summaryrefslogtreecommitdiffstats
path: root/drivers/scsi/cpqfcTSinit.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/scsi/cpqfcTSinit.c')
-rw-r--r--drivers/scsi/cpqfcTSinit.c2098
1 files changed, 2098 insertions, 0 deletions
diff --git a/drivers/scsi/cpqfcTSinit.c b/drivers/scsi/cpqfcTSinit.c
new file mode 100644
index 000000000000..2eeb493f5a2b
--- /dev/null
+++ b/drivers/scsi/cpqfcTSinit.c
@@ -0,0 +1,2098 @@
+/* Copyright(c) 2000, Compaq Computer Corporation
+ * Fibre Channel Host Bus Adapter
+ * 64-bit, 66MHz PCI
+ * Originally developed and tested on:
+ * (front): [chip] Tachyon TS HPFC-5166A/1.2 L2C1090 ...
+ * SP# P225CXCBFIEL6T, Rev XC
+ * SP# 161290-001, Rev XD
+ * (back): Board No. 010008-001 A/W Rev X5, FAB REV X5
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; either version 2, or (at your option) any
+ * later version.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ * Written by Don Zimmerman
+ * IOCTL and procfs added by Jouke Numan
+ * SMP testing by Chel Van Gennip
+ *
+ * portions copied from:
+ * QLogic CPQFCTS SCSI-FCP
+ * Written by Erik H. Moe, ehm@cris.com
+ * Copyright 1995, Erik H. Moe
+ * Renamed and updated to 1.3.x by Michael Griffith <grif@cs.ucr.edu>
+ * Chris Loveland <cwl@iol.unh.edu> to support the isp2100 and isp2200
+*/
+
+
+#define LinuxVersionCode(v, p, s) (((v)<<16)+((p)<<8)+(s))
+
+#include <linux/config.h>
+#include <linux/interrupt.h>
+#include <linux/module.h>
+#include <linux/version.h>
+#include <linux/blkdev.h>
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/delay.h>
+#include <linux/timer.h>
+#include <linux/init.h>
+#include <linux/ioport.h> // request_region() prototype
+#include <linux/completion.h>
+
+#include <asm/io.h>
+#include <asm/uaccess.h> // ioctl related
+#include <asm/irq.h>
+#include <linux/spinlock.h>
+#include "scsi.h"
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_ioctl.h>
+#include "cpqfcTSchip.h"
+#include "cpqfcTSstructs.h"
+#include "cpqfcTStrigger.h"
+
+#include "cpqfcTS.h"
+
+/* Embedded module documentation macros - see module.h */
+MODULE_AUTHOR("Compaq Computer Corporation");
+MODULE_DESCRIPTION("Driver for Compaq 64-bit/66Mhz PCI Fibre Channel HBA v. 2.5.4");
+MODULE_LICENSE("GPL");
+
+int cpqfcTS_TargetDeviceReset( Scsi_Device *ScsiDev, unsigned int reset_flags);
+
+// This struct was originally defined in
+// /usr/src/linux/include/linux/proc_fs.h
+// since it's only partially implemented, we only use first
+// few fields...
+// NOTE: proc_fs changes in 2.4 kernel
+
+#if LINUX_VERSION_CODE < LinuxVersionCode(2,3,27)
+static struct proc_dir_entry proc_scsi_cpqfcTS =
+{
+ PROC_SCSI_CPQFCTS, // ushort low_ino (enumerated list)
+ 7, // ushort namelen
+ DEV_NAME, // const char* name
+ S_IFDIR | S_IRUGO | S_IXUGO, // mode_t mode
+ 2 // nlink_t nlink
+ // etc. ...
+};
+
+
+#endif
+
+#if LINUX_VERSION_CODE >= LinuxVersionCode(2,4,7)
+# define CPQFC_DECLARE_COMPLETION(x) DECLARE_COMPLETION(x)
+# define CPQFC_WAITING waiting
+# define CPQFC_COMPLETE(x) complete(x)
+# define CPQFC_WAIT_FOR_COMPLETION(x) wait_for_completion(x);
+#else
+# define CPQFC_DECLARE_COMPLETION(x) DECLARE_MUTEX_LOCKED(x)
+# define CPQFC_WAITING sem
+# define CPQFC_COMPLETE(x) up(x)
+# define CPQFC_WAIT_FOR_COMPLETION(x) down(x)
+#endif
+
+static int cpqfc_alloc_private_data_pool(CPQFCHBA *hba);
+
+/* local function to load our per-HBA (local) data for chip
+ registers, FC link state, all FC exchanges, etc.
+
+ We allocate space and compute address offsets for the
+ most frequently accessed addresses; others (like World Wide
+ Name) are not necessary.
+*/
+static void Cpqfc_initHBAdata(CPQFCHBA *cpqfcHBAdata, struct pci_dev *PciDev )
+{
+
+ cpqfcHBAdata->PciDev = PciDev; // copy PCI info ptr
+
+ // since x86 port space is 64k, we only need the lower 16 bits
+ cpqfcHBAdata->fcChip.Registers.IOBaseL =
+ PciDev->resource[1].start & PCI_BASE_ADDRESS_IO_MASK;
+
+ cpqfcHBAdata->fcChip.Registers.IOBaseU =
+ PciDev->resource[2].start & PCI_BASE_ADDRESS_IO_MASK;
+
+ // 32-bit memory addresses
+ cpqfcHBAdata->fcChip.Registers.MemBase =
+ PciDev->resource[3].start & PCI_BASE_ADDRESS_MEM_MASK;
+
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase =
+ ioremap( PciDev->resource[3].start & PCI_BASE_ADDRESS_MEM_MASK,
+ 0x200);
+
+ cpqfcHBAdata->fcChip.Registers.RAMBase =
+ PciDev->resource[4].start;
+
+ cpqfcHBAdata->fcChip.Registers.SROMBase = // NULL for HP TS adapter
+ PciDev->resource[5].start;
+
+ // now the Tachlite chip registers
+ // the REGISTER struct holds both the physical address & last
+ // written value (some TL registers are WRITE ONLY)
+
+ cpqfcHBAdata->fcChip.Registers.SFQconsumerIndex.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_SFQ_CONSUMER_INDEX;
+
+ cpqfcHBAdata->fcChip.Registers.ERQproducerIndex.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_ERQ_PRODUCER_INDEX;
+
+ // TL Frame Manager
+ cpqfcHBAdata->fcChip.Registers.FMconfig.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_FM_CONFIG;
+ cpqfcHBAdata->fcChip.Registers.FMcontrol.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_FM_CONTROL;
+ cpqfcHBAdata->fcChip.Registers.FMstatus.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_FM_STATUS;
+ cpqfcHBAdata->fcChip.Registers.FMLinkStatus1.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_FM_LINK_STAT1;
+ cpqfcHBAdata->fcChip.Registers.FMLinkStatus2.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_FM_LINK_STAT2;
+ cpqfcHBAdata->fcChip.Registers.FMBB_CreditZero.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_FM_BB_CREDIT0;
+
+ // TL Control Regs
+ cpqfcHBAdata->fcChip.Registers.TYconfig.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_TACH_CONFIG;
+ cpqfcHBAdata->fcChip.Registers.TYcontrol.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_TACH_CONTROL;
+ cpqfcHBAdata->fcChip.Registers.TYstatus.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_TACH_STATUS;
+ cpqfcHBAdata->fcChip.Registers.rcv_al_pa.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_FM_RCV_AL_PA;
+ cpqfcHBAdata->fcChip.Registers.ed_tov.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + TL_MEM_FM_ED_TOV;
+
+
+ cpqfcHBAdata->fcChip.Registers.INTEN.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + IINTEN;
+ cpqfcHBAdata->fcChip.Registers.INTPEND.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + IINTPEND;
+ cpqfcHBAdata->fcChip.Registers.INTSTAT.address =
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase + IINTSTAT;
+
+ DEBUG_PCI(printk(" cpqfcHBAdata->fcChip.Registers. :\n"));
+ DEBUG_PCI(printk(" IOBaseL = %x\n",
+ cpqfcHBAdata->fcChip.Registers.IOBaseL));
+ DEBUG_PCI(printk(" IOBaseU = %x\n",
+ cpqfcHBAdata->fcChip.Registers.IOBaseU));
+
+ /* printk(" ioremap'd Membase: %p\n", cpqfcHBAdata->fcChip.Registers.ReMapMemBase); */
+
+ DEBUG_PCI(printk(" SFQconsumerIndex.address = %p\n",
+ cpqfcHBAdata->fcChip.Registers.SFQconsumerIndex.address));
+ DEBUG_PCI(printk(" ERQproducerIndex.address = %p\n",
+ cpqfcHBAdata->fcChip.Registers.ERQproducerIndex.address));
+ DEBUG_PCI(printk(" TYconfig.address = %p\n",
+ cpqfcHBAdata->fcChip.Registers.TYconfig.address));
+ DEBUG_PCI(printk(" FMconfig.address = %p\n",
+ cpqfcHBAdata->fcChip.Registers.FMconfig.address));
+ DEBUG_PCI(printk(" FMcontrol.address = %p\n",
+ cpqfcHBAdata->fcChip.Registers.FMcontrol.address));
+
+ // set default options for FC controller (chip)
+ cpqfcHBAdata->fcChip.Options.initiator = 1; // default: SCSI initiator
+ cpqfcHBAdata->fcChip.Options.target = 0; // default: SCSI target
+ cpqfcHBAdata->fcChip.Options.extLoopback = 0;// default: no loopback @GBIC
+ cpqfcHBAdata->fcChip.Options.intLoopback = 0;// default: no loopback inside chip
+
+ // set highest and lowest FC-PH version the adapter/driver supports
+ // (NOT strict compliance)
+ cpqfcHBAdata->fcChip.highest_FCPH_ver = FC_PH3;
+ cpqfcHBAdata->fcChip.lowest_FCPH_ver = FC_PH43;
+
+ // set function points for this controller / adapter
+ cpqfcHBAdata->fcChip.ResetTachyon = CpqTsResetTachLite;
+ cpqfcHBAdata->fcChip.FreezeTachyon = CpqTsFreezeTachlite;
+ cpqfcHBAdata->fcChip.UnFreezeTachyon = CpqTsUnFreezeTachlite;
+ cpqfcHBAdata->fcChip.CreateTachyonQues = CpqTsCreateTachLiteQues;
+ cpqfcHBAdata->fcChip.DestroyTachyonQues = CpqTsDestroyTachLiteQues;
+ cpqfcHBAdata->fcChip.InitializeTachyon = CpqTsInitializeTachLite;
+ cpqfcHBAdata->fcChip.LaserControl = CpqTsLaserControl;
+ cpqfcHBAdata->fcChip.ProcessIMQEntry = CpqTsProcessIMQEntry;
+ cpqfcHBAdata->fcChip.InitializeFrameManager = CpqTsInitializeFrameManager;
+ cpqfcHBAdata->fcChip.ReadWriteWWN = CpqTsReadWriteWWN;
+ cpqfcHBAdata->fcChip.ReadWriteNVRAM = CpqTsReadWriteNVRAM;
+
+ if (cpqfc_alloc_private_data_pool(cpqfcHBAdata) != 0) {
+ printk(KERN_WARNING
+ "cpqfc: unable to allocate pool for passthru ioctls. "
+ "Passthru ioctls disabled.\n");
+ }
+}
+
+
+/* (borrowed from linux/drivers/scsi/hosts.c) */
+static void launch_FCworker_thread(struct Scsi_Host *HostAdapter)
+{
+ DECLARE_MUTEX_LOCKED(sem);
+
+ CPQFCHBA *cpqfcHBAdata = (CPQFCHBA *)HostAdapter->hostdata;
+
+ ENTER("launch_FC_worker_thread");
+
+ cpqfcHBAdata->notify_wt = &sem;
+
+ /* must unlock before kernel_thread(), for it may cause a reschedule. */
+ spin_unlock_irq(HostAdapter->host_lock);
+ kernel_thread((int (*)(void *))cpqfcTSWorkerThread,
+ (void *) HostAdapter, 0);
+ /*
+ * Now wait for the kernel error thread to initialize itself
+
+ */
+ down (&sem);
+ spin_lock_irq(HostAdapter->host_lock);
+ cpqfcHBAdata->notify_wt = NULL;
+
+ LEAVE("launch_FC_worker_thread");
+
+}
+
+
+/* "Entry" point to discover if any supported PCI
+ bus adapter can be found
+*/
+/* We're supporting:
+ * Compaq 64-bit, 66MHz HBA with Tachyon TS
+ * Agilent XL2
+ * HP Tachyon
+ */
+#define HBA_TYPES 3
+
+#ifndef PCI_DEVICE_ID_COMPAQ_
+#define PCI_DEVICE_ID_COMPAQ_TACHYON 0xa0fc
+#endif
+
+static struct SupportedPCIcards cpqfc_boards[] __initdata = {
+ {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_TACHYON},
+ {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_TACHLITE},
+ {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_TACHYON},
+};
+
+
+int cpqfcTS_detect(Scsi_Host_Template *ScsiHostTemplate)
+{
+ int NumberOfAdapters=0; // how many of our PCI adapters are found?
