPI_DOORBELL_DATA_MASK);
+		panic("%s: IOC Fault (%04xh)!!!\n", ioc->name,
+			ioc_raw_state & MPI_DOORBELL_DATA_MASK);
+	} else {
+		CHIPREG_WRITE32(&ioc->chip->Doorbell, 0xC0FFEE00);
+		panic("%s: Firmware is halted due to command timeout\n",
+			ioc->name);
+	}
+}
+EXPORT_SYMBOL(mpt_halt_firmware);

Doing a panic() after we've already detected an error is plain nasty. Is there no way in which we can allow the kernel to continue?

Quoted 36 lines
drivers/message/fusion/mptbase.c +2 -0
 /*=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=*/
 /*
  *	Reset Handling
@@ -6345,6 +6404,8 @@ mpt_HardResetHandler(MPT_ADAPTER *ioc, int sleepFlag)
 	printk(MYIOC_s_INFO_FMT "HardResetHandler Entered!\n", ioc->name);
 	printk("MF count 0x%x !\n", ioc->mfcnt);
 #endif
+	if (mpt_fwfault_debug)
+		mpt_halt_firmware(ioc);
 
 	/* Reset the adapter. Prevent more than 1 call to
 	 * mpt_do_ioc_recovery at any instant in time.
drivers/message/fusion/mptbase.h +3 -0
index dff048c..b3e981d 100644
@@ -922,11 +922,14 @@ extern void	 mpt_free_fw_memory(MPT_ADAPTER *ioc);
 extern int	 mpt_findImVolumes(MPT_ADAPTER *ioc);
 extern int	 mptbase_sas_persist_operation(MPT_ADAPTER *ioc, u8 persist_opcode);
 extern int	 mpt_raid_phys_disk_pg0(MPT_ADAPTER *ioc, u8 phys_disk_num, pRaidPhysDiskPage0_t phys_disk);
+extern void     mpt_halt_firmware(MPT_ADAPTER *ioc);
+
 
 /*
  *  Public data decl's...
  */
 extern struct list_head	  ioc_list;
+extern int mpt_fwfault_debug;
 
 /*=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=*/
 #endif		/* } __KERNEL__ */
drivers/message/fusion/mptscsih.c +0 -0
index ee09041..e62c6bc 100644

drivers/message/fusion/mptscsih.c mptscsih_abort(struct scsi_cmnd * SCpnt) · line 1850

Quoted 5 lines
drivers/message/fusion/mptscsih.c +2 -0
@@ -1846,6 +1846,9 @@ mptscsih_abort(struct scsi_cmnd * SCpnt)
 	if (hd->timeouts < -1)
 		hd->timeouts++;
 
+	if (mpt_fwfault_debug)
+		mpt_halt_firmware(ioc);

So one single global flag affects all controllers in the system?

I guess that's OK if it's just a debug/diag thing.

Quoted 32 lines
drivers/message/fusion/mptscsih.c +0 -0
 	/* Most important!  Set TaskMsgContext to SCpnt's MsgContext!
 	 * (the IO to be ABORT'd)
 	 *
drivers/scsi/libiscsi_tcp.c +1 -2
index a745f91..e7705d3 100644
@@ -177,7 +177,6 @@ int iscsi_tcp_segment_done(struct iscsi_tcp_conn *tcp_conn,
 			   struct iscsi_segment *segment, int recv,
 			   unsigned copied)
 {
-	static unsigned char padbuf[ISCSI_PAD_LEN];
 	struct scatterlist sg;
 	unsigned int pad;
 
@@ -233,7 +232,7 @@ int iscsi_tcp_segment_done(struct iscsi_tcp_conn *tcp_conn,
 			debug_tcp("consume %d pad bytes\n", pad);
 			segment->total_size += pad;
 			segment->size = pad;
-			segment->data = padbuf;
+			segment->data = segment->padbuf;
 			return 0;
 		}
 	}
drivers/scsi/qla2xxx/qla_init.c +0 -0
index 2d4f32b..9ad4d09 100644
@@ -1258,35 +1258,48 @@ qla2x00_init_rings(scsi_qla_host_t *vha)
 {
 	int	rval;
 	unsigned long flags = 0;

drivers/scsi/qla2xxx/qla_init.c qla2x00_init_rings(scsi_qla_host_t *vha) · line 1267

