diff options
author | Linus Torvalds <torvalds@ppc970.osdl.org> | 2005-04-16 15:20:36 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@ppc970.osdl.org> | 2005-04-16 15:20:36 -0700 |
commit | 1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 (patch) | |
tree | 0bba044c4ce775e45a88a51686b5d9f90697ea9d /drivers/scsi/ips.c | |
download | linux-3.10-1da177e4c3f41524e886b7f1b8a0c1fc7321cac2.tar.gz linux-3.10-1da177e4c3f41524e886b7f1b8a0c1fc7321cac2.tar.bz2 linux-3.10-1da177e4c3f41524e886b7f1b8a0c1fc7321cac2.zip |
Linux-2.6.12-rc2v2.6.12-rc2
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
Diffstat (limited to 'drivers/scsi/ips.c')
-rw-r--r-- | drivers/scsi/ips.c | 7491 |
1 files changed, 7491 insertions, 0 deletions
diff --git a/drivers/scsi/ips.c b/drivers/scsi/ips.c new file mode 100644 index 00000000000..e46096da8db --- /dev/null +++ b/drivers/scsi/ips.c @@ -0,0 +1,7491 @@ +/*****************************************************************************/ +/* ips.c -- driver for the Adaptec / IBM ServeRAID controller */ +/* */ +/* Written By: Keith Mitchell, IBM Corporation */ +/* Jack Hammer, Adaptec, Inc. */ +/* David Jeffery, Adaptec, Inc. */ +/* */ +/* Copyright (C) 2000 IBM Corporation */ +/* Copyright (C) 2002,2003 Adaptec, Inc. */ +/* */ +/* 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 of the License, 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. */ +/* */ +/* NO WARRANTY */ +/* THE PROGRAM IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OR */ +/* CONDITIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED INCLUDING, WITHOUT */ +/* LIMITATION, ANY WARRANTIES OR CONDITIONS OF TITLE, NON-INFRINGEMENT, */ +/* MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Each Recipient is */ +/* solely responsible for determining the appropriateness of using and */ +/* distributing the Program and assumes all risks associated with its */ +/* exercise of rights under this Agreement, including but not limited to */ +/* the risks and costs of program errors, damage to or loss of data, */ +/* programs or equipment, and unavailability or interruption of operations. */ +/* */ +/* DISCLAIMER OF LIABILITY */ +/* NEITHER RECIPIENT NOR ANY CONTRIBUTORS SHALL HAVE ANY LIABILITY FOR ANY */ +/* DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL */ +/* DAMAGES (INCLUDING WITHOUT LIMITATION LOST PROFITS), HOWEVER CAUSED AND */ +/* ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR */ +/* TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE */ +/* USE OR DISTRIBUTION OF THE PROGRAM OR THE EXERCISE OF ANY RIGHTS GRANTED */ +/* HEREUNDER, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGES */ +/* */ +/* You should have received a copy of the GNU General Public License */ +/* along with this program; if not, write to the Free Software */ +/* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ +/* */ +/* Bugs/Comments/Suggestions about this driver should be mailed to: */ +/* ipslinux@adaptec.com */ +/* */ +/* For system support issues, contact your local IBM Customer support. */ +/* Directions to find IBM Customer Support for each country can be found at: */ +/* http://www.ibm.com/planetwide/ */ +/* */ +/*****************************************************************************/ + +/*****************************************************************************/ +/* Change Log */ +/* */ +/* 0.99.02 - Breakup commands that are bigger than 8 * the stripe size */ +/* 0.99.03 - Make interrupt routine handle all completed request on the */ +/* adapter not just the first one */ +/* - Make sure passthru commands get woken up if we run out of */ +/* SCBs */ +/* - Send all of the commands on the queue at once rather than */ +/* one at a time since the card will support it. */ +/* 0.99.04 - Fix race condition in the passthru mechanism -- this required */ +/* the interface to the utilities to change */ +/* - Fix error recovery code */ +/* 0.99.05 - Fix an oops when we get certain passthru commands */ +/* 1.00.00 - Initial Public Release */ +/* Functionally equivalent to 0.99.05 */ +/* 3.60.00 - Bump max commands to 128 for use with firmware 3.60 */ +/* - Change version to 3.60 to coincide with release numbering. */ +/* 3.60.01 - Remove bogus error check in passthru routine */ +/* 3.60.02 - Make DCDB direction based on lookup table */ +/* - Only allow one DCDB command to a SCSI ID at a time */ +/* 4.00.00 - Add support for ServeRAID 4 */ +/* 4.00.01 - Add support for First Failure Data Capture */ +/* 4.00.02 - Fix problem with PT DCDB with no buffer */ +/* 4.00.03 - Add alternative passthru interface */ +/* - Add ability to flash BIOS */ +/* 4.00.04 - Rename structures/constants to be prefixed with IPS_ */ +/* 4.00.05 - Remove wish_block from init routine */ +/* - Use linux/spinlock.h instead of asm/spinlock.h for kernels */ +/* 2.3.18 and later */ +/* - Sync with other changes from the 2.3 kernels */ +/* 4.00.06 - Fix timeout with initial FFDC command */ +/* 4.00.06a - Port to 2.4 (trivial) -- Christoph Hellwig <hch@infradead.org> */ +/* 4.10.00 - Add support for ServeRAID 4M/4L */ +/* 4.10.13 - Fix for dynamic unload and proc file system */ +/* 4.20.03 - Rename version to coincide with new release schedules */ +/* Performance fixes */ +/* Fix truncation of /proc files with cat */ +/* Merge in changes through kernel 2.4.0test1ac21 */ +/* 4.20.13 - Fix some failure cases / reset code */ +/* - Hook into the reboot_notifier to flush the controller cache */ +/* 4.50.01 - Fix problem when there is a hole in logical drive numbering */ +/* 4.70.09 - Use a Common ( Large Buffer ) for Flashing from the JCRM CD */ +/* - Add IPSSEND Flash Support */ +/* - Set Sense Data for Unknown SCSI Command */ +/* - Use Slot Number from NVRAM Page 5 */ +/* - Restore caller's DCDB Structure */ +/* 4.70.12 - Corrective actions for bad controller ( during initialization )*/ +/* 4.70.13 - Don't Send CDB's if we already know the device is not present */ +/* - Don't release HA Lock in ips_next() until SC taken off queue */ +/* - Unregister SCSI device in ips_release() */ +/* 4.70.15 - Fix Breakup for very large ( non-SG ) requests in ips_done() */ +/* 4.71.00 - Change all memory allocations to not use GFP_DMA flag */ +/* Code Clean-Up for 2.4.x kernel */ +/* 4.72.00 - Allow for a Scatter-Gather Element to exceed MAX_XFER Size */ +/* 4.72.01 - I/O Mapped Memory release ( so "insmod ips" does not Fail ) */ +/* - Don't Issue Internal FFDC Command if there are Active Commands */ +/* - Close Window for getting too many IOCTL's active */ +/* 4.80.00 - Make ia64 Safe */ +/* 4.80.04 - Eliminate calls to strtok() if 2.4.x or greater */ +/* - Adjustments to Device Queue Depth */ +/* 4.80.14 - Take all semaphores off stack */ +/* - Clean Up New_IOCTL path */ +/* 4.80.20 - Set max_sectors in Scsi_Host structure ( if >= 2.4.7 kernel ) */ +/* - 5 second delay needed after resetting an i960 adapter */ +/* 4.80.26 - Clean up potential code problems ( Arjan's recommendations ) */ +/* 4.90.01 - Version Matching for FirmWare, BIOS, and Driver */ +/* 4.90.05 - Use New PCI Architecture to facilitate Hot Plug Development */ +/* 4.90.08 - Increase Delays in Flashing ( Trombone Only - 4H ) */ +/* 4.90.08 - Data Corruption if First Scatter Gather Element is > 64K */ +/* 4.90.11 - Don't actually RESET unless it's physically required */ +/* - Remove unused compile options */ +/* 5.00.01 - Sarasota ( 5i ) adapters must always be scanned first */ +/* - Get rid on IOCTL_NEW_COMMAND code */ +/* - Add Extended DCDB Commands for Tape Support in 5I */ +/* 5.10.12 - use pci_dma interfaces, update for 2.5 kernel changes */ +/* 5.10.15 - remove unused code (sem, macros, etc.) */ +/* 5.30.00 - use __devexit_p() */ +/* 6.00.00 - Add 6x Adapters and Battery Flash */ +/* 6.10.00 - Remove 1G Addressing Limitations */ +/* 6.11.xx - Get VersionInfo buffer off the stack ! DDTS 60401 */ +/* 6.11.xx - Make Logical Drive Info structure safe for DMA DDTS 60639 */ +/* 7.10.xx - Add highmem_io flag in SCSI Templete for 2.4 kernels */ +/* - Fix path/name for scsi_hosts.h include for 2.6 kernels */ +/* - Fix sort order of 7k */ +/* - Remove 3 unused "inline" functions */ +/*****************************************************************************/ + +/* + * Conditional Compilation directives for this driver: + * + * IPS_DEBUG - Turn on debugging info + * + * Parameters: + * + * debug:<number> - Set debug level to <number> + * NOTE: only works when IPS_DEBUG compile directive is used. + * 1 - Normal debug messages + * 2 - Verbose debug messages + * 11 - Method trace (non interrupt) + * 12 - Method trace (includes interrupt) + * + * noi2o - Don't use I2O Queues (ServeRAID 4 only) + * nommap - Don't use memory mapped I/O + * ioctlsize - Initial size of the IOCTL buffer + */ + +#include <asm/io.h> +#include <asm/byteorder.h> +#include <asm/page.h> +#include <linux/stddef.h> +#include <linux/version.h> +#include <linux/string.h> +#include <linux/errno.h> +#include <linux/kernel.h> +#include <linux/ioport.h> +#include <linux/slab.h> +#include <linux/delay.h> +#include <linux/pci.h> +#include <linux/proc_fs.h> +#include <linux/reboot.h> +#include <linux/interrupt.h> + +#include <linux/blkdev.h> +#include <linux/types.h> + +#include <scsi/sg.h> + +#include "scsi.h" + +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,5,0) +#include "hosts.h" +#else +#include <scsi/scsi_host.h> +#endif + +#include "ips.h" + +#include <linux/module.h> + +#include <linux/stat.h> +#include <linux/config.h> + +#include <linux/spinlock.h> +#include <linux/init.h> + +#include <linux/smp.h> + +#ifdef MODULE +static char *ips = NULL; +module_param(ips, charp, 0); +#endif + +/* + * DRIVER_VER + */ +#define IPS_VERSION_HIGH "7.10" +#define IPS_VERSION_LOW ".18 " + +#if !defined(__i386__) && !defined(__ia64__) && !defined(__x86_64__) +#warning "This driver has only been tested on the x86/ia64/x86_64 platforms" +#endif + +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,5,0) +#include <linux/blk.h> +#include "sd.h" +#define IPS_SG_ADDRESS(sg) ((sg)->address) +#define IPS_LOCK_SAVE(lock,flags) spin_lock_irqsave(&io_request_lock,flags) +#define IPS_UNLOCK_RESTORE(lock,flags) spin_unlock_irqrestore(&io_request_lock,flags) +#ifndef __devexit_p +#define __devexit_p(x) x +#endif +#else +#define IPS_SG_ADDRESS(sg) (page_address((sg)->page) ? \ + page_address((sg)->page)+(sg)->offset : NULL) +#define IPS_LOCK_SAVE(lock,flags) do{spin_lock(lock);(void)flags;}while(0) +#define IPS_UNLOCK_RESTORE(lock,flags) do{spin_unlock(lock);(void)flags;}while(0) +#endif + +#define IPS_DMA_DIR(scb) ((!scb->scsi_cmd || ips_is_passthru(scb->scsi_cmd) || \ + SCSI_DATA_NONE == scb->scsi_cmd->sc_data_direction) ? \ + PCI_DMA_BIDIRECTIONAL : \ + scsi_to_pci_dma_dir(scb->scsi_cmd->sc_data_direction)) + +#ifdef IPS_DEBUG +#define METHOD_TRACE(s, i) if (ips_debug >= (i+10)) printk(KERN_NOTICE s "\n"); +#define DEBUG(i, s) if (ips_debug >= i) printk(KERN_NOTICE s "\n"); +#define DEBUG_VAR(i, s, v...) if (ips_debug >= i) printk(KERN_NOTICE s "\n", v); +#else +#define METHOD_TRACE(s, i) +#define DEBUG(i, s) +#define DEBUG_VAR(i, s, v...) +#endif + +/* + * Function prototypes + */ +static int ips_detect(Scsi_Host_Template *); +static int ips_release(struct Scsi_Host *); +static int ips_eh_abort(Scsi_Cmnd *); +static int ips_eh_reset(Scsi_Cmnd *); +static int ips_queue(Scsi_Cmnd *, void (*)(Scsi_Cmnd *)); +static const char *ips_info(struct Scsi_Host *); +static irqreturn_t do_ipsintr(int, void *, struct pt_regs *); +static int ips_hainit(ips_ha_t *); +static int ips_map_status(ips_ha_t *, ips_scb_t *, ips_stat_t *); +static int ips_send_wait(ips_ha_t *, ips_scb_t *, int, int); +static int ips_send_cmd(ips_ha_t *, ips_scb_t *); +static int ips_online(ips_ha_t *, ips_scb_t *); +static int ips_inquiry(ips_ha_t *, ips_scb_t *); +static int ips_rdcap(ips_ha_t *, ips_scb_t *); +static int ips_msense(ips_ha_t *, ips_scb_t *); +static int ips_reqsen(ips_ha_t *, ips_scb_t *); +static int ips_deallocatescbs(ips_ha_t *, int); +static int ips_allocatescbs(ips_ha_t *); +static int ips_reset_copperhead(ips_ha_t *); +static int ips_reset_copperhead_memio(ips_ha_t *); +static int ips_reset_morpheus(ips_ha_t *); +static int ips_issue_copperhead(ips_ha_t *, ips_scb_t *); +static int ips_issue_copperhead_memio(ips_ha_t *, ips_scb_t *); +static int ips_issue_i2o(ips_ha_t *, ips_scb_t *); +static int ips_issue_i2o_memio(ips_ha_t *, ips_scb_t *); +static int ips_isintr_copperhead(ips_ha_t *); +static int ips_isintr_copperhead_memio(ips_ha_t *); +static int ips_isintr_morpheus(ips_ha_t *); +static int ips_wait(ips_ha_t *, int, int); +static int ips_write_driver_status(ips_ha_t *, int); +static int ips_read_adapter_status(ips_ha_t *, int); +static int ips_read_subsystem_parameters(ips_ha_t *, int); +static int ips_read_config(ips_ha_t *, int); +static int ips_clear_adapter(ips_ha_t *, int); +static int ips_readwrite_page5(ips_ha_t *, int, int); +static int ips_init_copperhead(ips_ha_t *); +static int ips_init_copperhead_memio(ips_ha_t *); +static int ips_init_morpheus(ips_ha_t *); +static int ips_isinit_copperhead(ips_ha_t *); +static int ips_isinit_copperhead_memio(ips_ha_t *); +static int ips_isinit_morpheus(ips_ha_t *); +static int ips_erase_bios(ips_ha_t *); +static int ips_program_bios(ips_ha_t *, char *, uint32_t, uint32_t); +static int ips_verify_bios(ips_ha_t *, char *, uint32_t, uint32_t); +static int ips_erase_bios_memio(ips_ha_t *); +static int ips_program_bios_memio(ips_ha_t *, char *, uint32_t, uint32_t); +static int ips_verify_bios_memio(ips_ha_t *, char *, uint32_t, uint32_t); +static int ips_flash_copperhead(ips_ha_t *, ips_passthru_t *, ips_scb_t *); +static int ips_flash_bios(ips_ha_t *, ips_passthru_t *, ips_scb_t *); +static int ips_flash_firmware(ips_ha_t *, ips_passthru_t *, ips_scb_t *); +static void ips_free_flash_copperhead(ips_ha_t * ha); +static void ips_get_bios_version(ips_ha_t *, int); +static void ips_identify_controller(ips_ha_t *); +static void ips_chkstatus(ips_ha_t *, IPS_STATUS *); +static void ips_enable_int_copperhead(ips_ha_t *); +static void ips_enable_int_copperhead_memio(ips_ha_t *); +static void ips_enable_int_morpheus(ips_ha_t *); +static int ips_intr_copperhead(ips_ha_t *); +static int ips_intr_morpheus(ips_ha_t *); +static void ips_next(ips_ha_t *, int); +static void ipsintr_blocking(ips_ha_t *, struct ips_scb *); +static void ipsintr_done(ips_ha_t *, struct ips_scb *); +static void ips_done(ips_ha_t *, ips_scb_t *); +static void ips_free(ips_ha_t *); +static void ips_init_scb(ips_ha_t *, ips_scb_t *); +static void ips_freescb(ips_ha_t *, ips_scb_t *); +static void ips_setup_funclist(ips_ha_t *); +static void ips_statinit(ips_ha_t *); +static void ips_statinit_memio(ips_ha_t *); +static void ips_fix_ffdc_time(ips_ha_t *, ips_scb_t *, time_t); +static void ips_ffdc_reset(ips_ha_t *, int); +static void ips_ffdc_time(ips_ha_t *); +static uint32_t ips_statupd_copperhead(ips_ha_t *); +static uint32_t ips_statupd_copperhead_memio(ips_ha_t *); +static uint32_t ips_statupd_morpheus(ips_ha_t *); +static ips_scb_t *ips_getscb(ips_ha_t *); +static void ips_putq_scb_head(ips_scb_queue_t *, ips_scb_t *); +static void ips_putq_wait_tail(ips_wait_queue_t *, Scsi_Cmnd *); +static void ips_putq_copp_tail(ips_copp_queue_t *, + ips_copp_wait_item_t *); +static ips_scb_t *ips_removeq_scb_head(ips_scb_queue_t *); +static ips_scb_t *ips_removeq_scb(ips_scb_queue_t *, ips_scb_t *); +static Scsi_Cmnd *ips_removeq_wait_head(ips_wait_queue_t *); +static Scsi_Cmnd *ips_removeq_wait(ips_wait_queue_t *, Scsi_Cmnd *); +static ips_copp_wait_item_t *ips_removeq_copp(ips_copp_queue_t *, + ips_copp_wait_item_t *); +static ips_copp_wait_item_t *ips_removeq_copp_head(ips_copp_queue_t *); + +static int ips_is_passthru(Scsi_Cmnd *); +static int ips_make_passthru(ips_ha_t *, Scsi_Cmnd *, ips_scb_t *, int); +static int ips_usrcmd(ips_ha_t *, ips_passthru_t *, ips_scb_t *); +static void ips_cleanup_passthru(ips_ha_t *, ips_scb_t *); +static void ips_scmd_buf_write(Scsi_Cmnd * scmd, void *data, + unsigned int count); +static void ips_scmd_buf_read(Scsi_Cmnd * scmd, void *data, unsigned int count); + +static int ips_proc_info(struct Scsi_Host *, char *, char **, off_t, int, int); +static int ips_host_info(ips_ha_t *, char *, off_t, int); +static void copy_mem_info(IPS_INFOSTR *, char *, int); +static int copy_info(IPS_INFOSTR *, char *, ...); +static int ips_get_version_info(ips_ha_t * ha, dma_addr_t, int intr); +static void ips_version_check(ips_ha_t * ha, int intr); +static int ips_abort_init(ips_ha_t * ha, int index); +static int ips_init_phase2(int index); + +static int ips_init_phase1(struct pci_dev *pci_dev, int *indexPtr); +static int ips_register_scsi(int index); + +/* + * global variables + */ +static const char ips_name[] = "ips"; +static struct Scsi_Host *ips_sh[IPS_MAX_ADAPTERS]; /* Array of host controller structures */ +static ips_ha_t *ips_ha[IPS_MAX_ADAPTERS]; /* Array of HA structures */ +static unsigned int ips_next_controller; +static unsigned int ips_num_controllers; +static unsigned int ips_released_controllers; +static int ips_hotplug; +static int ips_cmd_timeout = 60; +static int ips_reset_timeout = 60 * 5; +static int ips_force_memio = 1; /* Always use Memory Mapped I/O */ +static int ips_force_i2o = 1; /* Always use I2O command delivery */ +static int ips_ioctlsize = IPS_IOCTL_SIZE; /* Size of the ioctl buffer */ +static int ips_cd_boot; /* Booting from Manager CD */ +static char *ips_FlashData = NULL; /* CD Boot - Flash Data Buffer */ +static dma_addr_t ips_flashbusaddr; +static long ips_FlashDataInUse; /* CD Boot - Flash Data In Use Flag */ +static uint32_t MaxLiteCmds = 32; /* Max Active Cmds for a Lite Adapter */ +static Scsi_Host_Template ips_driver_template = { + .detect = ips_detect, + .release = ips_release, + .info = ips_info, + .queuecommand = ips_queue, + .eh_abort_handler = ips_eh_abort, + .eh_host_reset_handler = ips_eh_reset, + .proc_name = "ips", +#if LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0) + .proc_info = ips_proc_info, + .slave_configure = ips_slave_configure, +#else + .proc_info = ips_proc24_info, + .select_queue_depths = ips_select_queue_depth, +#endif + .bios_param = ips_biosparam, + .this_id = -1, + .sg_tablesize = IPS_MAX_SG, + .cmd_per_lun = 3, + .use_clustering = ENABLE_CLUSTERING, +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0) + .use_new_eh_code = 1, +#endif +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,20) && LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0) + .highmem_io = 1, +#endif +}; + +static IPS_DEFINE_COMPAT_TABLE( Compatable ); /* Version Compatability Table */ + + +/* This table describes all ServeRAID Adapters */ +static struct pci_device_id ips_pci_table[] = { + { 0x1014, 0x002E, PCI_ANY_ID, PCI_ANY_ID, 0, 0 }, + { 0x1014, 0x01BD, PCI_ANY_ID, PCI_ANY_ID, 0, 0 }, + { 0x9005, 0x0250, PCI_ANY_ID, PCI_ANY_ID, 0, 0 }, + { 0, } +}; + +MODULE_DEVICE_TABLE( pci, ips_pci_table ); + +static char ips_hot_plug_name[] = "ips"; + +static int __devinit ips_insert_device(struct pci_dev *pci_dev, const struct pci_device_id *ent); +static void __devexit ips_remove_device(struct pci_dev *pci_dev); + +static struct pci_driver ips_pci_driver = { + .name = ips_hot_plug_name, + .id_table = ips_pci_table, + .probe = ips_insert_device, + .remove = __devexit_p(ips_remove_device), +}; + + +/* + * Necessary forward function protoypes + */ +static int ips_halt(struct notifier_block *nb, ulong event, void *buf); + +#define MAX_ADAPTER_NAME 15 + +static char ips_adapter_name[][30] = { + "ServeRAID", + "ServeRAID II", + "ServeRAID on motherboard", + "ServeRAID on motherboard", + "ServeRAID 3H", + "ServeRAID 3L", + "ServeRAID 4H", + "ServeRAID 4M", + "ServeRAID 4L", + "ServeRAID 4Mx", + "ServeRAID 4Lx", + "ServeRAID 5i", + "ServeRAID 5i", + "ServeRAID 6M", + "ServeRAID 6i", + "ServeRAID 7t", + "ServeRAID 7k", + "ServeRAID 7M" +}; + +static struct notifier_block ips_notifier = { + ips_halt, NULL, 0 +}; + +/* + * Direction table + */ +static char ips_command_direction[] = { + IPS_DATA_NONE, IPS_DATA_NONE, IPS_DATA_IN, IPS_DATA_IN, IPS_DATA_OUT, + IPS_DATA_IN, IPS_DATA_IN, IPS_DATA_OUT, IPS_DATA_IN, IPS_DATA_UNK, + IPS_DATA_OUT, IPS_DATA_OUT, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_IN, IPS_DATA_NONE, IPS_DATA_NONE, IPS_DATA_IN, IPS_DATA_OUT, + IPS_DATA_IN, IPS_DATA_OUT, IPS_DATA_NONE, IPS_DATA_NONE, IPS_DATA_OUT, + IPS_DATA_NONE, IPS_DATA_IN, IPS_DATA_NONE, IPS_DATA_IN, IPS_DATA_OUT, + IPS_DATA_NONE, IPS_DATA_UNK, IPS_DATA_IN, IPS_DATA_UNK, IPS_DATA_IN, + IPS_DATA_UNK, IPS_DATA_OUT, IPS_DATA_IN, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_IN, IPS_DATA_IN, IPS_DATA_OUT, IPS_DATA_NONE, IPS_DATA_UNK, + IPS_DATA_IN, IPS_DATA_OUT, IPS_DATA_OUT, IPS_DATA_OUT, IPS_DATA_OUT, + IPS_DATA_OUT, IPS_DATA_NONE, IPS_DATA_IN, IPS_DATA_NONE, IPS_DATA_NONE, + IPS_DATA_IN, IPS_DATA_OUT, IPS_DATA_OUT, IPS_DATA_OUT, IPS_DATA_OUT, + IPS_DATA_IN, IPS_DATA_OUT, IPS_DATA_IN, IPS_DATA_OUT, IPS_DATA_OUT, + IPS_DATA_OUT, IPS_DATA_IN, IPS_DATA_IN, IPS_DATA_IN, IPS_DATA_NONE, + IPS_DATA_UNK, IPS_DATA_NONE, IPS_DATA_NONE, IPS_DATA_NONE, IPS_DATA_UNK, + IPS_DATA_NONE, IPS_DATA_OUT, IPS_DATA_IN, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_OUT, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_IN, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_NONE, IPS_DATA_NONE, IPS_DATA_UNK, IPS_DATA_IN, IPS_DATA_NONE, + IPS_DATA_OUT, IPS_DATA_UNK, IPS_DATA_NONE, IPS_DATA_UNK, IPS_DATA_OUT, + IPS_DATA_OUT, IPS_DATA_OUT, IPS_DATA_OUT, IPS_DATA_OUT, IPS_DATA_NONE, + IPS_DATA_UNK, IPS_DATA_IN, IPS_DATA_OUT, IPS_DATA_IN, IPS_DATA_IN, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_OUT, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, + IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK, IPS_DATA_UNK +}; + + +/****************************************************************************/ +/* */ +/* Routine Name: ips_setup */ +/* */ +/* Routine Description: */ +/* */ +/* setup parameters to the driver */ +/* */ +/****************************************************************************/ +static int +ips_setup(char *ips_str) +{ + + int i; + char *key; + char *value; + IPS_OPTION options[] = { + {"noi2o", &ips_force_i2o, 0}, + {"nommap", &ips_force_memio, 0}, + {"ioctlsize", &ips_ioctlsize, IPS_IOCTL_SIZE}, + {"cdboot", &ips_cd_boot, 0}, + {"maxcmds", &MaxLiteCmds, 32}, + }; + + /* Don't use strtok() anymore ( if 2.4 Kernel or beyond ) */ + /* Search for value */ + while ((key = strsep(&ips_str, ",."))) { + if (!