d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
9aa2d5e4fa
freebsd-skq/sys/dev/iir/iir.c

1915 lines
66 KiB
C
Raw Normal View History

Start each of the license/copyright comments with /*-, minor shuffle of lines
2005-01-06 01:43:34 +00:00
/*-
sys/dev: further adoption of SPDX licensing ID tags. Mainly focus on files that use BSD 2-Clause license, however the tool I was using misidentified many licenses so this was mostly a manual - error prone - task. The Software Package Data Exchange (SPDX) group provides a specification to make it easier for automated tools to detect and summarize well known opensource licenses. We are gradually adopting the specification, noting that the tags are considered only advisory and do not, in any way, superceed or replace the license texts.
2017-11-27 14:52:40 +00:00
* SPDX-License-Identifier: BSD-3-Clause
*
Update from vendor. This also adds support for newer management tools. Submitted by: Achim Leubner
2004-05-19 17:46:34 +00:00
* Copyright (c) 2000-04 ICP vortex GmbH
* Copyright (c) 2002-04 Intel Corporation
* Copyright (c) 2003-04 Adaptec Inc.
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
* All Rights Reserved
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions, and the following disclaimer,
* without modification, immediately at the beginning of the file.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* 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 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
/*
sys/dev: minor spelling fixes. Most affect comments, very few have user-visible effects.
2016-05-03 03:41:25 +00:00
* iir.c: SCSI dependent code for the Intel Integrated RAID Controller driver
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
*
Update email addresses, copyrights, and tweak the management interface. Submitted by: "Leubner, Achim" <Achim_Leubner@adaptec.com>
2003-09-26 15:36:47 +00:00
* Written by: Achim Leubner <achim_leubner@adaptec.com>
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
* Fixes/Additions: Boji Tony Kannanthanam <boji.t.kannanthanam@intel.com>
*
* credits: Niklas Hallqvist; OpenBSD driver for the ICP Controllers.
* Mike Smith; Some driver source code.
* FreeBSD.ORG; Great O/S to work on and for.
*
Update from vendor. This also adds support for newer management tools. Submitted by: Achim Leubner
2004-05-19 17:46:34 +00:00
* $Id: iir.c 1.5 2004/03/30 10:17:53 achim Exp $"
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
*/
Use __FBSDID(). Also some minor style cleanups.
2003-08-24 17:55:58 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
#define _IIR_C_
/* #include "opt_iir.h" */
#include <sys/param.h>
#include <sys/systm.h>
Move the new byte order function prototypes from <sys/param.h> to <sys/endian.h>. This puts us in line with NetBSD and OpenBSD.
2002-04-26 22:48:23 +00:00
#include <sys/endian.h>
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
#include <sys/eventhandler.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
#include <sys/bus.h>
#include <machine/bus.h>
#include <machine/stdarg.h>
#include <cam/cam.h>
#include <cam/cam_ccb.h>
#include <cam/cam_sim.h>
#include <cam/cam_xpt_sim.h>
#include <cam/cam_debug.h>
#include <cam/scsi/scsi_all.h>
#include <cam/scsi/scsi_message.h>
#include <dev/iir/iir.h>
Mark MALLOC_DEFINEs static that have no corresponding MALLOC_DECLAREs. This means that their use is restricted to a single C file.
2011-11-07 06:44:47 +00:00
static MALLOC_DEFINE(M_GDTBUF, "iirbuf", "iir driver buffer");
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
#ifdef GDT_DEBUG
int gdt_debug = GDT_DEBUG;
#ifdef __SERIAL__
#define MAX_SERBUF 160
static void ser_init(void);
static void ser_puts(char *str);
static void ser_putc(int c);
static char strbuf[MAX_SERBUF+1];
#ifdef __COM2__
#define COM_BASE 0x2f8
#else
#define COM_BASE 0x3f8
#endif
static void ser_init()
{
unsigned port=COM_BASE;
outb(port+3, 0x80);
outb(port+1, 0);
/* 19200 Baud, if 9600: outb(12,port) */
outb(port, 6);
outb(port+3, 3);
outb(port+1, 0);
}
static void ser_puts(char *str)
{
char *ptr;
ser_init();
for (ptr=str;*ptr;++ptr)
ser_putc((int)(*ptr));
}
static void ser_putc(int c)
{
unsigned port=COM_BASE;
while ((inb(port+5) & 0x20)==0);
outb(port, c);
if (c==0x0a)
{
while ((inb(port+5) & 0x20)==0);
outb(port, 0x0d);
}
}
int ser_printf(const char *fmt, ...)
{
va_list args;
int i;
va_start(args,fmt);
i = vsprintf(strbuf,fmt,args);
ser_puts(strbuf);
va_end(args);
return i;
}
#endif
#endif
/* controller cnt. */
int gdt_cnt = 0;
/* event buffer */
static gdt_evt_str ebuffer[GDT_MAX_EVENTS];
static int elastidx, eoldidx;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
static struct mtx elock;
MTX_SYSINIT(iir_elock, &elock, "iir events", MTX_DEF);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/* statistics */
gdt_statist_t gdt_stat;
/* Definitions for our use of the SIM private CCB area */
#define ccb_sim_ptr spriv_ptr0
#define ccb_priority spriv_field1
static void iir_action(struct cam_sim *sim, union ccb *ccb);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
static int iir_intr_locked(struct gdt_softc *gdt);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
static void iir_poll(struct cam_sim *sim);
static void iir_shutdown(void *arg, int howto);
static void iir_timeout(void *arg);
static void gdt_eval_mapping(u_int32_t size, int *cyls, int *heads,
int *secs);
static int gdt_internal_cmd(struct gdt_softc *gdt, struct gdt_ccb *gccb,
u_int8_t service, u_int16_t opcode,
u_int32_t arg1, u_int32_t arg2, u_int32_t arg3);
static int gdt_wait(struct gdt_softc *gdt, struct gdt_ccb *ccb,
int timeout);
static struct gdt_ccb *gdt_get_ccb(struct gdt_softc *gdt);
static int gdt_sync_event(struct gdt_softc *gdt, int service,
u_int8_t index, struct gdt_ccb *gccb);
static int gdt_async_event(struct gdt_softc *gdt, int service);
static struct gdt_ccb *gdt_raw_cmd(struct gdt_softc *gdt,
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
union ccb *ccb);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
static struct gdt_ccb *gdt_cache_cmd(struct gdt_softc *gdt,
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
union ccb *ccb);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
static struct gdt_ccb *gdt_ioctl_cmd(struct gdt_softc *gdt,
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
gdt_ucmd_t *ucmd);
static void gdt_internal_cache_cmd(struct gdt_softc *gdt, union ccb *ccb);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
static void gdtmapmem(void *arg, bus_dma_segment_t *dm_segs,
int nseg, int error);
static void gdtexecuteccb(void *arg, bus_dma_segment_t *dm_segs,
int nseg, int error);
int
iir_init(struct gdt_softc *gdt)
{
u_int16_t cdev_cnt;
int i, id, drv_cyls, drv_hds, drv_secs;
struct gdt_ccb *gccb;
GDT_DPRINTF(GDT_D_DEBUG, ("iir_init()\n"));
gdt->sc_state = GDT_POLLING;
gdt_clear_events();
bzero(&gdt_stat, sizeof(gdt_statist_t));
SLIST_INIT(&gdt->sc_free_gccb);
SLIST_INIT(&gdt->sc_pending_gccb);
TAILQ_INIT(&gdt->sc_ccb_queue);
TAILQ_INIT(&gdt->sc_ucmd_queue);
/* DMA tag for mapping buffers into device visible space. */
if (bus_dma_tag_create(gdt->sc_parent_dmat, /*alignment*/1, /*boundary*/0,
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
/*lowaddr*/BUS_SPACE_MAXADDR_32BIT,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/*highaddr*/BUS_SPACE_MAXADDR,
/*filter*/NULL, /*filterarg*/NULL,
Remove MAXBSIZE use from drivers where it has nothing to do. In some cases limits are just not needed, in others -- DFLTPHYS is the right constant to use instead. MFC after: 1 month
2015-03-22 16:10:28 +00:00
/*maxsize*/DFLTPHYS,
/*nsegments*/GDT_MAXSG,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/*maxsegsz*/BUS_SPACE_MAXSIZE_32BIT,
/*flags*/BUS_DMA_ALLOCNOW,
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
/*lockfunc*/busdma_lock_mutex,
/*lockarg*/&gdt->sc_lock,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
&gdt->sc_buffer_dmat) != 0) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode,
"bus_dma_tag_create(..., gdt->sc_buffer_dmat) failed\n");
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
gdt->sc_init_level++;
/* DMA tag for our ccb structures */
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
if (bus_dma_tag_create(gdt->sc_parent_dmat,
/*alignment*/1,
/*boundary*/0,
/*lowaddr*/BUS_SPACE_MAXADDR_32BIT,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/*highaddr*/BUS_SPACE_MAXADDR,
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
/*filter*/NULL,
/*filterarg*/NULL,
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
GDT_MAXCMDS * GDT_SCRATCH_SZ, /* maxsize */
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/*nsegments*/1,
/*maxsegsz*/BUS_SPACE_MAXSIZE_32BIT,
Mega busdma API commit. Add two new arguments to bus_dma_tag_create(): lockfunc and lockfuncarg. Lockfunc allows a driver to provide a function for managing its locking semantics while using busdma. At the moment, this is used for the asynchronous busdma_swi and callback mechanism. Two lockfunc implementations are provided: busdma_lock_mutex() performs standard mutex operations on the mutex that is specified from lockfuncarg. dftl_lock() is a panic implementation and is defaulted to when NULL, NULL are passed to bus_dma_tag_create(). The only time that NULL, NULL should ever be used is when the driver ensures that bus_dmamap_load() will not be deferred. Drivers that do not provide their own locking can pass busdma_lock_mutex,&Giant args in order to preserve the former behaviour. sparc64 and powerpc do not provide real busdma_swi functions, so this is largely a noop on those platforms. The busdma_swi on is64 is not properly locked yet, so warnings will be emitted on this platform when busdma callback deferrals happen. If anyone gets panics or warnings from dflt_lock() being called, please let me know right away. Reviewed by: tmm, gibbs
2003-07-01 15:52:06 +00:00
/*flags*/0, /*lockfunc*/busdma_lock_mutex,
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
/*lockarg*/&gdt->sc_lock,
&gdt->sc_gcscratch_dmat) != 0) {
device_printf(gdt->sc_devnode,
"bus_dma_tag_create(...,gdt->sc_gcscratch_dmat) failed\n");
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
gdt->sc_init_level++;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
/* Allocation for our ccb scratch area */
if (bus_dmamem_alloc(gdt->sc_gcscratch_dmat, (void **)&gdt->sc_gcscratch,
BUS_DMA_NOWAIT, &gdt->sc_gcscratch_dmamap) != 0) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode,
"bus_dmamem_alloc(...,&gdt->sc_gccbs,...) failed\n");
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
gdt->sc_init_level++;
/* And permanently map them */
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bus_dmamap_load(gdt->sc_gcscratch_dmat, gdt->sc_gcscratch_dmamap,
gdt->sc_gcscratch, GDT_MAXCMDS * GDT_SCRATCH_SZ,
gdtmapmem, &gdt->sc_gcscratch_busbase, /*flags*/0);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_init_level++;
/* Clear them out. */