Quoted 10 lines
drivers/scsi/qla2xxx/qla_init.c +4 -3
-	int cnt;
+	int cnt, que;
 	struct qla_hw_data *ha = vha->hw;
-	struct req_que *req = ha->req_q_map[0];
-	struct rsp_que *rsp = ha->rsp_q_map[0];
+	struct req_que *req;
+	struct rsp_que *rsp;
+	struct scsi_qla_host *vp;
 	struct mid_init_cb_24xx *mid_init_cb =
 	    (struct mid_init_cb_24xx *) ha->init_cb;

This cast worries me. It's a cast between two complex data structures which appear to have nothing to do with each other.

Oh well. It _is_ a scsi driver :(

author maintainer
#2 lore raw
Quoted 38 lines · Andrew Morton
Quoted 7 lines · On Fri, 16 Jan 2009 09:56:11 -0500 James Bottomley…

This is acutally four bug fixes and part of a fusion update which was a bit late processing through the merge window, but which should be perfectly acceptable for -rc1.

...

Full diff below.

I like full diffs.

Quoted 26 lines
drivers/message/fusion/mptbase.c +16 -3
@@ -79,9 +79,22 @@ MODULE_VERSION(my_VERSION);
 /*
  *  cmd line parameters
  */
-static int mpt_msi_enable = -1;
-module_param(mpt_msi_enable, int, 0);
-MODULE_PARM_DESC(mpt_msi_enable, " MSI Support Enable (default=0)");
+
+static int mpt_msi_enable_spi;
+module_param(mpt_msi_enable_spi, int, 0);
+MODULE_PARM_DESC(mpt_msi_enable_spi, " Enable MSI Support for SPI \
+		controllers (default=0)");
+
+static int mpt_msi_enable_fc;
+module_param(mpt_msi_enable_fc, int, 0);
+MODULE_PARM_DESC(mpt_msi_enable_fc, " Enable MSI Support for FC \
+		controllers (default=0)");
+
+static int mpt_msi_enable_sas;
+module_param(mpt_msi_enable_sas, int, 1);
+MODULE_PARM_DESC(mpt_msi_enable_sas, " Enable MSI Support for SAS \
+		controllers (default=1)");
+
 
Quoted 4 lines
Quoted 1 line

...

Those MODULE_PARM_DESC strings are going to come out funny-looking, with extra tabs and stuff.

Yes ... the compiler complained about a few of the other updates, which I told them to redo ... I didn't notice these.

Quoted 75 lines

drivers/message/fusion/mptbase.c static int mpt_debug_level; · line 115

Quoted 9 lines
drivers/message/fusion/mptbase.c +9 -0
+MODULE_PARM_DESC(mpt_debug_level, " debug level - refer to mptdebug.h \
+	- (default=0)");
+
+int mpt_fwfault_debug;
+EXPORT_SYMBOL(mpt_fwfault_debug);
+module_param_call(mpt_fwfault_debug, param_set_int, param_get_int,
+	  &mpt_fwfault_debug, 0600);
+MODULE_PARM_DESC(mpt_fwfault_debug, "Enable detection of Firmware fault"
+	" and halt Firmware on fault - (default=0)");

That one got it right.

Quoted 63 lines

#ifdef MFCNT
static int mfcounter = 0;

drivers/message/fusion/mptbase.c +39 -9
@@ -1751,16 +1774,25 @@ mpt_attach(struct pci_dev *pdev, const struct pci_device_id *id)
 		ioc->bus_type = SAS;
 	}
 
-	if (mpt_msi_enable == -1) {
-		/* Enable on SAS, disable on FC and SPI */
-		if (ioc->bus_type == SAS)
-			ioc->msi_enable = 1;
-		else
-			ioc->msi_enable = 0;
-	} else
-		/* follow flag: 0 - disable; 1 - enable */
-		ioc->msi_enable = mpt_msi_enable;
 