*key) + continue; + value = strchr(key, ':'); + if (value) + *value++ = '\0'; + /* + * We now have key/value pairs. + * Update the variables + */ + for (i = 0; i < (sizeof (options) / sizeof (options[0])); i++) { + if (strnicmp + (key, options[i].option_name, + strlen(options[i].option_name)) == 0) { + if (value) + *options[i].option_flag = + simple_strtoul(value, NULL, 0); + else + *options[i].option_flag = + options[i].option_value; + break; + } + } + } + + return (1); +} + +__setup("ips=", ips_setup); + +/****************************************************************************/ +/* */ +/* Routine Name: ips_detect */ +/* */ +/* Routine Description: */ +/* */ +/* Detect and initialize the driver */ +/* */ +/* NOTE: this routine is called under the io_request_lock spinlock */ +/* */ +/****************************************************************************/ +static int +ips_detect(Scsi_Host_Template * SHT) +{ + int i; + + METHOD_TRACE("ips_detect", 1); + +#ifdef MODULE + if (ips) + ips_setup(ips); +#endif + + for (i = 0; i < ips_num_controllers; i++) { + if (ips_register_scsi(i)) + ips_free(ips_ha[i]); + ips_released_controllers++; + } + ips_hotplug = 1; + return (ips_num_controllers); +} + +/****************************************************************************/ +/* configure the function pointers to use the functions that will work */ +/* with the found version of the adapter */ +/****************************************************************************/ +static void +ips_setup_funclist(ips_ha_t * ha) +{ + + /* + * Setup Functions + */ + if (IPS_IS_MORPHEUS(ha) || IPS_IS_MARCO(ha)) { + /* morpheus / marco / sebring */ + ha->func.isintr = ips_isintr_morpheus; + ha->func.isinit = ips_isinit_morpheus; + ha->func.issue = ips_issue_i2o_memio; + ha->func.init = ips_init_morpheus; + ha->func.statupd = ips_statupd_morpheus; + ha->func.reset = ips_reset_morpheus; + ha->func.intr = ips_intr_morpheus; + ha->func.enableint = ips_enable_int_morpheus; + } else if (IPS_USE_MEMIO(ha)) { + /* copperhead w/MEMIO */ + ha->func.isintr = ips_isintr_copperhead_memio; + ha->func.isinit = ips_isinit_copperhead_memio; + ha->func.init = ips_init_copperhead_memio; + ha->func.statupd = ips_statupd_copperhead_memio; + ha->func.statinit = ips_statinit_memio; + ha->func.reset = ips_reset_copperhead_memio; + ha->func.intr = ips_intr_copperhead; + ha->func.erasebios = ips_erase_bios_memio; + ha->func.programbios = ips_program_bios_memio; + ha->func.verifybios = ips_verify_bios_memio; + ha->func.enableint = ips_enable_int_copperhead_memio; + if (IPS_USE_I2O_DELIVER(ha)) + ha->func.issue = ips_issue_i2o_memio; + else + ha->func.issue = ips_issue_copperhead_memio; + } else { + /* copperhead */ + ha->func.isintr = ips_isintr_copperhead; + ha->func.isinit = ips_isinit_copperhead; + ha->func.init = ips_init_copperhead; + ha->func.statupd = ips_statupd_copperhead; + ha->func.statinit = ips_statinit; + ha->func.reset = ips_reset_copperhead; + ha->func.intr = ips_intr_copperhead; + ha->func.erasebios = ips_erase_bios; + ha->func.programbios = ips_program_bios; + ha->func.verifybios = ips_verify_bios; + ha->func.enableint = ips_enable_int_copperhead; + + if (IPS_USE_I2O_DELIVER(ha)) + ha->func.issue = ips_issue_i2o; + else + ha->func.issue = ips_issue_copperhead; + } +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_release */ +/* */ +/* Routine Description: */ +/* */ +/* Remove a driver */ +/* */ +/****************************************************************************/ +static int +ips_release(struct Scsi_Host *sh) +{ + ips_scb_t *scb; + ips_ha_t *ha; + int i; + + METHOD_TRACE("ips_release", 1); + + for (i = 0; i < IPS_MAX_ADAPTERS && ips_sh[i] != sh; i++) ; + + if (i == IPS_MAX_ADAPTERS) { + printk(KERN_WARNING + "(%s) release, invalid Scsi_Host pointer.\n", ips_name); + BUG(); + return (FALSE); + } + + ha = IPS_HA(sh); + + if (!ha) + return (FALSE); + + /* flush the cache on the controller */ + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_FLUSH; + + scb->cmd.flush_cache.op_code = IPS_CMD_FLUSH; + scb->cmd.flush_cache.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.flush_cache.state = IPS_NORM_STATE; + scb->cmd.flush_cache.reserved = 0; + scb->cmd.flush_cache.reserved2 = 0; + scb->cmd.flush_cache.reserved3 = 0; + scb->cmd.flush_cache.reserved4 = 0; + + IPS_PRINTK(KERN_WARNING, ha->pcidev, "Flushing Cache.\n"); + + /* send command */ + if (ips_send_wait(ha, scb, ips_cmd_timeout, IPS_INTR_ON) == IPS_FAILURE) + IPS_PRINTK(KERN_WARNING, ha->pcidev, "Incomplete Flush.\n"); + + IPS_PRINTK(KERN_WARNING, ha->pcidev, "Flushing Complete.\n"); + + ips_sh[i] = NULL; + ips_ha[i] = NULL; + + /* free extra memory */ + ips_free(ha); + + /* Free I/O Region */ + if (ha->io_addr) + release_region(ha->io_addr, ha->io_len); + + /* free IRQ */ + free_irq(ha->irq, ha); + + IPS_REMOVE_HOST(sh); + scsi_host_put(sh); + + ips_released_controllers++; + + return (FALSE); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_halt */ +/* */ +/* Routine Description: */ +/* */ +/* Perform cleanup when the system reboots */ +/* */ +/****************************************************************************/ +static int +ips_halt(struct notifier_block *nb, ulong event, void *buf) +{ + ips_scb_t *scb; + ips_ha_t *ha; + int i; + + if ((event != SYS_RESTART) && (event != SYS_HALT) && + (event != SYS_POWER_OFF)) + return (NOTIFY_DONE); + + for (i = 0; i < ips_next_controller; i++) { + ha = (ips_ha_t *) ips_ha[i]; + + if (!ha) + continue; + + if (!ha->active) + continue; + + /* flush the cache on the controller */ + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_FLUSH; + + scb->cmd.flush_cache.op_code = IPS_CMD_FLUSH; + scb->cmd.flush_cache.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.flush_cache.state = IPS_NORM_STATE; + scb->cmd.flush_cache.reserved = 0; + scb->cmd.flush_cache.reserved2 = 0; + scb->cmd.flush_cache.reserved3 = 0; + scb->cmd.flush_cache.reserved4 = 0; + + IPS_PRINTK(KERN_WARNING, ha->pcidev, "Flushing Cache.\n"); + + /* send command */ + if (ips_send_wait(ha, scb, ips_cmd_timeout, IPS_INTR_ON) == + IPS_FAILURE) + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Incomplete Flush.\n"); + else + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Flushing Complete.\n"); + } + + return (NOTIFY_OK); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_eh_abort */ +/* */ +/* Routine Description: */ +/* */ +/* Abort a command (using the new error code stuff) */ +/* Note: this routine is called under the io_request_lock */ +/****************************************************************************/ +int +ips_eh_abort(Scsi_Cmnd * SC) +{ + ips_ha_t *ha; + ips_copp_wait_item_t *item; + int ret; + + METHOD_TRACE("ips_eh_abort", 1); + + if (!SC) + return (FAILED); + + ha = (ips_ha_t *) SC->device->host->hostdata; + + if (!ha) + return (FAILED); + + if (!ha->active) + return (FAILED); + + if (SC->serial_number != SC->serial_number_at_timeout) { + /* HMM, looks like a bogus command */ + DEBUG(1, "Abort called with bogus scsi command"); + + return (FAILED); + } + + /* See if the command is on the copp queue */ + item = ha->copp_waitlist.head; + while ((item) && (item->scsi_cmd != SC)) + item = item->next; + + if (item) { + /* Found it */ + ips_removeq_copp(&ha->copp_waitlist, item); + ret = (SUCCESS); + + /* See if the command is on the wait queue */ + } else if (ips_removeq_wait(&ha->scb_waitlist, SC)) { + /* command not sent yet */ + ret = (SUCCESS); + } else { + /* command must have already been sent */ + ret = (FAILED); + } + return ret; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_eh_reset */ +/* */ +/* Routine Description: */ +/* */ +/* Reset the controller (with new eh error code) */ +/* */ +/* NOTE: this routine is called under the io_request_lock spinlock */ +/* */ +/****************************************************************************/ +static int +ips_eh_reset(Scsi_Cmnd * SC) +{ + int ret; + int i; + ips_ha_t *ha; + ips_scb_t *scb; + ips_copp_wait_item_t *item; + + METHOD_TRACE("ips_eh_reset", 1); + +#ifdef NO_IPS_RESET + return (FAILED); +#else + + if (!SC) { + DEBUG(1, "Reset called with NULL scsi command"); + + return (FAILED); + } + + ha = (ips_ha_t *) SC->device->host->hostdata; + + if (!ha) { + DEBUG(1, "Reset called with NULL ha struct"); + + return (FAILED); + } + + if (!ha->active) + return (FAILED); + + /* See if the command is on the copp queue */ + item = ha->copp_waitlist.head; + while ((item) && (item->scsi_cmd != SC)) + item = item->next; + + if (item) { + /* Found it */ + ips_removeq_copp(&ha->copp_waitlist, item); + return (SUCCESS); + } + + /* See if the command is on the wait queue */ + if (ips_removeq_wait(&ha->scb_waitlist, SC)) { + /* command not sent yet */ + return (SUCCESS); + } + + /* An explanation for the casual observer: */ + /* Part of the function of a RAID controller is automatic error */ + /* detection and recovery. As such, the only problem that physically */ + /* resetting an adapter will ever fix is when, for some reason, */ + /* the driver is not successfully communicating with the adapter. */ + /* Therefore, we will attempt to flush this adapter. If that succeeds, */ + /* then there's no real purpose in a physical reset. This will complete */ + /* much faster and avoids any problems that might be caused by a */ + /* physical reset ( such as having to fail all the outstanding I/O's ). */ + + if (ha->ioctl_reset == 0) { /* IF Not an IOCTL Requested Reset */ + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_FLUSH; + + scb->cmd.flush_cache.op_code = IPS_CMD_FLUSH; + scb->cmd.flush_cache.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.flush_cache.state = IPS_NORM_STATE; + scb->cmd.flush_cache.reserved = 0; + scb->cmd.flush_cache.reserved2 = 0; + scb->cmd.flush_cache.reserved3 = 0; + scb->cmd.flush_cache.reserved4 = 0; + + /* Attempt the flush command */ + ret = ips_send_wait(ha, scb, ips_cmd_timeout, IPS_INTR_IORL); + if (ret == IPS_SUCCESS) { + IPS_PRINTK(KERN_NOTICE, ha->pcidev, + "Reset Request - Flushed Cache\n"); + return (SUCCESS); + } + } + + /* Either we can't communicate with the adapter or it's an IOCTL request */ + /* from a utility. A physical reset is needed at this point. */ + + ha->ioctl_reset = 0; /* Reset the IOCTL Requested Reset Flag */ + + /* + * command must have already been sent + * reset the controller + */ + IPS_PRINTK(KERN_NOTICE, ha->pcidev, "Resetting controller.\n"); + ret = (*ha->func.reset) (ha); + + if (!ret) { + Scsi_Cmnd *scsi_cmd; + + IPS_PRINTK(KERN_NOTICE, ha->pcidev, + "Controller reset failed - controller now offline.\n"); + + /* Now fail all of the active commands */ + DEBUG_VAR(1, "(%s%d) Failing active commands", + ips_name, ha->host_num); + + while ((scb = ips_removeq_scb_head(&ha->scb_activelist))) { + scb->scsi_cmd->result = DID_ERROR << 16; + scb->scsi_cmd->scsi_done(scb->scsi_cmd); + ips_freescb(ha, scb); + } + + /* Now fail all of the pending commands */ + DEBUG_VAR(1, "(%s%d) Failing pending commands", + ips_name, ha->host_num); + + while ((scsi_cmd = ips_removeq_wait_head(&ha->scb_waitlist))) { + scsi_cmd->result = DID_ERROR; + scsi_cmd->scsi_done(scsi_cmd); + } + + ha->active = FALSE; + return (FAILED); + } + + if (!ips_clear_adapter(ha, IPS_INTR_IORL)) { + Scsi_Cmnd *scsi_cmd; + + IPS_PRINTK(KERN_NOTICE, ha->pcidev, + "Controller reset failed - controller now offline.\n"); + + /* Now fail all of the active commands */ + DEBUG_VAR(1, "(%s%d) Failing active commands", + ips_name, ha->host_num); + + while ((scb = ips_removeq_scb_head(&ha->scb_activelist))) { + scb->scsi_cmd->result = DID_ERROR << 16; + scb->scsi_cmd->scsi_done(scb->scsi_cmd); + ips_freescb(ha, scb); + } + + /* Now fail all of the pending commands */ + DEBUG_VAR(1, "(%s%d) Failing pending commands", + ips_name, ha->host_num); + + while ((scsi_cmd = ips_removeq_wait_head(&ha->scb_waitlist))) { + scsi_cmd->result = DID_ERROR << 16; + scsi_cmd->scsi_done(scsi_cmd); + } + + ha->active = FALSE; + return (FAILED); + } + + /* FFDC */ + if (le32_to_cpu(ha->subsys->param[3]) & 0x300000) { + struct timeval tv; + + do_gettimeofday(&tv); + ha->last_ffdc = tv.tv_sec; + ha->reset_count++; + ips_ffdc_reset(ha, IPS_INTR_IORL); + } + + /* Now fail all of the active commands */ + DEBUG_VAR(1, "(%s%d) Failing active commands", ips_name, ha->host_num); + + while ((scb = ips_removeq_scb_head(&ha->scb_activelist))) { + scb->scsi_cmd->result = + (DID_RESET << 16) | (SUGGEST_RETRY << 24); + scb->scsi_cmd->scsi_done(scb->scsi_cmd); + ips_freescb(ha, scb); + } + + /* Reset DCDB active command bits */ + for (i = 1; i < ha->nbus; i++) + ha->dcdb_active[i - 1] = 0; + + /* Reset the number of active IOCTLs */ + ha->num_ioctl = 0; + + ips_next(ha, IPS_INTR_IORL); + + return (SUCCESS); +#endif /* NO_IPS_RESET */ + +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_queue */ +/* */ +/* Routine Description: */ +/* */ +/* Send a command to the controller */ +/* */ +/* NOTE: */ +/* Linux obtains io_request_lock before calling this function */ +/* */ +/****************************************************************************/ +static int +ips_queue(Scsi_Cmnd * SC, void (*done) (Scsi_Cmnd *)) +{ + ips_ha_t *ha; + ips_passthru_t *pt; + + METHOD_TRACE("ips_queue", 1); + + ha = (ips_ha_t *) SC->device->host->hostdata; + + if (!ha) + return (1); + + if (!ha->active) + return (DID_ERROR); + + if (ips_is_passthru(SC)) { + if (ha->copp_waitlist.count == IPS_MAX_IOCTL_QUEUE) { + SC->result = DID_BUS_BUSY << 16; + done(SC); + + return (0); + } + } else if (ha->scb_waitlist.count == IPS_MAX_QUEUE) { + SC->result = DID_BUS_BUSY << 16; + done(SC); + + return (0); + } + + SC->scsi_done = done; + + DEBUG_VAR(2, "(%s%d): ips_queue: cmd 0x%X (%d %d %d)", + ips_name, + ha->host_num, + SC->cmnd[0], + SC->device->channel, SC->device->id, SC->device->lun); + + /* Check for command to initiator IDs */ + if ((SC->device->channel > 0) + && (SC->device->id == ha->ha_id[SC->device->channel])) { + SC->result = DID_NO_CONNECT << 16; + done(SC); + + return (0); + } + + if (ips_is_passthru(SC)) { + + ips_copp_wait_item_t *scratch; + + /* A Reset IOCTL is only sent by the boot CD in extreme cases. */ + /* There can never be any system activity ( network or disk ), but check */ + /* anyway just as a good practice. */ + pt = (ips_passthru_t *) SC->request_buffer; + if ((pt->CoppCP.cmd.reset.op_code == IPS_CMD_RESET_CHANNEL) && + (pt->CoppCP.cmd.reset.adapter_flag == 1)) { + if (ha->scb_activelist.count != 0) { + SC->result = DID_BUS_BUSY << 16; + done(SC); + return (0); + } + ha->ioctl_reset = 1; /* This reset request is from an IOCTL */ + ips_eh_reset(SC); + SC->result = DID_OK << 16; + SC->scsi_done(SC); + return (0); + } + + /* allocate space for the scribble */ + scratch = kmalloc(sizeof (ips_copp_wait_item_t), GFP_ATOMIC); + + if (!scratch) { + SC->result = DID_ERROR << 16; + done(SC); + + return (0); + } + + scratch->scsi_cmd = SC; + scratch->next = NULL; + + ips_putq_copp_tail(&ha->copp_waitlist, scratch); + } else { + ips_putq_wait_tail(&ha->scb_waitlist, SC); + } + + ips_next(ha, IPS_INTR_IORL); + + return (0); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_biosparam */ +/* */ +/* Routine Description: */ +/* */ +/* Set bios geometry for the controller */ +/* */ +/****************************************************************************/ +static int +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0) +ips_biosparam(Disk * disk, kdev_t dev, int geom[]) +{ + ips_ha_t *ha = (ips_ha_t *) disk->device->host->hostdata; + unsigned long capacity = disk->capacity; +#else +ips_biosparam(struct scsi_device *sdev, struct block_device *bdev, + sector_t capacity, int geom[]) +{ + ips_ha_t *ha = (ips_ha_t *) sdev->host->hostdata; +#endif + int heads; + int sectors; + int cylinders; + + METHOD_TRACE("ips_biosparam", 1); + + if (!ha) + /* ?!?! host adater info invalid */ + return (0); + + if (!ha->active) + return (0); + + if (!ips_read_adapter_status(ha, IPS_INTR_ON)) + /* ?!?! Enquiry command failed */ + return (0); + + if ((capacity > 0x400000) && ((ha->enq->ucMiscFlag & 0x8) == 0)) { + heads = IPS_NORM_HEADS; + sectors = IPS_NORM_SECTORS; + } else { + heads = IPS_COMP_HEADS; + sectors = IPS_COMP_SECTORS; + } + + cylinders = (unsigned long) capacity / (heads * sectors); + + DEBUG_VAR(2, "Geometry: heads: %d, sectors: %d, cylinders: %d", + heads, sectors, cylinders); + + geom[0] = heads; + geom[1] = sectors; + geom[2] = cylinders; + + return (0); +} + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0) + +/* ips_proc24_info is a wrapper around ips_proc_info * + * for compatibility with the 2.4 scsi parameters */ +static int +ips_proc24_info(char *buffer, char **start, off_t offset, int length, + int hostno, int func) +{ + int i; + + for (i = 0; i < ips_next_controller; i++) { + if (ips_sh[i] && ips_sh[i]->host_no == hostno) { + return ips_proc_info(ips_sh[i], buffer, start, + offset, length, func); + } + } + return -EINVAL; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_select_queue_depth */ +/* */ +/* Routine Description: */ +/* */ +/* Select queue depths for the devices on the contoller */ +/* */ +/****************************************************************************/ +static void +ips_select_queue_depth(struct Scsi_Host *host, Scsi_Device * scsi_devs) +{ + Scsi_Device *device; + ips_ha_t *ha; + int count = 0; + int min; + + ha = IPS_HA(host); + min = ha->max_cmds / 4; + + for (device = scsi_devs; device; device = device->next) { + if (device->host == host) { + if ((device->channel == 0) && (device->type == 0)) + count++; + } + } + + for (device = scsi_devs; device; device = device->next) { + if (device->host == host) { + if ((device->channel == 0) && (device->type == 0)) { + device->queue_depth = + (ha->max_cmds - 1) / count; + if (device->queue_depth < min) + device->queue_depth = min; + } else { + device->queue_depth = 2; + } + + if (device->queue_depth < 2) + device->queue_depth = 2; + } + } +} + +#else +/****************************************************************************/ +/* */ +/* Routine Name: ips_slave_configure */ +/* */ +/* Routine Description: */ +/* */ +/* Set queue depths on devices once scan is complete */ +/* */ +/****************************************************************************/ +static int +ips_slave_configure(Scsi_Device * SDptr) +{ + ips_ha_t *ha; + int min; + + ha = IPS_HA(SDptr->host); + if (SDptr->tagged_supported && SDptr->type == TYPE_DISK) { + min = ha->max_cmds / 2; + if (ha->enq->ucLogDriveCount <= 2) + min = ha->max_cmds - 1; + scsi_adjust_queue_depth(SDptr, MSG_ORDERED_TAG, min); + } + return 0; +} +#endif + +/****************************************************************************/ +/* */ +/* Routine Name: do_ipsintr */ +/* */ +/* Routine Description: */ +/* */ +/* Wrapper for the interrupt handler */ +/* */ +/****************************************************************************/ +static irqreturn_t +do_ipsintr(int irq, void *dev_id, struct pt_regs * regs) +{ + ips_ha_t *ha; + unsigned long cpu_flags; + struct Scsi_Host *host; + int irqstatus; + + METHOD_TRACE("do_ipsintr", 2); + + ha = (ips_ha_t *) dev_id; + if (!ha) + return IRQ_NONE; + host = ips_sh[ha->host_num]; + /* interrupt during initialization */ + if (!host) { + (*ha->func.intr) (ha); + return IRQ_HANDLED; + } + + IPS_LOCK_SAVE(host->host_lock, cpu_flags); + + if (!ha->active) { + IPS_UNLOCK_RESTORE(host->host_lock, cpu_flags); + return IRQ_HANDLED; + } + + irqstatus = (*ha->func.intr) (ha); + + IPS_UNLOCK_RESTORE(host->host_lock, cpu_flags); + + /* start the next command */ + ips_next(ha, IPS_INTR_ON); + return IRQ_RETVAL(irqstatus); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_intr_copperhead */ +/* */ +/* Routine Description: */ +/* */ +/* Polling interrupt handler */ +/* */ +/* ASSUMES interrupts are disabled */ +/* */ +/****************************************************************************/ +int +ips_intr_copperhead(ips_ha_t * ha) +{ + ips_stat_t *sp; + ips_scb_t *scb; + IPS_STATUS cstatus; + int intrstatus; + + METHOD_TRACE("ips_intr", 2); + + if (!ha) + return 0; + + if (!ha->active) + return 0; + + intrstatus = (*ha->func.isintr) (ha); + + if (!intrstatus) { + /* + * Unexpected/Shared interrupt + */ + + return 0; + } + + while (TRUE) { + sp = &ha->sp; + + intrstatus = (*ha->func.isintr) (ha); + + if (!intrstatus) + break; + else + cstatus.value = (*ha->func.statupd) (ha); + + if (cstatus.fields.command_id > (IPS_MAX_CMDS - 1)) { + /* Spurious Interupt ? */ + continue; + } + + ips_chkstatus(ha, &cstatus); + scb = (ips_scb_t *) sp->scb_addr; + + /* + * use the callback function to finish things up + * NOTE: interrupts are OFF for this + */ + (*scb->callback) (ha, scb); + } /* end while */ + return 1; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_intr_morpheus */ +/* */ +/* Routine Description: */ +/* */ +/* Polling interrupt handler */ +/* */ +/* ASSUMES interrupts are disabled */ +/* */ +/****************************************************************************/ +int +ips_intr_morpheus(ips_ha_t * ha) +{ + ips_stat_t *sp; + ips_scb_t *scb; + IPS_STATUS cstatus; + int intrstatus; + + METHOD_TRACE("ips_intr_morpheus", 2); + + if (!ha) + return 0; + + if (!ha->active) + return 0; + + intrstatus = (*ha->func.isintr) (ha); + + if (!intrstatus) { + /* + * Unexpected/Shared interrupt + */ + + return 0; + } + + while (TRUE) { + sp = &ha->sp; + + intrstatus = (*ha->func.isintr) (ha); + + if (!intrstatus) + break; + else + cstatus.value = (*ha->func.statupd) (ha); + + if (cstatus.value == 0xffffffff) + /* No more to process */ + break; + + if (cstatus.fields.command_id > (IPS_MAX_CMDS - 1)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Spurious interrupt; no ccb.