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bzero(gdt->sc_gcscratch, GDT_MAXCMDS * GDT_SCRATCH_SZ);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/* Initialize the ccbs */
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt->sc_gccbs = malloc(sizeof(struct gdt_ccb) * GDT_MAXCMDS, M_GDTBUF,
M_NOWAIT | M_ZERO);
if (gdt->sc_gccbs == NULL) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode, "no memory for gccbs.\n");
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
return (1);
}
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
for (i = GDT_MAXCMDS-1; i >= 0; i--) {
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb = &gdt->sc_gccbs[i];
gccb->gc_cmd_index = i + 2;
gccb->gc_flags = GDT_GCF_UNUSED;
gccb->gc_map_flag = FALSE;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
if (bus_dmamap_create(gdt->sc_buffer_dmat, /*flags*/0,
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
&gccb->gc_dmamap) != 0)
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return(1);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_map_flag = TRUE;
Prevent overflow issues in timeout processing Previously, any timeout value for which (timeout * hz) will overflow the signed integer, will give weird results, since callout(9) routines will convert negative values of ticks to '1'. For unsigned integer overflow we will get sufficiently smaller timeout values than expected. Switch from callout_reset, which requires conversion to int based ticks to callout_reset_sbt to avoid this. Also correct isci to correctly resolve ccb timeout. This was based on the original work done by Eygene Ryabinkin <rea@freebsd.org> back in 5 Aug 2011 which used a macro to help avoid the overlow. Differential Revision: https://reviews.freebsd.org/D1157 Reviewed by: mav, davide MFC after: 1 month Sponsored by: Multiplay
2014-11-21 21:01:24 +00:00
gccb->gc_scratch = &gdt->sc_gcscratch[GDT_SCRATCH_SZ * i];
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_scratch_busbase = gdt->sc_gcscratch_busbase + GDT_SCRATCH_SZ * i;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
callout_init_mtx(&gccb->gc_timeout, &gdt->sc_lock, 0);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
SLIST_INSERT_HEAD(&gdt->sc_free_gccb, gccb, sle);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
gdt->sc_init_level++;
/* create the control device */
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
gdt->sc_dev = gdt_make_dev(gdt);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/* allocate ccb for gdt_internal_cmd() */
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_lock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb = gdt_get_ccb(gdt);
if (gccb == NULL) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
device_printf(gdt->sc_devnode, "No free command index found\n");
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bzero(gccb->gc_cmd, GDT_CMD_SZ);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
if (!gdt_internal_cmd(gdt, gccb, GDT_SCREENSERVICE, GDT_INIT,
0, 0, 0)) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode,
"Screen service initialization error %d\n", gdt->sc_status);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_free_ccb(gdt, gccb);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
gdt_internal_cmd(gdt, gccb, GDT_CACHESERVICE, GDT_UNFREEZE_IO,
0, 0, 0);
if (!gdt_internal_cmd(gdt, gccb, GDT_CACHESERVICE, GDT_INIT,
GDT_LINUX_OS, 0, 0)) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode, "Cache service initialization error %d\n",
gdt->sc_status);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_free_ccb(gdt, gccb);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
cdev_cnt = (u_int16_t)gdt->sc_info;
gdt->sc_fw_vers = gdt->sc_service;
/* Detect number of buses */
gdt_enc32(gccb->gc_scratch + GDT_IOC_VERSION, GDT_IOC_NEWEST);
gccb->gc_scratch[GDT_IOC_LIST_ENTRIES] = GDT_MAXBUS;
gccb->gc_scratch[GDT_IOC_FIRST_CHAN] = 0;
gccb->gc_scratch[GDT_IOC_LAST_CHAN] = GDT_MAXBUS - 1;
gdt_enc32(gccb->gc_scratch + GDT_IOC_LIST_OFFSET, GDT_IOC_HDR_SZ);
if (gdt_internal_cmd(gdt, gccb, GDT_CACHESERVICE, GDT_IOCTL,
GDT_IOCHAN_RAW_DESC, GDT_INVALID_CHANNEL,
GDT_IOC_HDR_SZ + GDT_MAXBUS * GDT_RAWIOC_SZ)) {
gdt->sc_bus_cnt = gccb->gc_scratch[GDT_IOC_CHAN_COUNT];
for (i = 0; i < gdt->sc_bus_cnt; i++) {
id = gccb->gc_scratch[GDT_IOC_HDR_SZ +
i * GDT_RAWIOC_SZ + GDT_RAWIOC_PROC_ID];
gdt->sc_bus_id[i] = id < GDT_MAXID_FC ? id : 0xff;
}
} else {
/* New method failed, use fallback. */
for (i = 0; i < GDT_MAXBUS; i++) {
gdt_enc32(gccb->gc_scratch + GDT_GETCH_CHANNEL_NO, i);
if (!gdt_internal_cmd(gdt, gccb, GDT_CACHESERVICE, GDT_IOCTL,
GDT_SCSI_CHAN_CNT | GDT_L_CTRL_PATTERN,
GDT_IO_CHANNEL | GDT_INVALID_CHANNEL,
GDT_GETCH_SZ)) {
if (i == 0) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode, "Cannot get channel count, "
"error %d\n", gdt->sc_status);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_free_ccb(gdt, gccb);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
break;
}
gdt->sc_bus_id[i] =
(gccb->gc_scratch[GDT_GETCH_SIOP_ID] < GDT_MAXID_FC) ?
gccb->gc_scratch[GDT_GETCH_SIOP_ID] : 0xff;
}
gdt->sc_bus_cnt = i;
}
/* add one "virtual" channel for the host drives */
gdt->sc_virt_bus = gdt->sc_bus_cnt;
gdt->sc_bus_cnt++;
if (!gdt_internal_cmd(gdt, gccb, GDT_SCSIRAWSERVICE, GDT_INIT,
0, 0, 0)) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode,
"Raw service initialization error %d\n", gdt->sc_status);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_free_ccb(gdt, gccb);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
/* Set/get features raw service (scatter/gather) */
gdt->sc_raw_feat = 0;
if (gdt_internal_cmd(gdt, gccb, GDT_SCSIRAWSERVICE, GDT_SET_FEAT,
GDT_SCATTER_GATHER, 0, 0)) {
if (gdt_internal_cmd(gdt, gccb, GDT_SCSIRAWSERVICE, GDT_GET_FEAT,
0, 0, 0)) {
gdt->sc_raw_feat = gdt->sc_info;
if (!(gdt->sc_info & GDT_SCATTER_GATHER)) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
panic("%s: Scatter/Gather Raw Service "
"required but not supported!\n",
device_get_nameunit(gdt->sc_devnode));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_free_ccb(gdt, gccb);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
}
}
/* Set/get features cache service (scatter/gather) */
gdt->sc_cache_feat = 0;
if (gdt_internal_cmd(gdt, gccb, GDT_CACHESERVICE, GDT_SET_FEAT,
0, GDT_SCATTER_GATHER, 0)) {
if (gdt_internal_cmd(gdt, gccb, GDT_CACHESERVICE, GDT_GET_FEAT,
0, 0, 0)) {
gdt->sc_cache_feat = gdt->sc_info;
if (!(gdt->sc_info & GDT_SCATTER_GATHER)) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
panic("%s: Scatter/Gather Cache Service "
"required but not supported!\n",
device_get_nameunit(gdt->sc_devnode));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_free_ccb(gdt, gccb);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
}
}
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
/* OEM */
gdt_enc32(gccb->gc_scratch + GDT_OEM_VERSION, 0x01);
gdt_enc32(gccb->gc_scratch + GDT_OEM_BUFSIZE, sizeof(gdt_oem_record_t));
if (gdt_internal_cmd(gdt, gccb, GDT_CACHESERVICE, GDT_IOCTL,
GDT_OEM_STR_RECORD, GDT_INVALID_CHANNEL,
sizeof(gdt_oem_str_record_t))) {
strncpy(gdt->oem_name, ((gdt_oem_str_record_t *)
gccb->gc_scratch)->text.scsi_host_drive_inquiry_vendor_id, 7);
gdt->oem_name[7]='\0';
} else {
/* Old method, based on PCI ID */
Avoid potential redefinition of the macro.
2013-08-15 20:03:22 +00:00
if (gdt->sc_vendor == INTEL_VENDOR_ID_IIR)
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
strcpy(gdt->oem_name,"Intel ");
else
strcpy(gdt->oem_name,"ICP ");
}
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/* Scan for cache devices */
for (i = 0; i < cdev_cnt && i < GDT_MAX_HDRIVES; i++) {
if (gdt_internal_cmd(gdt, gccb, GDT_CACHESERVICE, GDT_INFO,
i, 0, 0)) {
gdt->sc_hdr[i].hd_present = 1;
gdt->sc_hdr[i].hd_size = gdt->sc_info;
/*
* Evaluate mapping (sectors per head, heads per cyl)
*/
gdt->sc_hdr[i].hd_size &= ~GDT_SECS32;
if (gdt->sc_info2 == 0)
gdt_eval_mapping(gdt->sc_hdr[i].hd_size,
&drv_cyls, &drv_hds, &drv_secs);
else {
drv_hds = gdt->sc_info2 & 0xff;
drv_secs = (gdt->sc_info2 >> 8) & 0xff;
drv_cyls = gdt->sc_hdr[i].hd_size / drv_hds /
drv_secs;
}
gdt->sc_hdr[i].hd_heads = drv_hds;
gdt->sc_hdr[i].hd_secs = drv_secs;
/* Round the size */
gdt->sc_hdr[i].hd_size = drv_cyls * drv_hds * drv_secs;
if (gdt_internal_cmd(gdt, gccb, GDT_CACHESERVICE,
GDT_DEVTYPE, i, 0, 0))
gdt->sc_hdr[i].hd_devtype = gdt->sc_info;
}
}
GDT_DPRINTF(GDT_D_INIT, ("dpmem %x %d-bus %d cache device%s\n",
gdt->sc_dpmembase,
gdt->sc_bus_cnt, cdev_cnt,
cdev_cnt == 1 ? "" : "s"));
gdt_free_ccb(gdt, gccb);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
atomic_add_int(&gdt_cnt, 1);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (0);
}
void
iir_free(struct gdt_softc *gdt)
{
int i;
GDT_DPRINTF(GDT_D_INIT, ("iir_free()\n"));
switch (gdt->sc_init_level) {
default:
gdt_destroy_dev(gdt->sc_dev);
case 5:
for (i = GDT_MAXCMDS-1; i >= 0; i--)
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
if (gdt->sc_gccbs[i].gc_map_flag) {
callout_drain(&gdt->sc_gccbs[i].gc_timeout);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
bus_dmamap_destroy(gdt->sc_buffer_dmat,
gdt->sc_gccbs[i].gc_dmamap);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bus_dmamap_unload(gdt->sc_gcscratch_dmat, gdt->sc_gcscratch_dmamap);
free(gdt->sc_gccbs, M_GDTBUF);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
case 4:
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bus_dmamem_free(gdt->sc_gcscratch_dmat, gdt->sc_gcscratch, gdt->sc_gcscratch_dmamap);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
case 3:
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bus_dma_tag_destroy(gdt->sc_gcscratch_dmat);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
case 2:
bus_dma_tag_destroy(gdt->sc_buffer_dmat);
case 1:
bus_dma_tag_destroy(gdt->sc_parent_dmat);
case 0:
break;
}
}
void
iir_attach(struct gdt_softc *gdt)
{
struct cam_devq *devq;
int i;
GDT_DPRINTF(GDT_D_INIT, ("iir_attach()\n"));
/*
* Create the device queue for our SIM.
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
* XXX Throttle this down since the card has problems under load.
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
*/
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
devq = cam_simq_alloc(32);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
if (devq == NULL)
return;
for (i = 0; i < gdt->sc_bus_cnt; i++) {
/*
* Construct our SIM entry
*/
gdt->sims[i] = cam_sim_alloc(iir_action, iir_poll, "iir",
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
gdt, device_get_unit(gdt->sc_devnode), &gdt->sc_lock,
/*untagged*/1, /*tagged*/GDT_MAXCMDS, devq);
mtx_lock(&gdt->sc_lock);
Prepare for future integration between CAM and newbus. xpt_bus_register now takes a device_t to be the parent of the bus that is being created. Most SIMs have been updated with a reasonable argument, but a few exceptions just pass NULL for now. This argument isn't used yet and the newbus integration likely won't be ready until after 7.0-RELEASE.