+	switch (ioc->bus_type) {
+
+	case SAS:
+		ioc->msi_enable = mpt_msi_enable_sas;
+		break;
+
+	case SPI:
+		ioc->msi_enable = mpt_msi_enable_spi;
+		break;
+
+	case FC:
+		ioc->msi_enable = mpt_msi_enable_fc;
+		break;
+
+	default:
+		ioc->msi_enable = 0;
+		break;
+	}
 	if (ioc->errata_flag_1064)
 		pci_disable_io_access(pdev);
 
@@ -6313,6 +6345,33 @@ mpt_print_ioc_summary(MPT_ADAPTER *ioc, char *buffer, int *size, int len, int sh
 	*size = y;
 }
 
+
+/**
+ *	mpt_halt_firmware - Halts the firmware if it is operational and panic
+ *	the kernel
+ *	@ioc: Pointer to MPT_ADAPTER structure
+ *
+ **/
+void
+mpt_halt_firmware(MPT_ADAPTER *ioc)
+{
+	u32	 ioc_raw_state;
+
+	ioc_raw_state = mpt_GetIocState(ioc, 0);
+
+	if ((ioc_raw_state & MPI_IOC_STATE_MASK) == MPI_IOC_STATE_FAULT) {
+		printk(MYIOC_s_ERR_FMT "IOC is in FAULT state (%04xh)!!!\n",
+			ioc->name, ioc_raw_state & MPI_DOORBELL_DATA_MASK);
+		panic("%s: IOC Fault (%04xh)!!!\n", ioc->name,
+			ioc_raw_state & MPI_DOORBELL_DATA_MASK);
+	} else {
+		CHIPREG_WRITE32(&ioc->chip->Doorbell, 0xC0FFEE00);
Quoted 8 lines

drivers/message/fusion/mptbase.c mpt_print_ioc_summary(MPT_ADAPTER *ioc, char *buffer, int *size, int len, int sh · line 6373

Quoted 5 lines
drivers/message/fusion/mptbase.c +5 -0
+		panic("%s: Firmware is halted due to command timeout\n",
+			ioc->name);
+	}
+}
+EXPORT_SYMBOL(mpt_halt_firmware);

Doing a panic() after we've already detected an error is plain nasty. Is there no way in which we can allow the kernel to continue?

There's a long thread discussing this very point on linux-scsi. The short version is "no, it's only used for debugging firmware and if you've corrupted your firmware to this point you need the machine halting".

Quoted 40 lines
Quoted 40 lines
drivers/message/fusion/mptbase.c +2 -0
 /*=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=*/
 /*
  *	Reset Handling
@@ -6345,6 +6404,8 @@ mpt_HardResetHandler(MPT_ADAPTER *ioc, int sleepFlag)
 	printk(MYIOC_s_INFO_FMT "HardResetHandler Entered!\n", ioc->name);
 	printk("MF count 0x%x !\n", ioc->mfcnt);
 #endif
+	if (mpt_fwfault_debug)
+		mpt_halt_firmware(ioc);
 
 	/* Reset the adapter. Prevent more than 1 call to
 	 * mpt_do_ioc_recovery at any instant in time.
drivers/message/fusion/mptbase.h +3 -0
index dff048c..b3e981d 100644
@@ -922,11 +922,14 @@ extern void	 mpt_free_fw_memory(MPT_ADAPTER *ioc);
 extern int	 mpt_findImVolumes(MPT_ADAPTER *ioc);
 extern int	 mptbase_sas_persist_operation(MPT_ADAPTER *ioc, u8 persist_opcode);
 extern int	 mpt_raid_phys_disk_pg0(MPT_ADAPTER *ioc, u8 phys_disk_num, pRaidPhysDiskPage0_t phys_disk);
+extern void     mpt_halt_firmware(MPT_ADAPTER *ioc);
+
 
 /*
  *  Public data decl's...
  */
 extern struct list_head	  ioc_list;
+extern int mpt_fwfault_debug;
 
 /*=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=*/
 #endif		/* } __KERNEL__ */
drivers/message/fusion/mptscsih.c +0 -0
index ee09041..e62c6bc 100644
@@ -1846,6 +1846,9 @@ mptscsih_abort(struct scsi_cmnd * SCpnt)
 	if (hd->timeouts < -1)
 		hd->timeouts++;
 
Quoted 6 lines

drivers/message/fusion/mptscsih.c mptscsih_abort(struct scsi_cmnd * SCpnt) · line 1850

Quoted 2 lines
drivers/message/fusion/mptscsih.c +2 -0
+	if (mpt_fwfault_debug)
+		mpt_halt_firmware(ioc);

So one single global flag affects all controllers in the system?