\n"); + + continue; + } + + ips_chkstatus(ha, &cstatus); + scb = (ips_scb_t *) sp->scb_addr; + + /* + * use the callback function to finish things up + * NOTE: interrupts are OFF for this + */ + (*scb->callback) (ha, scb); + } /* end while */ + return 1; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_info */ +/* */ +/* Routine Description: */ +/* */ +/* Return info about the driver */ +/* */ +/****************************************************************************/ +static const char * +ips_info(struct Scsi_Host *SH) +{ + static char buffer[256]; + char *bp; + ips_ha_t *ha; + + METHOD_TRACE("ips_info", 1); + + ha = IPS_HA(SH); + + if (!ha) + return (NULL); + + bp = &buffer[0]; + memset(bp, 0, sizeof (buffer)); + + sprintf(bp, "%s%s%s Build %d", "IBM PCI ServeRAID ", + IPS_VERSION_HIGH, IPS_VERSION_LOW, IPS_BUILD_IDENT); + + if (ha->ad_type > 0 && ha->ad_type <= MAX_ADAPTER_NAME) { + strcat(bp, " <"); + strcat(bp, ips_adapter_name[ha->ad_type - 1]); + strcat(bp, ">"); + } + + return (bp); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_proc_info */ +/* */ +/* Routine Description: */ +/* */ +/* The passthru interface for the driver */ +/* */ +/****************************************************************************/ +static int +ips_proc_info(struct Scsi_Host *host, char *buffer, char **start, off_t offset, + int length, int func) +{ + int i; + int ret; + ips_ha_t *ha = NULL; + + METHOD_TRACE("ips_proc_info", 1); + + /* Find our host structure */ + for (i = 0; i < ips_next_controller; i++) { + if (ips_sh[i]) { + if (ips_sh[i] == host) { + ha = (ips_ha_t *) ips_sh[i]->hostdata; + break; + } + } + } + + if (!ha) + return (-EINVAL); + + if (func) { + /* write */ + return (0); + } else { + /* read */ + if (start) + *start = buffer; + + ret = ips_host_info(ha, buffer, offset, length); + + return (ret); + } +} + +/*--------------------------------------------------------------------------*/ +/* Helper Functions */ +/*--------------------------------------------------------------------------*/ + +/****************************************************************************/ +/* */ +/* Routine Name: ips_is_passthru */ +/* */ +/* Routine Description: */ +/* */ +/* Determine if the specified SCSI command is really a passthru command */ +/* */ +/****************************************************************************/ +static int +ips_is_passthru(Scsi_Cmnd * SC) +{ + METHOD_TRACE("ips_is_passthru", 1); + + if (!SC) + return (0); + + if ((SC->cmnd[0] == IPS_IOCTL_COMMAND) && + (SC->device->channel == 0) && + (SC->device->id == IPS_ADAPTER_ID) && + (SC->device->lun == 0) && SC->request_buffer) { + if ((!SC->use_sg) && SC->request_bufflen && + (((char *) SC->request_buffer)[0] == 'C') && + (((char *) SC->request_buffer)[1] == 'O') && + (((char *) SC->request_buffer)[2] == 'P') && + (((char *) SC->request_buffer)[3] == 'P')) + return 1; + else if (SC->use_sg) { + struct scatterlist *sg = SC->request_buffer; + char *buffer = IPS_SG_ADDRESS(sg); + if (buffer && buffer[0] == 'C' && buffer[1] == 'O' && + buffer[2] == 'P' && buffer[3] == 'P') + return 1; + } + } + return 0; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_alloc_passthru_buffer */ +/* */ +/* Routine Description: */ +/* allocate a buffer large enough for the ioctl data if the ioctl buffer */ +/* is too small or doesn't exist */ +/****************************************************************************/ +static int +ips_alloc_passthru_buffer(ips_ha_t * ha, int length) +{ + void *bigger_buf; + dma_addr_t dma_busaddr; + + if (ha->ioctl_data && length <= ha->ioctl_len) + return 0; + /* there is no buffer or it's not big enough, allocate a new one */ + bigger_buf = pci_alloc_consistent(ha->pcidev, length, &dma_busaddr); + if (bigger_buf) { + /* free the old memory */ + pci_free_consistent(ha->pcidev, ha->ioctl_len, ha->ioctl_data, + ha->ioctl_busaddr); + /* use the new memory */ + ha->ioctl_data = (char *) bigger_buf; + ha->ioctl_len = length; + ha->ioctl_busaddr = dma_busaddr; + } else { + return -1; + } + return 0; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_make_passthru */ +/* */ +/* Routine Description: */ +/* */ +/* Make a passthru command out of the info in the Scsi block */ +/* */ +/****************************************************************************/ +static int +ips_make_passthru(ips_ha_t * ha, Scsi_Cmnd * SC, ips_scb_t * scb, int intr) +{ + ips_passthru_t *pt; + int length = 0; + int ret; + + METHOD_TRACE("ips_make_passthru", 1); + + if (!SC->use_sg) { + length = SC->request_bufflen; + } else { + struct scatterlist *sg = SC->request_buffer; + int i; + for (i = 0; i < SC->use_sg; i++) + length += sg[i].length; + } + if (length < sizeof (ips_passthru_t)) { + /* wrong size */ + DEBUG_VAR(1, "(%s%d) Passthru structure wrong size", + ips_name, ha->host_num); + return (IPS_FAILURE); + } + if (ips_alloc_passthru_buffer(ha, length)) { + /* allocation failure! If ha->ioctl_data exists, use it to return + some error codes. Return a failed command to the scsi layer. */ + if (ha->ioctl_data) { + pt = (ips_passthru_t *) ha->ioctl_data; + ips_scmd_buf_read(SC, pt, sizeof (ips_passthru_t)); + pt->BasicStatus = 0x0B; + pt->ExtendedStatus = 0x00; + ips_scmd_buf_write(SC, pt, sizeof (ips_passthru_t)); + } + return IPS_FAILURE; + } + ha->ioctl_datasize = length; + + ips_scmd_buf_read(SC, ha->ioctl_data, ha->ioctl_datasize); + pt = (ips_passthru_t *) ha->ioctl_data; + + /* + * Some notes about the passthru interface used + * + * IF the scsi op_code == 0x0d then we assume + * that the data came along with/goes with the + * packet we received from the sg driver. In this + * case the CmdBSize field of the pt structure is + * used for the size of the buffer. + */ + + switch (pt->CoppCmd) { + case IPS_NUMCTRLS: + memcpy(ha->ioctl_data + sizeof (ips_passthru_t), + &ips_num_controllers, sizeof (int)); + ips_scmd_buf_write(SC, ha->ioctl_data, + sizeof (ips_passthru_t) + sizeof (int)); + SC->result = DID_OK << 16; + + return (IPS_SUCCESS_IMM); + + case IPS_COPPUSRCMD: + case IPS_COPPIOCCMD: + if (SC->cmnd[0] == IPS_IOCTL_COMMAND) { + if (length < (sizeof (ips_passthru_t) + pt->CmdBSize)) { + /* wrong size */ + DEBUG_VAR(1, + "(%s%d) Passthru structure wrong size", + ips_name, ha->host_num); + + return (IPS_FAILURE); + } + + if (ha->device_id == IPS_DEVICEID_COPPERHEAD && + pt->CoppCP.cmd.flashfw.op_code == + IPS_CMD_RW_BIOSFW) { + ret = ips_flash_copperhead(ha, pt, scb); + ips_scmd_buf_write(SC, ha->ioctl_data, + sizeof (ips_passthru_t)); + return ret; + } + if (ips_usrcmd(ha, pt, scb)) + return (IPS_SUCCESS); + else + return (IPS_FAILURE); + } + + break; + + } /* end switch */ + + return (IPS_FAILURE); +} + +/****************************************************************************/ +/* Routine Name: ips_flash_copperhead */ +/* Routine Description: */ +/* Flash the BIOS/FW on a Copperhead style controller */ +/****************************************************************************/ +static int +ips_flash_copperhead(ips_ha_t * ha, ips_passthru_t * pt, ips_scb_t * scb) +{ + int datasize; + + /* Trombone is the only copperhead that can do packet flash, but only + * for firmware. No one said it had to make sence. */ + if (IPS_IS_TROMBONE(ha) && pt->CoppCP.cmd.flashfw.type == IPS_FW_IMAGE) { + if (ips_usrcmd(ha, pt, scb)) + return IPS_SUCCESS; + else + return IPS_FAILURE; + } + pt->BasicStatus = 0x0B; + pt->ExtendedStatus = 0; + scb->scsi_cmd->result = DID_OK << 16; + /* IF it's OK to Use the "CD BOOT" Flash Buffer, then you can */ + /* avoid allocating a huge buffer per adapter ( which can fail ). */ + if (pt->CoppCP.cmd.flashfw.type == IPS_BIOS_IMAGE && + pt->CoppCP.cmd.flashfw.direction == IPS_ERASE_BIOS) { + pt->BasicStatus = 0; + return ips_flash_bios(ha, pt, scb); + } else if (pt->CoppCP.cmd.flashfw.packet_num == 0) { + if (ips_FlashData && !test_and_set_bit(0, &ips_FlashDataInUse)){ + ha->flash_data = ips_FlashData; + ha->flash_busaddr = ips_flashbusaddr; + ha->flash_len = PAGE_SIZE << 7; + ha->flash_datasize = 0; + } else if (!ha->flash_data) { + datasize = pt->CoppCP.cmd.flashfw.total_packets * + pt->CoppCP.cmd.flashfw.count; + ha->flash_data = pci_alloc_consistent(ha->pcidev, + datasize, + &ha->flash_busaddr); + if (!ha->flash_data){ + printk(KERN_WARNING "Unable to allocate a flash buffer\n"); + return IPS_FAILURE; + } + ha->flash_datasize = 0; + ha->flash_len = datasize; + } else + return IPS_FAILURE; + } else { + if (pt->CoppCP.cmd.flashfw.count + ha->flash_datasize > + ha->flash_len) { + ips_free_flash_copperhead(ha); + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "failed size sanity check\n"); + return IPS_FAILURE; + } + } + if (!ha->flash_data) + return IPS_FAILURE; + pt->BasicStatus = 0; + memcpy(&ha->flash_data[ha->flash_datasize], pt + 1, + pt->CoppCP.cmd.flashfw.count); + ha->flash_datasize += pt->CoppCP.cmd.flashfw.count; + if (pt->CoppCP.cmd.flashfw.packet_num == + pt->CoppCP.cmd.flashfw.total_packets - 1) { + if (pt->CoppCP.cmd.flashfw.type == IPS_BIOS_IMAGE) + return ips_flash_bios(ha, pt, scb); + else if (pt->CoppCP.cmd.flashfw.type == IPS_FW_IMAGE) + return ips_flash_firmware(ha, pt, scb); + } + return IPS_SUCCESS_IMM; +} + +/****************************************************************************/ +/* Routine Name: ips_flash_bios */ +/* Routine Description: */ +/* flashes the bios of a copperhead adapter */ +/****************************************************************************/ +static int +ips_flash_bios(ips_ha_t * ha, ips_passthru_t * pt, ips_scb_t * scb) +{ + + if (pt->CoppCP.cmd.flashfw.type == IPS_BIOS_IMAGE && + pt->CoppCP.cmd.flashfw.direction == IPS_WRITE_BIOS) { + if ((!ha->func.programbios) || (!ha->func.erasebios) || + (!ha->func.verifybios)) + goto error; + if ((*ha->func.erasebios) (ha)) { + DEBUG_VAR(1, + "(%s%d) flash bios failed - unable to erase flash", + ips_name, ha->host_num); + goto error; + } else + if ((*ha->func.programbios) (ha, + ha->flash_data + + IPS_BIOS_HEADER, + ha->flash_datasize - + IPS_BIOS_HEADER, 0)) { + DEBUG_VAR(1, + "(%s%d) flash bios failed - unable to flash", + ips_name, ha->host_num); + goto error; + } else + if ((*ha->func.verifybios) (ha, + ha->flash_data + + IPS_BIOS_HEADER, + ha->flash_datasize - + IPS_BIOS_HEADER, 0)) { + DEBUG_VAR(1, + "(%s%d) flash bios failed - unable to verify flash", + ips_name, ha->host_num); + goto error; + } + ips_free_flash_copperhead(ha); + return IPS_SUCCESS_IMM; + } else if (pt->CoppCP.cmd.flashfw.type == IPS_BIOS_IMAGE && + pt->CoppCP.cmd.flashfw.direction == IPS_ERASE_BIOS) { + if (!ha->func.erasebios) + goto error; + if ((*ha->func.erasebios) (ha)) { + DEBUG_VAR(1, + "(%s%d) flash bios failed - unable to erase flash", + ips_name, ha->host_num); + goto error; + } + return IPS_SUCCESS_IMM; + } + error: + pt->BasicStatus = 0x0B; + pt->ExtendedStatus = 0x00; + ips_free_flash_copperhead(ha); + return IPS_FAILURE; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_fill_scb_sg_single */ +/* */ +/* Routine Description: */ +/* Fill in a single scb sg_list element from an address */ +/* return a -1 if a breakup occurred */ +/****************************************************************************/ +static int +ips_fill_scb_sg_single(ips_ha_t * ha, dma_addr_t busaddr, + ips_scb_t * scb, int indx, unsigned int e_len) +{ + + int ret_val = 0; + + if ((scb->data_len + e_len) > ha->max_xfer) { + e_len = ha->max_xfer - scb->data_len; + scb->breakup = indx; + ++scb->sg_break; + ret_val = -1; + } else { + scb->breakup = 0; + scb->sg_break = 0; + } + if (IPS_USE_ENH_SGLIST(ha)) { + scb->sg_list.enh_list[indx].address_lo = + cpu_to_le32(pci_dma_lo32(busaddr)); + scb->sg_list.enh_list[indx].address_hi = + cpu_to_le32(pci_dma_hi32(busaddr)); + scb->sg_list.enh_list[indx].length = cpu_to_le32(e_len); + } else { + scb->sg_list.std_list[indx].address = + cpu_to_le32(pci_dma_lo32(busaddr)); + scb->sg_list.std_list[indx].length = cpu_to_le32(e_len); + } + + ++scb->sg_len; + scb->data_len += e_len; + return ret_val; +} + +/****************************************************************************/ +/* Routine Name: ips_flash_firmware */ +/* Routine Description: */ +/* flashes the firmware of a copperhead adapter */ +/****************************************************************************/ +static int +ips_flash_firmware(ips_ha_t * ha, ips_passthru_t * pt, ips_scb_t * scb) +{ + IPS_SG_LIST sg_list; + uint32_t cmd_busaddr; + + if (pt->CoppCP.cmd.flashfw.type == IPS_FW_IMAGE && + pt->CoppCP.cmd.flashfw.direction == IPS_WRITE_FW) { + memset(&pt->CoppCP.cmd, 0, sizeof (IPS_HOST_COMMAND)); + pt->CoppCP.cmd.flashfw.op_code = IPS_CMD_DOWNLOAD; + pt->CoppCP.cmd.flashfw.count = cpu_to_le32(ha->flash_datasize); + } else { + pt->BasicStatus = 0x0B; + pt->ExtendedStatus = 0x00; + ips_free_flash_copperhead(ha); + return IPS_FAILURE; + } + /* Save the S/G list pointer so it doesn't get clobbered */ + sg_list.list = scb->sg_list.list; + cmd_busaddr = scb->scb_busaddr; + /* copy in the CP */ + memcpy(&scb->cmd, &pt->CoppCP.cmd, sizeof (IPS_IOCTL_CMD)); + /* FIX stuff that might be wrong */ + scb->sg_list.list = sg_list.list; + scb->scb_busaddr = cmd_busaddr; + scb->bus = scb->scsi_cmd->device->channel; + scb->target_id = scb->scsi_cmd->device->id; + scb->lun = scb->scsi_cmd->device->lun; + scb->sg_len = 0; + scb->data_len = 0; + scb->flags = 0; + scb->op_code = 0; + scb->callback = ipsintr_done; + scb->timeout = ips_cmd_timeout; + + scb->data_len = ha->flash_datasize; + scb->data_busaddr = + pci_map_single(ha->pcidev, ha->flash_data, scb->data_len, + IPS_DMA_DIR(scb)); + scb->flags |= IPS_SCB_MAP_SINGLE; + scb->cmd.flashfw.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.flashfw.buffer_addr = cpu_to_le32(scb->data_busaddr); + if (pt->TimeOut) + scb->timeout = pt->TimeOut; + scb->scsi_cmd->result = DID_OK << 16; + return IPS_SUCCESS; +} + +/****************************************************************************/ +/* Routine Name: ips_free_flash_copperhead */ +/* Routine Description: */ +/* release the memory resources used to hold the flash image */ +/****************************************************************************/ +static void +ips_free_flash_copperhead(ips_ha_t * ha) +{ + if (ha->flash_data == ips_FlashData) + test_and_clear_bit(0, &ips_FlashDataInUse); + else if (ha->flash_data) + pci_free_consistent(ha->pcidev, ha->flash_len, ha->flash_data, + ha->flash_busaddr); + ha->flash_data = NULL; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_usrcmd */ +/* */ +/* Routine Description: */ +/* */ +/* Process a user command and make it ready to send */ +/* */ +/****************************************************************************/ +static int +ips_usrcmd(ips_ha_t * ha, ips_passthru_t * pt, ips_scb_t * scb) +{ + IPS_SG_LIST sg_list; + uint32_t cmd_busaddr; + + METHOD_TRACE("ips_usrcmd", 1); + + if ((!scb) || (!pt) || (!ha)) + return (0); + + /* Save the S/G list pointer so it doesn't get clobbered */ + sg_list.list = scb->sg_list.list; + cmd_busaddr = scb->scb_busaddr; + /* copy in the CP */ + memcpy(&scb->cmd, &pt->CoppCP.cmd, sizeof (IPS_IOCTL_CMD)); + memcpy(&scb->dcdb, &pt->CoppCP.dcdb, sizeof (IPS_DCDB_TABLE)); + + /* FIX stuff that might be wrong */ + scb->sg_list.list = sg_list.list; + scb->scb_busaddr = cmd_busaddr; + scb->bus = scb->scsi_cmd->device->channel; + scb->target_id = scb->scsi_cmd->device->id; + scb->lun = scb->scsi_cmd->device->lun; + scb->sg_len = 0; + scb->data_len = 0; + scb->flags = 0; + scb->op_code = 0; + scb->callback = ipsintr_done; + scb->timeout = ips_cmd_timeout; + scb->cmd.basic_io.command_id = IPS_COMMAND_ID(ha, scb); + + /* we don't support DCDB/READ/WRITE Scatter Gather */ + if ((scb->cmd.basic_io.op_code == IPS_CMD_READ_SG) || + (scb->cmd.basic_io.op_code == IPS_CMD_WRITE_SG) || + (scb->cmd.basic_io.op_code == IPS_CMD_DCDB_SG)) + return (0); + + if (pt->CmdBSize) { + scb->data_len = pt->CmdBSize; + scb->data_busaddr = ha->ioctl_busaddr + sizeof (ips_passthru_t); + } else { + scb->data_busaddr = 0L; + } + + if (scb->cmd.dcdb.op_code == IPS_CMD_DCDB) + scb->cmd.dcdb.dcdb_address = cpu_to_le32(scb->scb_busaddr + + (unsigned long) &scb-> + dcdb - + (unsigned long) scb); + + if (pt->CmdBSize) { + if (scb->cmd.dcdb.op_code == IPS_CMD_DCDB) + scb->dcdb.buffer_pointer = + cpu_to_le32(scb->data_busaddr); + else + scb->cmd.basic_io.sg_addr = + cpu_to_le32(scb->data_busaddr); + } + + /* set timeouts */ + if (pt->TimeOut) { + scb->timeout = pt->TimeOut; + + if (pt->TimeOut <= 10) + scb->dcdb.cmd_attribute |= IPS_TIMEOUT10; + else if (pt->TimeOut <= 60) + scb->dcdb.cmd_attribute |= IPS_TIMEOUT60; + else + scb->dcdb.cmd_attribute |= IPS_TIMEOUT20M; + } + + /* assume success */ + scb->scsi_cmd->result = DID_OK << 16; + + /* success */ + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_cleanup_passthru */ +/* */ +/* Routine Description: */ +/* */ +/* Cleanup after a passthru command */ +/* */ +/****************************************************************************/ +static void +ips_cleanup_passthru(ips_ha_t * ha, ips_scb_t * scb) +{ + ips_passthru_t *pt; + + METHOD_TRACE("ips_cleanup_passthru", 1); + + if ((!scb) || (!scb->scsi_cmd) || (!scb->scsi_cmd->request_buffer)) { + DEBUG_VAR(1, "(%s%d) couldn't cleanup after passthru", + ips_name, ha->host_num); + + return; + } + pt = (ips_passthru_t *) ha->ioctl_data; + + /* Copy data back to the user */ + if (scb->cmd.dcdb.op_code == IPS_CMD_DCDB) /* Copy DCDB Back to Caller's Area */ + memcpy(&pt->CoppCP.dcdb, &scb->dcdb, sizeof (IPS_DCDB_TABLE)); + + pt->BasicStatus = scb->basic_status; + pt->ExtendedStatus = scb->extended_status; + pt->AdapterType = ha->ad_type; + + if (ha->device_id == IPS_DEVICEID_COPPERHEAD && + (scb->cmd.flashfw.op_code == IPS_CMD_DOWNLOAD || + scb->cmd.flashfw.op_code == IPS_CMD_RW_BIOSFW)) + ips_free_flash_copperhead(ha); + + ips_scmd_buf_write(scb->scsi_cmd, ha->ioctl_data, ha->ioctl_datasize); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_host_info */ +/* */ +/* Routine Description: */ +/* */ +/* The passthru interface for the driver */ +/* */ +/****************************************************************************/ +static int +ips_host_info(ips_ha_t * ha, char *ptr, off_t offset, int len) +{ + IPS_INFOSTR info; + + METHOD_TRACE("ips_host_info", 1); + + info.buffer = ptr; + info.length = len; + info.offset = offset; + info.pos = 0; + info.localpos = 0; + + copy_info(&info, "\nIBM ServeRAID General Information:\n\n"); + + if ((le32_to_cpu(ha->nvram->signature) == IPS_NVRAM_P5_SIG) && + (le16_to_cpu(ha->nvram->adapter_type) != 0)) + copy_info(&info, "\tController Type : %s\n", + ips_adapter_name[ha->ad_type - 1]); + else + copy_info(&info, + "\tController Type : Unknown\n"); + + if (ha->io_addr) + copy_info(&info, + "\tIO region : 0x%lx (%d bytes)\n", + ha->io_addr, ha->io_len); + + if (ha->mem_addr) { + copy_info(&info, + "\tMemory region : 0x%lx (%d bytes)\n", + ha->mem_addr, ha->mem_len); + copy_info(&info, + "\tShared memory address : 0x%lx\n", + ha->mem_ptr); + } + + copy_info(&info, "\tIRQ number : %d\n", ha->irq); + + /* For the Next 3 lines Check for Binary 0 at the end and don't include it if it's there. */ + /* That keeps everything happy for "text" operations on the proc file. */ + + if (le32_to_cpu(ha->nvram->signature) == IPS_NVRAM_P5_SIG) { + if (ha->nvram->bios_low[3] == 0) { + copy_info(&info, + "\tBIOS Version : %c%c%c%c%c%c%c\n", + ha->nvram->bios_high[0], ha->nvram->bios_high[1], + ha->nvram->bios_high[2], ha->nvram->bios_high[3], + ha->nvram->bios_low[0], ha->nvram->bios_low[1], + ha->nvram->bios_low[2]); + + } else { + copy_info(&info, + "\tBIOS Version : %c%c%c%c%c%c%c%c\n", + ha->nvram->bios_high[0], ha->nvram->bios_high[1], + ha->nvram->bios_high[2], ha->nvram->bios_high[3], + ha->nvram->bios_low[0], ha->nvram->bios_low[1], + ha->nvram->bios_low[2], ha->nvram->bios_low[3]); + } + + } + + if (ha->enq->CodeBlkVersion[7] == 0) { + copy_info(&info, + "\tFirmware Version : %c%c%c%c%c%c%c\n", + ha->enq->CodeBlkVersion[0], ha->enq->CodeBlkVersion[1], + ha->enq->CodeBlkVersion[2], ha->enq->CodeBlkVersion[3], + ha->enq->CodeBlkVersion[4], ha->enq->CodeBlkVersion[5], + ha->enq->CodeBlkVersion[6]); + } else { + copy_info(&info, + "\tFirmware Version : %c%c%c%c%c%c%c%c\n", + ha->enq->CodeBlkVersion[0], ha->enq->CodeBlkVersion[1], + ha->enq->CodeBlkVersion[2], ha->enq->CodeBlkVersion[3], + ha->enq->CodeBlkVersion[4], ha->enq->CodeBlkVersion[5], + ha->enq->CodeBlkVersion[6], ha->enq->CodeBlkVersion[7]); + } + + if (ha->enq->BootBlkVersion[7] == 0) { + copy_info(&info, + "\tBoot Block Version : %c%c%c%c%c%c%c\n", + ha->enq->BootBlkVersion[0], ha->enq->BootBlkVersion[1], + ha->enq->BootBlkVersion[2], ha->enq->BootBlkVersion[3], + ha->enq->BootBlkVersion[4], ha->enq->BootBlkVersion[5], + ha->enq->BootBlkVersion[6]); + } else { + copy_info(&info, + "\tBoot Block Version : %c%c%c%c%c%c%c%c\n", + ha->enq->BootBlkVersion[0], ha->enq->BootBlkVersion[1], + ha->enq->BootBlkVersion[2], ha->enq->BootBlkVersion[3], + ha->enq->BootBlkVersion[4], ha->enq->BootBlkVersion[5], + ha->enq->BootBlkVersion[6], ha->enq->BootBlkVersion[7]); + } + + copy_info(&info, "\tDriver Version : %s%s\n", + IPS_VERSION_HIGH, IPS_VERSION_LOW); + + copy_info(&info, "\tDriver Build : %d\n", + IPS_BUILD_IDENT); + + copy_info(&info, "\tMax Physical Devices : %d\n", + ha->enq->ucMaxPhysicalDevices); + copy_info(&info, "\tMax Active Commands : %d\n", + ha->max_cmds); + copy_info(&info, "\tCurrent Queued Commands : %d\n", + ha->scb_waitlist.count); + copy_info(&info, "\tCurrent Active Commands : %d\n", + ha->scb_activelist.count - ha->num_ioctl); + copy_info(&info, "\tCurrent Queued PT Commands : %d\n", + ha->copp_waitlist.count); + copy_info(&info, "\tCurrent Active PT Commands : %d\n", + ha->num_ioctl); + + copy_info(&info, "\n"); + + return (info.localpos); +} + +/****************************************************************************/ +/* */ +/* Routine Name: copy_mem_info */ +/* */ +/* Routine Description: */ +/* */ +/* Copy data into an IPS_INFOSTR structure */ +/* */ +/****************************************************************************/ +static void +copy_mem_info(IPS_INFOSTR * info, char *data, int len) +{ + METHOD_TRACE("copy_mem_info", 1); + + if (info->pos + len < info->offset) { + info->pos += len; + return; + } + + if (info->pos < info->offset) { + data += (info->offset - info->pos); + len -= (info->offset - info->pos); + info->pos += (info->offset - info->pos); + } + + if (info->localpos + len > info->length) + len = info->length - info->localpos; + + if (len > 0) { + memcpy(info->buffer + info->localpos, data, len); + info->pos += len; + info->localpos += len; + } +} + +/****************************************************************************/ +/* */ +/* Routine Name: copy_info */ +/* */ +/* Routine Description: */ +/* */ +/* printf style wrapper for an info structure */ +/* */ +/****************************************************************************/ +static int +copy_info(IPS_INFOSTR * info, char *fmt, ...) +{ + va_list args; + char buf[128]; + int len; + + METHOD_TRACE("copy_info", 1); + + va_start(args, fmt); + len = vsprintf(buf, fmt, args); + va_end(args); + + copy_mem_info(info, buf, len); + + return (len); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_identify_controller */ +/* */ +/* Routine Description: */ +/* */ +/* Identify this controller */ +/* */ +/****************************************************************************/ +static void +ips_identify_controller(ips_ha_t * ha) +{ + METHOD_TRACE("ips_identify_controller", 1); + + switch (ha->device_id) { + case IPS_DEVICEID_COPPERHEAD: + if (ha->revision_id <= IPS_REVID_SERVERAID) { + ha->ad_type = IPS_ADTYPE_SERVERAID; + } else if (ha->revision_id == IPS_REVID_SERVERAID2) { + ha->ad_type = IPS_ADTYPE_SERVERAID2; + } else if (ha->revision_id == IPS_REVID_NAVAJO) { + ha->ad_type = IPS_ADTYPE_NAVAJO; + } else if ((ha->revision_id == IPS_REVID_SERVERAID2) + && (ha->slot_num == 0)) { + ha->ad_type = IPS_ADTYPE_KIOWA; + } else if ((ha->revision_id >= IPS_REVID_CLARINETP1) && + (ha->revision_id <= IPS_REVID_CLARINETP3)) { + if (ha->enq->ucMaxPhysicalDevices == 15) + ha->ad_type = IPS_ADTYPE_SERVERAID3L; + else + ha->ad_type = IPS_ADTYPE_SERVERAID3; + } else if ((ha->revision_id >= IPS_REVID_TROMBONE32) && + (ha->revision_id <= IPS_REVID_TROMBONE64)) { + ha->ad_type = IPS_ADTYPE_SERVERAID4H; + } + break; + + case IPS_DEVICEID_MORPHEUS: + switch (ha->subdevice_id) { + case IPS_SUBDEVICEID_4L: + ha->ad_type = IPS_ADTYPE_SERVERAID4L; + break; + + case IPS_SUBDEVICEID_4M: + ha->ad_type = IPS_ADTYPE_SERVERAID4M; + break; + + case IPS_SUBDEVICEID_4MX: + ha->ad_type = IPS_ADTYPE_SERVERAID4MX; + break; + + case IPS_SUBDEVICEID_4LX: + ha->ad_type = IPS_ADTYPE_SERVERAID4LX; + break; + + case IPS_SUBDEVICEID_5I2: + ha->ad_type = IPS_ADTYPE_SERVERAID5I2; + break; + + case IPS_SUBDEVICEID_5I1: + ha->ad_type = IPS_ADTYPE_SERVERAID5I1; + break; + } + + break; + + case IPS_DEVICEID_MARCO: + switch (ha->subdevice_id) { + case IPS_SUBDEVICEID_6M: + ha->ad_type = IPS_ADTYPE_SERVERAID6M; + break; + case IPS_SUBDEVICEID_6I: + ha->ad_type = IPS_ADTYPE_SERVERAID6I; + break; + case IPS_SUBDEVICEID_7k: + ha->ad_type = IPS_ADTYPE_SERVERAID7k; + break; + case IPS_SUBDEVICEID_7M: + ha->ad_type = IPS_ADTYPE_SERVERAID7M; + break; + } + break; + } +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_get_bios_version */ +/* */ +/* Routine Description: */ +/* */ +/* Get the BIOS revision number */ +/* */ +/****************************************************************************/ +static void +ips_get_bios_version(ips_ha_t * ha, int intr) +{ + ips_scb_t *scb; + int ret; + uint8_t major; + uint8_t minor; + uint8_t subminor; + uint8_t *buffer; + char hexDigits[] = + { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', + 'D', 'E', 'F' }; + + METHOD_TRACE("ips_get_bios_version", 1); + + major = 0; + minor = 0; + + strncpy(ha->bios_version, " ?", 8); + + if (ha->device_id == IPS_DEVICEID_COPPERHEAD) { + if (IPS_USE_MEMIO(ha)) { + /* Memory Mapped I/O */ + + /* test 1st byte */ + writel(0, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + if (readb(ha->mem_ptr + IPS_REG_FLDP) != 0x55) + return; + + writel(1, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + if (readb(ha->mem_ptr + IPS_REG_FLDP) != 0xAA) + return; + + /* Get Major version */ + writel(0x1FF, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + major = readb(ha->mem_ptr + IPS_REG_FLDP); + + /* Get Minor version */ + writel(0x1FE, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + minor = readb(ha->mem_ptr + IPS_REG_FLDP); + + /* Get SubMinor version */ + writel(0x1FD, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + subminor = readb(ha->mem_ptr + IPS_REG_FLDP); + + } else { + /* Programmed I/O */ + + /* test 1st byte */ + outl(0, ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + if (inb(ha->io_addr + IPS_REG_FLDP) != 0x55) + return; + + outl(cpu_to_le32(1), ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + if (inb(ha->io_addr + IPS_REG_FLDP) != 0xAA) + return; + + /* Get Major version */ + outl(cpu_to_le32(0x1FF), ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + major = inb(ha->io_addr + IPS_REG_FLDP); + + /* Get Minor version */ + outl(cpu_to_le32(0x1FE), ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + minor = inb(ha->io_addr + IPS_REG_FLDP); + + /* Get SubMinor version */ + outl(cpu_to_le32(0x1FD), ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + subminor = inb(ha->io_addr + IPS_REG_FLDP); + + } + } else { + /* Morpheus Family - Send Command to the card */ + + buffer = ha->ioctl_data; + + memset(buffer, 0, 0x1000); + + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_RW_BIOSFW; + + scb->cmd.flashfw.op_code = IPS_CMD_RW_BIOSFW; + scb->cmd.flashfw.