+ struct pci_dev *PciDev = NULL;
+ struct Scsi_Host *HostAdapter = NULL;
+ CPQFCHBA *cpqfcHBAdata = NULL;
+ struct timer_list *cpqfcTStimer = NULL;
+ int i;
+
+ ENTER("cpqfcTS_detect");
+
+#if LINUX_VERSION_CODE < LinuxVersionCode(2,3,27)
+ ScsiHostTemplate->proc_dir = &proc_scsi_cpqfcTS;
+#else
+ ScsiHostTemplate->proc_name = "cpqfcTS";
+#endif
+
+ for( i=0; i < HBA_TYPES; i++)
+ {
+ // look for all HBAs of each type
+
+ while((PciDev = pci_find_device(cpqfc_boards[i].vendor_id,
+ cpqfc_boards[i].device_id, PciDev)))
+ {
+
+ if (pci_enable_device(PciDev)) {
+ printk(KERN_ERR
+ "cpqfc: can't enable PCI device at %s\n", pci_name(PciDev));
+ goto err_continue;
+ }
+
+ if (pci_set_dma_mask(PciDev, CPQFCTS_DMA_MASK) != 0) {
+ printk(KERN_WARNING
+ "cpqfc: HBA cannot support required DMA mask, skipping.\n");
+ goto err_disable_dev;
+ }
+
+ // NOTE: (kernel 2.2.12-32) limits allocation to 128k bytes...
+ /* printk(" scsi_register allocating %d bytes for FC HBA\n",
+ (ULONG)sizeof(CPQFCHBA)); */
+
+ HostAdapter = scsi_register( ScsiHostTemplate, sizeof( CPQFCHBA ) );
+
+ if(HostAdapter == NULL) {
+ printk(KERN_WARNING
+ "cpqfc: can't register SCSI HBA, skipping.\n");
+ goto err_disable_dev;
+ }
+ DEBUG_PCI( printk(" HBA found!\n"));
+ DEBUG_PCI( printk(" HostAdapter->PciDev->irq = %u\n", PciDev->irq) );
+ DEBUG_PCI(printk(" PciDev->baseaddress[0]= %lx\n",
+ PciDev->resource[0].start));
+ DEBUG_PCI(printk(" PciDev->baseaddress[1]= %lx\n",
+ PciDev->resource[1].start));
+ DEBUG_PCI(printk(" PciDev->baseaddress[2]= %lx\n",
+ PciDev->resource[2].start));
+ DEBUG_PCI(printk(" PciDev->baseaddress[3]= %lx\n",
+ PciDev->resource[3].start));
+
+ scsi_set_device(HostAdapter, &PciDev->dev);
+ HostAdapter->irq = PciDev->irq; // copy for Scsi layers
+
+ // HP Tachlite uses two (255-byte) ranges of Port I/O (lower & upper),
+ // for a total I/O port address space of 512 bytes.
+ // mask out the I/O port address (lower) & record
+ HostAdapter->io_port = (unsigned int)
+ PciDev->resource[1].start & PCI_BASE_ADDRESS_IO_MASK;
+ HostAdapter->n_io_port = 0xff;
+
+ // i.e., expect 128 targets (arbitrary number), while the
+ // RA-4000 supports 32 LUNs
+ HostAdapter->max_id = 0; // incremented as devices log in
+ HostAdapter->max_lun = CPQFCTS_MAX_LUN; // LUNs per FC device
+ HostAdapter->max_channel = CPQFCTS_MAX_CHANNEL; // multiple busses?
+
+ // get the pointer to our HBA specific data... (one for
+ // each HBA on the PCI bus(ses)).
+ cpqfcHBAdata = (CPQFCHBA *)HostAdapter->hostdata;
+
+ // make certain our data struct is clear
+ memset( cpqfcHBAdata, 0, sizeof( CPQFCHBA ) );
+
+
+ // initialize our HBA info
+ cpqfcHBAdata->HBAnum = NumberOfAdapters;
+
+ cpqfcHBAdata->HostAdapter = HostAdapter; // back ptr
+ Cpqfc_initHBAdata( cpqfcHBAdata, PciDev ); // fill MOST fields
+
+ cpqfcHBAdata->HBAnum = NumberOfAdapters;
+ spin_lock_init(&cpqfcHBAdata->hba_spinlock);
+
+ // request necessary resources and check for conflicts
+ if( request_irq( HostAdapter->irq,
+ cpqfcTS_intr_handler,
+ SA_INTERRUPT | SA_SHIRQ,
+ DEV_NAME,
+ HostAdapter) )
+ {
+ printk(KERN_WARNING "cpqfc: IRQ %u already used\n", HostAdapter->irq);
+ goto err_unregister;
+ }
+
+ // Since we have two 256-byte I/O port ranges (upper
+ // and lower), check them both
+ if( !request_region( cpqfcHBAdata->fcChip.Registers.IOBaseU,
+ 0xff, DEV_NAME ) )
+ {
+ printk(KERN_WARNING "cpqfc: address in use: %x\n",
+ cpqfcHBAdata->fcChip.Registers.IOBaseU);
+ goto err_free_irq;
+ }
+
+ if( !request_region( cpqfcHBAdata->fcChip.Registers.IOBaseL,
+ 0xff, DEV_NAME ) )
+ {
+ printk(KERN_WARNING "cpqfc: address in use: %x\n",
+ cpqfcHBAdata->fcChip.Registers.IOBaseL);
+ goto err_release_region_U;
+ }
+
+ // OK, we have grabbed everything we need now.
+ DEBUG_PCI(printk(" Reserved 255 I/O addresses @ %x\n",
+ cpqfcHBAdata->fcChip.Registers.IOBaseL ));
+ DEBUG_PCI(printk(" Reserved 255 I/O addresses @ %x\n",
+ cpqfcHBAdata->fcChip.Registers.IOBaseU ));
+
+
+
+ // start our kernel worker thread
+
+ spin_lock_irq(HostAdapter->host_lock);
+ launch_FCworker_thread(HostAdapter);
+
+
+ // start our TimerTask...
+
+ cpqfcTStimer = &cpqfcHBAdata->cpqfcTStimer;
+
+ init_timer( cpqfcTStimer); // Linux clears next/prev values
+ cpqfcTStimer->expires = jiffies + HZ; // one second
+ cpqfcTStimer->data = (unsigned long)cpqfcHBAdata; // this adapter
+ cpqfcTStimer->function = cpqfcTSheartbeat; // handles timeouts, housekeeping
+
+ add_timer( cpqfcTStimer); // give it to Linux
+
+
+ // now initialize our hardware...
+ if (cpqfcHBAdata->fcChip.InitializeTachyon( cpqfcHBAdata, 1,1)) {
+ printk(KERN_WARNING "cpqfc: initialization of HBA hardware failed.\n");
+ goto err_release_region_L;
+ }
+
+ cpqfcHBAdata->fcStatsTime = jiffies; // (for FC Statistics delta)
+
+ // give our HBA time to initialize and login current devices...