I guess that's OK if it's just a debug/diag thing.

Yes, that's correct.

Quoted 37 lines
Quoted 37 lines
drivers/message/fusion/mptscsih.c +0 -0
 	/* Most important!  Set TaskMsgContext to SCpnt's MsgContext!
 	 * (the IO to be ABORT'd)
 	 *
drivers/scsi/libiscsi_tcp.c +1 -2
index a745f91..e7705d3 100644
@@ -177,7 +177,6 @@ int iscsi_tcp_segment_done(struct iscsi_tcp_conn *tcp_conn,
 			   struct iscsi_segment *segment, int recv,
 			   unsigned copied)
 {
-	static unsigned char padbuf[ISCSI_PAD_LEN];
 	struct scatterlist sg;
 	unsigned int pad;
 
@@ -233,7 +232,7 @@ int iscsi_tcp_segment_done(struct iscsi_tcp_conn *tcp_conn,
 			debug_tcp("consume %d pad bytes\n", pad);
 			segment->total_size += pad;
 			segment->size = pad;
-			segment->data = padbuf;
+			segment->data = segment->padbuf;
 			return 0;
 		}
 	}
drivers/scsi/qla2xxx/qla_init.c +1 -3
index 2d4f32b..9ad4d09 100644
@@ -1258,35 +1258,48 @@ qla2x00_init_rings(scsi_qla_host_t *vha)
 {
 	int	rval;
 	unsigned long flags = 0;
-	int cnt;
+	int cnt, que;
 	struct qla_hw_data *ha = vha->hw;
-	struct req_que *req = ha->req_q_map[0];
-	struct rsp_que *rsp = ha->rsp_q_map[0];
Quoted 8 lines

drivers/scsi/qla2xxx/qla_init.c qla2x00_init_rings(scsi_qla_host_t *vha) · line 1267

Quoted 5 lines
drivers/scsi/qla2xxx/qla_init.c +3 -0
+	struct req_que *req;
+	struct rsp_que *rsp;
+	struct scsi_qla_host *vp;
 	struct mid_init_cb_24xx *mid_init_cb =
 	    (struct mid_init_cb_24xx *) ha->init_cb;

This cast worries me. It's a cast between two complex data structures which appear to have nothing to do with each other.

Actually, it's a C++ type construct. ha->init_cb is of type init_cb_t, mid_init_cb actually contains this (via a second indirection) as the first element, so what it's doing is dynamically casting out based on the board type.

Quoted 1 line

Oh well. It _is_ a scsi driver :(

James

#3 lore raw
Quoted 19 lines · On Fri, 16 Jan 2009 13:48:53 -0500 James Bottomley…
Quoted 10 lines
Quoted 10 lines

+++ b/drivers/scsi/qla2xxx/qla_init.c

drivers/scsi/qla2xxx/qla_init.c +1 -3
@@ -1258,35 +1258,48 @@ qla2x00_init_rings(scsi_qla_host_t *vha)
 {
 	int	rval;
 	unsigned long flags = 0;
-	int cnt;
+	int cnt, que;
 	struct qla_hw_data *ha = vha->hw;
-	struct req_que *req = ha->req_q_map[0];
-	struct rsp_que *rsp = ha->rsp_q_map[0];
Quoted 8 lines

drivers/scsi/qla2xxx/qla_init.c qla2x00_init_rings(scsi_qla_host_t *vha) · line 1267

Quoted 5 lines
drivers/scsi/qla2xxx/qla_init.c +3 -0
+	struct req_que *req;
+	struct rsp_que *rsp;
+	struct scsi_qla_host *vp;
 	struct mid_init_cb_24xx *mid_init_cb =
 	    (struct mid_init_cb_24xx *) ha->init_cb;

This cast worries me. It's a cast between two complex data structures which appear to have nothing to do with each other.

Quoted 4 lines

Actually, it's a C++ type construct. ha->init_cb is of type init_cb_t, mid_init_cb actually contains this (via a second indirection) as the first element, so what it's doing is dynamically casting out based on the board type.