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.flashfw.type = 1; + scb->cmd.flashfw.direction = 0; + scb->cmd.flashfw.count = cpu_to_le32(0x800); + scb->cmd.flashfw.total_packets = 1; + scb->cmd.flashfw.packet_num = 0; + scb->data_len = 0x1000; + scb->cmd.flashfw.buffer_addr = ha->ioctl_busaddr; + + /* issue the command */ + if (((ret = + ips_send_wait(ha, scb, ips_cmd_timeout, + intr)) == IPS_FAILURE) + || (ret == IPS_SUCCESS_IMM) + || ((scb->basic_status & IPS_GSC_STATUS_MASK) > 1)) { + /* Error occurred */ + + return; + } + + if ((buffer[0xC0] == 0x55) && (buffer[0xC1] == 0xAA)) { + major = buffer[0x1ff + 0xC0]; /* Offset 0x1ff after the header (0xc0) */ + minor = buffer[0x1fe + 0xC0]; /* Offset 0x1fe after the header (0xc0) */ + subminor = buffer[0x1fd + 0xC0]; /* Offset 0x1fd after the header (0xc0) */ + } else { + return; + } + } + + ha->bios_version[0] = hexDigits[(major & 0xF0) >> 4]; + ha->bios_version[1] = '.'; + ha->bios_version[2] = hexDigits[major & 0x0F]; + ha->bios_version[3] = hexDigits[subminor]; + ha->bios_version[4] = '.'; + ha->bios_version[5] = hexDigits[(minor & 0xF0) >> 4]; + ha->bios_version[6] = hexDigits[minor & 0x0F]; + ha->bios_version[7] = 0; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_hainit */ +/* */ +/* Routine Description: */ +/* */ +/* Initialize the controller */ +/* */ +/* NOTE: Assumes to be called from with a lock */ +/* */ +/****************************************************************************/ +static int +ips_hainit(ips_ha_t * ha) +{ + int i; + struct timeval tv; + + METHOD_TRACE("ips_hainit", 1); + + if (!ha) + return (0); + + if (ha->func.statinit) + (*ha->func.statinit) (ha); + + if (ha->func.enableint) + (*ha->func.enableint) (ha); + + /* Send FFDC */ + ha->reset_count = 1; + do_gettimeofday(&tv); + ha->last_ffdc = tv.tv_sec; + ips_ffdc_reset(ha, IPS_INTR_IORL); + + if (!ips_read_config(ha, IPS_INTR_IORL)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "unable to read config from controller.\n"); + + return (0); + } + /* end if */ + if (!ips_read_adapter_status(ha, IPS_INTR_IORL)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "unable to read controller status.\n"); + + return (0); + } + + /* Identify this controller */ + ips_identify_controller(ha); + + if (!ips_read_subsystem_parameters(ha, IPS_INTR_IORL)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "unable to read subsystem parameters.\n"); + + return (0); + } + + /* write nvram user page 5 */ + if (!ips_write_driver_status(ha, IPS_INTR_IORL)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "unable to write driver info to controller.\n"); + + return (0); + } + + /* If there are Logical Drives and a Reset Occurred, then an EraseStripeLock is Needed */ + if ((ha->conf->ucLogDriveCount > 0) && (ha->requires_esl == 1)) + ips_clear_adapter(ha, IPS_INTR_IORL); + + /* set limits on SID, LUN, BUS */ + ha->ntargets = IPS_MAX_TARGETS + 1; + ha->nlun = 1; + ha->nbus = (ha->enq->ucMaxPhysicalDevices / IPS_MAX_TARGETS) + 1; + + switch (ha->conf->logical_drive[0].ucStripeSize) { + case 4: + ha->max_xfer = 0x10000; + break; + + case 5: + ha->max_xfer = 0x20000; + break; + + case 6: + ha->max_xfer = 0x40000; + break; + + case 7: + default: + ha->max_xfer = 0x80000; + break; + } + + /* setup max concurrent commands */ + if (le32_to_cpu(ha->subsys->param[4]) & 0x1) { + /* Use the new method */ + ha->max_cmds = ha->enq->ucConcurrentCmdCount; + } else { + /* use the old method */ + switch (ha->conf->logical_drive[0].ucStripeSize) { + case 4: + ha->max_cmds = 32; + break; + + case 5: + ha->max_cmds = 16; + break; + + case 6: + ha->max_cmds = 8; + break; + + case 7: + default: + ha->max_cmds = 4; + break; + } + } + + /* Limit the Active Commands on a Lite Adapter */ + if ((ha->ad_type == IPS_ADTYPE_SERVERAID3L) || + (ha->ad_type == IPS_ADTYPE_SERVERAID4L) || + (ha->ad_type == IPS_ADTYPE_SERVERAID4LX)) { + if ((ha->max_cmds > MaxLiteCmds) && (MaxLiteCmds)) + ha->max_cmds = MaxLiteCmds; + } + + /* set controller IDs */ + ha->ha_id[0] = IPS_ADAPTER_ID; + for (i = 1; i < ha->nbus; i++) { + ha->ha_id[i] = ha->conf->init_id[i - 1] & 0x1f; + ha->dcdb_active[i - 1] = 0; + } + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_next */ +/* */ +/* Routine Description: */ +/* */ +/* Take the next command off the queue and send it to the controller */ +/* */ +/****************************************************************************/ +static void +ips_next(ips_ha_t * ha, int intr) +{ + ips_scb_t *scb; + Scsi_Cmnd *SC; + Scsi_Cmnd *p; + Scsi_Cmnd *q; + ips_copp_wait_item_t *item; + int ret; + unsigned long cpu_flags = 0; + struct Scsi_Host *host; + METHOD_TRACE("ips_next", 1); + + if (!ha) + return; + host = ips_sh[ha->host_num]; + /* + * Block access to the queue function so + * this command won't time out + */ + if (intr == IPS_INTR_ON) + IPS_LOCK_SAVE(host->host_lock, cpu_flags); + + if ((ha->subsys->param[3] & 0x300000) + && (ha->scb_activelist.count == 0)) { + struct timeval tv; + + do_gettimeofday(&tv); + + if (tv.tv_sec - ha->last_ffdc > IPS_SECS_8HOURS) { + ha->last_ffdc = tv.tv_sec; + ips_ffdc_time(ha); + } + } + + /* + * Send passthru commands + * These have priority over normal I/O + * but shouldn't affect performance too much + * since we limit the number that can be active + * on the card at any one time + */ + while ((ha->num_ioctl < IPS_MAX_IOCTL) && + (ha->copp_waitlist.head) && (scb = ips_getscb(ha))) { + + item = ips_removeq_copp_head(&ha->copp_waitlist); + ha->num_ioctl++; + if (intr == IPS_INTR_ON) + IPS_UNLOCK_RESTORE(host->host_lock, cpu_flags); + scb->scsi_cmd = item->scsi_cmd; + kfree(item); + + ret = ips_make_passthru(ha, scb->scsi_cmd, scb, intr); + + if (intr == IPS_INTR_ON) + IPS_LOCK_SAVE(host->host_lock, cpu_flags); + switch (ret) { + case IPS_FAILURE: + if (scb->scsi_cmd) { + scb->scsi_cmd->result = DID_ERROR << 16; + scb->scsi_cmd->scsi_done(scb->scsi_cmd); + } + + ips_freescb(ha, scb); + break; + case IPS_SUCCESS_IMM: + if (scb->scsi_cmd) { + scb->scsi_cmd->result = DID_OK << 16; + scb->scsi_cmd->scsi_done(scb->scsi_cmd); + } + + ips_freescb(ha, scb); + break; + default: + break; + } /* end case */ + + if (ret != IPS_SUCCESS) { + ha->num_ioctl--; + continue; + } + + ret = ips_send_cmd(ha, scb); + + if (ret == IPS_SUCCESS) + ips_putq_scb_head(&ha->scb_activelist, scb); + else + ha->num_ioctl--; + + switch (ret) { + case IPS_FAILURE: + if (scb->scsi_cmd) { + scb->scsi_cmd->result = DID_ERROR << 16; + } + + ips_freescb(ha, scb); + break; + case IPS_SUCCESS_IMM: + ips_freescb(ha, scb); + break; + default: + break; + } /* end case */ + + } + + /* + * Send "Normal" I/O commands + */ + + p = ha->scb_waitlist.head; + while ((p) && (scb = ips_getscb(ha))) { + if ((p->device->channel > 0) + && (ha-> + dcdb_active[p->device->channel - + 1] & (1 << p->device->id))) { + ips_freescb(ha, scb); + p = (Scsi_Cmnd *) p->host_scribble; + continue; + } + + q = p; + SC = ips_removeq_wait(&ha->scb_waitlist, q); + + if (intr == IPS_INTR_ON) + IPS_UNLOCK_RESTORE(host->host_lock, cpu_flags); /* Unlock HA after command is taken off queue */ + + SC->result = DID_OK; + SC->host_scribble = NULL; + + memset(SC->sense_buffer, 0, sizeof (SC->sense_buffer)); + + scb->target_id = SC->device->id; + scb->lun = SC->device->lun; + scb->bus = SC->device->channel; + scb->scsi_cmd = SC; + scb->breakup = 0; + scb->data_len = 0; + scb->callback = ipsintr_done; + scb->timeout = ips_cmd_timeout; + memset(&scb->cmd, 0, 16); + + /* copy in the CDB */ + memcpy(scb->cdb, SC->cmnd, SC->cmd_len); + + /* Now handle the data buffer */ + if (SC->use_sg) { + struct scatterlist *sg; + int i; + + sg = SC->request_buffer; + scb->sg_count = pci_map_sg(ha->pcidev, sg, SC->use_sg, + scsi_to_pci_dma_dir(SC-> + sc_data_direction)); + scb->flags |= IPS_SCB_MAP_SG; + for (i = 0; i < scb->sg_count; i++) { + if (ips_fill_scb_sg_single + (ha, sg_dma_address(&sg[i]), scb, i, + sg_dma_len(&sg[i])) < 0) + break; + } + scb->dcdb.transfer_length = scb->data_len; + } else { + if (SC->request_bufflen) { + scb->data_busaddr = + pci_map_single(ha->pcidev, + SC->request_buffer, + SC->request_bufflen, + scsi_to_pci_dma_dir(SC-> + sc_data_direction)); + scb->flags |= IPS_SCB_MAP_SINGLE; + ips_fill_scb_sg_single(ha, scb->data_busaddr, + scb, 0, + SC->request_bufflen); + scb->dcdb.transfer_length = scb->data_len; + } else { + scb->data_busaddr = 0L; + scb->sg_len = 0; + scb->data_len = 0; + scb->dcdb.transfer_length = 0; + } + + } + + scb->dcdb.cmd_attribute = + ips_command_direction[scb->scsi_cmd->cmnd[0]]; + + /* Allow a WRITE BUFFER Command to Have no Data */ + /* This is Used by Tape Flash Utilites */ + if ((scb->scsi_cmd->cmnd[0] == WRITE_BUFFER) && (scb->data_len == 0)) + scb->dcdb.cmd_attribute = 0; + + if (!(scb->dcdb.cmd_attribute & 0x3)) + scb->dcdb.transfer_length = 0; + + if (scb->data_len >= IPS_MAX_XFER) { + scb->dcdb.cmd_attribute |= IPS_TRANSFER64K; + scb->dcdb.transfer_length = 0; + } + if (intr == IPS_INTR_ON) + IPS_LOCK_SAVE(host->host_lock, cpu_flags); + + ret = ips_send_cmd(ha, scb); + + switch (ret) { + case IPS_SUCCESS: + ips_putq_scb_head(&ha->scb_activelist, scb); + break; + case IPS_FAILURE: + if (scb->scsi_cmd) { + scb->scsi_cmd->result = DID_ERROR << 16; + scb->scsi_cmd->scsi_done(scb->scsi_cmd); + } + + if (scb->bus) + ha->dcdb_active[scb->bus - 1] &= + ~(1 << scb->target_id); + + ips_freescb(ha, scb); + break; + case IPS_SUCCESS_IMM: + if (scb->scsi_cmd) + scb->scsi_cmd->scsi_done(scb->scsi_cmd); + + if (scb->bus) + ha->dcdb_active[scb->bus - 1] &= + ~(1 << scb->target_id); + + ips_freescb(ha, scb); + break; + default: + break; + } /* end case */ + + p = (Scsi_Cmnd *) p->host_scribble; + + } /* end while */ + + if (intr == IPS_INTR_ON) + IPS_UNLOCK_RESTORE(host->host_lock, cpu_flags); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_putq_scb_head */ +/* */ +/* Routine Description: */ +/* */ +/* Add an item to the head of the queue */ +/* */ +/* ASSUMED to be called from within the HA lock */ +/* */ +/****************************************************************************/ +static void +ips_putq_scb_head(ips_scb_queue_t * queue, ips_scb_t * item) +{ + METHOD_TRACE("ips_putq_scb_head", 1); + + if (!item) + return; + + item->q_next = queue->head; + queue->head = item; + + if (!queue->tail) + queue->tail = item; + + queue->count++; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_removeq_scb_head */ +/* */ +/* Routine Description: */ +/* */ +/* Remove the head of the queue */ +/* */ +/* ASSUMED to be called from within the HA lock */ +/* */ +/****************************************************************************/ +static ips_scb_t * +ips_removeq_scb_head(ips_scb_queue_t * queue) +{ + ips_scb_t *item; + + METHOD_TRACE("ips_removeq_scb_head", 1); + + item = queue->head; + + if (!item) { + return (NULL); + } + + queue->head = item->q_next; + item->q_next = NULL; + + if (queue->tail == item) + queue->tail = NULL; + + queue->count--; + + return (item); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_removeq_scb */ +/* */ +/* Routine Description: */ +/* */ +/* Remove an item from a queue */ +/* */ +/* ASSUMED to be called from within the HA lock */ +/* */ +/****************************************************************************/ +static ips_scb_t * +ips_removeq_scb(ips_scb_queue_t * queue, ips_scb_t * item) +{ + ips_scb_t *p; + + METHOD_TRACE("ips_removeq_scb", 1); + + if (!item) + return (NULL); + + if (item == queue->head) { + return (ips_removeq_scb_head(queue)); + } + + p = queue->head; + + while ((p) && (item != p->q_next)) + p = p->q_next; + + if (p) { + /* found a match */ + p->q_next = item->q_next; + + if (!item->q_next) + queue->tail = p; + + item->q_next = NULL; + queue->count--; + + return (item); + } + + return (NULL); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_putq_wait_tail */ +/* */ +/* Routine Description: */ +/* */ +/* Add an item to the tail of the queue */ +/* */ +/* ASSUMED to be called from within the HA lock */ +/* */ +/****************************************************************************/ +static void +ips_putq_wait_tail(ips_wait_queue_t * queue, Scsi_Cmnd * item) +{ + METHOD_TRACE("ips_putq_wait_tail", 1); + + if (!item) + return; + + item->host_scribble = NULL; + + if (queue->tail) + queue->tail->host_scribble = (char *) item; + + queue->tail = item; + + if (!queue->head) + queue->head = item; + + queue->count++; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_removeq_wait_head */ +/* */ +/* Routine Description: */ +/* */ +/* Remove the head of the queue */ +/* */ +/* ASSUMED to be called from within the HA lock */ +/* */ +/****************************************************************************/ +static Scsi_Cmnd * +ips_removeq_wait_head(ips_wait_queue_t * queue) +{ + Scsi_Cmnd *item; + + METHOD_TRACE("ips_removeq_wait_head", 1); + + item = queue->head; + + if (!item) { + return (NULL); + } + + queue->head = (Scsi_Cmnd *) item->host_scribble; + item->host_scribble = NULL; + + if (queue->tail == item) + queue->tail = NULL; + + queue->count--; + + return (item); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_removeq_wait */ +/* */ +/* Routine Description: */ +/* */ +/* Remove an item from a queue */ +/* */ +/* ASSUMED to be called from within the HA lock */ +/* */ +/****************************************************************************/ +static Scsi_Cmnd * +ips_removeq_wait(ips_wait_queue_t * queue, Scsi_Cmnd * item) +{ + Scsi_Cmnd *p; + + METHOD_TRACE("ips_removeq_wait", 1); + + if (!item) + return (NULL); + + if (item == queue->head) { + return (ips_removeq_wait_head(queue)); + } + + p = queue->head; + + while ((p) && (item != (Scsi_Cmnd *) p->host_scribble)) + p = (Scsi_Cmnd *) p->host_scribble; + + if (p) { + /* found a match */ + p->host_scribble = item->host_scribble; + + if (!item->host_scribble) + queue->tail = p; + + item->host_scribble = NULL; + queue->count--; + + return (item); + } + + return (NULL); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_putq_copp_tail */ +/* */ +/* Routine Description: */ +/* */ +/* Add an item to the tail of the queue */ +/* */ +/* ASSUMED to be called from within the HA lock */ +/* */ +/****************************************************************************/ +static void +ips_putq_copp_tail(ips_copp_queue_t * queue, ips_copp_wait_item_t * item) +{ + METHOD_TRACE("ips_putq_copp_tail", 1); + + if (!item) + return; + + item->next = NULL; + + if (queue->tail) + queue->tail->next = item; + + queue->tail = item; + + if (!queue->head) + queue->head = item; + + queue->count++; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_removeq_copp_head */ +/* */ +/* Routine Description: */ +/* */ +/* Remove the head of the queue */ +/* */ +/* ASSUMED to be called from within the HA lock */ +/* */ +/****************************************************************************/ +static ips_copp_wait_item_t * +ips_removeq_copp_head(ips_copp_queue_t * queue) +{ + ips_copp_wait_item_t *item; + + METHOD_TRACE("ips_removeq_copp_head", 1); + + item = queue->head; + + if (!item) { + return (NULL); + } + + queue->head = item->next; + item->next = NULL; + + if (queue->tail == item) + queue->tail = NULL; + + queue->count--; + + return (item); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_removeq_copp */ +/* */ +/* Routine Description: */ +/* */ +/* Remove an item from a queue */ +/* */ +/* ASSUMED to be called from within the HA lock */ +/* */ +/****************************************************************************/ +static ips_copp_wait_item_t * +ips_removeq_copp(ips_copp_queue_t * queue, ips_copp_wait_item_t * item) +{ + ips_copp_wait_item_t *p; + + METHOD_TRACE("ips_removeq_copp", 1); + + if (!item) + return (NULL); + + if (item == queue->head) { + return (ips_removeq_copp_head(queue)); + } + + p = queue->head; + + while ((p) && (item != p->next)) + p = p->next; + + if (p) { + /* found a match */ + p->next = item->next; + + if (!item->next) + queue->tail = p; + + item->next = NULL; + queue->count--; + + return (item); + } + + return (NULL); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ipsintr_blocking */ +/* */ +/* Routine Description: */ +/* */ +/* Finalize an interrupt for internal commands */ +/* */ +/****************************************************************************/ +static void +ipsintr_blocking(ips_ha_t * ha, ips_scb_t * scb) +{ + METHOD_TRACE("ipsintr_blocking", 2); + + ips_freescb(ha, scb); + if ((ha->waitflag == TRUE) && (ha->cmd_in_progress == scb->cdb[0])) { + ha->waitflag = FALSE; + + return; + } +} + +/****************************************************************************/ +/* */ +/* Routine Name: ipsintr_done */ +/* */ +/* Routine Description: */ +/* */ +/* Finalize an interrupt for non-internal commands */ +/* */ +/****************************************************************************/ +static void +ipsintr_done(ips_ha_t * ha, ips_scb_t * scb) +{ + METHOD_TRACE("ipsintr_done", 2); + + if (!scb) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Spurious interrupt; scb NULL.\n"); + + return; + } + + if (scb->scsi_cmd == NULL) { + /* unexpected interrupt */ + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Spurious interrupt; scsi_cmd not set.\n"); + + return; + } + + ips_done(ha, scb); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_done */ +/* */ +/* Routine Description: */ +/* */ +/* Do housekeeping on completed commands */ +/* ASSUMED to be called form within the request lock */ +/****************************************************************************/ +static void +ips_done(ips_ha_t * ha, ips_scb_t * scb) +{ + int ret; + + METHOD_TRACE("ips_done", 1); + + if (!scb) + return; + + if ((scb->scsi_cmd) && (ips_is_passthru(scb->scsi_cmd))) { + ips_cleanup_passthru(ha, scb); + ha->num_ioctl--; + } else { + /* + * Check to see if this command had too much + * data and had to be broke up. If so, queue + * the rest of the data and continue. + */ + if ((scb->breakup) || (scb->sg_break)) { + /* we had a data breakup */ + scb->data_len = 0; + + if (scb->sg_count) { + /* S/G request */ + struct scatterlist *sg; + int ips_sg_index = 0; + int sg_dma_index; + + sg = scb->scsi_cmd->request_buffer; + + /* Spin forward to last dma chunk */ + sg_dma_index = scb->breakup; + + /* Take care of possible partial on last chunk */ + ips_fill_scb_sg_single(ha, + sg_dma_address(&sg + [sg_dma_index]), + scb, ips_sg_index++, + sg_dma_len(&sg + [sg_dma_index])); + + for (; sg_dma_index < scb->sg_count; + sg_dma_index++) { + if (ips_fill_scb_sg_single + (ha, + sg_dma_address(&sg[sg_dma_index]), + scb, ips_sg_index++, + sg_dma_len(&sg[sg_dma_index])) < 0) + break; + + } + + } else { + /* Non S/G Request */ + (void) ips_fill_scb_sg_single(ha, + scb-> + data_busaddr + + (scb->sg_break * + ha->max_xfer), + scb, 0, + scb->scsi_cmd-> + request_bufflen - + (scb->sg_break * + ha->max_xfer)); + } + + scb->dcdb.transfer_length = scb->data_len; + scb->dcdb.cmd_attribute |= + ips_command_direction[scb->scsi_cmd->cmnd[0]]; + + if (!