2007-06-17 05:55:54 +00:00
if (xpt_bus_register(gdt->sims[i], gdt->sc_devnode, i) != CAM_SUCCESS) {
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
cam_sim_free(gdt->sims[i], /*free_devq*/i == 0);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
break;
}
if (xpt_create_path(&gdt->paths[i], /*periph*/NULL,
cam_sim_path(gdt->sims[i]),
CAM_TARGET_WILDCARD,
CAM_LUN_WILDCARD) != CAM_REQ_CMP) {
xpt_bus_deregister(cam_sim_path(gdt->sims[i]));
cam_sim_free(gdt->sims[i], /*free_devq*/i == 0);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
break;
}
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
if (i > 0)
EVENTHANDLER_REGISTER(shutdown_final, iir_shutdown,
gdt, SHUTDOWN_PRI_DEFAULT);
/* iir_watchdog(gdt); */
gdt->sc_state = GDT_NORMAL;
}
static void
gdt_eval_mapping(u_int32_t size, int *cyls, int *heads, int *secs)
{
*cyls = size / GDT_HEADS / GDT_SECS;
if (*cyls < GDT_MAXCYLS) {
*heads = GDT_HEADS;
*secs = GDT_SECS;
} else {
/* Too high for 64 * 32 */
*cyls = size / GDT_MEDHEADS / GDT_MEDSECS;
if (*cyls < GDT_MAXCYLS) {
*heads = GDT_MEDHEADS;
*secs = GDT_MEDSECS;
} else {
/* Too high for 127 * 63 */
*cyls = size / GDT_BIGHEADS / GDT_BIGSECS;
*heads = GDT_BIGHEADS;
*secs = GDT_BIGSECS;
}
}
}
static int
gdt_wait(struct gdt_softc *gdt, struct gdt_ccb *gccb,
int timeout)
{
int rv = 0;
GDT_DPRINTF(GDT_D_INIT,
("gdt_wait(%p, %p, %d)\n", gdt, gccb, timeout));
gdt->sc_state |= GDT_POLL_WAIT;
do {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
if (iir_intr_locked(gdt) == gccb->gc_cmd_index) {
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
rv = 1;
break;
}
DELAY(1);
} while (--timeout);
gdt->sc_state &= ~GDT_POLL_WAIT;
while (gdt->sc_test_busy(gdt))
DELAY(1); /* XXX correct? */
return (rv);
}
static int
gdt_internal_cmd(struct gdt_softc *gdt, struct gdt_ccb *gccb,
u_int8_t service, u_int16_t opcode,
u_int32_t arg1, u_int32_t arg2, u_int32_t arg3)
{
int retries;
GDT_DPRINTF(GDT_D_CMD, ("gdt_internal_cmd(%p, %d, %d, %d, %d, %d)\n",
gdt, service, opcode, arg1, arg2, arg3));
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bzero(gccb->gc_cmd, GDT_CMD_SZ);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
for (retries = GDT_RETRIES; ; ) {
gccb->gc_service = service;
gccb->gc_flags = GDT_GCF_INTERNAL;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_COMMANDINDEX,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_cmd_index);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_OPCODE, opcode);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
switch (service) {
case GDT_CACHESERVICE:
if (opcode == GDT_IOCTL) {
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_IOCTL_SUBFUNC, arg1);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_IOCTL_CHANNEL, arg2);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_UNION +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_IOCTL_PARAM_SIZE, (u_int16_t)arg3);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_IOCTL_P_PARAM,
gccb->gc_scratch_busbase);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
} else {
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_UNION +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_CACHE_DEVICENO, (u_int16_t)arg1);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_CACHE_BLOCKNO, arg2);
}
break;
case GDT_SCSIRAWSERVICE:
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_RAW_DIRECTION, arg1);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd[GDT_CMD_UNION + GDT_RAW_BUS] =
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
(u_int8_t)arg2;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd[GDT_CMD_UNION + GDT_RAW_TARGET] =
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
(u_int8_t)arg3;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd[GDT_CMD_UNION + GDT_RAW_LUN] =
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
(u_int8_t)(arg3 >> 8);
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt->sc_set_sema0(gdt);
gccb->gc_cmd_len = GDT_CMD_SZ;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_cmd_off = 0;
gdt->sc_cmd_cnt = 0;
gdt->sc_copy_cmd(gdt, gccb);
gdt->sc_release_event(gdt);
DELAY(20);
if (!gdt_wait(gdt, gccb, GDT_POLL_TIMEOUT))
return (0);
if (gdt->sc_status != GDT_S_BSY || --retries == 0)
break;
DELAY(1);
}
return (gdt->sc_status == GDT_S_OK);
}
static struct gdt_ccb *
gdt_get_ccb(struct gdt_softc *gdt)
{
struct gdt_ccb *gccb;
GDT_DPRINTF(GDT_D_QUEUE, ("gdt_get_ccb(%p)\n", gdt));
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_assert(&gdt->sc_lock, MA_OWNED);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb = SLIST_FIRST(&gdt->sc_free_gccb);
if (gccb != NULL) {
SLIST_REMOVE_HEAD(&gdt->sc_free_gccb, sle);
SLIST_INSERT_HEAD(&gdt->sc_pending_gccb, gccb, sle);
++gdt_stat.cmd_index_act;
if (gdt_stat.cmd_index_act > gdt_stat.cmd_index_max)
gdt_stat.cmd_index_max = gdt_stat.cmd_index_act;
}
return (gccb);
}
void
gdt_free_ccb(struct gdt_softc *gdt, struct gdt_ccb *gccb)
{
GDT_DPRINTF(GDT_D_QUEUE, ("gdt_free_ccb(%p, %p)\n", gdt, gccb));
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_assert(&gdt->sc_lock, MA_OWNED);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_flags = GDT_GCF_UNUSED;
SLIST_REMOVE(&gdt->sc_pending_gccb, gccb, gdt_ccb, sle);
SLIST_INSERT_HEAD(&gdt->sc_free_gccb, gccb, sle);
--gdt_stat.cmd_index_act;
if (gdt->sc_state & GDT_SHUTDOWN)
wakeup(gccb);
}
void
gdt_next(struct gdt_softc *gdt)
{
union ccb *ccb;
gdt_ucmd_t *ucmd;
struct cam_sim *sim;
int bus, target, lun;
int next_cmd;
struct ccb_scsiio *csio;
struct ccb_hdr *ccbh;
struct gdt_ccb *gccb = NULL;
u_int8_t cmd;
GDT_DPRINTF(GDT_D_QUEUE, ("gdt_next(%p)\n", gdt));
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_assert(&gdt->sc_lock, MA_OWNED);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
if (gdt->sc_test_busy(gdt)) {
if (!(gdt->sc_state & GDT_POLLING)) {
return;
}
while (gdt->sc_test_busy(gdt))
DELAY(1);
}
gdt->sc_cmd_cnt = gdt->sc_cmd_off = 0;
next_cmd = TRUE;
for (;;) {
/* I/Os in queue? controller ready? */
if (!TAILQ_FIRST(&gdt->sc_ucmd_queue) &&
!TAILQ_FIRST(&gdt->sc_ccb_queue))
break;
/* 1.: I/Os without ccb (IOCTLs) */
ucmd = TAILQ_FIRST(&gdt->sc_ucmd_queue);
if (ucmd != NULL) {
TAILQ_REMOVE(&gdt->sc_ucmd_queue, ucmd, links);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
if ((gccb = gdt_ioctl_cmd(gdt, ucmd)) == NULL) {
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
TAILQ_INSERT_HEAD(&gdt->sc_ucmd_queue, ucmd, links);
break;
}
break;
/* wenn mehrere Kdos. zulassen: if (!gdt_polling) continue; */
}
/* 2.: I/Os with ccb */
ccb = (union ccb *)TAILQ_FIRST(&gdt->sc_ccb_queue);
/* ist dann immer != NULL, da oben getestet */
sim = (struct cam_sim *)ccb->ccb_h.ccb_sim_ptr;
bus = cam_sim_bus(sim);
target = ccb->ccb_h.target_id;
lun = ccb->ccb_h.target_lun;
TAILQ_REMOVE(&gdt->sc_ccb_queue, &ccb->ccb_h, sim_links.tqe);
--gdt_stat.req_queue_act;
/* ccb->ccb_h.func_code is XPT_SCSI_IO */
GDT_DPRINTF(GDT_D_QUEUE, ("XPT_SCSI_IO flags 0x%x)\n",
ccb->ccb_h.flags));
csio = &ccb->csio;
ccbh = &ccb->ccb_h;
Make sure we check for CAM_CDB_POINTER for all drivers. Also, for the drivers I've touched, filter out CAM_CDB_PHYS. Differential Revision: https://reviews.freebsd.org/D5585
2016-03-15 05:17:29 +00:00
cmd = scsiio_cdb_ptr(csio)[0];
/* Max CDB length is 12 bytes, can't be phys addr */
if (csio->cdb_len > 12 || (ccbh->flags & CAM_CDB_PHYS)) {
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccbh->status = CAM_REQ_INVALID;
--gdt_stat.io_count_act;
xpt_done(ccb);
} else if (bus != gdt->sc_virt_bus) {
/* raw service command */
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
if ((gccb = gdt_raw_cmd(gdt, ccb)) == NULL) {
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
TAILQ_INSERT_HEAD(&gdt->sc_ccb_queue, &ccb->ccb_h,
sim_links.tqe);
++gdt_stat.req_queue_act;
if (gdt_stat.req_queue_act > gdt_stat.req_queue_max)
gdt_stat.req_queue_max = gdt_stat.req_queue_act;
next_cmd = FALSE;
}
} else if (target >= GDT_MAX_HDRIVES ||
!gdt->sc_hdr[target].hd_present || lun != 0) {
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
ccbh->status = CAM_DEV_NOT_THERE;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
--gdt_stat.io_count_act;
xpt_done(ccb);
} else {
/* cache service command */
if (cmd == READ_6 || cmd == WRITE_6 ||
cmd == READ_10 || cmd == WRITE_10) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
if ((gccb = gdt_cache_cmd(gdt, ccb)) == NULL) {
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
TAILQ_INSERT_HEAD(&gdt->sc_ccb_queue, &ccb->ccb_h,
sim_links.tqe);
++gdt_stat.req_queue_act;
if (gdt_stat.req_queue_act > gdt_stat.req_queue_max)
gdt_stat.req_queue_max = gdt_stat.req_queue_act;
next_cmd = FALSE;
}
} else {
gdt_internal_cache_cmd(gdt, ccb);
}
}
if ((gdt->sc_state & GDT_POLLING) || !next_cmd)
break;
}
if (gdt->sc_cmd_cnt > 0)
gdt->sc_release_event(gdt);
if ((gdt->sc_state & GDT_POLLING) && gdt->sc_cmd_cnt > 0) {
gdt_wait(gdt, gccb, GDT_POLL_TIMEOUT);
}
}
static struct gdt_ccb *
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
gdt_raw_cmd(struct gdt_softc *gdt, union ccb *ccb)
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
{
struct gdt_ccb *gccb;
struct cam_sim *sim;
Reform the busdma API so that new types may be added without modifying every architecture's busdma_machdep.c. It is done by unifying the bus_dmamap_load_buffer() routines so that they may be called from MI code. The MD busdma is then given a chance to do any final processing in the complete() callback. The cam changes unify the bus_dmamap_load* handling in cam drivers. The arm and mips implementations are updated to track virtual addresses for sync(). Previously this was done in a type specific way. Now it is done in a generic way by recording the list of virtuals in the map. Submitted by: jeff (sponsored by EMC/Isilon) Reviewed by: kan (previous version), scottl, mjacob (isp(4), no objections for target mode changes) Discussed with: ian (arm changes) Tested by: marius (sparc64), mips (jmallet), isci(4) on x86 (jharris), amd64 (Fabian Keil <freebsd-listen@fabiankeil.de>)
2013-02-12 16:57:20 +00:00
int error;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_DPRINTF(GDT_D_CMD, ("gdt_raw_cmd(%p, %p)\n", gdt, ccb));
if (roundup(GDT_CMD_UNION + GDT_RAW_SZ, sizeof(u_int32_t)) +
gdt->sc_cmd_off + GDT_DPMEM_COMMAND_OFFSET >
gdt->sc_ic_all_size) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
GDT_DPRINTF(GDT_D_INVALID, ("%s: gdt_raw_cmd(): DPMEM overflow\n",
device_get_nameunit(gdt->sc_devnode)));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (NULL);
}
gccb = gdt_get_ccb(gdt);
if (gccb == NULL) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
GDT_DPRINTF(GDT_D_INVALID, ("%s: No free command index found\n",
device_get_nameunit(gdt->sc_devnode)));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (gccb);
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bzero(gccb->gc_cmd, GDT_CMD_SZ);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
sim = (struct cam_sim *)ccb->ccb_h.ccb_sim_ptr;
gccb->gc_ccb = ccb;
gccb->gc_service = GDT_SCSIRAWSERVICE;
gccb->gc_flags = GDT_GCF_SCSI;
if (gdt->sc_cmd_cnt == 0)
gdt->sc_set_sema0(gdt);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_COMMANDINDEX,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_cmd_index);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_OPCODE, GDT_WRITE);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_DIRECTION,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
(ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN ?