Ah, OK, I got lost amongst the typedefs and pointers.

This code could/should use container_of() for this operation?

author maintainer
#4 lore raw
Quoted 22 lines · Andrew Morton
Quoted 19 lines · On Fri, 16 Jan 2009 13:48:53 -0500 James Bottomley…
Quoted 18 lines
Quoted 5 lines

+++ b/drivers/scsi/qla2xxx/qla_init.c

drivers/scsi/qla2xxx/qla_init.c +0 -0
@@ -1258,35 +1258,48 @@ qla2x00_init_rings(scsi_qla_host_t *vha)
 {
 	int	rval;
 	unsigned long flags = 0;

drivers/scsi/qla2xxx/qla_init.c qla2x00_init_rings(scsi_qla_host_t *vha) · line 1267

Quoted 10 lines
drivers/scsi/qla2xxx/qla_init.c +4 -3
-	int cnt;
+	int cnt, que;
 	struct qla_hw_data *ha = vha->hw;
-	struct req_que *req = ha->req_q_map[0];
-	struct rsp_que *rsp = ha->rsp_q_map[0];
+	struct req_que *req;
+	struct rsp_que *rsp;
+	struct scsi_qla_host *vp;
 	struct mid_init_cb_24xx *mid_init_cb =
 	    (struct mid_init_cb_24xx *) ha->init_cb;

This cast worries me. It's a cast between two complex data structures which appear to have nothing to do with each other.

Quoted 8 lines
Quoted 4 lines

Actually, it's a C++ type construct. ha->init_cb is of type init_cb_t, mid_init_cb actually contains this (via a second indirection) as the first element, so what it's doing is dynamically casting out based on the board type.

Ah, OK, I got lost amongst the typedefs and pointers.

This code could/should use container_of() for this operation?

It could, but since it's a double cast out it would get a lot uglier. We don't really pick up any safety from this because the FW structure is shared with the device, so no-one can move the first element anyway.

James

#5 lore raw
In reply to James Bottomley

From: James Bottomley <[email protected]>
Date: Fri, 16 Jan 2009 13:48:53 -0500

Quoted 8 lines · Andrew Morton
Quoted 2 lines

Doing a panic() after we've already detected an error is plain nasty. Is there no way in which we can allow the kernel to continue?

There's a long thread discussing this very point on linux-scsi. The short version is "no, it's only used for debugging firmware and if you've corrupted your firmware to this point you need the machine halting".

Long thread or not, taking out one's entire system because one device hits a fail state is always wrong.

If I have other block devices which are working, all I want is for this specific controller and it's block devices to fail.

This way I can get the failure log message and actually do something with that data without rebooting.

Or I guess digital cameras and serial/net consoles are the only sanctioned way to record kernel failure log messages of this kind?

Give me a break. :)

#6 lore raw
In reply to Andrew Morton

On Fri, Jan 16, 2009 at 9:09 AM, Andrew Morton
<[email protected]> wrote:
...

Quoted 9 lines

drivers/message/fusion/mptbase.c mpt_print_ioc_summary(MPT_ADAPTER *ioc, char *buffer, int *size, int len, int sh · line 6373

Quoted 6 lines
drivers/message/fusion/mptbase.c +6 -0
+             CHIPREG_WRITE32(&ioc->chip->Doorbell, 0xC0FFEE00);
+             panic("%s: Firmware is halted due to command timeout\n",
+                     ioc->name);
+     }
+}
+EXPORT_SYMBOL(mpt_halt_firmware);

Doing a panic() after we've already detected an error is plain nasty. Is there no way in which we can allow the kernel to continue?

Basically what James said. It's not long though: only 6 emails. I started that thread thinking the same thing you did:

    http://marc.info/?t=123123569200003&r=1&w=2

TBH, I still wouldn't implement it the way LSI maintainers did. I'm happy they are the maintainers and it's their call.

grant

#7 lore raw

From: Grant Grundler <[email protected]>
Date: Fri, 16 Jan 2009 11:02:12 -0800

Quoted 13 lines

On Fri, Jan 16, 2009 at 9:09 AM, Andrew Morton
<[email protected]> wrote:
...