(scb->dcdb.cmd_attribute & 0x3)) + scb->dcdb.transfer_length = 0; + + if (scb->data_len >= IPS_MAX_XFER) { + scb->dcdb.cmd_attribute |= IPS_TRANSFER64K; + scb->dcdb.transfer_length = 0; + } + + ret = ips_send_cmd(ha, scb); + + switch (ret) { + case IPS_FAILURE: + if (scb->scsi_cmd) { + scb->scsi_cmd->result = DID_ERROR << 16; + scb->scsi_cmd->scsi_done(scb->scsi_cmd); + } + + ips_freescb(ha, scb); + break; + case IPS_SUCCESS_IMM: + if (scb->scsi_cmd) { + scb->scsi_cmd->result = DID_ERROR << 16; + scb->scsi_cmd->scsi_done(scb->scsi_cmd); + } + + ips_freescb(ha, scb); + break; + default: + break; + } /* end case */ + + return; + } + } /* end if passthru */ + + if (scb->bus) { + ha->dcdb_active[scb->bus - 1] &= ~(1 << scb->target_id); + } + + scb->scsi_cmd->scsi_done(scb->scsi_cmd); + + ips_freescb(ha, scb); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_map_status */ +/* */ +/* Routine Description: */ +/* */ +/* Map Controller Error codes to Linux Error Codes */ +/* */ +/****************************************************************************/ +static int +ips_map_status(ips_ha_t * ha, ips_scb_t * scb, ips_stat_t * sp) +{ + int errcode; + int device_error; + uint32_t transfer_len; + IPS_DCDB_TABLE_TAPE *tapeDCDB; + + METHOD_TRACE("ips_map_status", 1); + + if (scb->bus) { + DEBUG_VAR(2, + "(%s%d) Physical device error (%d %d %d): %x %x, Sense Key: %x, ASC: %x, ASCQ: %x", + ips_name, ha->host_num, + scb->scsi_cmd->device->channel, + scb->scsi_cmd->device->id, scb->scsi_cmd->device->lun, + scb->basic_status, scb->extended_status, + scb->extended_status == + IPS_ERR_CKCOND ? scb->dcdb.sense_info[2] & 0xf : 0, + scb->extended_status == + IPS_ERR_CKCOND ? scb->dcdb.sense_info[12] : 0, + scb->extended_status == + IPS_ERR_CKCOND ? scb->dcdb.sense_info[13] : 0); + } + + /* default driver error */ + errcode = DID_ERROR; + device_error = 0; + + switch (scb->basic_status & IPS_GSC_STATUS_MASK) { + case IPS_CMD_TIMEOUT: + errcode = DID_TIME_OUT; + break; + + case IPS_INVAL_OPCO: + case IPS_INVAL_CMD_BLK: + case IPS_INVAL_PARM_BLK: + case IPS_LD_ERROR: + case IPS_CMD_CMPLT_WERROR: + break; + + case IPS_PHYS_DRV_ERROR: + switch (scb->extended_status) { + case IPS_ERR_SEL_TO: + if (scb->bus) + errcode = DID_NO_CONNECT; + + break; + + case IPS_ERR_OU_RUN: + if ((scb->cmd.dcdb.op_code == IPS_CMD_EXTENDED_DCDB) || + (scb->cmd.dcdb.op_code == + IPS_CMD_EXTENDED_DCDB_SG)) { + tapeDCDB = (IPS_DCDB_TABLE_TAPE *) & scb->dcdb; + transfer_len = tapeDCDB->transfer_length; + } else { + transfer_len = + (uint32_t) scb->dcdb.transfer_length; + } + + if ((scb->bus) && (transfer_len < scb->data_len)) { + /* Underrun - set default to no error */ + errcode = DID_OK; + + /* Restrict access to physical DASD */ + if ((scb->scsi_cmd->cmnd[0] == INQUIRY) && + ((((char *) scb->scsi_cmd-> + buffer)[0] & 0x1f) == TYPE_DISK)) { + /* underflow -- no error */ + /* restrict access to physical DASD */ + errcode = DID_TIME_OUT; + break; + } + } else + errcode = DID_ERROR; + + break; + + case IPS_ERR_RECOVERY: + /* don't fail recovered errors */ + if (scb->bus) + errcode = DID_OK; + + break; + + case IPS_ERR_HOST_RESET: + case IPS_ERR_DEV_RESET: + errcode = DID_RESET; + break; + + case IPS_ERR_CKCOND: + if (scb->bus) { + if ((scb->cmd.dcdb.op_code == + IPS_CMD_EXTENDED_DCDB) + || (scb->cmd.dcdb.op_code == + IPS_CMD_EXTENDED_DCDB_SG)) { + tapeDCDB = + (IPS_DCDB_TABLE_TAPE *) & scb->dcdb; + memcpy(scb->scsi_cmd->sense_buffer, + tapeDCDB->sense_info, + sizeof (scb->scsi_cmd-> + sense_buffer)); + } else { + memcpy(scb->scsi_cmd->sense_buffer, + scb->dcdb.sense_info, + sizeof (scb->scsi_cmd-> + sense_buffer)); + } + device_error = 2; /* check condition */ + } + + errcode = DID_OK; + + break; + + default: + errcode = DID_ERROR; + break; + + } /* end switch */ + } /* end switch */ + + scb->scsi_cmd->result = device_error | (errcode << 16); + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_send_wait */ +/* */ +/* Routine Description: */ +/* */ +/* Send a command to the controller and wait for it to return */ +/* */ +/* The FFDC Time Stamp use this function for the callback, but doesn't */ +/* actually need to wait. */ +/****************************************************************************/ +static int +ips_send_wait(ips_ha_t * ha, ips_scb_t * scb, int timeout, int intr) +{ + int ret; + + METHOD_TRACE("ips_send_wait", 1); + + if (intr != IPS_FFDC) { /* Won't be Waiting if this is a Time Stamp */ + ha->waitflag = TRUE; + ha->cmd_in_progress = scb->cdb[0]; + } + scb->callback = ipsintr_blocking; + ret = ips_send_cmd(ha, scb); + + if ((ret == IPS_FAILURE) || (ret == IPS_SUCCESS_IMM)) + return (ret); + + if (intr != IPS_FFDC) /* Don't Wait around if this is a Time Stamp */ + ret = ips_wait(ha, timeout, intr); + + return (ret); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_scmd_buf_write */ +/* */ +/* Routine Description: */ +/* Write data to Scsi_Cmnd request_buffer at proper offsets */ +/****************************************************************************/ +static void +ips_scmd_buf_write(Scsi_Cmnd * scmd, void *data, unsigned + int count) +{ + if (scmd->use_sg) { + int i; + unsigned int min_cnt, xfer_cnt; + char *cdata = (char *) data; + struct scatterlist *sg = scmd->request_buffer; + for (i = 0, xfer_cnt = 0; + (i < scmd->use_sg) && (xfer_cnt < count); i++) { + if (!IPS_SG_ADDRESS(&sg[i])) + return; + min_cnt = min(count - xfer_cnt, sg[i].length); + memcpy(IPS_SG_ADDRESS(&sg[i]), &cdata[xfer_cnt], + min_cnt); + xfer_cnt += min_cnt; + } + + } else { + unsigned int min_cnt = min(count, scmd->request_bufflen); + memcpy(scmd->request_buffer, data, min_cnt); + } +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_scmd_buf_read */ +/* */ +/* Routine Description: */ +/* Copy data from a Scsi_Cmnd to a new, linear buffer */ +/****************************************************************************/ +static void +ips_scmd_buf_read(Scsi_Cmnd * scmd, void *data, unsigned + int count) +{ + if (scmd->use_sg) { + int i; + unsigned int min_cnt, xfer_cnt; + char *cdata = (char *) data; + struct scatterlist *sg = scmd->request_buffer; + for (i = 0, xfer_cnt = 0; + (i < scmd->use_sg) && (xfer_cnt < count); i++) { + if (!IPS_SG_ADDRESS(&sg[i])) + return; + min_cnt = min(count - xfer_cnt, sg[i].length); + memcpy(&cdata[xfer_cnt], IPS_SG_ADDRESS(&sg[i]), + min_cnt); + xfer_cnt += min_cnt; + } + + } else { + unsigned int min_cnt = min(count, scmd->request_bufflen); + memcpy(data, scmd->request_buffer, min_cnt); + } +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_send_cmd */ +/* */ +/* Routine Description: */ +/* */ +/* Map SCSI commands to ServeRAID commands for logical drives */ +/* */ +/****************************************************************************/ +static int +ips_send_cmd(ips_ha_t * ha, ips_scb_t * scb) +{ + int ret; + char *sp; + int device_error; + IPS_DCDB_TABLE_TAPE *tapeDCDB; + int TimeOut; + + METHOD_TRACE("ips_send_cmd", 1); + + ret = IPS_SUCCESS; + + if (!scb->scsi_cmd) { + /* internal command */ + + if (scb->bus > 0) { + /* Controller commands can't be issued */ + /* to real devices -- fail them */ + if ((ha->waitflag == TRUE) && + (ha->cmd_in_progress == scb->cdb[0])) { + ha->waitflag = FALSE; + } + + return (1); + } + } else if ((scb->bus == 0) && (!ips_is_passthru(scb->scsi_cmd))) { + /* command to logical bus -- interpret */ + ret = IPS_SUCCESS_IMM; + + switch (scb->scsi_cmd->cmnd[0]) { + case ALLOW_MEDIUM_REMOVAL: + case REZERO_UNIT: + case ERASE: + case WRITE_FILEMARKS: + case SPACE: + scb->scsi_cmd->result = DID_ERROR << 16; + break; + + case START_STOP: + scb->scsi_cmd->result = DID_OK << 16; + + case TEST_UNIT_READY: + case INQUIRY: + if (scb->target_id == IPS_ADAPTER_ID) { + /* + * Either we have a TUR + * or we have a SCSI inquiry + */ + if (scb->scsi_cmd->cmnd[0] == TEST_UNIT_READY) + scb->scsi_cmd->result = DID_OK << 16; + + if (scb->scsi_cmd->cmnd[0] == INQUIRY) { + IPS_SCSI_INQ_DATA inquiry; + + memset(&inquiry, 0, + sizeof (IPS_SCSI_INQ_DATA)); + + inquiry.DeviceType = + IPS_SCSI_INQ_TYPE_PROCESSOR; + inquiry.DeviceTypeQualifier = + IPS_SCSI_INQ_LU_CONNECTED; + inquiry.Version = IPS_SCSI_INQ_REV2; + inquiry.ResponseDataFormat = + IPS_SCSI_INQ_RD_REV2; + inquiry.AdditionalLength = 31; + inquiry.Flags[0] = + IPS_SCSI_INQ_Address16; + inquiry.Flags[1] = + IPS_SCSI_INQ_WBus16 | + IPS_SCSI_INQ_Sync; + strncpy(inquiry.VendorId, "IBM ", + 8); + strncpy(inquiry.ProductId, + "SERVERAID ", 16); + strncpy(inquiry.ProductRevisionLevel, + "1.00", 4); + + ips_scmd_buf_write(scb->scsi_cmd, + &inquiry, + sizeof (inquiry)); + + scb->scsi_cmd->result = DID_OK << 16; + } + } else { + scb->cmd.logical_info.op_code = IPS_CMD_GET_LD_INFO; + scb->cmd.logical_info.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.logical_info.reserved = 0; + scb->cmd.logical_info.reserved2 = 0; + scb->data_len = sizeof (IPS_LD_INFO); + scb->data_busaddr = ha->logical_drive_info_dma_addr; + scb->flags = 0; + scb->cmd.logical_info.buffer_addr = scb->data_busaddr; + ret = IPS_SUCCESS; + } + + break; + + case REQUEST_SENSE: + ips_reqsen(ha, scb); + scb->scsi_cmd->result = DID_OK << 16; + break; + + case READ_6: + case WRITE_6: + if (!scb->sg_len) { + scb->cmd.basic_io.op_code = + (scb->scsi_cmd->cmnd[0] == + READ_6) ? IPS_CMD_READ : IPS_CMD_WRITE; + scb->cmd.basic_io.enhanced_sg = 0; + scb->cmd.basic_io.sg_addr = + cpu_to_le32(scb->data_busaddr); + } else { + scb->cmd.basic_io.op_code = + (scb->scsi_cmd->cmnd[0] == + READ_6) ? IPS_CMD_READ_SG : + IPS_CMD_WRITE_SG; + scb->cmd.basic_io.enhanced_sg = + IPS_USE_ENH_SGLIST(ha) ? 0xFF : 0; + scb->cmd.basic_io.sg_addr = + cpu_to_le32(scb->sg_busaddr); + } + + scb->cmd.basic_io.segment_4G = 0; + scb->cmd.basic_io.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.basic_io.log_drv = scb->target_id; + scb->cmd.basic_io.sg_count = scb->sg_len; + + if (scb->cmd.basic_io.lba) + scb->cmd.basic_io.lba = + cpu_to_le32(le32_to_cpu + (scb->cmd.basic_io.lba) + + le16_to_cpu(scb->cmd.basic_io. + sector_count)); + else + scb->cmd.basic_io.lba = + (((scb->scsi_cmd-> + cmnd[1] & 0x1f) << 16) | (scb->scsi_cmd-> + cmnd[2] << 8) | + (scb->scsi_cmd->cmnd[3])); + + scb->cmd.basic_io.sector_count = + cpu_to_le16(scb->data_len / IPS_BLKSIZE); + + if (le16_to_cpu(scb->cmd.basic_io.sector_count) == 0) + scb->cmd.basic_io.sector_count = + cpu_to_le16(256); + + ret = IPS_SUCCESS; + break; + + case READ_10: + case WRITE_10: + if (!scb->sg_len) { + scb->cmd.basic_io.op_code = + (scb->scsi_cmd->cmnd[0] == + READ_10) ? IPS_CMD_READ : IPS_CMD_WRITE; + scb->cmd.basic_io.enhanced_sg = 0; + scb->cmd.basic_io.sg_addr = + cpu_to_le32(scb->data_busaddr); + } else { + scb->cmd.basic_io.op_code = + (scb->scsi_cmd->cmnd[0] == + READ_10) ? IPS_CMD_READ_SG : + IPS_CMD_WRITE_SG; + scb->cmd.basic_io.enhanced_sg = + IPS_USE_ENH_SGLIST(ha) ? 0xFF : 0; + scb->cmd.basic_io.sg_addr = + cpu_to_le32(scb->sg_busaddr); + } + + scb->cmd.basic_io.segment_4G = 0; + scb->cmd.basic_io.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.basic_io.log_drv = scb->target_id; + scb->cmd.basic_io.sg_count = scb->sg_len; + + if (scb->cmd.basic_io.lba) + scb->cmd.basic_io.lba = + cpu_to_le32(le32_to_cpu + (scb->cmd.basic_io.lba) + + le16_to_cpu(scb->cmd.basic_io. + sector_count)); + else + scb->cmd.basic_io.lba = + ((scb->scsi_cmd->cmnd[2] << 24) | (scb-> + scsi_cmd-> + cmnd[3] + << 16) | + (scb->scsi_cmd->cmnd[4] << 8) | scb-> + scsi_cmd->cmnd[5]); + + scb->cmd.basic_io.sector_count = + cpu_to_le16(scb->data_len / IPS_BLKSIZE); + + if (cpu_to_le16(scb->cmd.basic_io.sector_count) == 0) { + /* + * This is a null condition + * we don't have to do anything + * so just return + */ + scb->scsi_cmd->result = DID_OK << 16; + } else + ret = IPS_SUCCESS; + + break; + + case RESERVE: + case RELEASE: + scb->scsi_cmd->result = DID_OK << 16; + break; + + case MODE_SENSE: + scb->cmd.basic_io.op_code = IPS_CMD_ENQUIRY; + scb->cmd.basic_io.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.basic_io.segment_4G = 0; + scb->cmd.basic_io.enhanced_sg = 0; + scb->data_len = sizeof (*ha->enq); + scb->cmd.basic_io.sg_addr = ha->enq_busaddr; + ret = IPS_SUCCESS; + break; + + case READ_CAPACITY: + scb->cmd.logical_info.op_code = IPS_CMD_GET_LD_INFO; + scb->cmd.logical_info.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.logical_info.reserved = 0; + scb->cmd.logical_info.reserved2 = 0; + scb->cmd.logical_info.reserved3 = 0; + scb->data_len = sizeof (IPS_LD_INFO); + scb->data_busaddr = ha->logical_drive_info_dma_addr; + scb->flags = 0; + scb->cmd.logical_info.buffer_addr = scb->data_busaddr; + ret = IPS_SUCCESS; + break; + + case SEND_DIAGNOSTIC: + case REASSIGN_BLOCKS: + case FORMAT_UNIT: + case SEEK_10: + case VERIFY: + case READ_DEFECT_DATA: + case READ_BUFFER: + case WRITE_BUFFER: + scb->scsi_cmd->result = DID_OK << 16; + break; + + default: + /* Set the Return Info to appear like the Command was */ + /* attempted, a Check Condition occurred, and Sense */ + /* Data indicating an Invalid CDB OpCode is returned. */ + sp = (char *) scb->scsi_cmd->sense_buffer; + memset(sp, 0, sizeof (scb->scsi_cmd->sense_buffer)); + + sp[0] = 0x70; /* Error Code */ + sp[2] = ILLEGAL_REQUEST; /* Sense Key 5 Illegal Req. */ + sp[7] = 0x0A; /* Additional Sense Length */ + sp[12] = 0x20; /* ASC = Invalid OpCode */ + sp[13] = 0x00; /* ASCQ */ + + device_error = 2; /* Indicate Check Condition */ + scb->scsi_cmd->result = device_error | (DID_OK << 16); + break; + } /* end switch */ + } + /* end if */ + if (ret == IPS_SUCCESS_IMM) + return (ret); + + /* setup DCDB */ + if (scb->bus > 0) { + + /* If we already know the Device is Not there, no need to attempt a Command */ + /* This also protects an NT FailOver Controller from getting CDB's sent to it */ + if (ha->conf->dev[scb->bus - 1][scb->target_id].ucState == 0) { + scb->scsi_cmd->result = DID_NO_CONNECT << 16; + return (IPS_SUCCESS_IMM); + } + + ha->dcdb_active[scb->bus - 1] |= (1 << scb->target_id); + scb->cmd.dcdb.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.dcdb.dcdb_address = cpu_to_le32(scb->scb_busaddr + + (unsigned long) &scb-> + dcdb - + (unsigned long) scb); + scb->cmd.dcdb.reserved = 0; + scb->cmd.dcdb.reserved2 = 0; + scb->cmd.dcdb.reserved3 = 0; + scb->cmd.dcdb.segment_4G = 0; + scb->cmd.dcdb.enhanced_sg = 0; + + TimeOut = scb->scsi_cmd->timeout_per_command; + + if (ha->subsys->param[4] & 0x00100000) { /* If NEW Tape DCDB is Supported */ + if (!scb->sg_len) { + scb->cmd.dcdb.op_code = IPS_CMD_EXTENDED_DCDB; + } else { + scb->cmd.dcdb.op_code = + IPS_CMD_EXTENDED_DCDB_SG; + scb->cmd.dcdb.enhanced_sg = + IPS_USE_ENH_SGLIST(ha) ? 0xFF : 0; + } + + tapeDCDB = (IPS_DCDB_TABLE_TAPE *) & scb->dcdb; /* Use Same Data Area as Old DCDB Struct */ + tapeDCDB->device_address = + ((scb->bus - 1) << 4) | scb->target_id; + tapeDCDB->cmd_attribute |= IPS_DISCONNECT_ALLOWED; + tapeDCDB->cmd_attribute &= ~IPS_TRANSFER64K; /* Always Turn OFF 64K Size Flag */ + + if (TimeOut) { + if (TimeOut < (10 * HZ)) + tapeDCDB->cmd_attribute |= IPS_TIMEOUT10; /* TimeOut is 10 Seconds */ + else if (TimeOut < (60 * HZ)) + tapeDCDB->cmd_attribute |= IPS_TIMEOUT60; /* TimeOut is 60 Seconds */ + else if (TimeOut < (1200 * HZ)) + tapeDCDB->cmd_attribute |= IPS_TIMEOUT20M; /* TimeOut is 20 Minutes */ + } + + tapeDCDB->cdb_length = scb->scsi_cmd->cmd_len; + tapeDCDB->reserved_for_LUN = 0; + tapeDCDB->transfer_length = scb->data_len; + if (scb->cmd.dcdb.op_code == IPS_CMD_EXTENDED_DCDB_SG) + tapeDCDB->buffer_pointer = + cpu_to_le32(scb->sg_busaddr); + else + tapeDCDB->buffer_pointer = + cpu_to_le32(scb->data_busaddr); + tapeDCDB->sg_count = scb->sg_len; + tapeDCDB->sense_length = sizeof (tapeDCDB->sense_info); + tapeDCDB->scsi_status = 0; + tapeDCDB->reserved = 0; + memcpy(tapeDCDB->scsi_cdb, scb->scsi_cmd->cmnd, + scb->scsi_cmd->cmd_len); + } else { + if (!scb->sg_len) { + scb->cmd.dcdb.op_code = IPS_CMD_DCDB; + } else { + scb->cmd.dcdb.op_code = IPS_CMD_DCDB_SG; + scb->cmd.dcdb.enhanced_sg = + IPS_USE_ENH_SGLIST(ha) ? 0xFF : 0; + } + + scb->dcdb.device_address = + ((scb->bus - 1) << 4) | scb->target_id; + scb->dcdb.cmd_attribute |= IPS_DISCONNECT_ALLOWED; + + if (TimeOut) { + if (TimeOut < (10 * HZ)) + scb->dcdb.cmd_attribute |= IPS_TIMEOUT10; /* TimeOut is 10 Seconds */ + else if (TimeOut < (60 * HZ)) + scb->dcdb.cmd_attribute |= IPS_TIMEOUT60; /* TimeOut is 60 Seconds */ + else if (TimeOut < (1200 * HZ)) + scb->dcdb.cmd_attribute |= IPS_TIMEOUT20M; /* TimeOut is 20 Minutes */ + } + + scb->dcdb.transfer_length = scb->data_len; + if (scb->dcdb.cmd_attribute & IPS_TRANSFER64K) + scb->dcdb.transfer_length = 0; + if (scb->cmd.dcdb.op_code == IPS_CMD_DCDB_SG) + scb->dcdb.buffer_pointer = + cpu_to_le32(scb->sg_busaddr); + else + scb->dcdb.buffer_pointer = + cpu_to_le32(scb->data_busaddr); + scb->dcdb.cdb_length = scb->scsi_cmd->cmd_len; + scb->dcdb.sense_length = sizeof (scb->dcdb.sense_info); + scb->dcdb.sg_count = scb->sg_len; + scb->dcdb.reserved = 0; + memcpy(scb->dcdb.scsi_cdb, scb->scsi_cmd->cmnd, + scb->scsi_cmd->cmd_len); + scb->dcdb.scsi_status = 0; + scb->dcdb.reserved2[0] = 0; + scb->dcdb.reserved2[1] = 0; + scb->dcdb.reserved2[2] = 0; + } + } + + return ((*ha->func.issue) (ha, scb)); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_chk_status */ +/* */ +/* Routine Description: */ +/* */ +/* Check the status of commands to logical drives */ +/* Assumed to be called with the HA lock */ +/****************************************************************************/ +static void +ips_chkstatus(ips_ha_t * ha, IPS_STATUS * pstatus) +{ + ips_scb_t *scb; + ips_stat_t *sp; + uint8_t basic_status; + uint8_t ext_status; + int errcode; + + METHOD_TRACE("ips_chkstatus", 1); + + scb = &ha->scbs[pstatus->fields.command_id]; + scb->basic_status = basic_status = + pstatus->fields.basic_status & IPS_BASIC_STATUS_MASK; + scb->extended_status = ext_status = pstatus->fields.extended_status; + + sp = &ha->sp; + sp->residue_len = 0; + sp->scb_addr = (void *) scb; + + /* Remove the item from the active queue */ + ips_removeq_scb(&ha->scb_activelist, scb); + + if (!scb->scsi_cmd) + /* internal commands are handled in do_ipsintr */ + return; + + DEBUG_VAR(2, "(%s%d) ips_chkstatus: cmd 0x%X id %d (%d %d %d)", + ips_name, + ha->host_num, + scb->cdb[0], + scb->cmd.basic_io.command_id, + scb->bus, scb->target_id, scb->lun); + + if ((scb->scsi_cmd) && (ips_is_passthru(scb->scsi_cmd))) + /* passthru - just returns the raw result */ + return; + + errcode = DID_OK; + + if (((basic_status & IPS_GSC_STATUS_MASK) == IPS_CMD_SUCCESS) || + ((basic_status & IPS_GSC_STATUS_MASK) == IPS_CMD_RECOVERED_ERROR)) { + + if (scb->bus == 0) { + if ((basic_status & IPS_GSC_STATUS_MASK) == + IPS_CMD_RECOVERED_ERROR) { + DEBUG_VAR(1, + "(%s%d) Recovered Logical Drive Error OpCode: %x, BSB: %x, ESB: %x", + ips_name, ha->host_num, + scb->cmd.basic_io.op_code, + basic_status, ext_status); + } + + switch (scb->scsi_cmd->cmnd[0]) { + case ALLOW_MEDIUM_REMOVAL: + case REZERO_UNIT: + case ERASE: + case WRITE_FILEMARKS: + case SPACE: + errcode = DID_ERROR; + break; + + case START_STOP: + break; + + case TEST_UNIT_READY: + if (!ips_online(ha, scb)) { + errcode = DID_TIME_OUT; + } + break; + + case INQUIRY: + if (ips_online(ha, scb)) { + ips_inquiry(ha, scb); + } else { + errcode = DID_TIME_OUT; + } + break; + + case REQUEST_SENSE: + ips_reqsen(ha, scb); + break; + + case READ_6: + case WRITE_6: + case READ_10: + case WRITE_10: + case RESERVE: + case RELEASE: + break; + + case MODE_SENSE: + if (!ips_online(ha, scb) + || !ips_msense(ha, scb)) { + errcode = DID_ERROR; + } + break; + + case READ_CAPACITY: + if (ips_online(ha, scb)) + ips_rdcap(ha, scb); + else { + errcode = DID_TIME_OUT; + } + break; + + case SEND_DIAGNOSTIC: + case REASSIGN_BLOCKS: + break; + + case FORMAT_UNIT: + errcode = DID_ERROR; + break; + + case SEEK_10: + case VERIFY: + case READ_DEFECT_DATA: + case READ_BUFFER: + case WRITE_BUFFER: + break; + + default: + errcode = DID_ERROR; + } /* end switch */ + + scb->scsi_cmd->result = errcode << 16; + } else { /* bus == 0 */ + /* restrict access to physical drives */ + if ((scb->scsi_cmd->cmnd[0] == INQUIRY) && + ((((char *) scb->scsi_cmd->buffer)[0] & 0x1f) == + TYPE_DISK)) { + + scb->scsi_cmd->result = DID_TIME_OUT << 16; + } + } /* else */ + } else { /* recovered error / success */ + if (scb->bus == 0) { + DEBUG_VAR(1, + "(%s%d) Unrecovered Logical Drive Error OpCode: %x, BSB: %x, ESB: %x", + ips_name, ha->host_num, + scb->cmd.basic_io.op_code, basic_status, + ext_status); + } + + ips_map_status(ha, scb, sp); + } /* else */ +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_online */ +/* */ +/* Routine Description: */ +/* */ +/* Determine if a logical drive is online */ +/* */ +/****************************************************************************/ +static int +ips_online(ips_ha_t * ha, ips_scb_t * scb) +{ + METHOD_TRACE("ips_online", 1); + + if (scb->target_id >= IPS_MAX_LD) + return (0); + + if ((scb->basic_status & IPS_GSC_STATUS_MASK) > 1) { + memset(ha->logical_drive_info, 0, sizeof (IPS_LD_INFO)); + return (0); + } + + if (ha->logical_drive_info->drive_info[scb->target_id].state != + IPS_LD_OFFLINE + && ha->logical_drive_info->drive_info[scb->target_id].state != + IPS_LD_FREE + && ha->logical_drive_info->drive_info[scb->target_id].state != + IPS_LD_CRS + && ha->logical_drive_info->drive_info[scb->target_id].state != + IPS_LD_SYS) + return (1); + else + return (0); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_inquiry */ +/* */ +/* Routine Description: */ +/* */ +/* Simulate an inquiry command to a logical drive */ +/* */ +/****************************************************************************/ +static int +ips_inquiry(ips_ha_t * ha, ips_scb_t * scb) +{ + IPS_SCSI_INQ_DATA inquiry; + + METHOD_TRACE("ips_inquiry", 1); + + memset(&inquiry, 0, sizeof (IPS_SCSI_INQ_DATA)); + + inquiry.DeviceType = IPS_SCSI_INQ_TYPE_DASD; + inquiry.DeviceTypeQualifier = IPS_SCSI_INQ_LU_CONNECTED; + inquiry.Version = IPS_SCSI_INQ_REV2; + inquiry.ResponseDataFormat = IPS_SCSI_INQ_RD_REV2; + inquiry.AdditionalLength = 31; + inquiry.Flags[0] = IPS_SCSI_INQ_Address16; + inquiry.Flags[1] = + IPS_SCSI_INQ_WBus16 | IPS_SCSI_INQ_Sync | IPS_SCSI_INQ_CmdQue; + strncpy(inquiry.VendorId, "IBM ", 8); + strncpy(inquiry.ProductId, "SERVERAID ", 16); + strncpy(inquiry.ProductRevisionLevel, "1.00", 4); + + ips_scmd_buf_write(scb->scsi_cmd, &inquiry, sizeof (inquiry)); + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_rdcap */ +/* */ +/* Routine Description: */ +/* */ +/* Simulate a read capacity command to a logical drive */ +/* */ +/****************************************************************************/ +static int +ips_rdcap(ips_ha_t * ha, ips_scb_t * scb) +{ + IPS_SCSI_CAPACITY cap; + + METHOD_TRACE("ips_rdcap", 1); + + if (scb->scsi_cmd->bufflen < 8) + return (0); + + cap.lba = + cpu_to_be32(le32_to_cpu + (ha->logical_drive_info-> + drive_info[scb->target_id].sector_count) - 1); + cap.len = cpu_to_be32((uint32_t) IPS_BLKSIZE); + + ips_scmd_buf_write(scb->scsi_cmd, &cap, sizeof (cap)); + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_msense */ +/* */ +/* Routine Description: */ +/* */ +/* Simulate a mode sense command to a logical drive */ +/* */ +/****************************************************************************/ +static int +ips_msense(ips_ha_t * ha, ips_scb_t * scb) +{ + uint16_t heads; + uint16_t sectors; + uint32_t cylinders; + IPS_SCSI_MODE_PAGE_DATA mdata; + + METHOD_TRACE("ips_msense", 1); + + if (le32_to_cpu(ha->enq->ulDriveSize[scb->target_id]) > 0x400000 && + (ha->enq->ucMiscFlag & 0x8) == 0) { + heads = IPS_NORM_HEADS; + sectors = IPS_NORM_SECTORS; + } else { + heads = IPS_COMP_HEADS; + sectors = IPS_COMP_SECTORS; + } + + cylinders = + (le32_to_cpu(ha->enq->ulDriveSize[scb->target_id]) - + 1) / (heads * sectors); + + memset(&mdata, 0, sizeof (IPS_SCSI_MODE_PAGE_DATA)); + + mdata.hdr.BlockDescLength = 8; + + switch (scb->scsi_cmd->cmnd[2] & 0x3f) { + case 0x03: /* page 3 */ + mdata.pdata.pg3.PageCode = 3; + mdata.pdata.pg3.PageLength = sizeof (IPS_SCSI_MODE_PAGE3); + mdata.hdr.DataLength = + 3 + mdata.hdr.BlockDescLength + mdata.pdata.pg3.PageLength; + mdata.pdata.pg3.TracksPerZone = 0; + mdata.pdata.pg3.AltSectorsPerZone = 0; + mdata.pdata.pg3.AltTracksPerZone = 0; + mdata.pdata.pg3.AltTracksPerVolume = 0; + mdata.pdata.pg3.SectorsPerTrack = cpu_to_be16(sectors); + mdata.pdata.pg3.BytesPerSector = cpu_to_be16(IPS_BLKSIZE); + mdata.pdata.pg3.Interleave = cpu_to_be16(1); + mdata.pdata.pg3.TrackSkew = 0; + mdata.pdata.pg3.CylinderSkew = 0; + mdata.pdata.pg3.flags = IPS_SCSI_MP3_SoftSector; + break; + + case 0x4: + mdata.pdata.pg4.PageCode = 4; + mdata.pdata.pg4.PageLength = sizeof (IPS_SCSI_MODE_PAGE4); + mdata.hdr.DataLength = + 3 + mdata.hdr.BlockDescLength + mdata.pdata.pg4.PageLength; + mdata.pdata.pg4.CylindersHigh = + cpu_to_be16((cylinders >> 8) & 0xFFFF); + mdata.pdata.pg4.CylindersLow = (cylinders & 0xFF); + mdata.pdata.pg4.Heads = heads; + mdata.pdata.pg4.WritePrecompHigh = 0; + mdata.pdata.pg4.WritePrecompLow = 0; + mdata.pdata.pg4.ReducedWriteCurrentHigh = 0; + mdata.pdata.pg4.ReducedWriteCurrentLow = 0; + mdata.pdata.pg4.StepRate = cpu_to_be16(1); + mdata.pdata.pg4.LandingZoneHigh = 0; + mdata.pdata.pg4.LandingZoneLow = 0; + mdata.pdata.pg4.flags = 0; + mdata.pdata.pg4.RotationalOffset = 0; + mdata.pdata.pg4.MediumRotationRate = 0; + break; + case 0x8: + mdata.pdata.pg8.PageCode = 8; + mdata.pdata.pg8.PageLength = sizeof (IPS_SCSI_MODE_PAGE8); + mdata.hdr.DataLength = + 3 + mdata.hdr.BlockDescLength + mdata.pdata.pg8.PageLength; + /* everything else is left set to 0 */ + break; + + default: + return (0); + } /* end switch */ + + ips_scmd_buf_write(scb->scsi_cmd, &mdata, sizeof (mdata)); + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_reqsen */ +/* */ +/* Routine Description: */ +/* */ +/* Simulate a request sense command to a logical drive */ +/* */ +/****************************************************************************/ +static int +ips_reqsen(ips_ha_t * ha, ips_scb_t * scb) +{ + IPS_SCSI_REQSEN reqsen; + + METHOD_TRACE("ips_reqsen", 1); + + memset(&reqsen, 0, sizeof (IPS_SCSI_REQSEN)); + + reqsen.ResponseCode = + IPS_SCSI_REQSEN_VALID | IPS_SCSI_REQSEN_CURRENT_ERR; + reqsen.AdditionalLength = 10; + reqsen.AdditionalSenseCode = IPS_SCSI_REQSEN_NO_SENSE; + reqsen.AdditionalSenseCodeQual = IPS_SCSI_REQSEN_NO_SENSE; + + ips_scmd_buf_write(scb->scsi_cmd, &reqsen, sizeof (reqsen)); + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_free */ +/* */ +/* Routine Description: */ +/* */ +/* Free