GDT_DATA_IN : GDT_DATA_OUT);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SDLEN,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccb->csio.dxfer_len);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_CLEN,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccb->csio.cdb_len);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bcopy(ccb->csio.cdb_io.cdb_bytes, gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_CMD,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccb->csio.cdb_len);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd[GDT_CMD_UNION + GDT_RAW_TARGET] =
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccb->ccb_h.target_id;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd[GDT_CMD_UNION + GDT_RAW_LUN] =
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccb->ccb_h.target_lun;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd[GDT_CMD_UNION + GDT_RAW_BUS] =
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
cam_sim_bus(sim);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SENSE_LEN,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
sizeof(struct scsi_sense_data));
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SENSE_DATA,
gccb->gc_scratch_busbase);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
Reform the busdma API so that new types may be added without modifying every architecture's busdma_machdep.c. It is done by unifying the bus_dmamap_load_buffer() routines so that they may be called from MI code. The MD busdma is then given a chance to do any final processing in the complete() callback. The cam changes unify the bus_dmamap_load* handling in cam drivers. The arm and mips implementations are updated to track virtual addresses for sync(). Previously this was done in a type specific way. Now it is done in a generic way by recording the list of virtuals in the map. Submitted by: jeff (sponsored by EMC/Isilon) Reviewed by: kan (previous version), scottl, mjacob (isp(4), no objections for target mode changes) Discussed with: ian (arm changes) Tested by: marius (sparc64), mips (jmallet), isci(4) on x86 (jharris), amd64 (Fabian Keil <freebsd-listen@fabiankeil.de>)
2013-02-12 16:57:20 +00:00
error = bus_dmamap_load_ccb(gdt->sc_buffer_dmat,
gccb->gc_dmamap,
ccb,
gdtexecuteccb,
gccb, /*flags*/0);
if (error == EINPROGRESS) {
xpt_freeze_simq(sim, 1);
gccb->gc_ccb->ccb_h.status |= CAM_RELEASE_SIMQ;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
return (gccb);
}
static struct gdt_ccb *
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
gdt_cache_cmd(struct gdt_softc *gdt, union ccb *ccb)
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
{
struct gdt_ccb *gccb;
struct cam_sim *sim;
u_int8_t *cmdp;
u_int16_t opcode;
u_int32_t blockno, blockcnt;
Reform the busdma API so that new types may be added without modifying every architecture's busdma_machdep.c. It is done by unifying the bus_dmamap_load_buffer() routines so that they may be called from MI code. The MD busdma is then given a chance to do any final processing in the complete() callback. The cam changes unify the bus_dmamap_load* handling in cam drivers. The arm and mips implementations are updated to track virtual addresses for sync(). Previously this was done in a type specific way. Now it is done in a generic way by recording the list of virtuals in the map. Submitted by: jeff (sponsored by EMC/Isilon) Reviewed by: kan (previous version), scottl, mjacob (isp(4), no objections for target mode changes) Discussed with: ian (arm changes) Tested by: marius (sparc64), mips (jmallet), isci(4) on x86 (jharris), amd64 (Fabian Keil <freebsd-listen@fabiankeil.de>)
2013-02-12 16:57:20 +00:00
int error;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_DPRINTF(GDT_D_CMD, ("gdt_cache_cmd(%p, %p)\n", gdt, ccb));
if (roundup(GDT_CMD_UNION + GDT_CACHE_SZ, sizeof(u_int32_t)) +
gdt->sc_cmd_off + GDT_DPMEM_COMMAND_OFFSET >
gdt->sc_ic_all_size) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
GDT_DPRINTF(GDT_D_INVALID, ("%s: gdt_cache_cmd(): DPMEM overflow\n",
device_get_nameunit(gdt->sc_devnode)));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (NULL);
}
gccb = gdt_get_ccb(gdt);
if (gccb == NULL) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
GDT_DPRINTF(GDT_D_DEBUG, ("%s: No free command index found\n",
device_get_nameunit(gdt->sc_devnode)));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (gccb);
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bzero(gccb->gc_cmd, GDT_CMD_SZ);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
sim = (struct cam_sim *)ccb->ccb_h.ccb_sim_ptr;
gccb->gc_ccb = ccb;
gccb->gc_service = GDT_CACHESERVICE;
gccb->gc_flags = GDT_GCF_SCSI;
if (gdt->sc_cmd_cnt == 0)
gdt->sc_set_sema0(gdt);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_COMMANDINDEX,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_cmd_index);
cmdp = ccb->csio.cdb_io.cdb_bytes;
opcode = (*cmdp == WRITE_6 || *cmdp == WRITE_10) ? GDT_WRITE : GDT_READ;
if ((gdt->sc_state & GDT_SHUTDOWN) && opcode == GDT_WRITE)
opcode = GDT_WRITE_THR;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_OPCODE, opcode);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_DEVICENO,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccb->ccb_h.target_id);
if (ccb->csio.cdb_len == 6) {
struct scsi_rw_6 *rw = (struct scsi_rw_6 *)cmdp;
blockno = scsi_3btoul(rw->addr) & ((SRW_TOPADDR<<16) | 0xffff);
blockcnt = rw->length ? rw->length : 0x100;
} else {
struct scsi_rw_10 *rw = (struct scsi_rw_10 *)cmdp;
blockno = scsi_4btoul(rw->addr);
blockcnt = scsi_2btoul(rw->length);
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_BLOCKNO,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
blockno);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_BLOCKCNT,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
blockcnt);
Reform the busdma API so that new types may be added without modifying every architecture's busdma_machdep.c. It is done by unifying the bus_dmamap_load_buffer() routines so that they may be called from MI code. The MD busdma is then given a chance to do any final processing in the complete() callback. The cam changes unify the bus_dmamap_load* handling in cam drivers. The arm and mips implementations are updated to track virtual addresses for sync(). Previously this was done in a type specific way. Now it is done in a generic way by recording the list of virtuals in the map. Submitted by: jeff (sponsored by EMC/Isilon) Reviewed by: kan (previous version), scottl, mjacob (isp(4), no objections for target mode changes) Discussed with: ian (arm changes) Tested by: marius (sparc64), mips (jmallet), isci(4) on x86 (jharris), amd64 (Fabian Keil <freebsd-listen@fabiankeil.de>)
2013-02-12 16:57:20 +00:00
error = bus_dmamap_load_ccb(gdt->sc_buffer_dmat,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_dmamap,
Reform the busdma API so that new types may be added without modifying every architecture's busdma_machdep.c. It is done by unifying the bus_dmamap_load_buffer() routines so that they may be called from MI code. The MD busdma is then given a chance to do any final processing in the complete() callback. The cam changes unify the bus_dmamap_load* handling in cam drivers. The arm and mips implementations are updated to track virtual addresses for sync(). Previously this was done in a type specific way. Now it is done in a generic way by recording the list of virtuals in the map. Submitted by: jeff (sponsored by EMC/Isilon) Reviewed by: kan (previous version), scottl, mjacob (isp(4), no objections for target mode changes) Discussed with: ian (arm changes) Tested by: marius (sparc64), mips (jmallet), isci(4) on x86 (jharris), amd64 (Fabian Keil <freebsd-listen@fabiankeil.de>)
2013-02-12 16:57:20 +00:00
ccb,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdtexecuteccb,
gccb, /*flags*/0);
Reform the busdma API so that new types may be added without modifying every architecture's busdma_machdep.c. It is done by unifying the bus_dmamap_load_buffer() routines so that they may be called from MI code. The MD busdma is then given a chance to do any final processing in the complete() callback. The cam changes unify the bus_dmamap_load* handling in cam drivers. The arm and mips implementations are updated to track virtual addresses for sync(). Previously this was done in a type specific way. Now it is done in a generic way by recording the list of virtuals in the map. Submitted by: jeff (sponsored by EMC/Isilon) Reviewed by: kan (previous version), scottl, mjacob (isp(4), no objections for target mode changes) Discussed with: ian (arm changes) Tested by: marius (sparc64), mips (jmallet), isci(4) on x86 (jharris), amd64 (Fabian Keil <freebsd-listen@fabiankeil.de>)
2013-02-12 16:57:20 +00:00
if (error == EINPROGRESS) {
xpt_freeze_simq(sim, 1);
gccb->gc_ccb->ccb_h.status |= CAM_RELEASE_SIMQ;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
return (gccb);
}
static struct gdt_ccb *
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
gdt_ioctl_cmd(struct gdt_softc *gdt, gdt_ucmd_t *ucmd)
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
{
struct gdt_ccb *gccb;
u_int32_t cnt;
GDT_DPRINTF(GDT_D_DEBUG, ("gdt_ioctl_cmd(%p, %p)\n", gdt, ucmd));
gccb = gdt_get_ccb(gdt);
if (gccb == NULL) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
GDT_DPRINTF(GDT_D_DEBUG, ("%s: No free command index found\n",
device_get_nameunit(gdt->sc_devnode)));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (gccb);
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bzero(gccb->gc_cmd, GDT_CMD_SZ);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_ucmd = ucmd;
gccb->gc_service = ucmd->service;
gccb->gc_flags = GDT_GCF_IOCTL;
/* check DPMEM space, copy data buffer from user space */
if (ucmd->service == GDT_CACHESERVICE) {
if (ucmd->OpCode == GDT_IOCTL) {
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd_len = roundup(GDT_CMD_UNION + GDT_IOCTL_SZ,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
sizeof(u_int32_t));
cnt = ucmd->u.ioctl.param_size;
if (cnt > GDT_SCRATCH_SZ) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode,
"Scratch buffer too small (%d/%d)\n", GDT_SCRATCH_SZ, cnt);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_free_ccb(gdt, gccb);
return (NULL);
}
} else {
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd_len = roundup(GDT_CMD_UNION + GDT_CACHE_SG_LST +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_SG_SZ, sizeof(u_int32_t));
cnt = ucmd->u.cache.BlockCnt * GDT_SECTOR_SIZE;
if (cnt > GDT_SCRATCH_SZ) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode,
"Scratch buffer too small (%d/%d)\n", GDT_SCRATCH_SZ, cnt);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_free_ccb(gdt, gccb);
return (NULL);
}
}
} else {
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd_len = roundup(GDT_CMD_UNION + GDT_RAW_SG_LST +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_SG_SZ, sizeof(u_int32_t));
cnt = ucmd->u.raw.sdlen;
if (cnt + ucmd->u.raw.sense_len > GDT_SCRATCH_SZ) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode, "Scratch buffer too small (%d/%d)\n",
GDT_SCRATCH_SZ, cnt + ucmd->u.raw.sense_len);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_free_ccb(gdt, gccb);
return (NULL);
}
}
if (cnt != 0)
bcopy(ucmd->data, gccb->gc_scratch, cnt);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
if (gdt->sc_cmd_off + gccb->gc_cmd_len + GDT_DPMEM_COMMAND_OFFSET >
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_ic_all_size) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
GDT_DPRINTF(GDT_D_INVALID, ("%s: gdt_ioctl_cmd(): DPMEM overflow\n",
device_get_nameunit(gdt->sc_devnode)));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_free_ccb(gdt, gccb);
return (NULL);
}
if (gdt->sc_cmd_cnt == 0)
gdt->sc_set_sema0(gdt);
/* fill cmd structure */
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_COMMANDINDEX,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_cmd_index);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_OPCODE,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->OpCode);
if (ucmd->service == GDT_CACHESERVICE) {
if (ucmd->OpCode == GDT_IOCTL) {
/* IOCTL */
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_UNION + GDT_IOCTL_PARAM_SIZE,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.ioctl.param_size);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_IOCTL_SUBFUNC,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.ioctl.subfunc);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_IOCTL_CHANNEL,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.ioctl.channel);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_IOCTL_P_PARAM,
gccb->gc_scratch_busbase);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
} else {
/* cache service command */
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_DEVICENO,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.cache.DeviceNo);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_BLOCKNO,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.cache.BlockNo);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_BLOCKCNT,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.cache.BlockCnt);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_DESTADDR,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
0xffffffffUL);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_SG_CANZ,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
1);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_SG_LST +
GDT_SG_PTR, gccb->gc_scratch_busbase);
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_SG_LST +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_SG_LEN, ucmd->u.cache.BlockCnt * GDT_SECTOR_SIZE);
}
} else {
/* raw service command */
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_DIRECTION,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.raw.direction);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SDATA,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
0xffffffffUL);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SDLEN,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.raw.sdlen);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_CLEN,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.raw.clen);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bcopy(ucmd->u.raw.cmd, gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_CMD,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
12);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd[GDT_CMD_UNION + GDT_RAW_TARGET] =
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.raw.target;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd[GDT_CMD_UNION + GDT_RAW_LUN] =
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.raw.lun;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd[GDT_CMD_UNION + GDT_RAW_BUS] =
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.raw.bus;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SENSE_LEN,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ucmd->u.raw.sense_len);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SENSE_DATA,
gccb->gc_scratch_busbase + ucmd->u.raw.sdlen);
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SG_RANZ,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
1);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SG_LST +
GDT_SG_PTR, gccb->gc_scratch_busbase);
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SG_LST +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_SG_LEN, ucmd->u.raw.sdlen);
}
gdt_stat.sg_count_act = 1;
gdt->sc_copy_cmd(gdt, gccb);
return (gccb);
}
static void
gdt_internal_cache_cmd(struct gdt_softc *gdt,union ccb *ccb)
{
int t;
t = ccb->ccb_h.target_id;
GDT_DPRINTF(GDT_D_CMD, ("gdt_internal_cache_cmd(%p, %p, 0x%x, %d)\n",
gdt, ccb, ccb->csio.cdb_io.cdb_bytes[0], t));
switch (ccb->csio.cdb_io.cdb_bytes[0]) {
case TEST_UNIT_READY:
case START_STOP:
break;
case REQUEST_SENSE:
GDT_DPRINTF(GDT_D_MISC, ("REQUEST_SENSE\n"));
break;
case INQUIRY:
{
Fix a buffer overflow in the processing of various SCSI commands. This likely fixes a an great number of weird problems that have been reported with this driver. PR: 119114 Submitted by: Jens Rehsack
2008-01-09 20:02:56 +00:00
struct scsi_inquiry_data inq;
size_t copylen = MIN(sizeof(inq), ccb->csio.dxfer_len);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
Fix a buffer overflow in the processing of various SCSI commands. This likely fixes a an great number of weird problems that have been reported with this driver. PR: 119114 Submitted by: Jens Rehsack
2008-01-09 20:02:56 +00:00
bzero(&inq, sizeof(inq));
inq.device = (gdt->sc_hdr[t].hd_devtype & 4) ?