Quoted 9 lines

drivers/message/fusion/mptbase.c mpt_print_ioc_summary(MPT_ADAPTER *ioc, char *buffer, int *size, int len, int sh · line 6373

Quoted 6 lines
drivers/message/fusion/mptbase.c +6 -0
+             CHIPREG_WRITE32(&ioc->chip->Doorbell, 0xC0FFEE00);
+             panic("%s: Firmware is halted due to command timeout\n",
+                     ioc->name);
+     }
+}
+EXPORT_SYMBOL(mpt_halt_firmware);

Doing a panic() after we've already detected an error is plain nasty. Is there no way in which we can allow the kernel to continue?

Quoted 6 lines

Basically what James said. It's not long though: only 6 emails. I started that thread thinking the same thing you did:

    http://marc.info/?t=123123569200003&r=1&w=2

TBH, I still wouldn't implement it the way LSI maintainers did. I'm happy they are the maintainers and it's their call.

They are doing things which effect the entire system, not just their driver. So this doesn't fall under "maintainer's rights"

If they fail locally, that's fine and perfectly expected, but to take the whole machine out is another matter altogether.

#8 lore raw

On Fri, Jan 16, 2009 at 11:10 AM, David Miller <[email protected]> wrote:
...

Quoted 6 lines
Quoted 3 lines

TBH, I still wouldn't implement it the way LSI maintainers did. I'm happy they are the maintainers and it's their call.

They are doing things which effect the entire system, not just their driver. So this doesn't fall under "maintainer's rights"

Agreed. This case is about a debug option which is normally disabled. So most people will never see this.

The thread was more about how to make the failure visible in a sane way WITHOUT the debug flag enabled. Then when the debug flag is enabled, panic the machine as well.

Quoted 2 lines

If they fail locally, that's fine and perfectly expected, but to take the whole machine out is another matter altogether.

Agreed.

thanks,
grant

author maintainer
#9 lore raw
Quoted 12 lines · Grant Grundler

On Fri, Jan 16, 2009 at 11:10 AM, David Miller <[email protected]> wrote:
...

Quoted 6 lines
Quoted 3 lines

TBH, I still wouldn't implement it the way LSI maintainers did. I'm happy they are the maintainers and it's their call.

They are doing things which effect the entire system, not just their driver. So this doesn't fall under "maintainer's rights"

Agreed. This case is about a debug option which is normally disabled. So most people will never see this.

Quoted 8 lines

The thread was more about how to make the failure visible in a sane way WITHOUT the debug flag enabled. Then when the debug flag is enabled, panic the machine as well.

Quoted 2 lines

If they fail locally, that's fine and perfectly expected, but to take the whole machine out is another matter altogether.

Agreed.

Me too, but I rejected half of the LSI update because of other problems. Since this is isolated to actual firmware debugging, which no normal user will see, I let it go after a bit of arguing.

James

#10 lore raw
In reply to James Bottomley
Quoted 1 line · James Bottomley

mpt fusion: Add separate msi enable disable for FC,SPI,SAS

I thought we rejected this one as being completely bogus? I certainly don't remember seeing even an excuse for it after I challenged it.

Signature · 4 lines
Matthew Wilcox				Intel Open Source Technology Centre
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."
#11 lore raw
Quoted 5 lines · Matthew Wilcox
Quoted 1 line · James Bottomley

mpt fusion: Add separate msi enable disable for FC,SPI,SAS

I thought we rejected this one as being completely bogus? I certainly don't remember seeing even an excuse for it after I challenged it.

... no response?

Signature · 4 lines
Matthew Wilcox				Intel Open Source Technology Centre
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."
author maintainer
#12 lore raw
Quoted 10 lines · Matthew Wilcox
Quoted 5 lines · Matthew Wilcox
Quoted 1 line · James Bottomley

mpt fusion: Add separate msi enable disable for FC,SPI,SAS

I thought we rejected this one as being completely bogus? I certainly don't remember seeing even an excuse for it after I challenged it.

... no response?

From whom? It's no less bogus than what they were doing previously. I

happen to think the whole idea of drivers chosing msi/not msi by parameters is bogus ... we just don't have a good platform way of sorting it out at the moment. In that absence, the driver maintainer gets to fix up.

James