+ {
+ // The Brocade switch (e.g. 2400, 2010, etc.) as of March 2000,
+ // has the following algorithm for FL_Port startup:
+ // Time(sec) Action
+ // 0: Device Plugin and LIP(F7,F7) transmission
+ // 1.0 LIP incoming
+ // 1.027 LISA incoming, no CLS! (link not up)
+ // 1.028 NOS incoming (switch test for N_Port)
+ // 1.577 ED_TOV expired, transmit LIPs again
+ // 3.0 LIP(F8,F7) incoming (switch passes Tach Prim.Sig)
+ // 3.028 LILP received, link up, FLOGI starts
+ // slowest(worst) case, measured on 1Gb Finisar GT analyzer
+
+ unsigned long stop_time;
+
+ spin_unlock_irq(HostAdapter->host_lock);
+ stop_time = jiffies + 4*HZ;
+ while ( time_before(jiffies, stop_time) )
+ schedule(); // (our worker task needs to run)
+
+ }
+
+ spin_lock_irq(HostAdapter->host_lock);
+ NumberOfAdapters++;
+ spin_unlock_irq(HostAdapter->host_lock);
+
+ continue;
+
+err_release_region_L:
+ release_region( cpqfcHBAdata->fcChip.Registers.IOBaseL, 0xff );
+err_release_region_U:
+ release_region( cpqfcHBAdata->fcChip.Registers.IOBaseU, 0xff );
+err_free_irq:
+ free_irq( HostAdapter->irq, HostAdapter);
+err_unregister:
+ scsi_unregister( HostAdapter);
+err_disable_dev:
+ pci_disable_device( PciDev );
+err_continue:
+ continue;
+ } // end of while()
+ }
+
+ LEAVE("cpqfcTS_detect");
+
+ return NumberOfAdapters;
+}
+
+#ifdef SUPPORT_RESET
+static void my_ioctl_done (Scsi_Cmnd * SCpnt)
+{
+ struct request * req;
+
+ req = SCpnt->request;
+ req->rq_status = RQ_SCSI_DONE; /* Busy, but indicate request done */
+
+ if (req->CPQFC_WAITING != NULL)
+ CPQFC_COMPLETE(req->CPQFC_WAITING);
+}
+#endif
+
+static int cpqfc_alloc_private_data_pool(CPQFCHBA *hba)
+{
+ hba->private_data_bits = NULL;
+ hba->private_data_pool = NULL;
+ hba->private_data_bits =
+ kmalloc(((CPQFC_MAX_PASSTHRU_CMDS+BITS_PER_LONG-1) /
+ BITS_PER_LONG)*sizeof(unsigned long),
+ GFP_KERNEL);
+ if (hba->private_data_bits == NULL)
+ return -1;
+ memset(hba->private_data_bits, 0,
+ ((CPQFC_MAX_PASSTHRU_CMDS+BITS_PER_LONG-1) /
+ BITS_PER_LONG)*sizeof(unsigned long));
+ hba->private_data_pool = kmalloc(sizeof(cpqfc_passthru_private_t) *
+ CPQFC_MAX_PASSTHRU_CMDS, GFP_KERNEL);
+ if (hba->private_data_pool == NULL) {
+ kfree(hba->private_data_bits);
+ hba->private_data_bits = NULL;
+ return -1;
+ }
+ return 0;
+}
+
+static void cpqfc_free_private_data_pool(CPQFCHBA *hba)
+{
+ kfree(hba->private_data_bits);
+ kfree(hba->private_data_pool);
+}
+
+int is_private_data_of_cpqfc(CPQFCHBA *hba, void *pointer)
+{
+ /* Is pointer within our private data pool?
+ We use Scsi_Request->upper_private_data (normally
+ reserved for upper layer drivers, e.g. the sg driver)
+ We check to see if the pointer is ours by looking at
+ its address. Is this ok? Hmm, it occurs to me that
+ a user app might do something bad by using sg to send
+ a cpqfc passthrough ioctl with upper_data_private
+ forged to be somewhere in our pool..., though they'd
+ normally have to be root already to do this. */
+
+ return (pointer != NULL &&
+ pointer >= (void *) hba->private_data_pool &&
+ pointer < (void *) hba->private_data_pool +
+ sizeof(*hba->private_data_pool) *
+ CPQFC_MAX_PASSTHRU_CMDS);
+}
+
+cpqfc_passthru_private_t *cpqfc_alloc_private_data(CPQFCHBA *hba)
+{
+ int i;
+
+ do {
+ i = find_first_zero_bit(hba->private_data_bits,
+ CPQFC_MAX_PASSTHRU_CMDS);
+ if (i == CPQFC_MAX_PASSTHRU_CMDS)
+ return NULL;
+ } while ( test_and_set_bit(i & (BITS_PER_LONG - 1),
+ hba->private_data_bits+(i/BITS_PER_LONG)) != 0);
+ return &hba->private_data_pool[i];
+}
+
+void cpqfc_free_private_data(CPQFCHBA *hba, cpqfc_passthru_private_t *data)
+{
+ int i;
+ i = data - hba->private_data_pool;
+ clear_bit(i&(BITS_PER_LONG-1),
+ hba->private_data_bits+(i/BITS_PER_LONG));
+}
+
+int cpqfcTS_ioctl( struct scsi_device *ScsiDev, int Cmnd, void *arg)
+{
+ int result = 0;
+ struct Scsi_Host *HostAdapter = ScsiDev->host;
+ CPQFCHBA *cpqfcHBAdata = (CPQFCHBA *)HostAdapter->hostdata;
+ PTACHYON fcChip = &cpqfcHBAdata->fcChip;
+ PFC_LOGGEDIN_PORT pLoggedInPort = NULL;
+ struct scsi_cmnd *DumCmnd;
+ int i, j;
+ VENDOR_IOCTL_REQ ioc;
+ cpqfc_passthru_t *vendor_cmd;
+ Scsi_Device *SDpnt;
+ Scsi_Request *ScsiPassThruReq;
+ cpqfc_passthru_private_t *privatedata;
+
+ ENTER("cpqfcTS_ioctl ");
+
+ // printk("ioctl CMND %d", Cmnd);
+ switch (Cmnd) {
+ // Passthrough provides a mechanism to bypass the RAID
+ // or other controller and talk directly to the devices
+ // (e.g. physical disk drive)
+ // Passthrough commands, unfortunately, tend to be vendor
+ // specific; this is tailored to COMPAQ's RAID (RA4x00)
+ case CPQFCTS_SCSI_PASSTHRU:
+ {
+ void *buf = NULL; // for kernel space buffer for user data
+
+ /* Check that our pool got allocated ok. */
+ if (cpqfcHBAdata->private_data_pool == NULL)
+ return -ENOMEM;
+
+ if( !arg)
+ return -EINVAL;
+
+ // must be super user to send stuff directly to the
+ // controller and/or physical drives...
+ if( !capable(CAP_SYS_RAWIO) )
+ return -EPERM;
+
+ // copy the caller's struct to our space.
+ if( copy_from_user( &ioc, arg, sizeof( VENDOR_IOCTL_REQ)))
+ return( -EFAULT);
+
+ vendor_cmd = ioc.argp; // i.e., CPQ specific command struct
+
+ // If necessary, grab a kernel/DMA buffer
+ if( vendor_cmd->len)
+ {
+ buf = kmalloc( vendor_cmd->len, GFP_KERNEL);
+ if( !buf)
+ return -ENOMEM;
+ }
+ // Now build a Scsi_Request to pass down...
+ ScsiPassThruReq = scsi_allocate_request(ScsiDev, GFP_KERNEL);
+ if (ScsiPassThruReq == NULL) {
+ kfree(buf);
+ return -ENOMEM;
+ }
+ ScsiPassThruReq->upper_private_data =
+ cpqfc_alloc_private_data(cpqfcHBAdata);
+ if (ScsiPassThruReq->upper_private_data == NULL) {
+ kfree(buf);
+ scsi_release_request(ScsiPassThruReq); // "de-allocate"
+ return -ENOMEM;
+ }
+
+ if (vendor_cmd->rw_flag == VENDOR_WRITE_OPCODE) {
+ if (vendor_cmd->len) { // Need data from user?
+ if (copy_from_user(buf, vendor_cmd->bufp,
+ vendor_cmd->len)) {
+ kfree(buf);
+ cpqfc_free_private_data(cpqfcHBAdata,
+ ScsiPassThruReq->upper_private_data);
+ scsi_release_request(ScsiPassThruReq);
+ return( -EFAULT);
+ }
+ }
+ ScsiPassThruReq->sr_data_direction = SCSI_DATA_WRITE;
+ } else if (vendor_cmd->rw_flag == VENDOR_READ_OPCODE) {
+ ScsiPassThruReq->sr_data_direction = SCSI_DATA_READ;
+ } else
+ // maybe this means a bug in the user app
+ ScsiPassThruReq->sr_data_direction = SCSI_DATA_NONE;
+
+ ScsiPassThruReq->sr_cmd_len = 0; // set correctly by scsi_do_req()
+ ScsiPassThruReq->sr_sense_buffer[0] = 0;
+ ScsiPassThruReq->sr_sense_buffer[2] = 0;
+
+ // We copy the scheme used by sd.c:spinup_disk() to submit commands
+ // to our own HBA. We do this in order to stall the
+ // thread calling the IOCTL until it completes, and use
+ // the same "_quecommand" function for synchronizing
+ // FC Link events with our "worker thread".
+
+ privatedata = ScsiPassThruReq->upper_private_data;
+ privatedata->bus = vendor_cmd->bus;
+ privatedata->pdrive = vendor_cmd->pdrive;
+
+ // eventually gets us to our own _quecommand routine
+ scsi_wait_req(ScsiPassThruReq,
+ &vendor_cmd->cdb[0], buf, vendor_cmd->len,
+ 10*HZ, // timeout
+ 1); // retries
+ result = ScsiPassThruReq->sr_result;
+
+ // copy any sense data back to caller
+ if( result != 0 )
+ {
+ memcpy( vendor_cmd->sense_data, // see struct def - size=40
+ ScsiPassThruReq->sr_sense_buffer,
+ sizeof(ScsiPassThruReq->sr_sense_buffer) <
+ sizeof(vendor_cmd->sense_data) ?
+ sizeof(ScsiPassThruReq->sr_sense_buffer) :
+ sizeof(vendor_cmd->sense_data)
+ );
+ }
+ SDpnt = ScsiPassThruReq->sr_device;
+ /* upper_private_data is already freed in call_scsi_done() */
+ scsi_release_request(ScsiPassThruReq); // "de-allocate"
+ ScsiPassThruReq = NULL;
+
+ // need to pass data back to user (space)?
+ if( (vendor_cmd->rw_flag == VENDOR_READ_OPCODE) &&
+ vendor_cmd->len )
+ if( copy_to_user( vendor_cmd->bufp, buf, vendor_cmd->len))
+ result = -EFAULT;
+
+ if( buf)
+ kfree( buf);
+
+ return result;
+ }
+
+ case CPQFCTS_GETPCIINFO:
+ {
+ cpqfc_pci_info_struct pciinfo;
+
+ if( !arg)
+ return -EINVAL;
+
+
+
+ pciinfo.bus = cpqfcHBAdata->PciDev->bus->number;
+ pciinfo.dev_fn = cpqfcHBAdata->PciDev->devfn;
+ pciinfo.board_id = cpqfcHBAdata->PciDev->device |
+ (cpqfcHBAdata->PciDev->vendor <<16);
+
+ if(copy_to_user( arg, &pciinfo, sizeof(cpqfc_pci_info_struct)))
+ return( -EFAULT);
+ return 0;
+ }
+
+ case CPQFCTS_GETDRIVVER:
+ {
+ DriverVer_type DriverVer =
+ CPQFCTS_DRIVER_VER( VER_MAJOR,VER_MINOR,VER_SUBMINOR);
+
+ if( !arg)
+ return -EINVAL;
+
+ if(copy_to_user( arg, &DriverVer, sizeof(DriverVer)))
+ return( -EFAULT);
+ return 0;
+ }
+
+
+
+ case CPQFC_IOCTL_FC_TARGET_ADDRESS:
+ // can we find an FC device mapping to this SCSI target?