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
T_CDROM : T_DIRECT;
Fix a buffer overflow in the processing of various SCSI commands. This likely fixes a an great number of weird problems that have been reported with this driver. PR: 119114 Submitted by: Jens Rehsack
2008-01-09 20:02:56 +00:00
inq.dev_qual2 = (gdt->sc_hdr[t].hd_devtype & 1) ? 0x80 : 0;
inq.version = SCSI_REV_2;
inq.response_format = 2;
inq.additional_length = 32;
inq.flags = SID_CmdQue | SID_Sync;
strncpy(inq.vendor, gdt->oem_name, sizeof(inq.vendor));
snprintf(inq.product, sizeof(inq.product),
"Host Drive #%02d", t);
strncpy(inq.revision, " ", sizeof(inq.revision));
bcopy(&inq, ccb->csio.data_ptr, copylen );
if( ccb->csio.dxfer_len > copylen )
bzero( ccb->csio.data_ptr+copylen,
ccb->csio.dxfer_len - copylen );
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
break;
}
case MODE_SENSE_6:
{
struct mpd_data {
struct scsi_mode_hdr_6 hd;
struct scsi_mode_block_descr bd;
struct scsi_control_page cp;
Fix a buffer overflow in the processing of various SCSI commands. This likely fixes a an great number of weird problems that have been reported with this driver. PR: 119114 Submitted by: Jens Rehsack
2008-01-09 20:02:56 +00:00
} mpd;
size_t copylen = MIN(sizeof(mpd), ccb->csio.dxfer_len);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
u_int8_t page;
Fix a buffer overflow in the processing of various SCSI commands. This likely fixes a an great number of weird problems that have been reported with this driver. PR: 119114 Submitted by: Jens Rehsack
2008-01-09 20:02:56 +00:00
/*mpd = (struct mpd_data *)ccb->csio.data_ptr;*/
bzero(&mpd, sizeof(mpd));
mpd.hd.datalen = sizeof(struct scsi_mode_hdr_6) +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
sizeof(struct scsi_mode_block_descr);
Fix a buffer overflow in the processing of various SCSI commands. This likely fixes a an great number of weird problems that have been reported with this driver. PR: 119114 Submitted by: Jens Rehsack
2008-01-09 20:02:56 +00:00
mpd.hd.dev_specific = (gdt->sc_hdr[t].hd_devtype & 2) ? 0x80 : 0;
mpd.hd.block_descr_len = sizeof(struct scsi_mode_block_descr);
mpd.bd.block_len[0] = (GDT_SECTOR_SIZE & 0x00ff0000) >> 16;
mpd.bd.block_len[1] = (GDT_SECTOR_SIZE & 0x0000ff00) >> 8;
mpd.bd.block_len[2] = (GDT_SECTOR_SIZE & 0x000000ff);
bcopy(&mpd, ccb->csio.data_ptr, copylen );
if( ccb->csio.dxfer_len > copylen )
bzero( ccb->csio.data_ptr+copylen,
ccb->csio.dxfer_len - copylen );
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
page=((struct scsi_mode_sense_6 *)ccb->csio.cdb_io.cdb_bytes)->page;
switch (page) {
default:
GDT_DPRINTF(GDT_D_MISC, ("MODE_SENSE_6: page 0x%x\n", page));
break;
}
break;
}
case READ_CAPACITY:
{
Fix a buffer overflow in the processing of various SCSI commands. This likely fixes a an great number of weird problems that have been reported with this driver. PR: 119114 Submitted by: Jens Rehsack
2008-01-09 20:02:56 +00:00
struct scsi_read_capacity_data rcd;
size_t copylen = MIN(sizeof(rcd), ccb->csio.dxfer_len);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
Fix a buffer overflow in the processing of various SCSI commands. This likely fixes a an great number of weird problems that have been reported with this driver. PR: 119114 Submitted by: Jens Rehsack
2008-01-09 20:02:56 +00:00
/*rcd = (struct scsi_read_capacity_data *)ccb->csio.data_ptr;*/
bzero(&rcd, sizeof(rcd));
scsi_ulto4b(gdt->sc_hdr[t].hd_size - 1, rcd.addr);
scsi_ulto4b(GDT_SECTOR_SIZE, rcd.length);
bcopy(&rcd, ccb->csio.data_ptr, copylen );
if( ccb->csio.dxfer_len > copylen )
bzero( ccb->csio.data_ptr+copylen,
ccb->csio.dxfer_len - copylen );
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
break;
}
default:
GDT_DPRINTF(GDT_D_MISC, ("gdt_internal_cache_cmd(%d) unknown\n",
ccb->csio.cdb_io.cdb_bytes[0]));
break;
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
ccb->ccb_h.status |= CAM_REQ_CMP;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
--gdt_stat.io_count_act;
xpt_done(ccb);
}
static void
gdtmapmem(void *arg, bus_dma_segment_t *dm_segs, int nseg, int error)
{
bus_addr_t *busaddrp;
busaddrp = (bus_addr_t *)arg;
*busaddrp = dm_segs->ds_addr;
}
static void
gdtexecuteccb(void *arg, bus_dma_segment_t *dm_segs, int nseg, int error)
{
struct gdt_ccb *gccb;
union ccb *ccb;
struct gdt_softc *gdt;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
int i;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb = (struct gdt_ccb *)arg;
ccb = gccb->gc_ccb;
gdt = cam_sim_softc((struct cam_sim *)ccb->ccb_h.ccb_sim_ptr);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_assert(&gdt->sc_lock, MA_OWNED);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_DPRINTF(GDT_D_CMD, ("gdtexecuteccb(%p, %p, %p, %d, %d)\n",
gdt, gccb, dm_segs, nseg, error));
gdt_stat.sg_count_act = nseg;
if (nseg > gdt_stat.sg_count_max)
gdt_stat.sg_count_max = nseg;
/* Copy the segments into our SG list */
if (gccb->gc_service == GDT_CACHESERVICE) {
for (i = 0; i < nseg; ++i) {
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_SG_LST +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
i * GDT_SG_SZ + GDT_SG_PTR, dm_segs->ds_addr);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_SG_LST +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
i * GDT_SG_SZ + GDT_SG_LEN, dm_segs->ds_len);
dm_segs++;
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_SG_CANZ,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
nseg);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_CACHE_DESTADDR,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
0xffffffffUL);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd_len = roundup(GDT_CMD_UNION + GDT_CACHE_SG_LST +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
nseg * GDT_SG_SZ, sizeof(u_int32_t));
} else {
for (i = 0; i < nseg; ++i) {
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SG_LST +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
i * GDT_SG_SZ + GDT_SG_PTR, dm_segs->ds_addr);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SG_LST +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
i * GDT_SG_SZ + GDT_SG_LEN, dm_segs->ds_len);
dm_segs++;
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SG_RANZ,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
nseg);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_RAW_SDATA,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
0xffffffffUL);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd_len = roundup(GDT_CMD_UNION + GDT_RAW_SG_LST +
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
nseg * GDT_SG_SZ, sizeof(u_int32_t));
}
if (nseg != 0) {
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
bus_dmamap_sync(gdt->sc_buffer_dmat, gccb->gc_dmamap,
(ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN ?
BUS_DMASYNC_PREREAD : BUS_DMASYNC_PREWRITE);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
/* We must NOT abort the command here if CAM_REQ_INPROG is not set,
* because command semaphore is already set!
*/
ccb->ccb_h.status |= CAM_SIM_QUEUED;
/* timeout handling */
Prevent overflow issues in timeout processing Previously, any timeout value for which (timeout * hz) will overflow the signed integer, will give weird results, since callout(9) routines will convert negative values of ticks to '1'. For unsigned integer overflow we will get sufficiently smaller timeout values than expected. Switch from callout_reset, which requires conversion to int based ticks to callout_reset_sbt to avoid this. Also correct isci to correctly resolve ccb timeout. This was based on the original work done by Eygene Ryabinkin <rea@freebsd.org> back in 5 Aug 2011 which used a macro to help avoid the overlow. Differential Revision: https://reviews.freebsd.org/D1157 Reviewed by: mav, davide MFC after: 1 month Sponsored by: Multiplay
2014-11-21 21:01:24 +00:00
callout_reset_sbt(&gccb->gc_timeout, SBT_1MS * ccb->ccb_h.timeout, 0,
iir_timeout, (caddr_t)gccb, 0);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_copy_cmd(gdt, gccb);
}
static void
iir_action( struct cam_sim *sim, union ccb *ccb )
{
struct gdt_softc *gdt;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
int bus, target, lun;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt = (struct gdt_softc *)cam_sim_softc( sim );
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_assert(&gdt->sc_lock, MA_OWNED);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccb->ccb_h.ccb_sim_ptr = sim;
bus = cam_sim_bus(sim);
target = ccb->ccb_h.target_id;
lun = ccb->ccb_h.target_lun;
GDT_DPRINTF(GDT_D_CMD,
("iir_action(%p) func 0x%x cmd 0x%x bus %d target %d lun %d\n",
gdt, ccb->ccb_h.func_code, ccb->csio.cdb_io.cdb_bytes[0],
bus, target, lun));
++gdt_stat.io_count_act;
if (gdt_stat.io_count_act > gdt_stat.io_count_max)
gdt_stat.io_count_max = gdt_stat.io_count_act;
switch (ccb->ccb_h.func_code) {
case XPT_SCSI_IO:
TAILQ_INSERT_TAIL(&gdt->sc_ccb_queue, &ccb->ccb_h, sim_links.tqe);
++gdt_stat.req_queue_act;
if (gdt_stat.req_queue_act > gdt_stat.req_queue_max)
gdt_stat.req_queue_max = gdt_stat.req_queue_act;
gdt_next(gdt);
break;
case XPT_RESET_DEV: /* Bus Device Reset the specified SCSI device */
case XPT_ABORT: /* Abort the specified CCB */
/* XXX Implement */
ccb->ccb_h.status = CAM_REQ_INVALID;
--gdt_stat.io_count_act;
xpt_done(ccb);
break;
case XPT_SET_TRAN_SETTINGS:
ccb->ccb_h.status = CAM_FUNC_NOTAVAIL;
--gdt_stat.io_count_act;
xpt_done(ccb);
break;
case XPT_GET_TRAN_SETTINGS:
/* Get default/user set transfer settings for the target */
{
The first of 3 major steps to move the CAM layer forward to using the CAM_NEW_TRAN_CODE that has been in the tree for some years now. This first step consists solely of adding to or correcting CAM_NEW_TRAN_CODE pieces in the kernel source tree such that a both a GENERIC (at least on i386) and a LINT build with CAM_NEW_TRAN_CODE as an option will compile correctly and run (at least with some the h/w I have). After a short settle time, the other pieces (making CAM_NEW_TRAN_CODE the default and updating libcam and camcontrol) will be brought in. This will be an incompatible change in that the size of structures related to XPT_PATH_INQ and XPT_{GET,SET}_TRAN_SETTINGS change in both size and content. However, basic system operation and basic system utilities work well enough with this change. Reviewed by: freebsd-scsi and specific stakeholders
2006-10-31 05:53:29 +00:00
struct ccb_trans_settings *cts = &ccb->cts;
struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi;
struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi;