any allocated space for this controller */ +/* */ +/****************************************************************************/ +static void +ips_free(ips_ha_t * ha) +{ + + METHOD_TRACE("ips_free", 1); + + if (ha) { + if (ha->enq) { + pci_free_consistent(ha->pcidev, sizeof(IPS_ENQ), + ha->enq, ha->enq_busaddr); + ha->enq = NULL; + } + + if (ha->conf) { + kfree(ha->conf); + ha->conf = NULL; + } + + if (ha->adapt) { + pci_free_consistent(ha->pcidev, + sizeof (IPS_ADAPTER) + + sizeof (IPS_IO_CMD), ha->adapt, + ha->adapt->hw_status_start); + ha->adapt = NULL; + } + + if (ha->logical_drive_info) { + pci_free_consistent(ha->pcidev, + sizeof (IPS_LD_INFO), + ha->logical_drive_info, + ha->logical_drive_info_dma_addr); + ha->logical_drive_info = NULL; + } + + if (ha->nvram) { + kfree(ha->nvram); + ha->nvram = NULL; + } + + if (ha->subsys) { + kfree(ha->subsys); + ha->subsys = NULL; + } + + if (ha->ioctl_data) { + pci_free_consistent(ha->pcidev, ha->ioctl_len, + ha->ioctl_data, ha->ioctl_busaddr); + ha->ioctl_data = NULL; + ha->ioctl_datasize = 0; + ha->ioctl_len = 0; + } + ips_deallocatescbs(ha, ha->max_cmds); + + /* free memory mapped (if applicable) */ + if (ha->mem_ptr) { + iounmap(ha->ioremap_ptr); + ha->ioremap_ptr = NULL; + ha->mem_ptr = NULL; + } + + if (ha->mem_addr) + release_mem_region(ha->mem_addr, ha->mem_len); + ha->mem_addr = 0; + + } +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_deallocatescbs */ +/* */ +/* Routine Description: */ +/* */ +/* Free the command blocks */ +/* */ +/****************************************************************************/ +static int +ips_deallocatescbs(ips_ha_t * ha, int cmds) +{ + if (ha->scbs) { + pci_free_consistent(ha->pcidev, + IPS_SGLIST_SIZE(ha) * IPS_MAX_SG * cmds, + ha->scbs->sg_list.list, + ha->scbs->sg_busaddr); + pci_free_consistent(ha->pcidev, sizeof (ips_scb_t) * cmds, + ha->scbs, ha->scbs->scb_busaddr); + ha->scbs = NULL; + } /* end if */ + return 1; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_allocatescbs */ +/* */ +/* Routine Description: */ +/* */ +/* Allocate the command blocks */ +/* */ +/****************************************************************************/ +static int +ips_allocatescbs(ips_ha_t * ha) +{ + ips_scb_t *scb_p; + IPS_SG_LIST ips_sg; + int i; + dma_addr_t command_dma, sg_dma; + + METHOD_TRACE("ips_allocatescbs", 1); + + /* Allocate memory for the SCBs */ + ha->scbs = + pci_alloc_consistent(ha->pcidev, ha->max_cmds * sizeof (ips_scb_t), + &command_dma); + if (ha->scbs == NULL) + return 0; + ips_sg.list = + pci_alloc_consistent(ha->pcidev, + IPS_SGLIST_SIZE(ha) * IPS_MAX_SG * + ha->max_cmds, &sg_dma); + if (ips_sg.list == NULL) { + pci_free_consistent(ha->pcidev, + ha->max_cmds * sizeof (ips_scb_t), ha->scbs, + command_dma); + return 0; + } + + memset(ha->scbs, 0, ha->max_cmds * sizeof (ips_scb_t)); + + for (i = 0; i < ha->max_cmds; i++) { + scb_p = &ha->scbs[i]; + scb_p->scb_busaddr = command_dma + sizeof (ips_scb_t) * i; + /* set up S/G list */ + if (IPS_USE_ENH_SGLIST(ha)) { + scb_p->sg_list.enh_list = + ips_sg.enh_list + i * IPS_MAX_SG; + scb_p->sg_busaddr = + sg_dma + IPS_SGLIST_SIZE(ha) * IPS_MAX_SG * i; + } else { + scb_p->sg_list.std_list = + ips_sg.std_list + i * IPS_MAX_SG; + scb_p->sg_busaddr = + sg_dma + IPS_SGLIST_SIZE(ha) * IPS_MAX_SG * i; + } + + /* add to the free list */ + if (i < ha->max_cmds - 1) { + scb_p->q_next = ha->scb_freelist; + ha->scb_freelist = scb_p; + } + } + + /* success */ + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_init_scb */ +/* */ +/* Routine Description: */ +/* */ +/* Initialize a CCB to default values */ +/* */ +/****************************************************************************/ +static void +ips_init_scb(ips_ha_t * ha, ips_scb_t * scb) +{ + IPS_SG_LIST sg_list; + uint32_t cmd_busaddr, sg_busaddr; + METHOD_TRACE("ips_init_scb", 1); + + if (scb == NULL) + return; + + sg_list.list = scb->sg_list.list; + cmd_busaddr = scb->scb_busaddr; + sg_busaddr = scb->sg_busaddr; + /* zero fill */ + memset(scb, 0, sizeof (ips_scb_t)); + memset(ha->dummy, 0, sizeof (IPS_IO_CMD)); + + /* Initialize dummy command bucket */ + ha->dummy->op_code = 0xFF; + ha->dummy->ccsar = cpu_to_le32(ha->adapt->hw_status_start + + sizeof (IPS_ADAPTER)); + ha->dummy->command_id = IPS_MAX_CMDS; + + /* set bus address of scb */ + scb->scb_busaddr = cmd_busaddr; + scb->sg_busaddr = sg_busaddr; + scb->sg_list.list = sg_list.list; + + /* Neptune Fix */ + scb->cmd.basic_io.cccr = cpu_to_le32((uint32_t) IPS_BIT_ILE); + scb->cmd.basic_io.ccsar = cpu_to_le32(ha->adapt->hw_status_start + + sizeof (IPS_ADAPTER)); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_get_scb */ +/* */ +/* Routine Description: */ +/* */ +/* Initialize a CCB to default values */ +/* */ +/* ASSUMED to be callled from within a lock */ +/* */ +/****************************************************************************/ +static ips_scb_t * +ips_getscb(ips_ha_t * ha) +{ + ips_scb_t *scb; + + METHOD_TRACE("ips_getscb", 1); + + if ((scb = ha->scb_freelist) == NULL) { + + return (NULL); + } + + ha->scb_freelist = scb->q_next; + scb->flags = 0; + scb->q_next = NULL; + + ips_init_scb(ha, scb); + + return (scb); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_free_scb */ +/* */ +/* Routine Description: */ +/* */ +/* Return an unused CCB back to the free list */ +/* */ +/* ASSUMED to be called from within a lock */ +/* */ +/****************************************************************************/ +static void +ips_freescb(ips_ha_t * ha, ips_scb_t * scb) +{ + + METHOD_TRACE("ips_freescb", 1); + if (scb->flags & IPS_SCB_MAP_SG) + pci_unmap_sg(ha->pcidev, scb->scsi_cmd->request_buffer, + scb->scsi_cmd->use_sg, IPS_DMA_DIR(scb)); + else if (scb->flags & IPS_SCB_MAP_SINGLE) + pci_unmap_single(ha->pcidev, scb->data_busaddr, scb->data_len, + IPS_DMA_DIR(scb)); + + /* check to make sure this is not our "special" scb */ + if (IPS_COMMAND_ID(ha, scb) < (ha->max_cmds - 1)) { + scb->q_next = ha->scb_freelist; + ha->scb_freelist = scb; + } +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_isinit_copperhead */ +/* */ +/* Routine Description: */ +/* */ +/* Is controller initialized ? */ +/* */ +/****************************************************************************/ +static int +ips_isinit_copperhead(ips_ha_t * ha) +{ + uint8_t scpr; + uint8_t isr; + + METHOD_TRACE("ips_isinit_copperhead", 1); + + isr = inb(ha->io_addr + IPS_REG_HISR); + scpr = inb(ha->io_addr + IPS_REG_SCPR); + + if (((isr & IPS_BIT_EI) == 0) && ((scpr & IPS_BIT_EBM) == 0)) + return (0); + else + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_isinit_copperhead_memio */ +/* */ +/* Routine Description: */ +/* */ +/* Is controller initialized ? */ +/* */ +/****************************************************************************/ +static int +ips_isinit_copperhead_memio(ips_ha_t * ha) +{ + uint8_t isr = 0; + uint8_t scpr; + + METHOD_TRACE("ips_is_init_copperhead_memio", 1); + + isr = readb(ha->mem_ptr + IPS_REG_HISR); + scpr = readb(ha->mem_ptr + IPS_REG_SCPR); + + if (((isr & IPS_BIT_EI) == 0) && ((scpr & IPS_BIT_EBM) == 0)) + return (0); + else + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_isinit_morpheus */ +/* */ +/* Routine Description: */ +/* */ +/* Is controller initialized ? */ +/* */ +/****************************************************************************/ +static int +ips_isinit_morpheus(ips_ha_t * ha) +{ + uint32_t post; + uint32_t bits; + + METHOD_TRACE("ips_is_init_morpheus", 1); + + post = readl(ha->mem_ptr + IPS_REG_I960_MSG0); + bits = readl(ha->mem_ptr + IPS_REG_I2O_HIR); + + if (post == 0) + return (0); + else if (bits & 0x3) + return (0); + else + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_enable_int_copperhead */ +/* */ +/* Routine Description: */ +/* Turn on interrupts */ +/* */ +/****************************************************************************/ +static void +ips_enable_int_copperhead(ips_ha_t * ha) +{ + METHOD_TRACE("ips_enable_int_copperhead", 1); + + outb(ha->io_addr + IPS_REG_HISR, IPS_BIT_EI); + inb(ha->io_addr + IPS_REG_HISR); /*Ensure PCI Posting Completes*/ +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_enable_int_copperhead_memio */ +/* */ +/* Routine Description: */ +/* Turn on interrupts */ +/* */ +/****************************************************************************/ +static void +ips_enable_int_copperhead_memio(ips_ha_t * ha) +{ + METHOD_TRACE("ips_enable_int_copperhead_memio", 1); + + writeb(IPS_BIT_EI, ha->mem_ptr + IPS_REG_HISR); + readb(ha->mem_ptr + IPS_REG_HISR); /*Ensure PCI Posting Completes*/ +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_enable_int_morpheus */ +/* */ +/* Routine Description: */ +/* Turn on interrupts */ +/* */ +/****************************************************************************/ +static void +ips_enable_int_morpheus(ips_ha_t * ha) +{ + uint32_t Oimr; + + METHOD_TRACE("ips_enable_int_morpheus", 1); + + Oimr = readl(ha->mem_ptr + IPS_REG_I960_OIMR); + Oimr &= ~0x08; + writel(Oimr, ha->mem_ptr + IPS_REG_I960_OIMR); + readl(ha->mem_ptr + IPS_REG_I960_OIMR); /*Ensure PCI Posting Completes*/ +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_init_copperhead */ +/* */ +/* Routine Description: */ +/* */ +/* Initialize a copperhead controller */ +/* */ +/****************************************************************************/ +static int +ips_init_copperhead(ips_ha_t * ha) +{ + uint8_t Isr; + uint8_t Cbsp; + uint8_t PostByte[IPS_MAX_POST_BYTES]; + uint8_t ConfigByte[IPS_MAX_CONFIG_BYTES]; + int i, j; + + METHOD_TRACE("ips_init_copperhead", 1); + + for (i = 0; i < IPS_MAX_POST_BYTES; i++) { + for (j = 0; j < 45; j++) { + Isr = inb(ha->io_addr + IPS_REG_HISR); + if (Isr & IPS_BIT_GHI) + break; + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + } + + if (j >= 45) + /* error occurred */ + return (0); + + PostByte[i] = inb(ha->io_addr + IPS_REG_ISPR); + outb(Isr, ha->io_addr + IPS_REG_HISR); + } + + if (PostByte[0] < IPS_GOOD_POST_STATUS) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "reset controller fails (post status %x %x).\n", + PostByte[0], PostByte[1]); + + return (0); + } + + for (i = 0; i < IPS_MAX_CONFIG_BYTES; i++) { + for (j = 0; j < 240; j++) { + Isr = inb(ha->io_addr + IPS_REG_HISR); + if (Isr & IPS_BIT_GHI) + break; + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + } + + if (j >= 240) + /* error occurred */ + return (0); + + ConfigByte[i] = inb(ha->io_addr + IPS_REG_ISPR); + outb(Isr, ha->io_addr + IPS_REG_HISR); + } + + for (i = 0; i < 240; i++) { + Cbsp = inb(ha->io_addr + IPS_REG_CBSP); + + if ((Cbsp & IPS_BIT_OP) == 0) + break; + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + } + + if (i >= 240) + /* reset failed */ + return (0); + + /* setup CCCR */ + outl(cpu_to_le32(0x1010), ha->io_addr + IPS_REG_CCCR); + + /* Enable busmastering */ + outb(IPS_BIT_EBM, ha->io_addr + IPS_REG_SCPR); + + if (ha->revision_id == IPS_REVID_TROMBONE64) + /* fix for anaconda64 */ + outl(0, ha->io_addr + IPS_REG_NDAE); + + /* Enable interrupts */ + outb(IPS_BIT_EI, ha->io_addr + IPS_REG_HISR); + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_init_copperhead_memio */ +/* */ +/* Routine Description: */ +/* */ +/* Initialize a copperhead controller with memory mapped I/O */ +/* */ +/****************************************************************************/ +static int +ips_init_copperhead_memio(ips_ha_t * ha) +{ + uint8_t Isr = 0; + uint8_t Cbsp; + uint8_t PostByte[IPS_MAX_POST_BYTES]; + uint8_t ConfigByte[IPS_MAX_CONFIG_BYTES]; + int i, j; + + METHOD_TRACE("ips_init_copperhead_memio", 1); + + for (i = 0; i < IPS_MAX_POST_BYTES; i++) { + for (j = 0; j < 45; j++) { + Isr = readb(ha->mem_ptr + IPS_REG_HISR); + if (Isr & IPS_BIT_GHI) + break; + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + } + + if (j >= 45) + /* error occurred */ + return (0); + + PostByte[i] = readb(ha->mem_ptr + IPS_REG_ISPR); + writeb(Isr, ha->mem_ptr + IPS_REG_HISR); + } + + if (PostByte[0] < IPS_GOOD_POST_STATUS) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "reset controller fails (post status %x %x).\n", + PostByte[0], PostByte[1]); + + return (0); + } + + for (i = 0; i < IPS_MAX_CONFIG_BYTES; i++) { + for (j = 0; j < 240; j++) { + Isr = readb(ha->mem_ptr + IPS_REG_HISR); + if (Isr & IPS_BIT_GHI) + break; + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + } + + if (j >= 240) + /* error occurred */ + return (0); + + ConfigByte[i] = readb(ha->mem_ptr + IPS_REG_ISPR); + writeb(Isr, ha->mem_ptr + IPS_REG_HISR); + } + + for (i = 0; i < 240; i++) { + Cbsp = readb(ha->mem_ptr + IPS_REG_CBSP); + + if ((Cbsp & IPS_BIT_OP) == 0) + break; + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + } + + if (i >= 240) + /* error occurred */ + return (0); + + /* setup CCCR */ + writel(0x1010, ha->mem_ptr + IPS_REG_CCCR); + + /* Enable busmastering */ + writeb(IPS_BIT_EBM, ha->mem_ptr + IPS_REG_SCPR); + + if (ha->revision_id == IPS_REVID_TROMBONE64) + /* fix for anaconda64 */ + writel(0, ha->mem_ptr + IPS_REG_NDAE); + + /* Enable interrupts */ + writeb(IPS_BIT_EI, ha->mem_ptr + IPS_REG_HISR); + + /* if we get here then everything went OK */ + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_init_morpheus */ +/* */ +/* Routine Description: */ +/* */ +/* Initialize a morpheus controller */ +/* */ +/****************************************************************************/ +static int +ips_init_morpheus(ips_ha_t * ha) +{ + uint32_t Post; + uint32_t Config; + uint32_t Isr; + uint32_t Oimr; + int i; + + METHOD_TRACE("ips_init_morpheus", 1); + + /* Wait up to 45 secs for Post */ + for (i = 0; i < 45; i++) { + Isr = readl(ha->mem_ptr + IPS_REG_I2O_HIR); + + if (Isr & IPS_BIT_I960_MSG0I) + break; + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + } + + if (i >= 45) { + /* error occurred */ + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "timeout waiting for post.\n"); + + return (0); + } + + Post = readl(ha->mem_ptr + IPS_REG_I960_MSG0); + + if (Post == 0x4F00) { /* If Flashing the Battery PIC */ + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Flashing Battery PIC, Please wait ...\n"); + + /* Clear the interrupt bit */ + Isr = (uint32_t) IPS_BIT_I960_MSG0I; + writel(Isr, ha->mem_ptr + IPS_REG_I2O_HIR); + + for (i = 0; i < 120; i++) { /* Wait Up to 2 Min. for Completion */ + Post = readl(ha->mem_ptr + IPS_REG_I960_MSG0); + if (Post != 0x4F00) + break; + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + } + + if (i >= 120) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "timeout waiting for Battery PIC Flash\n"); + return (0); + } + + } + + /* Clear the interrupt bit */ + Isr = (uint32_t) IPS_BIT_I960_MSG0I; + writel(Isr, ha->mem_ptr + IPS_REG_I2O_HIR); + + if (Post < (IPS_GOOD_POST_STATUS << 8)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "reset controller fails (post status %x).\n", Post); + + return (0); + } + + /* Wait up to 240 secs for config bytes */ + for (i = 0; i < 240; i++) { + Isr = readl(ha->mem_ptr + IPS_REG_I2O_HIR); + + if (Isr & IPS_BIT_I960_MSG1I) + break; + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + } + + if (i >= 240) { + /* error occurred */ + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "timeout waiting for config.\n"); + + return (0); + } + + Config = readl(ha->mem_ptr + IPS_REG_I960_MSG1); + + /* Clear interrupt bit */ + Isr = (uint32_t) IPS_BIT_I960_MSG1I; + writel(Isr, ha->mem_ptr + IPS_REG_I2O_HIR); + + /* Turn on the interrupts */ + Oimr = readl(ha->mem_ptr + IPS_REG_I960_OIMR); + Oimr &= ~0x8; + writel(Oimr, ha->mem_ptr + IPS_REG_I960_OIMR); + + /* if we get here then everything went OK */ + + /* Since we did a RESET, an EraseStripeLock may be needed */ + if (Post == 0xEF10) { + if ((Config == 0x000F) || (Config == 0x0009)) + ha->requires_esl = 1; + } + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_reset_copperhead */ +/* */ +/* Routine Description: */ +/* */ +/* Reset the controller */ +/* */ +/****************************************************************************/ +static int +ips_reset_copperhead(ips_ha_t * ha) +{ + int reset_counter; + + METHOD_TRACE("ips_reset_copperhead", 1); + + DEBUG_VAR(1, "(%s%d) ips_reset_copperhead: io addr: %x, irq: %d", + ips_name, ha->host_num, ha->io_addr, ha->irq); + + reset_counter = 0; + + while (reset_counter < 2) { + reset_counter++; + + outb(IPS_BIT_RST, ha->io_addr + IPS_REG_SCPR); + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + + outb(0, ha->io_addr + IPS_REG_SCPR); + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + + if ((*ha->func.init) (ha)) + break; + else if (reset_counter >= 2) { + + return (0); + } + } + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_reset_copperhead_memio */ +/* */ +/* Routine Description: */ +/* */ +/* Reset the controller */ +/* */ +/****************************************************************************/ +static int +ips_reset_copperhead_memio(ips_ha_t * ha) +{ + int reset_counter; + + METHOD_TRACE("ips_reset_copperhead_memio", 1); + + DEBUG_VAR(1, "(%s%d) ips_reset_copperhead_memio: mem addr: %x, irq: %d", + ips_name, ha->host_num, ha->mem_addr, ha->irq); + + reset_counter = 0; + + while (reset_counter < 2) { + reset_counter++; + + writeb(IPS_BIT_RST, ha->mem_ptr + IPS_REG_SCPR); + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + + writeb(0, ha->mem_ptr + IPS_REG_SCPR); + + /* Delay for 1 Second */ + MDELAY(IPS_ONE_SEC); + + if ((*ha->func.init) (ha)) + break; + else if (reset_counter >= 2) { + + return (0); + } + } + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_reset_morpheus */ +/* */ +/* Routine Description: */ +/* */ +/* Reset the controller */ +/* */ +/****************************************************************************/ +static int +ips_reset_morpheus(ips_ha_t * ha) +{ + int reset_counter; + uint8_t junk; + + METHOD_TRACE("ips_reset_morpheus", 1); + + DEBUG_VAR(1, "(%s%d) ips_reset_morpheus: mem addr: %x, irq: %d", + ips_name, ha->host_num, ha->mem_addr, ha->irq); + + reset_counter = 0; + + while (reset_counter < 2) { + reset_counter++; + + writel(0x80000000, ha->mem_ptr + IPS_REG_I960_IDR); + + /* Delay for 5 Seconds */ + MDELAY(5 * IPS_ONE_SEC); + + /* Do a PCI config read to wait for adapter */ + pci_read_config_byte(ha->pcidev, 4, &junk); + + if ((*ha->func.init) (ha)) + break; + else if (reset_counter >= 2) { + + return (0); + } + } + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_statinit */ +/* */ +/* Routine Description: */ +/* */ +/* Initialize the status queues on the controller */ +/* */ +/****************************************************************************/ +static void +ips_statinit(ips_ha_t * ha) +{ + uint32_t phys_status_start; + + METHOD_TRACE("ips_statinit", 1); + + ha->adapt->p_status_start = ha->adapt->status; + ha->adapt->p_status_end = ha->adapt->status + IPS_MAX_CMDS; + ha->adapt->p_status_tail = ha->adapt->status; + + phys_status_start = ha->adapt->hw_status_start; + outl(cpu_to_le32(phys_status_start), ha->io_addr + IPS_REG_SQSR); + outl(cpu_to_le32(phys_status_start + IPS_STATUS_Q_SIZE), + ha->io_addr + IPS_REG_SQER); + outl(cpu_to_le32(phys_status_start + IPS_STATUS_SIZE), + ha->io_addr + IPS_REG_SQHR); + outl(cpu_to_le32(phys_status_start), ha->io_addr + IPS_REG_SQTR); + + ha->adapt->hw_status_tail = phys_status_start; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_statinit_memio */ +/* */ +/* Routine Description: */ +/* */ +/* Initialize the status queues on the controller */ +/* */ +/****************************************************************************/ +static void +ips_statinit_memio(ips_ha_t * ha) +{ + uint32_t phys_status_start; + + METHOD_TRACE("ips_statinit_memio", 1); + + ha->adapt->p_status_start = ha->adapt->status; + ha->adapt->p_status_end = ha->adapt->status + IPS_MAX_CMDS; + ha->adapt->p_status_tail = ha->adapt->status; + + phys_status_start = ha->adapt->hw_status_start; + writel(phys_status_start, ha->mem_ptr + IPS_REG_SQSR); + writel(phys_status_start + IPS_STATUS_Q_SIZE, + ha->mem_ptr + IPS_REG_SQER); + writel(phys_status_start + IPS_STATUS_SIZE, ha->mem_ptr + IPS_REG_SQHR); + writel(phys_status_start, ha->mem_ptr + IPS_REG_SQTR); + + ha->adapt->hw_status_tail = phys_status_start; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_statupd_copperhead */ +/* */ +/* Routine Description: */ +/* */ +/* Remove an element from the status queue */ +/* */ +/****************************************************************************/ +static uint32_t +ips_statupd_copperhead(ips_ha_t * ha) +{ + METHOD_TRACE("ips_statupd_copperhead", 1); + + if (ha->adapt->p_status_tail != ha->adapt->p_status_end) { + ha->adapt->p_status_tail++; + ha->adapt->hw_status_tail += sizeof (IPS_STATUS); + } else { + ha->adapt->p_status_tail = ha->adapt->p_status_start; + ha->adapt->hw_status_tail = ha->adapt->hw_status_start; + } + + outl(cpu_to_le32(ha->adapt->hw_status_tail), + ha->io_addr + IPS_REG_SQTR); + + return (ha->adapt->p_status_tail->value); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_statupd_copperhead_memio */ +/* */ +/* Routine Description: */ +/* */ +/* Remove an element from the status queue */ +/* */ +/****************************************************************************/ +static uint32_t +ips_statupd_copperhead_memio(ips_ha_t * ha) +{ + METHOD_TRACE("ips_statupd_copperhead_memio", 1); + + if (ha->adapt->p_status_tail != ha->adapt->p_status_end) { + ha->adapt->p_status_tail++; + ha->adapt->hw_status_tail += sizeof (IPS_STATUS); + } else { + ha->adapt->p_status_tail = ha->adapt->p_status_start; + ha->adapt->hw_status_tail = ha->adapt->hw_status_start; + } + + writel(ha->adapt->hw_status_tail, ha->mem_ptr + IPS_REG_SQTR); + + return (ha->adapt->p_status_tail->value); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_statupd_morpheus */ +/* */ +/* Routine Description: */ +/* */ +/* Remove an element from the status queue */ +/* */ +/****************************************************************************/ +static uint32_t +ips_statupd_morpheus(ips_ha_t * ha) +{ + uint32_t val; + + METHOD_TRACE("ips_statupd_morpheus", 1); + + val = readl(ha->mem_ptr + IPS_REG_I2O_OUTMSGQ); + + return (val); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_issue_copperhead */ +/* */ +/* Routine Description: */ +/* */ +/* Send a command down to the controller */ +/* */ +/****************************************************************************/ +static int +ips_issue_copperhead(ips_ha_t * ha, ips_scb_t * scb) +{ + uint32_t TimeOut; + uint32_t val; + + METHOD_TRACE("ips_issue_copperhead", 1); + + if (scb->scsi_cmd) { + DEBUG_VAR(2, "(%s%d) ips_issue: cmd 0x%X id %d (%d %d %d)", + ips_name, + ha->host_num, + scb->cdb[0], + scb->cmd.basic_io.command_id, + scb->bus, scb->target_id, scb->lun); + } else { + DEBUG_VAR(2, KERN_NOTICE "(%s%d) ips_issue: logical cmd id %d", + ips_name, ha->host_num, scb->cmd.basic_io.command_id); + } + + TimeOut = 0; + + while ((val = + le32_to_cpu(inl(ha->io_addr + IPS_REG_CCCR))) & IPS_BIT_SEM) { + udelay(1000); + + if (++TimeOut >= IPS_SEM_TIMEOUT) { + if (!(val & IPS_BIT_START_STOP)) + break; + + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "ips_issue val [0x%x].\n", val); + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "ips_issue semaphore chk timeout.