+/* DumCmnd.channel = ScsiDev->channel; */ // For searching
+/* DumCmnd.target = ScsiDev->id; */
+/* DumCmnd.lun = ScsiDev->lun; */
+
+ DumCmnd = scsi_get_command (ScsiDev, GFP_KERNEL);
+ if (!DumCmnd)
+ return -ENOMEM;
+
+ pLoggedInPort = fcFindLoggedInPort( fcChip,
+ DumCmnd, // search Scsi Nexus
+ 0, // DON'T search linked list for FC port id
+ NULL, // DON'T search linked list for FC WWN
+ NULL); // DON'T care about end of list
+ scsi_put_command (DumCmnd);
+ if (pLoggedInPort == NULL) {
+ result = -ENXIO;
+ break;
+ }
+ result = access_ok(VERIFY_WRITE, arg, sizeof(Scsi_FCTargAddress)) ? 0 : -EFAULT;
+ if (result) break;
+
+ put_user(pLoggedInPort->port_id,
+ &((Scsi_FCTargAddress *) arg)->host_port_id);
+
+ for( i=3,j=0; i>=0; i--) // copy the LOGIN port's WWN
+ put_user(pLoggedInPort->u.ucWWN[i],
+ &((Scsi_FCTargAddress *) arg)->host_wwn[j++]);
+ for( i=7; i>3; i--) // copy the LOGIN port's WWN
+ put_user(pLoggedInPort->u.ucWWN[i],
+ &((Scsi_FCTargAddress *) arg)->host_wwn[j++]);
+ break;
+
+
+ case CPQFC_IOCTL_FC_TDR:
+
+ result = cpqfcTS_TargetDeviceReset( ScsiDev, 0);
+
+ break;
+
+
+
+
+ default:
+ result = -EINVAL;
+ break;
+ }
+
+ LEAVE("cpqfcTS_ioctl");
+ return result;
+}
+
+
+/* "Release" the Host Bus Adapter...
+ disable interrupts, stop the HBA, release the interrupt,
+ and free all resources */
+
+int cpqfcTS_release(struct Scsi_Host *HostAdapter)
+{
+ CPQFCHBA *cpqfcHBAdata = (CPQFCHBA *)HostAdapter->hostdata;
+
+
+ ENTER("cpqfcTS_release");
+
+ DEBUG_PCI( printk(" cpqfcTS: delete timer...\n"));
+ del_timer( &cpqfcHBAdata->cpqfcTStimer);
+
+ // disable the hardware...
+ DEBUG_PCI( printk(" disable hardware, destroy queues, free mem\n"));
+ cpqfcHBAdata->fcChip.ResetTachyon( cpqfcHBAdata, CLEAR_FCPORTS);
+
+ // kill kernel thread
+ if( cpqfcHBAdata->worker_thread ) // (only if exists)
+ {
+ DECLARE_MUTEX_LOCKED(sem); // synchronize thread kill
+
+ cpqfcHBAdata->notify_wt = &sem;
+ DEBUG_PCI( printk(" killing kernel thread\n"));
+ send_sig( SIGKILL, cpqfcHBAdata->worker_thread, 1);
+ down( &sem);
+ cpqfcHBAdata->notify_wt = NULL;
+
+ }
+
+ cpqfc_free_private_data_pool(cpqfcHBAdata);
+ // free Linux resources
+ DEBUG_PCI( printk(" cpqfcTS: freeing resources...\n"));
+ free_irq( HostAdapter->irq, HostAdapter);
+ scsi_unregister( HostAdapter);
+ release_region( cpqfcHBAdata->fcChip.Registers.IOBaseL, 0xff);
+ release_region( cpqfcHBAdata->fcChip.Registers.IOBaseU, 0xff);
+ /* we get "vfree: bad address" executing this - need to investigate...
+ if( (void*)((unsigned long)cpqfcHBAdata->fcChip.Registers.MemBase) !=
+ cpqfcHBAdata->fcChip.Registers.ReMapMemBase)
+ vfree( cpqfcHBAdata->fcChip.Registers.ReMapMemBase);
+*/
+ pci_disable_device( cpqfcHBAdata->PciDev);
+
+ LEAVE("cpqfcTS_release");
+ return 0;
+}
+
+
+const char * cpqfcTS_info(struct Scsi_Host *HostAdapter)
+{
+ static char buf[300];
+ CPQFCHBA *cpqfcHBA;
+ int BusSpeed, BusWidth;
+
+ // get the pointer to our Scsi layer HBA buffer
+ cpqfcHBA = (CPQFCHBA *)HostAdapter->hostdata;
+
+ BusWidth = (cpqfcHBA->fcChip.Registers.PCIMCTR &0x4) > 0 ?
+ 64 : 32;
+
+ if( cpqfcHBA->fcChip.Registers.TYconfig.value & 0x80000000)
+ BusSpeed = 66;
+ else
+ BusSpeed = 33;
+
+ sprintf(buf,
+"%s: WWN %08X%08X\n on PCI bus %d device 0x%02x irq %d IObaseL 0x%x, MEMBASE 0x%x\nPCI bus width %d bits, bus speed %d MHz\nFCP-SCSI Driver v%d.%d.%d",
+ cpqfcHBA->fcChip.Name,
+ cpqfcHBA->fcChip.Registers.wwn_hi,
+ cpqfcHBA->fcChip.Registers.wwn_lo,
+ cpqfcHBA->PciDev->bus->number,
+ cpqfcHBA->PciDev->device,
+ HostAdapter->irq,
+ cpqfcHBA->fcChip.Registers.IOBaseL,
+ cpqfcHBA->fcChip.Registers.MemBase,
+ BusWidth,
+ BusSpeed,
+ VER_MAJOR, VER_MINOR, VER_SUBMINOR
+);
+
+
+ cpqfcTSDecodeGBICtype( &cpqfcHBA->fcChip, &buf[ strlen(buf)]);
+ cpqfcTSGetLPSM( &cpqfcHBA->fcChip, &buf[ strlen(buf)]);
+ return buf;
+}
+
+//
+// /proc/scsi support. The following routines allow us to do 'normal'
+// sprintf like calls to return the currently requested piece (buflenght
+// chars, starting at bufoffset) of the file. Although procfs allows for
+// a 1 Kb bytes overflow after te supplied buffer, I consider it bad
+// programming to use it to make programming a little simpler. This piece
+// of coding is borrowed from ncr53c8xx.c with some modifications
+//
+struct info_str
+{
+ char *buffer; // Pointer to output buffer
+ int buflength; // It's length
+ int bufoffset; // File offset corresponding with buf[0]
+ int buffillen; // Current filled length
+ int filpos; // Current file offset
+};
+
+static void copy_mem_info(struct info_str *info, char *data, int datalen)
+{
+
+ if (info->filpos < info->bufoffset) { // Current offset before buffer offset
+ if (info->filpos + datalen <= info->bufoffset) {
+ info->filpos += datalen; // Discard if completely before buffer
+ return;
+ } else { // Partial copy, set to begin
+ data += (info->bufoffset - info->filpos);
+ datalen -= (info->bufoffset - info->filpos);
+ info->filpos = info->bufoffset;
+ }
+ }
+
+ info->filpos += datalen; // Update current offset
+
+ if (info->buffillen == info->buflength) // Buffer full, discard
+ return;
+
+ if (info->buflength - info->buffillen < datalen) // Overflows buffer ?
+ datalen = info->buflength - info->buffillen;
+
+ memcpy(info->buffer + info->buffillen, data, datalen);
+ info->buffillen += datalen;
+}
+
+static int copy_info(struct info_str *info, char *fmt, ...)
+{
+ va_list args;
+ char buf[400];
+ int len;
+
+ va_start(args, fmt);
+ len = vsprintf(buf, fmt, args);
+ va_end(args);
+
+ copy_mem_info(info, buf, len);
+ return len;
+}
+
+
+// Routine to get data for /proc RAM filesystem
+//
+int cpqfcTS_proc_info (struct Scsi_Host *host, char *buffer, char **start, off_t offset, int length,
+ int inout)
+{
+ struct scsi_cmnd *DumCmnd;
+ struct scsi_device *ScsiDev;
+ int Chan, Targ, i;
+ struct info_str info;
+ CPQFCHBA *cpqfcHBA;
+ PTACHYON fcChip;
+ PFC_LOGGEDIN_PORT pLoggedInPort;
+ char buf[81];
+
+ if (inout) return -EINVAL;
+
+ // get the pointer to our Scsi layer HBA buffer
+ cpqfcHBA = (CPQFCHBA *)host->hostdata;
+ fcChip = &cpqfcHBA->fcChip;
+
+ *start = buffer;
+
+ info.buffer = buffer;
+ info.buflength = length;
+ info.bufoffset = offset;
+ info.filpos = 0;
+ info.buffillen = 0;
+ copy_info(&info, "Driver version = %d.%d.%d", VER_MAJOR, VER_MINOR, VER_SUBMINOR);
+ cpqfcTSDecodeGBICtype( &cpqfcHBA->fcChip, &buf[0]);
+ cpqfcTSGetLPSM( &cpqfcHBA->fcChip, &buf[ strlen(buf)]);
+ copy_info(&info, "%s\n", buf);
+
+#define DISPLAY_WWN_INFO
+#ifdef DISPLAY_WWN_INFO
+ ScsiDev = scsi_get_host_dev (host);
+ if (!ScsiDev)
+ return -ENOMEM;
+ DumCmnd = scsi_get_command (ScsiDev, GFP_KERNEL);
+ if (!DumCmnd) {
+ scsi_free_host_dev (ScsiDev);
+ return -ENOMEM;
+ }
+ copy_info(&info, "WWN database: (\"port_id: 000000\" means disconnected)\n");
+ for ( Chan=0; Chan <= host->max_channel; Chan++) {
+ DumCmnd->device->channel = Chan;
+ for (Targ=0; Targ <= host->max_id; Targ++) {
+ DumCmnd->device->id = Targ;
+ if ((pLoggedInPort = fcFindLoggedInPort( fcChip,
+ DumCmnd, // search Scsi Nexus
+ 0, // DON'T search list for FC port id
+ NULL, // DON'T search list for FC WWN
+ NULL))){ // DON'T care about end of list
+ copy_info(&info, "Host: scsi%d Channel: %02d TargetId: %02d -> WWN: ",
+ host->host_no, Chan, Targ);
+ for( i=3; i>=0; i--) // copy the LOGIN port's WWN
+ copy_info(&info, "%02X", pLoggedInPort->u.ucWWN[i]);
+ for( i=7; i>3; i--) // copy the LOGIN port's WWN
+ copy_info(&info, "%02X", pLoggedInPort->u.ucWWN[i]);
+ copy_info(&info, " port_id: %06X\n", pLoggedInPort->port_id);
+ }
+ }
+ }
+
+ scsi_put_command (DumCmnd);
+ scsi_free_host_dev (ScsiDev);
+#endif
+
+
+
+
+
+// Unfortunately, the proc_info buffer isn't big enough
+// for everything we would like...