cts->protocol = PROTO_SCSI;
cts->protocol_version = SCSI_REV_2;
cts->transport = XPORT_SPI;
cts->transport_version = 2;
if (cts->type == CTS_TYPE_USER_SETTINGS) {
spi->flags = CTS_SPI_FLAGS_DISC_ENB;
scsi->flags = CTS_SCSI_FLAGS_TAG_ENB;
spi->bus_width = MSG_EXT_WDTR_BUS_16_BIT;
spi->sync_period = 25; /* 10MHz */
if (spi->sync_period != 0)
spi->sync_offset = 15;
spi->valid = CTS_SPI_VALID_SYNC_RATE
| CTS_SPI_VALID_SYNC_OFFSET
| CTS_SPI_VALID_BUS_WIDTH
| CTS_SPI_VALID_DISC;
scsi->valid = CTS_SCSI_VALID_TQ;
ccb->ccb_h.status = CAM_REQ_CMP;
} else {
ccb->ccb_h.status = CAM_FUNC_NOTAVAIL;
}
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
--gdt_stat.io_count_act;
xpt_done(ccb);
break;
}
case XPT_CALC_GEOMETRY:
{
struct ccb_calc_geometry *ccg;
u_int32_t secs_per_cylinder;
ccg = &ccb->ccg;
ccg->heads = gdt->sc_hdr[target].hd_heads;
ccg->secs_per_track = gdt->sc_hdr[target].hd_secs;
secs_per_cylinder = ccg->heads * ccg->secs_per_track;
ccg->cylinders = ccg->volume_size / secs_per_cylinder;
ccb->ccb_h.status = CAM_REQ_CMP;
--gdt_stat.io_count_act;
xpt_done(ccb);
break;
}
case XPT_RESET_BUS: /* Reset the specified SCSI bus */
{
/* XXX Implement */
ccb->ccb_h.status = CAM_REQ_CMP;
--gdt_stat.io_count_act;
xpt_done(ccb);
break;
}
case XPT_TERM_IO: /* Terminate the I/O process */
/* XXX Implement */
ccb->ccb_h.status = CAM_REQ_INVALID;
--gdt_stat.io_count_act;
xpt_done(ccb);
break;
case XPT_PATH_INQ: /* Path routing inquiry */
{
struct ccb_pathinq *cpi = &ccb->cpi;
cpi->version_num = 1;
cpi->hba_inquiry = PI_SDTR_ABLE|PI_TAG_ABLE;
cpi->hba_inquiry |= PI_WIDE_16;
cpi->target_sprt = 1;
cpi->hba_misc = 0;
cpi->hba_eng_cnt = 0;
if (bus == gdt->sc_virt_bus)
cpi->max_target = GDT_MAX_HDRIVES - 1;
else if (gdt->sc_class & GDT_FC)
cpi->max_target = GDT_MAXID_FC - 1;
else
cpi->max_target = GDT_MAXID - 1;
cpi->max_lun = 7;
cpi->unit_number = cam_sim_unit(sim);
cpi->bus_id = bus;
cpi->initiator_id =
(bus == gdt->sc_virt_bus ? 127 : gdt->sc_bus_id[bus]);
cpi->base_transfer_speed = 3300;
Always null-terminate ccb_pathinq.(sim_vid|hba_vid|dev_name) The sim_vid, hba_vid, and dev_name fields of struct ccb_pathinq are fixed-length strings. AFAICT the only place they're read is in sbin/camcontrol/camcontrol.c, which assumes they'll be null-terminated. However, the kernel doesn't null-terminate them. A bunch of copy-pasted code uses strncpy to write them, and doesn't guarantee null-termination. For at least 4 drivers (mpr, mps, ciss, and hyperv), the hba_vid field actually overflows. You can see the result by doing "camcontrol negotiate da0 -v". This change null-terminates those fields everywhere they're set in the kernel. It also shortens a few strings to ensure they'll fit within the 16-character field. PR: 215474 Reported by: Coverity CID: 1009997 1010000 1010001 1010002 1010003 1010004 1010005 CID: 1331519 1010006 1215097 1010007 1288967 1010008 1306000 CID: 1211924 1010009 1010010 1010011 1010012 1010013 1010014 CID: 1147190 1010017 1010016 1010018 1216435 1010020 1010021 CID: 1010022 1009666 1018185 1010023 1010025 1010026 1010027 CID: 1010028 1010029 1010030 1010031 1010033 1018186 1018187 CID: 1010035 1010036 1010042 1010041 1010040 1010039 Reviewed by: imp, sephe, slm MFC after: 4 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9037 Differential Revision: https://reviews.freebsd.org/D9038
2017-01-04 20:26:42 +00:00
strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN);
Avoid potential redefinition of the macro.
2013-08-15 20:03:22 +00:00
if (gdt->sc_vendor == INTEL_VENDOR_ID_IIR)
Always null-terminate ccb_pathinq.(sim_vid|hba_vid|dev_name) The sim_vid, hba_vid, and dev_name fields of struct ccb_pathinq are fixed-length strings. AFAICT the only place they're read is in sbin/camcontrol/camcontrol.c, which assumes they'll be null-terminated. However, the kernel doesn't null-terminate them. A bunch of copy-pasted code uses strncpy to write them, and doesn't guarantee null-termination. For at least 4 drivers (mpr, mps, ciss, and hyperv), the hba_vid field actually overflows. You can see the result by doing "camcontrol negotiate da0 -v". This change null-terminates those fields everywhere they're set in the kernel. It also shortens a few strings to ensure they'll fit within the 16-character field. PR: 215474 Reported by: Coverity CID: 1009997 1010000 1010001 1010002 1010003 1010004 1010005 CID: 1331519 1010006 1215097 1010007 1288967 1010008 1306000 CID: 1211924 1010009 1010010 1010011 1010012 1010013 1010014 CID: 1147190 1010017 1010016 1010018 1216435 1010020 1010021 CID: 1010022 1009666 1018185 1010023 1010025 1010026 1010027 CID: 1010028 1010029 1010030 1010031 1010033 1018186 1018187 CID: 1010035 1010036 1010042 1010041 1010040 1010039 Reviewed by: imp, sephe, slm MFC after: 4 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9037 Differential Revision: https://reviews.freebsd.org/D9038
2017-01-04 20:26:42 +00:00
strlcpy(cpi->hba_vid, "Intel Corp.", HBA_IDLEN);
Update email addresses, copyrights, and tweak the management interface. Submitted by: "Leubner, Achim" <Achim_Leubner@adaptec.com>
2003-09-26 15:36:47 +00:00
else
Always null-terminate ccb_pathinq.(sim_vid|hba_vid|dev_name) The sim_vid, hba_vid, and dev_name fields of struct ccb_pathinq are fixed-length strings. AFAICT the only place they're read is in sbin/camcontrol/camcontrol.c, which assumes they'll be null-terminated. However, the kernel doesn't null-terminate them. A bunch of copy-pasted code uses strncpy to write them, and doesn't guarantee null-termination. For at least 4 drivers (mpr, mps, ciss, and hyperv), the hba_vid field actually overflows. You can see the result by doing "camcontrol negotiate da0 -v". This change null-terminates those fields everywhere they're set in the kernel. It also shortens a few strings to ensure they'll fit within the 16-character field. PR: 215474 Reported by: Coverity CID: 1009997 1010000 1010001 1010002 1010003 1010004 1010005 CID: 1331519 1010006 1215097 1010007 1288967 1010008 1306000 CID: 1211924 1010009 1010010 1010011 1010012 1010013 1010014 CID: 1147190 1010017 1010016 1010018 1216435 1010020 1010021 CID: 1010022 1009666 1018185 1010023 1010025 1010026 1010027 CID: 1010028 1010029 1010030 1010031 1010033 1018186 1018187 CID: 1010035 1010036 1010042 1010041 1010040 1010039 Reviewed by: imp, sephe, slm MFC after: 4 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9037 Differential Revision: https://reviews.freebsd.org/D9038
2017-01-04 20:26:42 +00:00
strlcpy(cpi->hba_vid, "ICP vortex ", HBA_IDLEN);
strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN);
The first of 3 major steps to move the CAM layer forward to using the CAM_NEW_TRAN_CODE that has been in the tree for some years now. This first step consists solely of adding to or correcting CAM_NEW_TRAN_CODE pieces in the kernel source tree such that a both a GENERIC (at least on i386) and a LINT build with CAM_NEW_TRAN_CODE as an option will compile correctly and run (at least with some the h/w I have). After a short settle time, the other pieces (making CAM_NEW_TRAN_CODE the default and updating libcam and camcontrol) will be brought in. This will be an incompatible change in that the size of structures related to XPT_PATH_INQ and XPT_{GET,SET}_TRAN_SETTINGS change in both size and content. However, basic system operation and basic system utilities work well enough with this change. Reviewed by: freebsd-scsi and specific stakeholders
2006-10-31 05:53:29 +00:00
cpi->transport = XPORT_SPI;
cpi->transport_version = 2;
cpi->protocol = PROTO_SCSI;
cpi->protocol_version = SCSI_REV_2;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
cpi->ccb_h.status = CAM_REQ_CMP;
--gdt_stat.io_count_act;
xpt_done(ccb);
break;
}
default:
GDT_DPRINTF(GDT_D_INVALID, ("gdt_next(%p) cmd 0x%x invalid\n",
gdt, ccb->ccb_h.func_code));
ccb->ccb_h.status = CAM_REQ_INVALID;
--gdt_stat.io_count_act;
xpt_done(ccb);
break;
}
}
static void
iir_poll( struct cam_sim *sim )
{
struct gdt_softc *gdt;
gdt = (struct gdt_softc *)cam_sim_softc( sim );
GDT_DPRINTF(GDT_D_CMD, ("iir_poll sim %p gdt %p\n", sim, gdt));
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
iir_intr_locked(gdt);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
static void
iir_timeout(void *arg)
{
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
GDT_DPRINTF(GDT_D_TIMEOUT, ("iir_timeout(%p)\n", gccb));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
static void
iir_shutdown( void *arg, int howto )
{
struct gdt_softc *gdt;
struct gdt_ccb *gccb;
gdt_ucmd_t *ucmd;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
int i;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt = (struct gdt_softc *)arg;
GDT_DPRINTF(GDT_D_CMD, ("iir_shutdown(%p, %d)\n", gdt, howto));
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode,
"Flushing all Host Drives. Please wait ... ");
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/* allocate ucmd buffer */
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
ucmd = malloc(sizeof(gdt_ucmd_t), M_GDTBUF, M_NOWAIT);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
if (ucmd == NULL) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
printf("\n");
device_printf(gdt->sc_devnode,
"iir_shutdown(): Cannot allocate resource\n");
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return;
}
bzero(ucmd, sizeof(gdt_ucmd_t));
/* wait for pending IOs */
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_lock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_state = GDT_SHUTDOWN;
if ((gccb = SLIST_FIRST(&gdt->sc_pending_gccb)) != NULL)
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_sleep(gccb, &gdt->sc_lock, PCATCH | PRIBIO, "iirshw", 100 * hz);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/* flush */
for (i = 0; i < GDT_MAX_HDRIVES; ++i) {
if (gdt->sc_hdr[i].hd_present) {
ucmd->service = GDT_CACHESERVICE;
ucmd->OpCode = GDT_FLUSH;
ucmd->u.cache.DeviceNo = i;
TAILQ_INSERT_TAIL(&gdt->sc_ucmd_queue, ucmd, links);
ucmd->complete_flag = FALSE;
gdt_next(gdt);
if (!ucmd->complete_flag)
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_sleep(ucmd, &gdt->sc_lock, PCATCH | PRIBIO, "iirshw",
10 * hz);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
}
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&gdt->sc_lock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
free(ucmd, M_DEVBUF);
printf("Done.\n");
}
void
iir_intr(void *arg)
{
struct gdt_softc *gdt = arg;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_lock(&gdt->sc_lock);
iir_intr_locked(gdt);