\n"); + + return (IPS_FAILURE); + } /* end if */ + } /* end while */ + + outl(cpu_to_le32(scb->scb_busaddr), ha->io_addr + IPS_REG_CCSAR); + outw(cpu_to_le32(IPS_BIT_START_CMD), ha->io_addr + IPS_REG_CCCR); + + return (IPS_SUCCESS); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_issue_copperhead_memio */ +/* */ +/* Routine Description: */ +/* */ +/* Send a command down to the controller */ +/* */ +/****************************************************************************/ +static int +ips_issue_copperhead_memio(ips_ha_t * ha, ips_scb_t * scb) +{ + uint32_t TimeOut; + uint32_t val; + + METHOD_TRACE("ips_issue_copperhead_memio", 1); + + if (scb->scsi_cmd) { + DEBUG_VAR(2, "(%s%d) ips_issue: cmd 0x%X id %d (%d %d %d)", + ips_name, + ha->host_num, + scb->cdb[0], + scb->cmd.basic_io.command_id, + scb->bus, scb->target_id, scb->lun); + } else { + DEBUG_VAR(2, "(%s%d) ips_issue: logical cmd id %d", + ips_name, ha->host_num, scb->cmd.basic_io.command_id); + } + + TimeOut = 0; + + while ((val = readl(ha->mem_ptr + IPS_REG_CCCR)) & IPS_BIT_SEM) { + udelay(1000); + + if (++TimeOut >= IPS_SEM_TIMEOUT) { + if (!(val & IPS_BIT_START_STOP)) + break; + + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "ips_issue val [0x%x].\n", val); + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "ips_issue semaphore chk timeout.\n"); + + return (IPS_FAILURE); + } /* end if */ + } /* end while */ + + writel(scb->scb_busaddr, ha->mem_ptr + IPS_REG_CCSAR); + writel(IPS_BIT_START_CMD, ha->mem_ptr + IPS_REG_CCCR); + + return (IPS_SUCCESS); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_issue_i2o */ +/* */ +/* Routine Description: */ +/* */ +/* Send a command down to the controller */ +/* */ +/****************************************************************************/ +static int +ips_issue_i2o(ips_ha_t * ha, ips_scb_t * scb) +{ + + METHOD_TRACE("ips_issue_i2o", 1); + + if (scb->scsi_cmd) { + DEBUG_VAR(2, "(%s%d) ips_issue: cmd 0x%X id %d (%d %d %d)", + ips_name, + ha->host_num, + scb->cdb[0], + scb->cmd.basic_io.command_id, + scb->bus, scb->target_id, scb->lun); + } else { + DEBUG_VAR(2, "(%s%d) ips_issue: logical cmd id %d", + ips_name, ha->host_num, scb->cmd.basic_io.command_id); + } + + outl(cpu_to_le32(scb->scb_busaddr), ha->io_addr + IPS_REG_I2O_INMSGQ); + + return (IPS_SUCCESS); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_issue_i2o_memio */ +/* */ +/* Routine Description: */ +/* */ +/* Send a command down to the controller */ +/* */ +/****************************************************************************/ +static int +ips_issue_i2o_memio(ips_ha_t * ha, ips_scb_t * scb) +{ + + METHOD_TRACE("ips_issue_i2o_memio", 1); + + if (scb->scsi_cmd) { + DEBUG_VAR(2, "(%s%d) ips_issue: cmd 0x%X id %d (%d %d %d)", + ips_name, + ha->host_num, + scb->cdb[0], + scb->cmd.basic_io.command_id, + scb->bus, scb->target_id, scb->lun); + } else { + DEBUG_VAR(2, "(%s%d) ips_issue: logical cmd id %d", + ips_name, ha->host_num, scb->cmd.basic_io.command_id); + } + + writel(scb->scb_busaddr, ha->mem_ptr + IPS_REG_I2O_INMSGQ); + + return (IPS_SUCCESS); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_isintr_copperhead */ +/* */ +/* Routine Description: */ +/* */ +/* Test to see if an interrupt is for us */ +/* */ +/****************************************************************************/ +static int +ips_isintr_copperhead(ips_ha_t * ha) +{ + uint8_t Isr; + + METHOD_TRACE("ips_isintr_copperhead", 2); + + Isr = inb(ha->io_addr + IPS_REG_HISR); + + if (Isr == 0xFF) + /* ?!?! Nothing really there */ + return (0); + + if (Isr & IPS_BIT_SCE) + return (1); + else if (Isr & (IPS_BIT_SQO | IPS_BIT_GHI)) { + /* status queue overflow or GHI */ + /* just clear the interrupt */ + outb(Isr, ha->io_addr + IPS_REG_HISR); + } + + return (0); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_isintr_copperhead_memio */ +/* */ +/* Routine Description: */ +/* */ +/* Test to see if an interrupt is for us */ +/* */ +/****************************************************************************/ +static int +ips_isintr_copperhead_memio(ips_ha_t * ha) +{ + uint8_t Isr; + + METHOD_TRACE("ips_isintr_memio", 2); + + Isr = readb(ha->mem_ptr + IPS_REG_HISR); + + if (Isr == 0xFF) + /* ?!?! Nothing really there */ + return (0); + + if (Isr & IPS_BIT_SCE) + return (1); + else if (Isr & (IPS_BIT_SQO | IPS_BIT_GHI)) { + /* status queue overflow or GHI */ + /* just clear the interrupt */ + writeb(Isr, ha->mem_ptr + IPS_REG_HISR); + } + + return (0); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_isintr_morpheus */ +/* */ +/* Routine Description: */ +/* */ +/* Test to see if an interrupt is for us */ +/* */ +/****************************************************************************/ +static int +ips_isintr_morpheus(ips_ha_t * ha) +{ + uint32_t Isr; + + METHOD_TRACE("ips_isintr_morpheus", 2); + + Isr = readl(ha->mem_ptr + IPS_REG_I2O_HIR); + + if (Isr & IPS_BIT_I2O_OPQI) + return (1); + else + return (0); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_wait */ +/* */ +/* Routine Description: */ +/* */ +/* Wait for a command to complete */ +/* */ +/****************************************************************************/ +static int +ips_wait(ips_ha_t * ha, int time, int intr) +{ + int ret; + int done; + + METHOD_TRACE("ips_wait", 1); + + ret = IPS_FAILURE; + done = FALSE; + + time *= IPS_ONE_SEC; /* convert seconds */ + + while ((time > 0) && (!done)) { + if (intr == IPS_INTR_ON) { + if (ha->waitflag == FALSE) { + ret = IPS_SUCCESS; + done = TRUE; + break; + } + } else if (intr == IPS_INTR_IORL) { + if (ha->waitflag == FALSE) { + /* + * controller generated an interrupt to + * acknowledge completion of the command + * and ips_intr() has serviced the interrupt. + */ + ret = IPS_SUCCESS; + done = TRUE; + break; + } + + /* + * NOTE: we already have the io_request_lock so + * even if we get an interrupt it won't get serviced + * until after we finish. + */ + + (*ha->func.intr) (ha); + } + + /* This looks like a very evil loop, but it only does this during start-up */ + udelay(1000); + time--; + } + + return (ret); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_write_driver_status */ +/* */ +/* Routine Description: */ +/* */ +/* Write OS/Driver version to Page 5 of the nvram on the controller */ +/* */ +/****************************************************************************/ +static int +ips_write_driver_status(ips_ha_t * ha, int intr) +{ + METHOD_TRACE("ips_write_driver_status", 1); + + if (!ips_readwrite_page5(ha, FALSE, intr)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "unable to read NVRAM page 5.\n"); + + return (0); + } + + /* check to make sure the page has a valid */ + /* signature */ + if (le32_to_cpu(ha->nvram->signature) != IPS_NVRAM_P5_SIG) { + DEBUG_VAR(1, + "(%s%d) NVRAM page 5 has an invalid signature: %X.", + ips_name, ha->host_num, ha->nvram->signature); + ha->nvram->signature = IPS_NVRAM_P5_SIG; + } + + DEBUG_VAR(2, + "(%s%d) Ad Type: %d, Ad Slot: %d, BIOS: %c%c%c%c %c%c%c%c.", + ips_name, ha->host_num, le16_to_cpu(ha->nvram->adapter_type), + ha->nvram->adapter_slot, ha->nvram->bios_high[0], + ha->nvram->bios_high[1], ha->nvram->bios_high[2], + ha->nvram->bios_high[3], ha->nvram->bios_low[0], + ha->nvram->bios_low[1], ha->nvram->bios_low[2], + ha->nvram->bios_low[3]); + + ips_get_bios_version(ha, intr); + + /* change values (as needed) */ + ha->nvram->operating_system = IPS_OS_LINUX; + ha->nvram->adapter_type = ha->ad_type; + strncpy((char *) ha->nvram->driver_high, IPS_VERSION_HIGH, 4); + strncpy((char *) ha->nvram->driver_low, IPS_VERSION_LOW, 4); + strncpy((char *) ha->nvram->bios_high, ha->bios_version, 4); + strncpy((char *) ha->nvram->bios_low, ha->bios_version + 4, 4); + + ips_version_check(ha, intr); /* Check BIOS/FW/Driver Versions */ + + /* now update the page */ + if (!ips_readwrite_page5(ha, TRUE, intr)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "unable to write NVRAM page 5.\n"); + + return (0); + } + + /* IF NVRAM Page 5 is OK, Use it for Slot Number Info Because Linux Doesn't Do Slots */ + ha->slot_num = ha->nvram->adapter_slot; + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_read_adapter_status */ +/* */ +/* Routine Description: */ +/* */ +/* Do an Inquiry command to the adapter */ +/* */ +/****************************************************************************/ +static int +ips_read_adapter_status(ips_ha_t * ha, int intr) +{ + ips_scb_t *scb; + int ret; + + METHOD_TRACE("ips_read_adapter_status", 1); + + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_ENQUIRY; + + scb->cmd.basic_io.op_code = IPS_CMD_ENQUIRY; + scb->cmd.basic_io.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.basic_io.sg_count = 0; + scb->cmd.basic_io.lba = 0; + scb->cmd.basic_io.sector_count = 0; + scb->cmd.basic_io.log_drv = 0; + scb->data_len = sizeof (*ha->enq); + scb->cmd.basic_io.sg_addr = ha->enq_busaddr; + + /* send command */ + if (((ret = + ips_send_wait(ha, scb, ips_cmd_timeout, intr)) == IPS_FAILURE) + || (ret == IPS_SUCCESS_IMM) + || ((scb->basic_status & IPS_GSC_STATUS_MASK) > 1)) + return (0); + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_read_subsystem_parameters */ +/* */ +/* Routine Description: */ +/* */ +/* Read subsystem parameters from the adapter */ +/* */ +/****************************************************************************/ +static int +ips_read_subsystem_parameters(ips_ha_t * ha, int intr) +{ + ips_scb_t *scb; + int ret; + + METHOD_TRACE("ips_read_subsystem_parameters", 1); + + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_GET_SUBSYS; + + scb->cmd.basic_io.op_code = IPS_CMD_GET_SUBSYS; + scb->cmd.basic_io.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.basic_io.sg_count = 0; + scb->cmd.basic_io.lba = 0; + scb->cmd.basic_io.sector_count = 0; + scb->cmd.basic_io.log_drv = 0; + scb->data_len = sizeof (*ha->subsys); + scb->cmd.basic_io.sg_addr = ha->ioctl_busaddr; + + /* send command */ + if (((ret = + ips_send_wait(ha, scb, ips_cmd_timeout, intr)) == IPS_FAILURE) + || (ret == IPS_SUCCESS_IMM) + || ((scb->basic_status & IPS_GSC_STATUS_MASK) > 1)) + return (0); + + memcpy(ha->subsys, ha->ioctl_data, sizeof(*ha->subsys)); + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_read_config */ +/* */ +/* Routine Description: */ +/* */ +/* Read the configuration on the adapter */ +/* */ +/****************************************************************************/ +static int +ips_read_config(ips_ha_t * ha, int intr) +{ + ips_scb_t *scb; + int i; + int ret; + + METHOD_TRACE("ips_read_config", 1); + + /* set defaults for initiator IDs */ + for (i = 0; i < 4; i++) + ha->conf->init_id[i] = 7; + + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_READ_CONF; + + scb->cmd.basic_io.op_code = IPS_CMD_READ_CONF; + scb->cmd.basic_io.command_id = IPS_COMMAND_ID(ha, scb); + scb->data_len = sizeof (*ha->conf); + scb->cmd.basic_io.sg_addr = ha->ioctl_busaddr; + + /* send command */ + if (((ret = + ips_send_wait(ha, scb, ips_cmd_timeout, intr)) == IPS_FAILURE) + || (ret == IPS_SUCCESS_IMM) + || ((scb->basic_status & IPS_GSC_STATUS_MASK) > 1)) { + + memset(ha->conf, 0, sizeof (IPS_CONF)); + + /* reset initiator IDs */ + for (i = 0; i < 4; i++) + ha->conf->init_id[i] = 7; + + /* Allow Completed with Errors, so JCRM can access the Adapter to fix the problems */ + if ((scb->basic_status & IPS_GSC_STATUS_MASK) == + IPS_CMD_CMPLT_WERROR) + return (1); + + return (0); + } + + memcpy(ha->conf, ha->ioctl_data, sizeof(*ha->conf)); + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_readwrite_page5 */ +/* */ +/* Routine Description: */ +/* */ +/* Read nvram page 5 from the adapter */ +/* */ +/****************************************************************************/ +static int +ips_readwrite_page5(ips_ha_t * ha, int write, int intr) +{ + ips_scb_t *scb; + int ret; + + METHOD_TRACE("ips_readwrite_page5", 1); + + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_RW_NVRAM_PAGE; + + scb->cmd.nvram.op_code = IPS_CMD_RW_NVRAM_PAGE; + scb->cmd.nvram.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.nvram.page = 5; + scb->cmd.nvram.write = write; + scb->cmd.nvram.reserved = 0; + scb->cmd.nvram.reserved2 = 0; + scb->data_len = sizeof (*ha->nvram); + scb->cmd.nvram.buffer_addr = ha->ioctl_busaddr; + if (write) + memcpy(ha->ioctl_data, ha->nvram, sizeof(*ha->nvram)); + + /* issue the command */ + if (((ret = + ips_send_wait(ha, scb, ips_cmd_timeout, intr)) == IPS_FAILURE) + || (ret == IPS_SUCCESS_IMM) + || ((scb->basic_status & IPS_GSC_STATUS_MASK) > 1)) { + + memset(ha->nvram, 0, sizeof (IPS_NVRAM_P5)); + + return (0); + } + if (!write) + memcpy(ha->nvram, ha->ioctl_data, sizeof(*ha->nvram)); + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_clear_adapter */ +/* */ +/* Routine Description: */ +/* */ +/* Clear the stripe lock tables */ +/* */ +/****************************************************************************/ +static int +ips_clear_adapter(ips_ha_t * ha, int intr) +{ + ips_scb_t *scb; + int ret; + + METHOD_TRACE("ips_clear_adapter", 1); + + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_reset_timeout; + scb->cdb[0] = IPS_CMD_CONFIG_SYNC; + + scb->cmd.config_sync.op_code = IPS_CMD_CONFIG_SYNC; + scb->cmd.config_sync.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.config_sync.channel = 0; + scb->cmd.config_sync.source_target = IPS_POCL; + scb->cmd.config_sync.reserved = 0; + scb->cmd.config_sync.reserved2 = 0; + scb->cmd.config_sync.reserved3 = 0; + + /* issue command */ + if (((ret = + ips_send_wait(ha, scb, ips_reset_timeout, intr)) == IPS_FAILURE) + || (ret == IPS_SUCCESS_IMM) + || ((scb->basic_status & IPS_GSC_STATUS_MASK) > 1)) + return (0); + + /* send unlock stripe command */ + ips_init_scb(ha, scb); + + scb->cdb[0] = IPS_CMD_ERROR_TABLE; + scb->timeout = ips_reset_timeout; + + scb->cmd.unlock_stripe.op_code = IPS_CMD_ERROR_TABLE; + scb->cmd.unlock_stripe.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.unlock_stripe.log_drv = 0; + scb->cmd.unlock_stripe.control = IPS_CSL; + scb->cmd.unlock_stripe.reserved = 0; + scb->cmd.unlock_stripe.reserved2 = 0; + scb->cmd.unlock_stripe.reserved3 = 0; + + /* issue command */ + if (((ret = + ips_send_wait(ha, scb, ips_cmd_timeout, intr)) == IPS_FAILURE) + || (ret == IPS_SUCCESS_IMM) + || ((scb->basic_status & IPS_GSC_STATUS_MASK) > 1)) + return (0); + + return (1); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_ffdc_reset */ +/* */ +/* Routine Description: */ +/* */ +/* FFDC: write reset info */ +/* */ +/****************************************************************************/ +static void +ips_ffdc_reset(ips_ha_t * ha, int intr) +{ + ips_scb_t *scb; + + METHOD_TRACE("ips_ffdc_reset", 1); + + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_FFDC; + scb->cmd.ffdc.op_code = IPS_CMD_FFDC; + scb->cmd.ffdc.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.ffdc.reset_count = ha->reset_count; + scb->cmd.ffdc.reset_type = 0x80; + + /* convert time to what the card wants */ + ips_fix_ffdc_time(ha, scb, ha->last_ffdc); + + /* issue command */ + ips_send_wait(ha, scb, ips_cmd_timeout, intr); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_ffdc_time */ +/* */ +/* Routine Description: */ +/* */ +/* FFDC: write time info */ +/* */ +/****************************************************************************/ +static void +ips_ffdc_time(ips_ha_t * ha) +{ + ips_scb_t *scb; + + METHOD_TRACE("ips_ffdc_time", 1); + + DEBUG_VAR(1, "(%s%d) Sending time update.", ips_name, ha->host_num); + + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_FFDC; + scb->cmd.ffdc.op_code = IPS_CMD_FFDC; + scb->cmd.ffdc.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.ffdc.reset_count = 0; + scb->cmd.ffdc.reset_type = 0; + + /* convert time to what the card wants */ + ips_fix_ffdc_time(ha, scb, ha->last_ffdc); + + /* issue command */ + ips_send_wait(ha, scb, ips_cmd_timeout, IPS_FFDC); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_fix_ffdc_time */ +/* */ +/* Routine Description: */ +/* Adjust time_t to what the card wants */ +/* */ +/****************************************************************************/ +static void +ips_fix_ffdc_time(ips_ha_t * ha, ips_scb_t * scb, time_t current_time) +{ + long days; + long rem; + int i; + int year; + int yleap; + int year_lengths[2] = { IPS_DAYS_NORMAL_YEAR, IPS_DAYS_LEAP_YEAR }; + int month_lengths[12][2] = { {31, 31}, + {28, 29}, + {31, 31}, + {30, 30}, + {31, 31}, + {30, 30}, + {31, 31}, + {31, 31}, + {30, 30}, + {31, 31}, + {30, 30}, + {31, 31} + }; + + METHOD_TRACE("ips_fix_ffdc_time", 1); + + days = current_time / IPS_SECS_DAY; + rem = current_time % IPS_SECS_DAY; + + scb->cmd.ffdc.hour = (rem / IPS_SECS_HOUR); + rem = rem % IPS_SECS_HOUR; + scb->cmd.ffdc.minute = (rem / IPS_SECS_MIN); + scb->cmd.ffdc.second = (rem % IPS_SECS_MIN); + + year = IPS_EPOCH_YEAR; + while (days < 0 || days >= year_lengths[yleap = IPS_IS_LEAP_YEAR(year)]) { + int newy; + + newy = year + (days / IPS_DAYS_NORMAL_YEAR); + if (days < 0) + --newy; + days -= (newy - year) * IPS_DAYS_NORMAL_YEAR + + IPS_NUM_LEAP_YEARS_THROUGH(newy - 1) - + IPS_NUM_LEAP_YEARS_THROUGH(year - 1); + year = newy; + } + + scb->cmd.ffdc.yearH = year / 100; + scb->cmd.ffdc.yearL = year % 100; + + for (i = 0; days >= month_lengths[i][yleap]; ++i) + days -= month_lengths[i][yleap]; + + scb->cmd.ffdc.month = i + 1; + scb->cmd.ffdc.day = days + 1; +} + +/**************************************************************************** + * BIOS Flash Routines * + ****************************************************************************/ + +/****************************************************************************/ +/* */ +/* Routine Name: ips_erase_bios */ +/* */ +/* Routine Description: */ +/* Erase the BIOS on the adapter */ +/* */ +/****************************************************************************/ +static int +ips_erase_bios(ips_ha_t * ha) +{ + int timeout; + uint8_t status = 0; + + METHOD_TRACE("ips_erase_bios", 1); + + status = 0; + + /* Clear the status register */ + outl(0, ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + outb(0x50, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* Erase Setup */ + outb(0x20, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* Erase Confirm */ + outb(0xD0, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* Erase Status */ + outb(0x70, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + timeout = 80000; /* 80 seconds */ + + while (timeout > 0) { + if (ha->revision_id == IPS_REVID_TROMBONE64) { + outl(0, ha->io_addr + IPS_REG_FLAP); + udelay(25); /* 25 us */ + } + + status = inb(ha->io_addr + IPS_REG_FLDP); + + if (status & 0x80) + break; + + MDELAY(1); + timeout--; + } + + /* check for timeout */ + if (timeout <= 0) { + /* timeout */ + + /* try to suspend the erase */ + outb(0xB0, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* wait for 10 seconds */ + timeout = 10000; + while (timeout > 0) { + if (ha->revision_id == IPS_REVID_TROMBONE64) { + outl(0, ha->io_addr + IPS_REG_FLAP); + udelay(25); /* 25 us */ + } + + status = inb(ha->io_addr + IPS_REG_FLDP); + + if (status & 0xC0) + break; + + MDELAY(1); + timeout--; + } + + return (1); + } + + /* check for valid VPP */ + if (status & 0x08) + /* VPP failure */ + return (1); + + /* check for succesful flash */ + if (status & 0x30) + /* sequence error */ + return (1); + + /* Otherwise, we were successful */ + /* clear status */ + outb(0x50, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* enable reads */ + outb(0xFF, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + return (0); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_erase_bios_memio */ +/* */ +/* Routine Description: */ +/* Erase the BIOS on the adapter */ +/* */ +/****************************************************************************/ +static int +ips_erase_bios_memio(ips_ha_t * ha) +{ + int timeout; + uint8_t status; + + METHOD_TRACE("ips_erase_bios_memio", 1); + + status = 0; + + /* Clear the status register */ + writel(0, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + writeb(0x50, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* Erase Setup */ + writeb(0x20, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* Erase Confirm */ + writeb(0xD0, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* Erase Status */ + writeb(0x70, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + timeout = 80000; /* 80 seconds */ + + while (timeout > 0) { + if (ha->revision_id == IPS_REVID_TROMBONE64) { + writel(0, ha->mem_ptr + IPS_REG_FLAP); + udelay(25); /* 25 us */ + } + + status = readb(ha->mem_ptr + IPS_REG_FLDP); + + if (status & 0x80) + break; + + MDELAY(1); + timeout--; + } + + /* check for timeout */ + if (timeout <= 0) { + /* timeout */ + + /* try to suspend the erase */ + writeb(0xB0, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* wait for 10 seconds */ + timeout = 10000; + while (timeout > 0) { + if (ha->revision_id == IPS_REVID_TROMBONE64) { + writel(0, ha->mem_ptr + IPS_REG_FLAP); + udelay(25); /* 25 us */ + } + + status = readb(ha->mem_ptr + IPS_REG_FLDP); + + if (status & 0xC0) + break; + + MDELAY(1); + timeout--; + } + + return (1); + } + + /* check for valid VPP */ + if (status & 0x08) + /* VPP failure */ + return (1); + + /* check for succesful flash */ + if (status & 0x30) + /* sequence error */ + return (1); + + /* Otherwise, we were successful */ + /* clear status */ + writeb(0x50, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* enable reads */ + writeb(0xFF, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + return (0); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_program_bios */ +/* */ +/* Routine Description: */ +/* Program the BIOS on the adapter */ +/* */ +/****************************************************************************/ +static int +ips_program_bios(ips_ha_t * ha, char *buffer, uint32_t buffersize, + uint32_t offset) +{ + int i; + int timeout; + uint8_t status = 0; + + METHOD_TRACE("ips_program_bios", 1); + + status = 0; + + for (i = 0; i < buffersize; i++) { + /* write a byte */ + outl(cpu_to_le32(i + offset), ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + outb(0x40, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + outb(buffer[i], ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* wait up to one second */ + timeout = 1000; + while (timeout > 0) { + if (ha->revision_id == IPS_REVID_TROMBONE64) { + outl(0, ha->io_addr + IPS_REG_FLAP); + udelay(25); /* 25 us */ + } + + status = inb(ha->io_addr + IPS_REG_FLDP); + + if (status & 0x80) + break; + + MDELAY(1); + timeout--; + } + + if (timeout == 0) { + /* timeout error */ + outl(0, ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + outb(0xFF, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + return (1); + } + + /* check the status */ + if (status & 0x18) { + /* programming error */ + outl(0, ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + outb(0xFF, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + return (1); + } + } /* end for */ + + /* Enable reading */ + outl(0, ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + outb(0xFF, ha->io_addr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + return (0); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_program_bios_memio */ +/* */ +/* Routine Description: */ +/* Program the BIOS on the adapter */ +/* */ +/****************************************************************************/ +static int +ips_program_bios_memio(ips_ha_t * ha, char *buffer, uint32_t buffersize, + uint32_t offset) +{ + int i; + int timeout; + uint8_t status = 0; + + METHOD_TRACE("ips_program_bios_memio", 1); + + status = 0; + + for (i = 0; i < buffersize; i++) { + /* write a byte */ + writel(i + offset, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + writeb(0x40, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + writeb(buffer[i], ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + /* wait up to one second */ + timeout = 1000; + while (timeout > 0) { + if (ha->revision_id == IPS_REVID_TROMBONE64) { + writel(0, ha->mem_ptr + IPS_REG_FLAP); + udelay(25); /* 25 us */ + } + + status = readb(ha->mem_ptr + IPS_REG_FLDP); + + if (status & 0x80) + break; + + MDELAY(1); + timeout--; + } + + if (timeout == 0) { + /* timeout error */ + writel(0, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + writeb(0xFF, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + return (1); + } + + /* check the status */ + if (status & 0x18) { + /* programming error */ + writel(0, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + writeb(0xFF, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + return (1); + } + } /* end for */ + + /* Enable reading */ + writel(0, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + writeb(0xFF, ha->mem_ptr + IPS_REG_FLDP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + return (0); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_verify_bios */ +/* */ +/* Routine Description: */ +/* Verify the BIOS on the adapter */ +/* */ +/****************************************************************************/ +static int +ips_verify_bios(ips_ha_t * ha, char *buffer, uint32_t buffersize, + uint32_t offset) +{ + uint8_t checksum; + int i; + + METHOD_TRACE("ips_verify_bios", 1); + + /* test 1st byte */ + outl(0, ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + if (inb(ha->io_addr + IPS_REG_FLDP) != 0x55) + return (1); + + outl(cpu_to_le32(1), ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + if (inb(ha->io_addr + IPS_REG_FLDP) != 0xAA) + return (1); + + checksum = 0xff; + for (i = 2; i < buffersize; i++) { + + outl(cpu_to_le32(i + offset), ha->io_addr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + checksum = (uint8_t) checksum + inb(ha->io_addr + IPS_REG_FLDP); + } + + if (checksum != 0) + /* failure */ + return (1); + else + /* success */ + return (0); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_verify_bios_memio */ +/* */ +/* Routine Description: */ +/* Verify the BIOS on the adapter */ +/* */ +/****************************************************************************/ +static int +ips_verify_bios_memio(ips_ha_t * ha, char *buffer, uint32_t buffersize, + uint32_t offset) +{ + uint8_t checksum; + int i; + + METHOD_TRACE("ips_verify_bios_memio", 1); + + /* test 1st byte */ + writel(0, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + if (readb(ha->mem_ptr + IPS_REG_FLDP) != 0x55) + return (1); + + writel(1, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + if (readb(ha->mem_ptr + IPS_REG_FLDP) != 0xAA) + return (1); + + checksum = 0xff; + for (i = 2; i < buffersize; i++) { + + writel(i + offset, ha->mem_ptr + IPS_REG_FLAP); + if (ha->revision_id == IPS_REVID_TROMBONE64) + udelay(25); /* 25 us */ + + checksum = + (uint8_t) checksum + readb(ha->mem_ptr + IPS_REG_FLDP); + } + + if (checksum != 0) + /* failure */ + return (1); + else + /* success */ + return (0); +} + +/*---------------------------------------------------------------------------*/ +/* Routine Name: ips_version_check */ +/* */ +/* Dependencies: */ +/* Assumes that ips_read_adapter_status() is called first filling in */ +/* the data for SubSystem Parameters. */ +/* Called from ips_write_driver_status() so it also assumes NVRAM Page 5 */ +/* Data is available. */ +/* */ +/*---------------------------------------------------------------------------*/ +static void +ips_version_check(ips_ha_t * ha, int intr) +{ + IPS_VERSION_DATA *VersionInfo; + uint8_t FirmwareVersion[IPS_COMPAT_ID_LENGTH + 1]; + uint8_t BiosVersion[IPS_COMPAT_ID_LENGTH + 1]; + int MatchError; + int rc; + char BiosString[10]; + char FirmwareString[10]; + + METHOD_TRACE("ips_version_check", 1); + + VersionInfo = ( IPS_VERSION_DATA * ) ha->ioctl_data; + + memset(FirmwareVersion, 0, IPS_COMPAT_ID_LENGTH + 1); + memset(BiosVersion, 0, IPS_COMPAT_ID_LENGTH + 1); + + /* Get the Compatible BIOS Version from NVRAM Page 5 */ + memcpy(BiosVersion, ha->nvram->BiosCompatibilityID, + IPS_COMPAT_ID_LENGTH); + + rc = IPS_FAILURE; + if (ha->subsys->param[4] & IPS_GET_VERSION_SUPPORT) { /* If Versioning is Supported */ + /* Get the Version Info with a Get Version Command */ + memset( VersionInfo, 0, sizeof (IPS_VERSION_DATA)); + rc = ips_get_version_info(ha, ha->ioctl_busaddr, intr); + if (rc == IPS_SUCCESS) + memcpy(FirmwareVersion, VersionInfo->compatibilityId, + IPS_COMPAT_ID_LENGTH); + } + + if (rc != IPS_SUCCESS) { /* If Data Not Obtainable from a GetVersion Command */ + /* Get the Firmware Version from Enquiry Data */ + memcpy(FirmwareVersion, ha->enq->CodeBlkVersion, + IPS_COMPAT_ID_LENGTH); + } + + /* printk(KERN_WARNING "Adapter's BIOS Version = %s\n", BiosVersion); */ + /* printk(KERN_WARNING "BIOS Compatible Version = %s\n", IPS_COMPAT_BIOS); */ + /* printk(KERN_WARNING "Adapter's Firmware Version = %s\n", FirmwareVersion); */ + /* printk(KERN_WARNING "Firmware Compatible Version = %s \n", Compatable[ ha->nvram->adapter_type ]); */ + + MatchError = 0; + + if (strncmp + (FirmwareVersion, Compatable[ha->nvram->adapter_type], + IPS_COMPAT_ID_LENGTH) != 0) + MatchError = 1; + + if (strncmp(BiosVersion, IPS_COMPAT_BIOS, IPS_COMPAT_ID_LENGTH) != 0) + MatchError = 1; + + ha->nvram->versioning = 1; /* Indicate the Driver Supports Versioning */ + + if (MatchError) { + ha->nvram->version_mismatch = 1; + if (ips_cd_boot == 0) { + strncpy(&BiosString[0], ha->nvram->bios_high, 4); + strncpy(&BiosString[4], ha->nvram->bios_low, 4); + BiosString[8] = 0; + + strncpy(&FirmwareString[0], ha->enq->CodeBlkVersion, 8); + FirmwareString[8] = 0; + + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Warning ! ! ! ServeRAID Version Mismatch\n"); + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Bios = %s, Firmware = %s, Device Driver = %s%s\n", + BiosString, FirmwareString, IPS_VERSION_HIGH, + IPS_VERSION_LOW); + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "These levels should match to avoid possible compatibility problems.\n"); + } + } else { + ha->nvram->version_mismatch = 0; + } + + return; +} + +/*---------------------------------------------------------------------------*/ +/* Routine Name: ips_get_version_info */ +/* */ +/* Routine Description: */ +/* Issue an internal GETVERSION Command */ +/* */ +/* Return Value: */ +/* 0 if Successful, else non-zero */ +/*---------------------------------------------------------------------------*/ +static int +ips_get_version_info(ips_ha_t * ha, dma_addr_t Buffer, int intr) +{ + ips_scb_t *scb; + int rc; + + METHOD_TRACE("ips_get_version_info", 1); + + scb = &ha->scbs[ha->max_cmds - 1]; + + ips_init_scb(ha, scb); + + scb->timeout = ips_cmd_timeout; + scb->cdb[0] = IPS_CMD_GET_VERSION_INFO; + scb->cmd.version_info.op_code = IPS_CMD_GET_VERSION_INFO; + scb->cmd.version_info.command_id = IPS_COMMAND_ID(ha, scb); + scb->cmd.version_info.reserved = 0; + scb->cmd.version_info.count = sizeof (IPS_VERSION_DATA); + scb->cmd.version_info.reserved2 = 0; + scb->data_len = sizeof (IPS_VERSION_DATA); + scb->data_busaddr = Buffer; + scb->cmd.version_info.buffer_addr = Buffer; + scb->flags = 0; + + /* issue command */ + rc = ips_send_wait(ha, scb, ips_cmd_timeout, intr); + return (rc); +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_abort_init */ +/* */ +/* Routine Description: */ +/* cleanup routine for a failed adapter initialization */ +/****************************************************************************/ +static int +ips_abort_init(ips_ha_t * ha, int index) +{ + ha->active = 0; + ips_free(ha); + ips_ha[index] = NULL; + ips_sh[index] = NULL; + return -1; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_shift_controllers */ +/* */ +/* Routine Description: */ +/* helper function for ordering adapters */ +/****************************************************************************/ +static void +ips_shift_controllers(int lowindex, int highindex) +{ + ips_ha_t *ha_sav = ips_ha[highindex]; + struct Scsi_Host *sh_sav = ips_sh[highindex]; + int i; + + for (i = highindex; i > lowindex; i--) { + ips_ha[i] = ips_ha[i - 1]; + ips_sh[i] = ips_sh[i - 1]; + ips_ha[i]->host_num = i; + } + ha_sav->host_num = lowindex; + ips_ha[lowindex] = ha_sav; + ips_sh[lowindex] = sh_sav; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_order_controllers */ +/* */ +/* Routine Description: */ +/* place controllers is the "proper" boot order */ +/****************************************************************************/ +static void +ips_order_controllers(void) +{ + int i, j, tmp, position = 0; + IPS_NVRAM_P5 *nvram; + if (!ips_ha[0]) + return; + nvram = ips_ha[0]->nvram; + + if (nvram->adapter_order[0]) { + for (i = 1; i <= nvram->adapter_order[0]; i++) { + for (j = position; j < ips_num_controllers; j++) { + switch (ips_ha[j]->ad_type) { + case IPS_ADTYPE_SERVERAID6M: + case IPS_ADTYPE_SERVERAID7M: + if (nvram->adapter_order[i] == 'M') { + ips_shift_controllers(position, + j); + position++; + } + break; + case IPS_ADTYPE_SERVERAID4L: + case IPS_ADTYPE_SERVERAID4M: + case IPS_ADTYPE_SERVERAID4MX: + case IPS_ADTYPE_SERVERAID4LX: + if (nvram->adapter_order[i] == 'N') { + ips_shift_controllers(position, + j); + position++; + } + break; + case IPS_ADTYPE_SERVERAID6I: + case IPS_ADTYPE_SERVERAID5I2: + case IPS_ADTYPE_SERVERAID5I1: + case IPS_ADTYPE_SERVERAID7k: + if (nvram->adapter_order[i] == 'S') { + ips_shift_controllers(position, + j); + position++; + } + break; + case IPS_ADTYPE_SERVERAID: + case IPS_ADTYPE_SERVERAID2: + case IPS_ADTYPE_NAVAJO: + case IPS_ADTYPE_KIOWA: + case IPS_ADTYPE_SERVERAID3L: + case IPS_ADTYPE_SERVERAID3: + case IPS_ADTYPE_SERVERAID4H: + if (nvram->adapter_order[i] == 'A') { + ips_shift_controllers(position, + j); + position++; + } + break; + default: + break; + } + } + } + /* if adapter_order[0], then ordering is complete */ + return; + } + /* old bios, use older ordering */ + tmp = 0; + for (i = position; i < ips_num_controllers; i++) { + if (ips_ha[i]->ad_type == IPS_ADTYPE_SERVERAID5I2 || + ips_ha[i]->ad_type == IPS_ADTYPE_SERVERAID5I1) { + ips_shift_controllers(position, i); + position++; + tmp = 1; + } + } + /* if there were no 5I cards, then don't do any extra ordering */ + if (!tmp) + return; + for (i = position; i < ips_num_controllers; i++) { + if (ips_ha[i]->ad_type == IPS_ADTYPE_SERVERAID4L || + ips_ha[i]->ad_type == IPS_ADTYPE_SERVERAID4M || + ips_ha[i]->ad_type == IPS_ADTYPE_SERVERAID4LX || + ips_ha[i]->ad_type == IPS_ADTYPE_SERVERAID4MX) { + ips_shift_controllers(position, i); + position++; + } + } + + return; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_register_scsi */ +/* */ +/* Routine Description: */ +/* perform any registration and setup with the scsi layer */ +/****************************************************************************/ +static int +ips_register_scsi(int index) +{ + struct Scsi_Host *sh; + ips_ha_t *ha, *oldha = ips_ha[index]; + sh = scsi_host_alloc(&ips_driver_template, sizeof (ips_ha_t)); + if (!sh) { + IPS_PRINTK(KERN_WARNING, oldha->pcidev, + "Unable to register controller with SCSI subsystem\n"); + return -1; + } + ha = IPS_HA(sh); + memcpy(ha, oldha, sizeof (ips_ha_t)); + free_irq(oldha->irq, oldha); + /* Install the interrupt handler with the new ha */ + if (request_irq(ha->irq, do_ipsintr, SA_SHIRQ, ips_name, ha)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Unable to install interrupt handler\n"); + scsi_host_put(sh); + return -1; + } + + kfree(oldha); + ips_sh[index] = sh; + ips_ha[index] = ha; + IPS_SCSI_SET_DEVICE(sh, ha); + + /* Store away needed values for later use */ + sh->io_port = ha->io_addr; + sh->n_io_port = ha->io_addr ? 255 : 0; + sh->unique_id = (ha->io_addr) ? ha->io_addr : ha->mem_addr; + sh->irq = ha->irq; + sh->sg_tablesize = sh->hostt->sg_tablesize; + sh->can_queue = sh->hostt->can_queue; + sh->cmd_per_lun = sh->hostt->cmd_per_lun; + sh->unchecked_isa_dma = sh->hostt->unchecked_isa_dma; + sh->use_clustering = sh->hostt->use_clustering; + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,7) + sh->max_sectors = 128; +#endif + + sh->max_id = ha->ntargets; + sh->max_lun = ha->nlun; + sh->max_channel = ha->nbus - 1; + sh->can_queue = ha->max_cmds - 1; + + IPS_ADD_HOST(sh, NULL); + return 0; +} + +/*---------------------------------------------------------------------------*/ +/* Routine Name: ips_remove_device */ +/* */ +/* Routine Description: */ +/* Remove one Adapter ( Hot Plugging ) */ +/*---------------------------------------------------------------------------*/ +static void __devexit +ips_remove_device(struct pci_dev *pci_dev) +{ + int i; + struct Scsi_Host *sh; + ips_ha_t *ha; + + for (i = 0; i < IPS_MAX_ADAPTERS; i++) { + ha = ips_ha[i]; + if (ha) { + if ((pci_dev->bus->number == ha->pcidev->bus->number) && + (pci_dev->devfn == ha->pcidev->devfn)) { + sh = ips_sh[i]; + ips_release(sh); + } + } + } +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_module_init */ +/* */ +/* Routine Description: */ +/* function called on module load */ +/****************************************************************************/ +static int __init +ips_module_init(void) +{ + if (pci_module_init(&ips_pci_driver) < 0) + return -ENODEV; + ips_driver_template.module = THIS_MODULE; + ips_order_controllers(); + if (IPS_REGISTER_HOSTS(&ips_driver_template)) { + pci_unregister_driver(&ips_pci_driver); + return -ENODEV; + } + register_reboot_notifier(&ips_notifier); + return 0; +} + +/****************************************************************************/ +/* */ +/* Routine Name: ips_module_exit */ +/* */ +/* Routine Description: */ +/* function called on module unload */ +/****************************************************************************/ +static void __exit +ips_module_exit(void) +{ + IPS_UNREGISTER_HOSTS(&ips_driver_template); + pci_unregister_driver(&ips_pci_driver); + unregister_reboot_notifier(&ips_notifier); +} + +module_init(ips_module_init); +module_exit(ips_module_exit); + +/*---------------------------------------------------------------------------*/ +/* Routine Name: ips_insert_device */ +/* */ +/* Routine Description: */ +/* Add One Adapter ( Hot Plug ) */ +/* */ +/* Return Value: */ +/* 0 if Successful, else non-zero */ +/*---------------------------------------------------------------------------*/ +static int __devinit +ips_insert_device(struct pci_dev *pci_dev, const struct pci_device_id *ent) +{ + int index; + int rc; + + METHOD_TRACE("ips_insert_device", 1); + if (pci_enable_device(pci_dev)) + return -1; + + rc = ips_init_phase1(pci_dev, &index); + if (rc == SUCCESS) + rc = ips_init_phase2(index); + + if (ips_hotplug) + if (ips_register_scsi(index)) { + ips_free(ips_ha[index]); + rc = -1; + } + + if (rc == SUCCESS) + ips_num_controllers++; + + ips_next_controller = ips_num_controllers; + return rc; +} + +/*---------------------------------------------------------------------------*/ +/* Routine Name: ips_init_phase1 */ +/* */ +/* Routine Description: */ +/* Adapter Initialization */ +/* */ +/* Return Value: */ +/* 0 if Successful, else non-zero */ +/*---------------------------------------------------------------------------*/ +static int +ips_init_phase1(struct pci_dev *pci_dev, int *indexPtr) +{ + ips_ha_t *ha; + uint32_t io_addr; + uint32_t mem_addr; + uint32_t io_len; + uint32_t mem_len; + uint8_t revision_id; + uint8_t bus; + uint8_t func; + uint8_t irq; + uint16_t subdevice_id; + int j; + int index; + dma_addr_t dma_address; + char __iomem *ioremap_ptr; + char __iomem *mem_ptr; + uint32_t IsDead; + + METHOD_TRACE("ips_init_phase1", 1); + index = IPS_MAX_ADAPTERS; + for (j = 0; j < IPS_MAX_ADAPTERS; j++) { + if (ips_ha[j] == 0) { + index = j; + break; + } + } + + if (index >= IPS_MAX_ADAPTERS) + return -1; + + /* stuff that we get in dev */ + irq = pci_dev->irq; + bus = pci_dev->bus->number; + func = pci_dev->devfn; + + /* Init MEM/IO addresses to 0 */ + mem_addr = 0; + io_addr = 0; + mem_len = 0; + io_len = 0; + + for (j = 0; j < 2; j++) { + if (!pci_resource_start(pci_dev, j)) + break; + + if (pci_resource_flags(pci_dev, j) & IORESOURCE_IO) { + io_addr = pci_resource_start(pci_dev, j); + io_len = pci_resource_len(pci_dev, j); + } else { + mem_addr = pci_resource_start(pci_dev, j); + mem_len = pci_resource_len(pci_dev, j); + } + } + + /* setup memory mapped area (if applicable) */ + if (mem_addr) { + uint32_t base; + uint32_t offs; + + if (!request_mem_region(mem_addr, mem_len, "ips")) { + IPS_PRINTK(KERN_WARNING, pci_dev, + "Couldn't allocate IO Memory space %x len %d.\n", + mem_addr, mem_len); + return -1; + } + + base = mem_addr & PAGE_MASK; + offs = mem_addr - base; + ioremap_ptr = ioremap(base, PAGE_SIZE); + mem_ptr = ioremap_ptr + offs; + } else { + ioremap_ptr = NULL; + mem_ptr = NULL; + } + + /* setup I/O mapped area (if applicable) */ + if (io_addr) { + if (!request_region(io_addr, io_len, "ips")) { + IPS_PRINTK(KERN_WARNING, pci_dev, + "Couldn't allocate IO space %x len %d.\n", + io_addr, io_len); + return -1; + } + } + + /* get the revision ID */ + if (pci_read_config_byte(pci_dev, PCI_REVISION_ID, &revision_id)) { + IPS_PRINTK(KERN_WARNING, pci_dev, "Can't get revision id.\n"); + return -1; + } + + subdevice_id = pci_dev->subsystem_device; + + /* found a controller */ + ha = kmalloc(sizeof (ips_ha_t), GFP_KERNEL); + if (ha == NULL) { + IPS_PRINTK(KERN_WARNING, pci_dev, + "Unable to allocate temporary ha struct\n"); + return -1; + } + + memset(ha, 0, sizeof (ips_ha_t)); + + ips_sh[index] = NULL; + ips_ha[index] = ha; + ha->active = 1; + + /* Store info in HA structure */ + ha->irq = irq; + ha->io_addr = io_addr; + ha->io_len = io_len; + ha->mem_addr = mem_addr; + ha->mem_len = mem_len; + ha->mem_ptr = mem_ptr; + ha->ioremap_ptr = ioremap_ptr; + ha->host_num = (uint32_t) index; + ha->revision_id = revision_id; + ha->slot_num = PCI_SLOT(pci_dev->devfn); + ha->device_id = pci_dev->device; + ha->subdevice_id = subdevice_id; + ha->pcidev = pci_dev; + + /* + * Set the pci_dev's dma_mask. Not all adapters support 64bit + * addressing so don't enable it if the adapter can't support + * it! Also, don't use 64bit addressing if dma addresses + * are guaranteed to be < 4G. + */ + if (IPS_ENABLE_DMA64 && IPS_HAS_ENH_SGLIST(ha) && + !pci_set_dma_mask(ha->pcidev, 0xffffffffffffffffULL)) { + (ha)->flags |= IPS_HA_ENH_SG; + } else { + if (pci_set_dma_mask(ha->pcidev, 0xffffffffULL) != 0) { + printk(KERN_WARNING "Unable to set DMA Mask\n"); + return ips_abort_init(ha, index); + } + } + if(ips_cd_boot && !ips_FlashData){ + ips_FlashData = pci_alloc_consistent(pci_dev, PAGE_SIZE << 7, + &ips_flashbusaddr); + } + + ha->enq = pci_alloc_consistent(pci_dev, sizeof (IPS_ENQ), + &ha->enq_busaddr); + if (!ha->enq) { + IPS_PRINTK(KERN_WARNING, pci_dev, + "Unable to allocate host inquiry structure\n"); + return ips_abort_init(ha, index); + } + + ha->adapt = pci_alloc_consistent(pci_dev, sizeof (IPS_ADAPTER) + + sizeof (IPS_IO_CMD), &dma_address); + if (!ha->adapt) { + IPS_PRINTK(KERN_WARNING, pci_dev, + "Unable to allocate host adapt & dummy structures\n"); + return ips_abort_init(ha, index); + } + ha->adapt->hw_status_start = dma_address; + ha->dummy = (void *) (ha->adapt + 1); + + + + ha->logical_drive_info = pci_alloc_consistent(pci_dev, sizeof (IPS_LD_INFO), &dma_address); + if (!ha->logical_drive_info) { + IPS_PRINTK(KERN_WARNING, pci_dev, + "Unable to allocate logical drive info structure\n"); + return ips_abort_init(ha, index); + } + ha->logical_drive_info_dma_addr = dma_address; + + + ha->conf = kmalloc(sizeof (IPS_CONF), GFP_KERNEL); + + if (!ha->conf) { + IPS_PRINTK(KERN_WARNING, pci_dev, + "Unable to allocate host conf structure\n"); + return ips_abort_init(ha, index); + } + + ha->nvram = kmalloc(sizeof (IPS_NVRAM_P5), GFP_KERNEL); + + if (!ha->nvram) { + IPS_PRINTK(KERN_WARNING, pci_dev, + "Unable to allocate host NVRAM structure\n"); + return ips_abort_init(ha, index); + } + + ha->subsys = kmalloc(sizeof (IPS_SUBSYS), GFP_KERNEL); + + if (!ha->subsys) { + IPS_PRINTK(KERN_WARNING, pci_dev, + "Unable to allocate host subsystem structure\n"); + return ips_abort_init(ha, index); + } + + /* the ioctl buffer is now used during adapter initialization, so its + * successful allocation is now required */ + if (ips_ioctlsize < PAGE_SIZE) + ips_ioctlsize = PAGE_SIZE; + + ha->ioctl_data = pci_alloc_consistent(pci_dev, ips_ioctlsize, + &ha->ioctl_busaddr); + ha->ioctl_len = ips_ioctlsize; + if (!ha->ioctl_data) { + IPS_PRINTK(KERN_WARNING, pci_dev, + "Unable to allocate IOCTL data\n"); + return ips_abort_init(ha, index); + } + + /* + * Setup Functions + */ + ips_setup_funclist(ha); + + if ((IPS_IS_MORPHEUS(ha)) || (IPS_IS_MARCO(ha))) { + /* If Morpheus appears dead, reset it */ + IsDead = readl(ha->mem_ptr + IPS_REG_I960_MSG1); + if (IsDead == 0xDEADBEEF) { + ips_reset_morpheus(ha); + } + } + + /* + * Initialize the card if it isn't already + */ + + if (!(*ha->func.isinit) (ha)) { + if (!(*ha->func.init) (ha)) { + /* + * Initialization failed + */ + IPS_PRINTK(KERN_WARNING, pci_dev, + "Unable to initialize controller\n"); + return ips_abort_init(ha, index); + } + } + + *indexPtr = index; + return SUCCESS; +} + +/*---------------------------------------------------------------------------*/ +/* Routine Name: ips_init_phase2 */ +/* */ +/* Routine Description: */ +/* Adapter Initialization Phase 2 */ +/* */ +/* Return Value: */ +/* 0 if Successful, else non-zero */ +/*---------------------------------------------------------------------------*/ +static int +ips_init_phase2(int index) +{ + ips_ha_t *ha; + + ha = ips_ha[index]; + + METHOD_TRACE("ips_init_phase2", 1); + if (!ha->active) { + ips_ha[index] = NULL; + return -1; + } + + /* Install the interrupt handler */ + if (request_irq(ha->irq, do_ipsintr, SA_SHIRQ, ips_name, ha)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Unable to install interrupt handler\n"); + return ips_abort_init(ha, index); + } + + /* + * Allocate a temporary SCB for initialization + */ + ha->max_cmds = 1; + if (!ips_allocatescbs(ha)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Unable to allocate a CCB\n"); + free_irq(ha->irq, ha); + return ips_abort_init(ha, index); + } + + if (!ips_hainit(ha)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Unable to initialize controller\n"); + free_irq(ha->irq, ha); + return ips_abort_init(ha, index); + } + /* Free the temporary SCB */ + ips_deallocatescbs(ha, 1); + + /* allocate CCBs */ + if (!ips_allocatescbs(ha)) { + IPS_PRINTK(KERN_WARNING, ha->pcidev, + "Unable to allocate CCBs\n"); + free_irq(ha->irq, ha); + return ips_abort_init(ha, index); + } + + return SUCCESS; +} + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,9) +MODULE_LICENSE("GPL"); +#endif + +MODULE_DESCRIPTION("IBM ServeRAID Adapter Driver " IPS_VER_STRING); + +#ifdef MODULE_VERSION +MODULE_VERSION(IPS_VER_STRING); +#endif + + +/* + * Overrides for Emacs so that we almost follow Linus's tabbing style. + * Emacs will notice this stuff at the end of the file and automatically + * adjust the settings for this buffer only. This must remain at the end + * of the file. + * --------------------------------------------------------------------------- + * Local variables: + * c-indent-level: 2 + * c-brace-imaginary-offset: 0 + * c-brace-offset: -2 + * c-argdecl-indent: 2 + * c-label-offset: -2 + * c-continued-statement-offset: 2 + * c-continued-brace-offset: 0 + * indent-tabs-mode: nil + * tab-width: 8 + * End: + */ |