+// For FC stats, compile this and turn off WWN stuff above
+//#define DISPLAY_FC_STATS
+#ifdef DISPLAY_FC_STATS
+// get the Fibre Channel statistics
+ {
+ int DeltaSecs = (jiffies - cpqfcHBA->fcStatsTime) / HZ;
+ int days,hours,minutes,secs;
+
+ days = DeltaSecs / (3600*24); // days
+ hours = (DeltaSecs% (3600*24)) / 3600; // hours
+ minutes = (DeltaSecs%3600 /60); // minutes
+ secs = DeltaSecs%60; // secs
+copy_info( &info, "Fibre Channel Stats (time dd:hh:mm:ss %02u:%02u:%02u:%02u\n",
+ days, hours, minutes, secs);
+ }
+
+ cpqfcHBA->fcStatsTime = jiffies; // (for next delta)
+
+ copy_info( &info, " LinkUp %9u LinkDown %u\n",
+ fcChip->fcStats.linkUp, fcChip->fcStats.linkDown);
+
+ copy_info( &info, " Loss of Signal %9u Loss of Sync %u\n",
+ fcChip->fcStats.LossofSignal, fcChip->fcStats.LossofSync);
+
+ copy_info( &info, " Discarded Frames %9u Bad CRC Frame %u\n",
+ fcChip->fcStats.Dis_Frm, fcChip->fcStats.Bad_CRC);
+
+ copy_info( &info, " TACH LinkFailTX %9u TACH LinkFailRX %u\n",
+ fcChip->fcStats.linkFailTX, fcChip->fcStats.linkFailRX);
+
+ copy_info( &info, " TACH RxEOFa %9u TACH Elastic Store %u\n",
+ fcChip->fcStats.Rx_EOFa, fcChip->fcStats.e_stores);
+
+ copy_info( &info, " BufferCreditWait %9uus TACH FM Inits %u\n",
+ fcChip->fcStats.BB0_Timer*10, fcChip->fcStats.FMinits );
+
+ copy_info( &info, " FC-2 Timeouts %9u FC-2 Logouts %u\n",
+ fcChip->fcStats.timeouts, fcChip->fcStats.logouts);
+
+ copy_info( &info, " FC-2 Aborts %9u FC-4 Aborts %u\n",
+ fcChip->fcStats.FC2aborted, fcChip->fcStats.FC4aborted);
+
+ // clear the counters
+ cpqfcTSClearLinkStatusCounters( fcChip);
+#endif
+
+ return info.buffillen;
+}
+
+
+#if DEBUG_CMND
+
+UCHAR *ScsiToAscii( UCHAR ScsiCommand)
+{
+
+/*++
+
+Routine Description:
+
+ Converts a SCSI command to a text string for debugging purposes.
+
+
+Arguments:
+
+ ScsiCommand -- hex value SCSI Command
+
+
+Return Value:
+
+ An ASCII, null-terminated string if found, else returns NULL.
+
+Original code from M. McGowen, Compaq
+--*/
+
+
+ switch (ScsiCommand)
+ {
+ case 0x00:
+ return( "Test Unit Ready" );
+
+ case 0x01:
+ return( "Rezero Unit or Rewind" );
+
+ case 0x02:
+ return( "Request Block Address" );
+
+ case 0x03:
+ return( "Requese Sense" );
+
+ case 0x04:
+ return( "Format Unit" );
+
+ case 0x05:
+ return( "Read Block Limits" );
+
+ case 0x07:
+ return( "Reassign Blocks" );
+
+ case 0x08:
+ return( "Read (6)" );
+
+ case 0x0a:
+ return( "Write (6)" );
+
+ case 0x0b:
+ return( "Seek (6)" );
+
+ case 0x12:
+ return( "Inquiry" );
+
+ case 0x15:
+ return( "Mode Select (6)" );
+
+ case 0x16:
+ return( "Reserve" );
+
+ case 0x17:
+ return( "Release" );
+
+ case 0x1a:
+ return( "ModeSen(6)" );
+
+ case 0x1b:
+ return( "Start/Stop Unit" );
+
+ case 0x1c:
+ return( "Receive Diagnostic Results" );
+
+ case 0x1d:
+ return( "Send Diagnostic" );
+
+ case 0x25:
+ return( "Read Capacity" );
+
+ case 0x28:
+ return( "Read (10)" );
+
+ case 0x2a:
+ return( "Write (10)" );
+
+ case 0x2b:
+ return( "Seek (10)" );
+
+ case 0x2e:
+ return( "Write and Verify" );
+
+ case 0x2f:
+ return( "Verify" );
+
+ case 0x34:
+ return( "Pre-Fetch" );
+
+ case 0x35:
+ return( "Synchronize Cache" );
+
+ case 0x37:
+ return( "Read Defect Data (10)" );
+
+ case 0x3b:
+ return( "Write Buffer" );
+
+ case 0x3c:
+ return( "Read Buffer" );
+
+ case 0x3e:
+ return( "Read Long" );
+
+ case 0x3f:
+ return( "Write Long" );
+
+ case 0x41:
+ return( "Write Same" );
+
+ case 0x4c:
+ return( "Log Select" );
+
+ case 0x4d:
+ return( "Log Sense" );
+
+ case 0x56:
+ return( "Reserve (10)" );
+
+ case 0x57:
+ return( "Release (10)" );
+
+ case 0xa0:
+ return( "ReportLuns" );
+
+ case 0xb7:
+ return( "Read Defect Data (12)" );
+
+ case 0xca:
+ return( "Peripheral Device Addressing SCSI Passthrough" );
+
+ case 0xcb:
+ return( "Compaq Array Firmware Passthrough" );
+
+ default:
+ return( NULL );
+ }
+
+} // end ScsiToAscii()
+
+void cpqfcTS_print_scsi_cmd(Scsi_Cmnd * cmd)
+{
+
+printk("cpqfcTS: (%s) chnl 0x%02x, trgt = 0x%02x, lun = 0x%02x, cmd_len = 0x%02x\n",
+ ScsiToAscii( cmd->cmnd[0]), cmd->channel, cmd->target, cmd->lun, cmd->cmd_len);
+
+if( cmd->cmnd[0] == 0) // Test Unit Ready?
+{
+ int i;
+
+ printk("Cmnd->request_bufflen = 0x%X, ->use_sg = %d, ->bufflen = %d\n",
+ cmd->request_bufflen, cmd->use_sg, cmd->bufflen);
+ printk("Cmnd->request_buffer = %p, ->sglist_len = %d, ->buffer = %p\n",
+ cmd->request_buffer, cmd->sglist_len, cmd->buffer);
+ for (i = 0; i < cmd->cmd_len; i++)
+ printk("0x%02x ", cmd->cmnd[i]);
+ printk("\n");
+}
+
+}
+
+#endif /* DEBUG_CMND */
+
+
+
+
+static void QueCmndOnBoardLock( CPQFCHBA *cpqfcHBAdata, Scsi_Cmnd *Cmnd)
+{
+ int i;
+
+ for( i=0; i< CPQFCTS_REQ_QUEUE_LEN; i++)
+ { // find spare slot
+ if( cpqfcHBAdata->BoardLockCmnd[i] == NULL )
+ {
+ cpqfcHBAdata->BoardLockCmnd[i] = Cmnd;
+// printk(" BoardLockCmnd[%d] %p Queued, chnl/target/lun %d/%d/%d\n",
+// i,Cmnd, Cmnd->channel, Cmnd->target, Cmnd->lun);
+ break;
+ }
+ }
+ if( i >= CPQFCTS_REQ_QUEUE_LEN)
+ {
+ printk(" cpqfcTS WARNING: Lost Cmnd %p on BoardLock Q full!", Cmnd);
+ }
+
+}
+
+
+static void QueLinkDownCmnd( CPQFCHBA *cpqfcHBAdata, Scsi_Cmnd *Cmnd)
+{
+ int indx;
+
+ // Remember the command ptr so we can return; we'll complete when
+ // the device comes back, causing immediate retry
+ for( indx=0; indx < CPQFCTS_REQ_QUEUE_LEN; indx++)//, SCptr++)
+ {
+ if( cpqfcHBAdata->LinkDnCmnd[indx] == NULL ) // available?
+ {
+#ifdef DUMMYCMND_DBG
+ printk(" @add Cmnd %p to LnkDnCmnd[%d]@ ", Cmnd,indx);
+#endif
+ cpqfcHBAdata->LinkDnCmnd[indx] = Cmnd;
+ break;
+ }
+ }
+
+ if( indx >= CPQFCTS_REQ_QUEUE_LEN ) // no space for Cmnd??
+ {
+ // this will result in an _abort call later (with possible trouble)
+ printk("no buffer for LinkDnCmnd!! %p\n", Cmnd);
+ }
+}
+
+
+
+
+
+// The file <scsi/scsi_host.h> says not to call scsi_done from
+// inside _queuecommand, so we'll do it from the heartbeat timer
+// (clarification: Turns out it's ok to call scsi_done from queuecommand
+// for cases that don't go to the hardware like scsi cmds destined
+// for LUNs we know don't exist, so this code might be simplified...)