mtx_unlock(&gdt->sc_lock);
}
int
iir_intr_locked(struct gdt_softc *gdt)
{
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
struct gdt_intr_ctx ctx;
struct gdt_ccb *gccb;
gdt_ucmd_t *ucmd;
u_int32_t cnt;
GDT_DPRINTF(GDT_D_INTR, ("gdt_intr(%p)\n", gdt));
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_assert(&gdt->sc_lock, MA_OWNED);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
/* If polling and we were not called from gdt_wait, just return */
if ((gdt->sc_state & GDT_POLLING) &&
!(gdt->sc_state & GDT_POLL_WAIT))
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
return (0);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ctx.istatus = gdt->sc_get_status(gdt);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
if (ctx.istatus == 0x00) {
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_status = GDT_S_NO_STATUS;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
return (ctx.istatus);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
gdt->sc_intr(gdt, &ctx);
gdt->sc_status = ctx.cmd_status;
gdt->sc_service = ctx.service;
gdt->sc_info = ctx.info;
gdt->sc_info2 = ctx.info2;
if (ctx.istatus == GDT_ASYNCINDEX) {
gdt_async_event(gdt, ctx.service);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
return (ctx.istatus);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
if (ctx.istatus == GDT_SPEZINDEX) {
GDT_DPRINTF(GDT_D_INVALID,
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
("%s: Service unknown or not initialized!\n",
device_get_nameunit(gdt->sc_devnode)));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_dvr.size = sizeof(gdt->sc_dvr.eu.driver);
gdt->sc_dvr.eu.driver.ionode = gdt->sc_hanum;
gdt_store_event(GDT_ES_DRIVER, 4, &gdt->sc_dvr);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
return (ctx.istatus);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
gccb = &gdt->sc_gccbs[ctx.istatus - 2];
ctx.service = gccb->gc_service;
switch (gccb->gc_flags) {
case GDT_GCF_UNUSED:
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
GDT_DPRINTF(GDT_D_INVALID, ("%s: Index (%d) to unused command!\n",
device_get_nameunit(gdt->sc_devnode), ctx.istatus));
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_dvr.size = sizeof(gdt->sc_dvr.eu.driver);
gdt->sc_dvr.eu.driver.ionode = gdt->sc_hanum;
gdt->sc_dvr.eu.driver.index = ctx.istatus;
gdt_store_event(GDT_ES_DRIVER, 1, &gdt->sc_dvr);
gdt_free_ccb(gdt, gccb);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
break;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
case GDT_GCF_INTERNAL:
break;
case GDT_GCF_IOCTL:
ucmd = gccb->gc_ucmd;
if (gdt->sc_status == GDT_S_BSY) {
GDT_DPRINTF(GDT_D_DEBUG, ("iir_intr(%p) ioctl: gccb %p busy\n",
gdt, gccb));
TAILQ_INSERT_HEAD(&gdt->sc_ucmd_queue, ucmd, links);
} else {
ucmd->status = gdt->sc_status;
ucmd->info = gdt->sc_info;
ucmd->complete_flag = TRUE;
if (ucmd->service == GDT_CACHESERVICE) {
if (ucmd->OpCode == GDT_IOCTL) {
cnt = ucmd->u.ioctl.param_size;
if (cnt != 0)
bcopy(gccb->gc_scratch, ucmd->data, cnt);
} else {
cnt = ucmd->u.cache.BlockCnt * GDT_SECTOR_SIZE;
if (cnt != 0)
bcopy(gccb->gc_scratch, ucmd->data, cnt);
}
} else {
cnt = ucmd->u.raw.sdlen;
if (cnt != 0)
bcopy(gccb->gc_scratch, ucmd->data, cnt);
if (ucmd->u.raw.sense_len != 0)
bcopy(gccb->gc_scratch, ucmd->data, cnt);
}
gdt_free_ccb(gdt, gccb);
/* wakeup */
wakeup(ucmd);
}
gdt_next(gdt);
break;
default:
gdt_free_ccb(gdt, gccb);
gdt_sync_event(gdt, ctx.service, ctx.istatus, gccb);
gdt_next(gdt);
break;
}
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
return (ctx.istatus);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
int
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_async_event(struct gdt_softc *gdt, int service)
{
struct gdt_ccb *gccb;
GDT_DPRINTF(GDT_D_INTR, ("gdt_async_event(%p, %d)\n", gdt, service));
if (service == GDT_SCREENSERVICE) {
if (gdt->sc_status == GDT_MSG_REQUEST) {
while (gdt->sc_test_busy(gdt))
DELAY(1);
gccb = gdt_get_ccb(gdt);
if (gccb == NULL) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode, "No free command index found\n");
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bzero(gccb->gc_cmd, GDT_CMD_SZ);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_service = service;
gccb->gc_flags = GDT_GCF_SCREEN;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_COMMANDINDEX,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_cmd_index);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_OPCODE, GDT_READ);
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_SCREEN_MSG_HANDLE,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_MSG_INV_HANDLE);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_SCREEN_MSG_ADDR,
gccb->gc_scratch_busbase);
gdt->sc_set_sema0(gdt);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_cmd_off = 0;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd_len = roundup(GDT_CMD_UNION + GDT_SCREEN_SZ,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
sizeof(u_int32_t));
gdt->sc_cmd_cnt = 0;
gdt->sc_copy_cmd(gdt, gccb);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode, "[PCI %d/%d] ", gdt->sc_bus,
gdt->sc_slot);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_release_event(gdt);
}
} else {
if ((gdt->sc_fw_vers & 0xff) >= 0x1a) {
gdt->sc_dvr.size = 0;
gdt->sc_dvr.eu.async.ionode = gdt->sc_hanum;
gdt->sc_dvr.eu.async.status = gdt->sc_status;
/* severity and event_string already set! */
} else {
gdt->sc_dvr.size = sizeof(gdt->sc_dvr.eu.async);
gdt->sc_dvr.eu.async.ionode = gdt->sc_hanum;
gdt->sc_dvr.eu.async.service = service;
gdt->sc_dvr.eu.async.status = gdt->sc_status;
gdt->sc_dvr.eu.async.info = gdt->sc_info;
*(u_int32_t *)gdt->sc_dvr.eu.async.scsi_coord = gdt->sc_info2;
}
gdt_store_event(GDT_ES_ASYNC, service, &gdt->sc_dvr);
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode, "%s\n", gdt->sc_dvr.event_string);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
return (0);
}
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
int
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_sync_event(struct gdt_softc *gdt, int service,
u_int8_t index, struct gdt_ccb *gccb)
{
union ccb *ccb;
GDT_DPRINTF(GDT_D_INTR,
("gdt_sync_event(%p, %d, %d, %p)\n", gdt,service,index,gccb));
ccb = gccb->gc_ccb;
if (service == GDT_SCREENSERVICE) {
u_int32_t msg_len;
msg_len = gdt_dec32(gccb->gc_scratch + GDT_SCR_MSG_LEN);
if (msg_len)
if (!(gccb->gc_scratch[GDT_SCR_MSG_ANSWER] &&
gccb->gc_scratch[GDT_SCR_MSG_EXT])) {
gccb->gc_scratch[GDT_SCR_MSG_TEXT + msg_len] = '\0';
printf("%s",&gccb->gc_scratch[GDT_SCR_MSG_TEXT]);
}
if (gccb->gc_scratch[GDT_SCR_MSG_EXT] &&
!gccb->gc_scratch[GDT_SCR_MSG_ANSWER]) {
while (gdt->sc_test_busy(gdt))
DELAY(1);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bzero(gccb->gc_cmd, GDT_CMD_SZ);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb = gdt_get_ccb(gdt);
if (gccb == NULL) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode, "No free command index found\n");
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
gccb->gc_service = service;
gccb->gc_flags = GDT_GCF_SCREEN;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_COMMANDINDEX,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_cmd_index);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_OPCODE, GDT_READ);
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_SCREEN_MSG_HANDLE,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_scratch[GDT_SCR_MSG_HANDLE]);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_SCREEN_MSG_ADDR,
gccb->gc_scratch_busbase);
gdt->sc_set_sema0(gdt);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_cmd_off = 0;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd_len = roundup(GDT_CMD_UNION + GDT_SCREEN_SZ,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
sizeof(u_int32_t));
gdt->sc_cmd_cnt = 0;
gdt->sc_copy_cmd(gdt, gccb);
gdt->sc_release_event(gdt);
return (0);
}
if (gccb->gc_scratch[GDT_SCR_MSG_ANSWER] &&
gdt_dec32(gccb->gc_scratch + GDT_SCR_MSG_ALEN)) {
/* default answers (getchar() not possible) */
if (gdt_dec32(gccb->gc_scratch + GDT_SCR_MSG_ALEN) == 1) {
gdt_enc32(gccb->gc_scratch + GDT_SCR_MSG_ALEN, 0);
gdt_enc32(gccb->gc_scratch + GDT_SCR_MSG_LEN, 1);
gccb->gc_scratch[GDT_SCR_MSG_TEXT] = 0;
} else {
gdt_enc32(gccb->gc_scratch + GDT_SCR_MSG_ALEN,
gdt_dec32(gccb->gc_scratch + GDT_SCR_MSG_ALEN) - 2);
gdt_enc32(gccb->gc_scratch + GDT_SCR_MSG_LEN, 2);
gccb->gc_scratch[GDT_SCR_MSG_TEXT] = 1;
gccb->gc_scratch[GDT_SCR_MSG_TEXT + 1] = 0;
}
gccb->gc_scratch[GDT_SCR_MSG_EXT] = 0;
gccb->gc_scratch[GDT_SCR_MSG_ANSWER] = 0;
while (gdt->sc_test_busy(gdt))
DELAY(1);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bzero(gccb->gc_cmd, GDT_CMD_SZ);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb = gdt_get_ccb(gdt);
if (gccb == NULL) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
device_printf(gdt->sc_devnode, "No free command index found\n");
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return (1);
}
gccb->gc_service = service;
gccb->gc_flags = GDT_GCF_SCREEN;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_COMMANDINDEX,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_cmd_index);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc16(gccb->gc_cmd + GDT_CMD_OPCODE, GDT_WRITE);
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_SCREEN_MSG_HANDLE,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gccb->gc_scratch[GDT_SCR_MSG_HANDLE]);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gdt_enc32(gccb->gc_cmd + GDT_CMD_UNION + GDT_SCREEN_MSG_ADDR,
gccb->gc_scratch_busbase);
gdt->sc_set_sema0(gdt);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_cmd_off = 0;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
gccb->gc_cmd_len = roundup(GDT_CMD_UNION + GDT_SCREEN_SZ,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
sizeof(u_int32_t));
gdt->sc_cmd_cnt = 0;
gdt->sc_copy_cmd(gdt, gccb);
gdt->sc_release_event(gdt);
return (0);
}
printf("\n");
return (0);
} else {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
callout_stop(&gccb->gc_timeout);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
if (gdt->sc_status == GDT_S_BSY) {
GDT_DPRINTF(GDT_D_DEBUG, ("gdt_sync_event(%p) gccb %p busy\n",
gdt, gccb));
TAILQ_INSERT_HEAD(&gdt->sc_ccb_queue, &ccb->ccb_h, sim_links.tqe);
++gdt_stat.req_queue_act;
if (gdt_stat.req_queue_act > gdt_stat.req_queue_max)
gdt_stat.req_queue_max = gdt_stat.req_queue_act;
return (2);
}
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
bus_dmamap_sync(gdt->sc_buffer_dmat, gccb->gc_dmamap,
(ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN ?