+
+static void QueBadTargetCmnd( CPQFCHBA *cpqfcHBAdata, Scsi_Cmnd *Cmnd)
+{
+ int i;
+ // printk(" can't find target %d\n", Cmnd->target);
+
+ for( i=0; i< CPQFCTS_MAX_TARGET_ID; i++)
+ { // find spare slot
+ if( cpqfcHBAdata->BadTargetCmnd[i] == NULL )
+ {
+ cpqfcHBAdata->BadTargetCmnd[i] = Cmnd;
+// printk(" BadTargetCmnd[%d] %p Queued, chnl/target/lun %d/%d/%d\n",
+// i,Cmnd, Cmnd->channel, Cmnd->target, Cmnd->lun);
+ break;
+ }
+ }
+}
+
+
+// This is the "main" entry point for Linux Scsi commands --
+// it all starts here.
+
+int cpqfcTS_queuecommand(Scsi_Cmnd *Cmnd, void (* done)(Scsi_Cmnd *))
+{
+ struct Scsi_Host *HostAdapter = Cmnd->device->host;
+ CPQFCHBA *cpqfcHBAdata = (CPQFCHBA *)HostAdapter->hostdata;
+ PTACHYON fcChip = &cpqfcHBAdata->fcChip;
+ TachFCHDR_GCMND fchs; // only use for FC destination id field
+ PFC_LOGGEDIN_PORT pLoggedInPort;
+ ULONG ulStatus, SESTtype;
+ LONG ExchangeID;
+
+
+
+
+ ENTER("cpqfcTS_queuecommand");
+
+ PCI_TRACEO( (ULONG)Cmnd, 0x98)
+
+
+ Cmnd->scsi_done = done;
+#ifdef DEBUG_CMND
+ cpqfcTS_print_scsi_cmd( Cmnd);
+#endif
+
+ // prevent board contention with kernel thread...
+
+ if( cpqfcHBAdata->BoardLock )
+ {
+// printk(" @BrdLck Hld@ ");
+ QueCmndOnBoardLock( cpqfcHBAdata, Cmnd);
+ }
+
+ else
+ {
+
+ // in the current system (2.2.12), this routine is called
+ // after spin_lock_irqsave(), so INTs are disabled. However,
+ // we might have something pending in the LinkQ, which
+ // might cause the WorkerTask to run. In case that
+ // happens, make sure we lock it out.
+
+
+
+ PCI_TRACE( 0x98)
+ CPQ_SPINLOCK_HBA( cpqfcHBAdata)
+ PCI_TRACE( 0x98)
+
+ // can we find an FC device mapping to this SCSI target?
+ pLoggedInPort = fcFindLoggedInPort( fcChip,
+ Cmnd, // search Scsi Nexus
+ 0, // DON'T search linked list for FC port id
+ NULL, // DON'T search linked list for FC WWN
+ NULL); // DON'T care about end of list
+
+ if( pLoggedInPort == NULL ) // not found!
+ {
+// printk(" @Q bad targ cmnd %p@ ", Cmnd);
+ QueBadTargetCmnd( cpqfcHBAdata, Cmnd);
+ }
+ else if (Cmnd->device->lun >= CPQFCTS_MAX_LUN)
+ {
+ printk(KERN_WARNING "cpqfc: Invalid LUN: %d\n", Cmnd->device->lun);
+ QueBadTargetCmnd( cpqfcHBAdata, Cmnd);
+ }
+
+ else // we know what FC device to send to...
+ {
+
+ // does this device support FCP target functions?
+ // (determined by PRLI field)
+
+ if( !(pLoggedInPort->fcp_info & TARGET_FUNCTION) )
+ {
+ printk(" Doesn't support TARGET functions port_id %Xh\n",
+ pLoggedInPort->port_id );
+ QueBadTargetCmnd( cpqfcHBAdata, Cmnd);
+ }
+
+ // In this case (previous login OK), the device is temporarily
+ // unavailable waiting for re-login, in which case we expect it
+ // to be back in between 25 - 500ms.
+ // If the FC port doesn't log back in within several seconds
+ // (i.e. implicit "logout"), or we get an explicit logout,
+ // we set "device_blocked" in Scsi_Device struct; in this
+ // case 30 seconds will elapse before Linux/Scsi sends another
+ // command to the device.
+ else if( pLoggedInPort->prli != TRUE )
+ {
+// printk("Device (Chnl/Target %d/%d) invalid PRLI, port_id %06lXh\n",
+// Cmnd->channel, Cmnd->target, pLoggedInPort->port_id);
+ QueLinkDownCmnd( cpqfcHBAdata, Cmnd);
+// Need to use "blocked" flag??
+// Cmnd->device->device_blocked = TRUE; // just let it timeout
+ }
+ else // device supports TARGET functions, and is logged in...
+ {
+ // (context of fchs is to "reply" to...)
+ fchs.s_id = pLoggedInPort->port_id; // destination FC address
+
+ // what is the data direction? For data TO the device,
+ // we need IWE (Intiator Write Entry). Otherwise, IRE.
+
+ if( Cmnd->cmnd[0] == WRITE_10 ||
+ Cmnd->cmnd[0] == WRITE_6 ||
+ Cmnd->cmnd[0] == WRITE_BUFFER ||
+ Cmnd->cmnd[0] == VENDOR_WRITE_OPCODE || // CPQ specific
+ Cmnd->cmnd[0] == MODE_SELECT )
+ {
+ SESTtype = SCSI_IWE; // data from HBA to Device
+ }
+ else
+ SESTtype = SCSI_IRE; // data from Device to HBA
+
+ ulStatus = cpqfcTSBuildExchange(
+ cpqfcHBAdata,
+ SESTtype, // e.g. Initiator Read Entry (IRE)
+ &fchs, // we are originator; only use d_id
+ Cmnd, // Linux SCSI command (with scatter/gather list)
+ &ExchangeID );// fcController->fcExchanges index, -1 if failed
+
+ if( !ulStatus ) // Exchange setup?
+
+ {
+ if( cpqfcHBAdata->BoardLock )
+ {
+ TriggerHBA( fcChip->Registers.ReMapMemBase, 0);
+ printk(" @bl! %d, xID %Xh@ ", current->pid, ExchangeID);
+ }
+
+ ulStatus = cpqfcTSStartExchange( cpqfcHBAdata, ExchangeID );
+ if( !ulStatus )
+ {
+ PCI_TRACEO( ExchangeID, 0xB8)
+ // submitted to Tach's Outbound Que (ERQ PI incremented)
+ // waited for completion for ELS type (Login frames issued
+ // synchronously)
+ }
+ else
+ // check reason for Exchange not being started - we might
+ // want to Queue and start later, or fail with error
+ {
+ printk("quecommand: cpqfcTSStartExchange failed: %Xh\n", ulStatus );
+ }
+ } // end good BuildExchange status
+
+ else // SEST table probably full -- why? hardware hang?
+ {
+ printk("quecommand: cpqfcTSBuildExchange faild: %Xh\n", ulStatus);
+ }
+ } // end can't do FCP-SCSI target functions
+ } // end can't find target (FC device)
+
+ CPQ_SPINUNLOCK_HBA( cpqfcHBAdata)
+ }
+
+ PCI_TRACEO( (ULONG)Cmnd, 0x9C)
+ LEAVE("cpqfcTS_queuecommand");
+ return 0;
+}
+
+
+// Entry point for upper Scsi layer intiated abort. Typically
+// this is called if the command (for hard disk) fails to complete
+// in 30 seconds. This driver intends to complete all disk commands
+// within Exchange ".timeOut" seconds (now 7) with target status, or
+// in case of ".timeOut" expiration, a DID_SOFT_ERROR which causes
+// immediate retry.
+// If any disk commands get the _abort call, except for the case that
+// the physical device was removed or unavailable due to hardware
+// errors, it should be considered a driver error and reported to
+// the author.
+
+int cpqfcTS_abort(Scsi_Cmnd *Cmnd)
+{
+// printk(" cpqfcTS_abort called?? \n");
+ return 0;
+}
+
+int cpqfcTS_eh_abort(Scsi_Cmnd *Cmnd)
+{
+
+ struct Scsi_Host *HostAdapter = Cmnd->device->host;
+ // get the pointer to our Scsi layer HBA buffer
+ CPQFCHBA *cpqfcHBAdata = (CPQFCHBA *)HostAdapter->hostdata;
+ PTACHYON fcChip = &cpqfcHBAdata->fcChip;
+ FC_EXCHANGES *Exchanges = fcChip->Exchanges;
+ int i;
+ ENTER("cpqfcTS_eh_abort");
+
+ Cmnd->result = DID_ABORT <<16; // assume we'll find it
+
+ printk(" @Linux _abort Scsi_Cmnd %p ", Cmnd);
+ // See if we can find a Cmnd pointer that matches...
+ // The most likely case is we accepted the command
+ // from Linux Scsi (e.g. ceated a SEST entry) and it
+ // got lost somehow. If we can't find any reference
+ // to the passed pointer, we can only presume it
+ // got completed as far as our driver is concerned.
+ // If we found it, we will try to abort it through
+ // common mechanism. If FC ABTS is successful (ACC)
+ // or is rejected (RJT) by target, we will call
+ // Scsi "done" quickly. Otherwise, the ABTS will timeout
+ // and we'll call "done" later.
+
+ // Search the SEST exchanges for a matching Cmnd ptr.
+ for( i=0; i< TACH_SEST_LEN; i++)
+ {
+ if( Exchanges->fcExchange[i].Cmnd == Cmnd )
+ {
+
+ // found it!
+ printk(" x_ID %Xh, type %Xh\n", i, Exchanges->fcExchange[i].type);
+
+ Exchanges->fcExchange[i].status = INITIATOR_ABORT; // seconds default
+ Exchanges->fcExchange[i].timeOut = 10; // seconds default (changed later)
+
+ // Since we need to immediately return the aborted Cmnd to Scsi
+ // upper layers, we can't make future reference to any of its
+ // fields (e.g the Nexus).
+
+ cpqfcTSPutLinkQue( cpqfcHBAdata, BLS_ABTS, &i);
+
+ break;
+ }
+ }
+
+ if( i >= TACH_SEST_LEN ) // didn't find Cmnd ptr in chip's SEST?
+ {
+ // now search our non-SEST buffers (i.e. Cmnd waiting to
+ // start on the HBA or waiting to complete with error for retry).
+
+ // first check BadTargetCmnd
+ for( i=0; i< CPQFCTS_MAX_TARGET_ID; i++)
+ {
+ if( cpqfcHBAdata->BadTargetCmnd[i] == Cmnd )
+ {
+ cpqfcHBAdata->BadTargetCmnd[i] = NULL;
+ printk("in BadTargetCmnd Q\n");
+ goto Done; // exit
+ }
+ }
+
+ // if not found above...
+
+ for( i=0; i < CPQFCTS_REQ_QUEUE_LEN; i++)
+ {
+ if( cpqfcHBAdata->LinkDnCmnd[i] == Cmnd )
+ {
+ cpqfcHBAdata->LinkDnCmnd[i] = NULL;
+ printk("in LinkDnCmnd Q\n");
+ goto Done;
+ }
+ }
+
+
+ for( i=0; i< CPQFCTS_REQ_QUEUE_LEN; i++)
+ { // find spare slot
+ if( cpqfcHBAdata->BoardLockCmnd[i] == Cmnd )
+ {
+ cpqfcHBAdata->BoardLockCmnd[i] = NULL;
+ printk("in BoardLockCmnd Q\n");
+ goto Done;
+ }
+ }
+
+ Cmnd->result = DID_ERROR <<16; // Hmmm...
+ printk("Not found! ");
+// panic("_abort");
+ }
+
+Done:
+
+// panic("_abort");
+ LEAVE("cpqfcTS_eh_abort");
+ return 0; // (see scsi.h)
+}
+
+
+// FCP-SCSI Target Device Reset
+// See dpANS Fibre Channel Protocol for SCSI
+// X3.269-199X revision 12, pg 25
+
+#ifdef SUPPORT_RESET
+
+int cpqfcTS_TargetDeviceReset( Scsi_Device *ScsiDev,
+ unsigned int reset_flags)
+{
+ int timeout = 10*HZ;
+ int retries = 1;
+ char scsi_cdb[12];
+ int result;
+ Scsi_Cmnd * SCpnt;
+ Scsi_Device * SDpnt;
+
+// FIXME, cpqfcTS_TargetDeviceReset needs to be fixed
+// similarly to how the passthrough ioctl was fixed
+// around the 2.5.30 kernel. Scsi_Cmnd replaced with
+// Scsi_Request, etc.
+// For now, so people don't fall into a hole...
+
+ // printk(" ENTERING cpqfcTS_TargetDeviceReset() - flag=%d \n",reset_flags);
+
+ if (ScsiDev->host->eh_active) return FAILED;
+
+ memset( scsi_cdb, 0, sizeof( scsi_cdb));
+
+ scsi_cdb[0] = RELEASE;
+
+ SCpnt = scsi_get_command(ScsiDev, GFP_KERNEL);
+ {
+ CPQFC_DECLARE_COMPLETION(wait);
+
+ SCpnt->SCp.buffers_residual = FCP_TARGET_RESET;
+
+ // FIXME: this would panic, SCpnt->request would be NULL.
+ SCpnt->request->CPQFC_WAITING = &wait;
+ scsi_do_cmd(SCpnt, scsi_cdb, NULL, 0, my_ioctl_done, timeout, retries);
+ CPQFC_WAIT_FOR_COMPLETION(&wait);
+ SCpnt->request->CPQFC_WAITING = NULL;
+ }
+
+
+ if(driver_byte(SCpnt->result) != 0)
+ switch(SCpnt->sense_buffer[2] & 0xf) {
+ case ILLEGAL_REQUEST:
+ if(cmd[0] == ALLOW_MEDIUM_REMOVAL) dev->lockable = 0;
+ else printk("SCSI device (ioctl) reports ILLEGAL REQUEST.\n");
+ break;
+ case NOT_READY: // This happens if there is no disc in drive
+ if(dev->removable && (cmd[0] != TEST_UNIT_READY)){
+ printk(KERN_INFO "Device not ready. Make sure there is a disc in the drive.\n");
+ break;
+ }
+ case UNIT_ATTENTION:
+ if (dev->removable){
+ dev->changed = 1;
+ SCpnt->result = 0; // This is no longer considered an error
+ // gag this error, VFS will log it anyway /axboe
+ // printk(KERN_INFO "Disc change detected.\n");
+ break;
+ };
+ default: // Fall through for non-removable media
+ printk("SCSI error: host %d id %d lun %d return code = %x\n",
+ dev->host->host_no,
+ dev->id,
+ dev->lun,
+ SCpnt->result);
+ printk("\tSense class %x, sense error %x, extended sense %x\n",
+ sense_class(SCpnt->sense_buffer[0]),
+ sense_error(SCpnt->sense_buffer[0]),
+ SCpnt->sense_buffer[2] & 0xf);
+
+ };
+ result = SCpnt->result;
+
+ SDpnt = SCpnt->device;
+ scsi_put_command(SCpnt);
+ SCpnt = NULL;
+
+ // printk(" LEAVING cpqfcTS_TargetDeviceReset() - return SUCCESS \n");
+ return SUCCESS;
+}
+
+#else
+int cpqfcTS_TargetDeviceReset( Scsi_Device *ScsiDev,
+ unsigned int reset_flags)
+{
+ return -ENOTSUPP;
+}
+
+#endif /* SUPPORT_RESET */
+
+int cpqfcTS_eh_device_reset(Scsi_Cmnd *Cmnd)
+{
+ int retval;
+ Scsi_Device *SDpnt = Cmnd->device;
+ // printk(" ENTERING cpqfcTS_eh_device_reset() \n");
+ spin_unlock_irq(Cmnd->device->host->host_lock);
+ retval = cpqfcTS_TargetDeviceReset( SDpnt, 0);
+ spin_lock_irq(Cmnd->device->host->host_lock);
+ return retval;
+}
+
+
+int cpqfcTS_reset(Scsi_Cmnd *Cmnd, unsigned int reset_flags)
+{
+
+ ENTER("cpqfcTS_reset");
+
+ LEAVE("cpqfcTS_reset");
+ return SCSI_RESET_ERROR; /* Bus Reset Not supported */
+}
+
+/* This function determines the bios parameters for a given
+ harddisk. These tend to be numbers that are made up by the
+ host adapter. Parameters:
+ size, device number, list (heads, sectors,cylinders).
+ (from hosts.h)
+*/
+
+int cpqfcTS_biosparam(struct scsi_device *sdev, struct block_device *n,
+ sector_t capacity, int ip[])
+{
+ int size = capacity;
+
+ ENTER("cpqfcTS_biosparam");
+ ip[0] = 64;
+ ip[1] = 32;
+ ip[2] = size >> 11;
+
+ if( ip[2] > 1024 )
+ {
+ ip[0] = 255;
+ ip[1] = 63;
+ ip[2] = size / (ip[0] * ip[1]);
+ }
+
+ LEAVE("cpqfcTS_biosparam");
+ return 0;
+}
+
+
+
+irqreturn_t cpqfcTS_intr_handler( int irq,
+ void *dev_id,
+ struct pt_regs *regs)
+{
+
+ unsigned long flags, InfLoopBrk=0;
+ struct Scsi_Host *HostAdapter = dev_id;
+ CPQFCHBA *cpqfcHBA = (CPQFCHBA *)HostAdapter->hostdata;
+ int MoreMessages = 1; // assume we have something to do
+ UCHAR IntPending;
+ int handled = 0;
+
+ ENTER("intr_handler");
+ spin_lock_irqsave( HostAdapter->host_lock, flags);
+ // is this our INT?
+ IntPending = readb( cpqfcHBA->fcChip.Registers.INTPEND.address);
+
+ // broken boards can generate messages forever, so
+ // prevent the infinite loop
+#define INFINITE_IMQ_BREAK 10000
+ if( IntPending )
+ {
+ handled = 1;
+ // mask our HBA interrupts until we handle it...
+ writeb( 0, cpqfcHBA->fcChip.Registers.INTEN.address);
+
+ if( IntPending & 0x4) // "INT" - Tach wrote to IMQ
+ {
+ while( (++InfLoopBrk < INFINITE_IMQ_BREAK) && (MoreMessages ==1) )
+ {
+ MoreMessages = CpqTsProcessIMQEntry( HostAdapter); // ret 0 when done
+ }
+ if( InfLoopBrk >= INFINITE_IMQ_BREAK )
+ {
+ printk("WARNING: Compaq FC adapter generating excessive INTs -REPLACE\n");
+ printk("or investigate alternate causes (e.g. physical FC layer)\n");
+ }
+
+ else // working normally - re-enable INTs and continue
+ writeb( 0x1F, cpqfcHBA->fcChip.Registers.INTEN.address);
+
+ } // (...ProcessIMQEntry() clears INT by writing IMQ consumer)
+ else // indications of errors or problems...
+ // these usually indicate critical system hardware problems.
+ {
+ if( IntPending & 0x10 )
+ printk(" cpqfcTS adapter external memory parity error detected\n");
+ if( IntPending & 0x8 )
+ printk(" cpqfcTS adapter PCI master address crossed 45-bit boundary\n");
+ if( IntPending & 0x2 )
+ printk(" cpqfcTS adapter DMA error detected\n");
+ if( IntPending & 0x1 ) {
+ UCHAR IntStat;
+ printk(" cpqfcTS adapter PCI error detected\n");
+ IntStat = readb( cpqfcHBA->fcChip.Registers.INTSTAT.address);
+ printk("cpqfc: ISR = 0x%02x\n", IntStat);
+ if (IntStat & 0x1) {
+ __u16 pcistat;
+ /* read the pci status register */
+ pci_read_config_word(cpqfcHBA->PciDev, 0x06, &pcistat);
+ printk("PCI status register is 0x%04x\n", pcistat);
+ if (pcistat & 0x8000) printk("Parity Error Detected.\n");
+ if (pcistat & 0x4000) printk("Signalled System Error\n");
+ if (pcistat & 0x2000) printk("Received Master Abort\n");
+ if (pcistat & 0x1000) printk("Received Target Abort\n");
+ if (pcistat & 0x0800) printk("Signalled Target Abort\n");
+ }
+ if (IntStat & 0x4) printk("(INT)\n");
+ if (IntStat & 0x8)
+ printk("CRS: PCI master address crossed 46 bit bouandary\n");
+ if (IntStat & 0x10) printk("MRE: external memory parity error.\n");
+ }
+ }
+ }
+ spin_unlock_irqrestore( HostAdapter->host_lock, flags);
+ LEAVE("intr_handler");
+ return IRQ_RETVAL(handled);
+}
+
+
+
+
+int cpqfcTSDecodeGBICtype( PTACHYON fcChip, char cErrorString[])
+{
+ // Verify GBIC type (if any) and correct Tachyon Port State Machine
+ // (GBIC) module definition is:
+ // GPIO1, GPIO0, GPIO4 for MD2, MD1, MD0. The input states appear
+ // to be inverted -- i.e., a setting of 111 is read when there is NO
+ // GBIC present. The Module Def (MD) spec says 000 is "no GBIC"
+ // Hard code the bit states to detect Copper,
+ // Long wave (single mode), Short wave (multi-mode), and absent GBIC
+
+ ULONG ulBuff;
+
+ sprintf( cErrorString, "\nGBIC detected: ");
+
+ ulBuff = fcChip->Registers.TYstatus.value & 0x13;
+ switch( ulBuff )
+ {
+ case 0x13: // GPIO4, GPIO1, GPIO0 = 111; no GBIC!
+ sprintf( &cErrorString[ strlen( cErrorString)],
+ "NONE! ");
+ return FALSE;
+
+
+ case 0x11: // Copper GBIC detected
+ sprintf( &cErrorString[ strlen( cErrorString)],
+ "Copper. ");
+ break;
+
+ case 0x10: // Long-wave (single mode) GBIC detected
+ sprintf( &cErrorString[ strlen( cErrorString)],
+ "Long-wave. ");
+ break;
+ case 0x1: // Short-wave (multi mode) GBIC detected
+ sprintf( &cErrorString[ strlen( cErrorString)],
+ "Short-wave. ");
+ break;
+ default: // unknown GBIC - presumably it will work (?)
+ sprintf( &cErrorString[ strlen( cErrorString)],
+ "Unknown. ");
+
+ break;
+ } // end switch GBIC detection
+
+ return TRUE;
+}
+
+
+
+
+
+
+int cpqfcTSGetLPSM( PTACHYON fcChip, char cErrorString[])
+{
+ // Tachyon's Frame Manager LPSM in LinkDown state?
+ // (For non-loop port, check PSM instead.)
+ // return string with state and FALSE is Link Down
+
+ int LinkUp;
+
+ if( fcChip->Registers.FMstatus.value & 0x80 )
+ LinkUp = FALSE;
+ else
+ LinkUp = TRUE;
+
+ sprintf( &cErrorString[ strlen( cErrorString)],
+ " LPSM %Xh ",
+ (fcChip->Registers.FMstatus.value >>4) & 0xf );
+
+
+ switch( fcChip->Registers.FMstatus.value & 0xF0)
+ {
+ // bits set in LPSM
+ case 0x10:
+ sprintf( &cErrorString[ strlen( cErrorString)], "ARB");
+ break;
+ case 0x20:
+ sprintf( &cErrorString[ strlen( cErrorString)], "ARBwon");
+ break;
+ case 0x30:
+ sprintf( &cErrorString[ strlen( cErrorString)], "OPEN");
+ break;
+ case 0x40:
+ sprintf( &cErrorString[ strlen( cErrorString)], "OPENed");
+ break;
+ case 0x50:
+ sprintf( &cErrorString[ strlen( cErrorString)], "XmitCLS");
+ break;
+ case 0x60:
+ sprintf( &cErrorString[ strlen( cErrorString)], "RxCLS");
+ break;
+ case 0x70:
+ sprintf( &cErrorString[ strlen( cErrorString)], "Xfer");
+ break;
+ case 0x80:
+ sprintf( &cErrorString[ strlen( cErrorString)], "Init");
+ break;
+ case 0x90:
+ sprintf( &cErrorString[ strlen( cErrorString)], "O-IInitFin");
+ break;
+ case 0xa0:
+ sprintf( &cErrorString[ strlen( cErrorString)], "O-IProtocol");
+ break;
+ case 0xb0:
+ sprintf( &cErrorString[ strlen( cErrorString)], "O-ILipRcvd");
+ break;
+ case 0xc0:
+ sprintf( &cErrorString[ strlen( cErrorString)], "HostControl");
+ break;
+ case 0xd0:
+ sprintf( &cErrorString[ strlen( cErrorString)], "LoopFail");
+ break;
+ case 0xe0:
+ sprintf( &cErrorString[ strlen( cErrorString)], "Offline");
+ break;
+ case 0xf0:
+ sprintf( &cErrorString[ strlen( cErrorString)], "OldPort");
+ break;
+ case 0:
+ default:
+ sprintf( &cErrorString[ strlen( cErrorString)], "Monitor");
+ break;
+
+ }
+
+ return LinkUp;
+}
+
+
+
+
+#include "linux/slab.h"
+
+// Dynamic memory allocation alignment routines
+// HP's Tachyon Fibre Channel Controller chips require
+// certain memory queues and register pointers to be aligned
+// on various boundaries, usually the size of the Queue in question.
+// Alignment might be on 2, 4, 8, ... or even 512 byte boundaries.
+// Since most O/Ss don't allow this (usually only Cache aligned -
+// 32-byte boundary), these routines provide generic alignment (after
+// O/S allocation) at any boundary, and store the original allocated
+// pointer for deletion (O/S free function). Typically, we expect
+// these functions to only be called at HBA initialization and
+// removal time (load and unload times)
+// ALGORITHM notes:
+// Memory allocation varies by compiler and platform. In the worst case,
+// we are only assured BYTE alignment, but in the best case, we can
+// request allocation on any desired boundary. Our strategy: pad the
+// allocation request size (i.e. waste memory) so that we are assured
+// of passing desired boundary near beginning of contiguous space, then
+// mask out lower address bits.
+// We define the following algorithm:
+// allocBoundary - compiler/platform specific address alignment
+// in number of bytes (default is single byte; i.e. 1)
+// n_alloc - number of bytes application wants @ aligned address
+// ab - alignment boundary, in bytes (e.g. 4, 32, ...)
+// t_alloc - total allocation needed to ensure desired boundary
+// mask - to clear least significant address bits for boundary
+// Compute:
+// t_alloc = n_alloc + (ab - allocBoundary)
+// allocate t_alloc bytes @ alloc_address
+// mask = NOT (ab - 1)
+// (e.g. if ab=32 _0001 1111 -> _1110 0000
+// aligned_address = alloc_address & mask
+// set n_alloc bytes to 0
+// return aligned_address (NULL if failed)
+//
+// If u32_AlignedAddress is non-zero, then search for BaseAddress (stored
+// from previous allocation). If found, invoke call to FREE the memory.
+// Return NULL if BaseAddress not found
+
+// we need about 8 allocations per HBA. Figuring at most 10 HBAs per server
+// size the dynamic_mem array at 80.
+
+void* fcMemManager( struct pci_dev *pdev, ALIGNED_MEM *dynamic_mem,
+ ULONG n_alloc, ULONG ab, ULONG u32_AlignedAddress,
+ dma_addr_t *dma_handle)
+{
+ USHORT allocBoundary=1; // compiler specific - worst case 1
+ // best case - replace malloc() call
+ // with function that allocates exactly
+ // at desired boundary
+
+ unsigned long ulAddress;
+ ULONG t_alloc, i;
+ void *alloc_address = 0; // def. error code / address not found
+ LONG mask; // must be 32-bits wide!
+
+ ENTER("fcMemManager");
+ if( u32_AlignedAddress ) // are we freeing existing memory?
+ {
+// printk(" freeing AlignedAddress %Xh\n", u32_AlignedAddress);
+ for( i=0; i<DYNAMIC_ALLOCATIONS; i++) // look for the base address
+ {
+// printk("dynamic_mem[%u].AlignedAddress %lX\n", i, dynamic_mem[i].AlignedAddress);
+ if( dynamic_mem[i].AlignedAddress == u32_AlignedAddress )
+ {
+ alloc_address = dynamic_mem[i].BaseAllocated; // 'success' status
+ pci_free_consistent(pdev,dynamic_mem[i].size,
+ alloc_address,
+ dynamic_mem[i].dma_handle);
+ dynamic_mem[i].BaseAllocated = 0; // clear for next use
+ dynamic_mem[i].AlignedAddress = 0;
+ dynamic_mem[i].size = 0;
+ break; // quit for loop; done
+ }
+ }
+ }
+ else if( n_alloc ) // want new memory?
+ {
+ dma_addr_t handle;
+ t_alloc = n_alloc + (ab - allocBoundary); // pad bytes for alignment
+// printk("pci_alloc_consistent() for Tach alignment: %ld bytes\n", t_alloc);
+
+// (would like to) allow thread block to free pages
+ alloc_address = // total bytes (NumberOfBytes)
+ pci_alloc_consistent(pdev, t_alloc, &handle);
+
+ // now mask off least sig. bits of address
+ if( alloc_address ) // (only if non-NULL)
+ {
+ // find place to store ptr, so we
+ // can free it later...
+
+ mask = (LONG)(ab - 1); // mask all low-order bits
+ mask = ~mask; // invert bits
+ for( i=0; i<DYNAMIC_ALLOCATIONS; i++) // look for free slot
+ {
+ if( dynamic_mem[i].BaseAllocated == 0) // take 1st available
+ {
+ dynamic_mem[i].BaseAllocated = alloc_address;// address from O/S
+ dynamic_mem[i].dma_handle = handle;
+ if (dma_handle != NULL)
+ {
+// printk("handle = %p, ab=%d, boundary = %d, mask=0x%08x\n",
+// handle, ab, allocBoundary, mask);
+ *dma_handle = (dma_addr_t)
+ ((((ULONG)handle) + (ab - allocBoundary)) & mask);
+ }
+ dynamic_mem[i].size = t_alloc;
+ break;
+ }
+ }
+ ulAddress = (unsigned long)alloc_address;
+
+ ulAddress += (ab - allocBoundary); // add the alignment bytes-
+ // then truncate address...
+ alloc_address = (void*)(ulAddress & mask);
+
+ dynamic_mem[i].AlignedAddress =
+ (ULONG)(ulAddress & mask); // 32bit Tach address
+ memset( alloc_address, 0, n_alloc ); // clear new memory
+ }
+ else // O/S dynamic mem alloc failed!
+ alloc_address = 0; // (for debugging breakpt)
+
+ }
+
+ LEAVE("fcMemManager");
+ return alloc_address; // good (or NULL) address
+}
+
+
+static Scsi_Host_Template driver_template = {
+ .detect = cpqfcTS_detect,
+ .release = cpqfcTS_release,
+ .info = cpqfcTS_info,
+ .proc_info = cpqfcTS_proc_info,
+ .ioctl = cpqfcTS_ioctl,
+ .queuecommand = cpqfcTS_queuecommand,
+ .eh_device_reset_handler = cpqfcTS_eh_device_reset,
+ .eh_abort_handler = cpqfcTS_eh_abort,
+ .bios_param = cpqfcTS_biosparam,
+ .can_queue = CPQFCTS_REQ_QUEUE_LEN,
+ .this_id = -1,
+ .sg_tablesize = SG_ALL,
+ .cmd_per_lun = CPQFCTS_CMD_PER_LUN,
+ .use_clustering = ENABLE_CLUSTERING,
+};
+#include "scsi_module.c"
+