BUS_DMASYNC_POSTREAD : BUS_DMASYNC_POSTWRITE);
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
bus_dmamap_unload(gdt->sc_buffer_dmat, gccb->gc_dmamap);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccb->csio.resid = 0;
if (gdt->sc_status == GDT_S_OK) {
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
ccb->ccb_h.status |= CAM_REQ_CMP;
ccb->ccb_h.status &= ~CAM_SIM_QUEUED;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
} else {
/* error */
if (gccb->gc_service == GDT_CACHESERVICE) {
Add descriptor sense support to CAM, and honor sense residuals properly in CAM. Desriptor sense is a new sense data format that originated in SPC-3. Among other things, it allows for an 8-byte info field, which is necessary to pass back block numbers larger than 4 bytes. This change adds a number of new functions to scsi_all.c (and therefore libcam) that abstract out most access to sense data. This includes a bump of CAM_VERSION, because the CCB ABI has changed. Userland programs that use the CAM pass(4) driver will need to be recompiled. camcontrol.c: Change uses of scsi_extract_sense() to use scsi_extract_sense_len(). Use scsi_get_sks() instead of accessing sense key specific data directly. scsi_modes: Update the control mode page to the latest version (SPC-4). scsi_cmds.c, scsi_target.c: Change references to struct scsi_sense_data to struct scsi_sense_data_fixed. This should be changed to allow the user to specify fixed or descriptor sense, and then use scsi_set_sense_data() to build the sense data. ps3cdrom.c: Use scsi_set_sense_data() instead of setting sense data manually. cam_periph.c: Use scsi_extract_sense_len() instead of using scsi_extract_sense() or accessing sense data directly. cam_ccb.h: Bump the CAM_VERSION from 0x15 to 0x16. The change of struct scsi_sense_data from 32 to 252 bytes changes the size of struct ccb_scsiio, but not the size of union ccb. So the version must be bumped to prevent structure mis-matches. scsi_all.h: Lots of updated SCSI sense data and other structures. Add function prototypes for the new sense data functions. Take out the inline implementation of scsi_extract_sense(). It is now too large to put in a header file. Add macros to calculate whether fields are present and filled in fixed and descriptor sense data scsi_all.c: In scsi_op_desc(), allow the user to pass in NULL inquiry data, and we'll assume a direct access device in that case. Changed the SCSI RESERVED sense key name and description to COMPLETED, as it is now defined in the spec. Change the error recovery action for a number of read errors to prevent lots of retries when the drive has said that the block isn't accessible. This speeds up reconstruction of the block by any RAID software running on top of the drive (e.g. ZFS). In scsi_sense_desc(), allow for invalid sense key numbers. This allows calling this routine without checking the input values first. Change scsi_error_action() to use scsi_extract_sense_len(), and handle things when invalid asc/ascq values are encountered. Add a new routine, scsi_desc_iterate(), that will call the supplied function for every descriptor in descriptor format sense data. Add scsi_set_sense_data(), and scsi_set_sense_data_va(), which build descriptor and fixed format sense data. They currently default to fixed format sense data. Add a number of scsi_get_*() functions, which get different types of sense data fields from either fixed or descriptor format sense data, if the data is present. Add a number of scsi_*_sbuf() functions, which print formatted versions of various sense data fields. These functions work for either fixed or descriptor sense. Add a number of scsi_sense_*_sbuf() functions, which have a standard calling interface and print the indicated field. These functions take descriptors only. Add scsi_sense_desc_sbuf(), which will print a formatted version of the given sense descriptor. Pull out a majority of the scsi_sense_sbuf() function and put it into scsi_sense_only_sbuf(). This allows callers that don't use struct ccb_scsiio to easily utilize the printing routines. Revamp that function to handle descriptor sense and use the new sense fetching and printing routines. Move scsi_extract_sense() into scsi_all.c, and implement it in terms of the new function, scsi_extract_sense_len(). The _len() version takes a length (which should be the sense length - residual) and can indicate which fields are present and valid in the sense data. Add a couple of new scsi_get_*() routines to get the sense key, asc, and ascq only. mly.c: Rename struct scsi_sense_data to struct scsi_sense_data_fixed. sbp_targ.c: Use the new sense fetching routines to get sense data instead of accessing it directly. sbp.c: Change the firewire/SCSI sense data transformation code to use struct scsi_sense_data_fixed instead of struct scsi_sense_data. This should be changed later to use scsi_set_sense_data(). ciss.c: Calculate the sense residual properly. Use scsi_get_sense_key() to fetch the sense key. mps_sas.c, mpt_cam.c: Set the sense residual properly. iir.c: Use scsi_set_sense_data() instead of building sense data by hand. iscsi_subr.c: Use scsi_extract_sense_len() instead of grabbing sense data directly. umass.c: Use scsi_set_sense_data() to build sense data. Grab the sense key using scsi_get_sense_key(). Calculate the sense residual properly. isp_freebsd.h: Use scsi_get_*() routines to grab asc, ascq, and sense key values. Calculate and set the sense residual. MFC after: 3 days Sponsored by: Spectra Logic Corporation
2011-10-03 20:32:55 +00:00
struct scsi_sense_data *sense;
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
ccb->ccb_h.status |= CAM_SCSI_STATUS_ERROR | CAM_AUTOSNS_VALID;
ccb->ccb_h.status &= ~CAM_SIM_QUEUED;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccb->csio.scsi_status = SCSI_STATUS_CHECK_COND;
bzero(&ccb->csio.sense_data, ccb->csio.sense_len);
Add descriptor sense support to CAM, and honor sense residuals properly in CAM. Desriptor sense is a new sense data format that originated in SPC-3. Among other things, it allows for an 8-byte info field, which is necessary to pass back block numbers larger than 4 bytes. This change adds a number of new functions to scsi_all.c (and therefore libcam) that abstract out most access to sense data. This includes a bump of CAM_VERSION, because the CCB ABI has changed. Userland programs that use the CAM pass(4) driver will need to be recompiled. camcontrol.c: Change uses of scsi_extract_sense() to use scsi_extract_sense_len(). Use scsi_get_sks() instead of accessing sense key specific data directly. scsi_modes: Update the control mode page to the latest version (SPC-4). scsi_cmds.c, scsi_target.c: Change references to struct scsi_sense_data to struct scsi_sense_data_fixed. This should be changed to allow the user to specify fixed or descriptor sense, and then use scsi_set_sense_data() to build the sense data. ps3cdrom.c: Use scsi_set_sense_data() instead of setting sense data manually. cam_periph.c: Use scsi_extract_sense_len() instead of using scsi_extract_sense() or accessing sense data directly. cam_ccb.h: Bump the CAM_VERSION from 0x15 to 0x16. The change of struct scsi_sense_data from 32 to 252 bytes changes the size of struct ccb_scsiio, but not the size of union ccb. So the version must be bumped to prevent structure mis-matches. scsi_all.h: Lots of updated SCSI sense data and other structures. Add function prototypes for the new sense data functions. Take out the inline implementation of scsi_extract_sense(). It is now too large to put in a header file. Add macros to calculate whether fields are present and filled in fixed and descriptor sense data scsi_all.c: In scsi_op_desc(), allow the user to pass in NULL inquiry data, and we'll assume a direct access device in that case. Changed the SCSI RESERVED sense key name and description to COMPLETED, as it is now defined in the spec. Change the error recovery action for a number of read errors to prevent lots of retries when the drive has said that the block isn't accessible. This speeds up reconstruction of the block by any RAID software running on top of the drive (e.g. ZFS). In scsi_sense_desc(), allow for invalid sense key numbers. This allows calling this routine without checking the input values first. Change scsi_error_action() to use scsi_extract_sense_len(), and handle things when invalid asc/ascq values are encountered. Add a new routine, scsi_desc_iterate(), that will call the supplied function for every descriptor in descriptor format sense data. Add scsi_set_sense_data(), and scsi_set_sense_data_va(), which build descriptor and fixed format sense data. They currently default to fixed format sense data. Add a number of scsi_get_*() functions, which get different types of sense data fields from either fixed or descriptor format sense data, if the data is present. Add a number of scsi_*_sbuf() functions, which print formatted versions of various sense data fields. These functions work for either fixed or descriptor sense. Add a number of scsi_sense_*_sbuf() functions, which have a standard calling interface and print the indicated field. These functions take descriptors only. Add scsi_sense_desc_sbuf(), which will print a formatted version of the given sense descriptor. Pull out a majority of the scsi_sense_sbuf() function and put it into scsi_sense_only_sbuf(). This allows callers that don't use struct ccb_scsiio to easily utilize the printing routines. Revamp that function to handle descriptor sense and use the new sense fetching and printing routines. Move scsi_extract_sense() into scsi_all.c, and implement it in terms of the new function, scsi_extract_sense_len(). The _len() version takes a length (which should be the sense length - residual) and can indicate which fields are present and valid in the sense data. Add a couple of new scsi_get_*() routines to get the sense key, asc, and ascq only. mly.c: Rename struct scsi_sense_data to struct scsi_sense_data_fixed. sbp_targ.c: Use the new sense fetching routines to get sense data instead of accessing it directly. sbp.c: Change the firewire/SCSI sense data transformation code to use struct scsi_sense_data_fixed instead of struct scsi_sense_data. This should be changed later to use scsi_set_sense_data(). ciss.c: Calculate the sense residual properly. Use scsi_get_sense_key() to fetch the sense key. mps_sas.c, mpt_cam.c: Set the sense residual properly. iir.c: Use scsi_set_sense_data() instead of building sense data by hand. iscsi_subr.c: Use scsi_extract_sense_len() instead of grabbing sense data directly. umass.c: Use scsi_set_sense_data() to build sense data. Grab the sense key using scsi_get_sense_key(). Calculate the sense residual properly. isp_freebsd.h: Use scsi_get_*() routines to grab asc, ascq, and sense key values. Calculate and set the sense residual. MFC after: 3 days Sponsored by: Spectra Logic Corporation
2011-10-03 20:32:55 +00:00
sense = &ccb->csio.sense_data;
scsi_set_sense_data(sense,
/*sense_format*/ SSD_TYPE_NONE,
/*current_error*/ 1,
/*sense_key*/ SSD_KEY_NOT_READY,
/*asc*/ 0x4,
/*ascq*/ 0x01,
SSD_ELEM_NONE);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt->sc_dvr.size = sizeof(gdt->sc_dvr.eu.sync);
gdt->sc_dvr.eu.sync.ionode = gdt->sc_hanum;
gdt->sc_dvr.eu.sync.service = service;
gdt->sc_dvr.eu.sync.status = gdt->sc_status;
gdt->sc_dvr.eu.sync.info = gdt->sc_info;
gdt->sc_dvr.eu.sync.hostdrive = ccb->ccb_h.target_id;
if (gdt->sc_status >= 0x8000)
gdt_store_event(GDT_ES_SYNC, 0, &gdt->sc_dvr);
else
gdt_store_event(GDT_ES_SYNC, service, &gdt->sc_dvr);
} else {
/* raw service */
if (gdt->sc_status != GDT_S_RAW_SCSI || gdt->sc_info >= 0x100) {
Import latest changes from the vendor. This also is reported to fix the driver at long last! Many thanks to vaidas.damosevicius@if.lt for keeping this issue alive and pursuing Intel for a fix, Intel/ICP for working on the driver, and Sergey Osokin for bringing the original patches up to 5-CURRENT.
2003-04-25 05:37:04 +00:00
ccb->ccb_h.status = CAM_DEV_NOT_THERE;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
} else {
Big update to the iir driver: - Don't use a common buffer in the softc to store per-command data. Reserve a buffer in the command itself. - Don't allocate DMA memory for the kernel command structures when all you really need is DMA memory for the scratch buffer embedded in them. Instead allocate a slab for the scratch buffers and divide it up as needed. - Call bus_dmamap_unload() at the completion of commands. - Preserve and clear the CAM CCB status flags at completion. - Reorder some low-level command operations to try to close races. - Limit the simq to 32 commands for now. There are some serious problems with the driver under load that are not well understood, so keeping the simq lower helps avoid this. It has been tested at a higher value, but this is a safe value that doesn't show much performance degredation. These changes allow the driver to work reliably with >4GB of memory on i386 and amd64, and also work around deadlocks seen under very high load in certain situations. The work-around is far from ideal, but without and documentation it is hard to know what the right fix is. MFC candidate
2006-03-01 07:24:39 +00:00
ccb->ccb_h.status |= CAM_SCSI_STATUS_ERROR|CAM_AUTOSNS_VALID;
ccb->ccb_h.status &= ~CAM_SIM_QUEUED;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
ccb->csio.scsi_status = gdt->sc_info;
bcopy(gccb->gc_scratch, &ccb->csio.sense_data,
ccb->csio.sense_len);
}
}
}
--gdt_stat.io_count_act;
xpt_done(ccb);
}
return (0);
}
/* Controller event handling functions */
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
void gdt_store_event(u_int16_t source, u_int16_t idx,
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
gdt_evt_data *evt)
{
gdt_evt_str *e;
struct timeval tv;
GDT_DPRINTF(GDT_D_MISC, ("gdt_store_event(%d, %d)\n", source, idx));
if (source == 0) /* no source -> no event */
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
return;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_lock(&elock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
if (ebuffer[elastidx].event_source == source &&
ebuffer[elastidx].event_idx == idx &&
((evt->size != 0 && ebuffer[elastidx].event_data.size != 0 &&
!memcmp((char *)&ebuffer[elastidx].event_data.eu,
(char *)&evt->eu, evt->size)) ||
(evt->size == 0 && ebuffer[elastidx].event_data.size == 0 &&
!strcmp((char *)&ebuffer[elastidx].event_data.event_string,
(char *)&evt->event_string)))) {
e = &ebuffer[elastidx];
getmicrotime(&tv);
e->last_stamp = tv.tv_sec;
++e->same_count;
} else {
if (ebuffer[elastidx].event_source != 0) { /* entry not free ? */
++elastidx;
if (elastidx == GDT_MAX_EVENTS)
elastidx = 0;
if (elastidx == eoldidx) { /* reached mark ? */
++eoldidx;
if (eoldidx == GDT_MAX_EVENTS)
eoldidx = 0;
}
}
e = &ebuffer[elastidx];
e->event_source = source;
e->event_idx = idx;
getmicrotime(&tv);
e->first_stamp = e->last_stamp = tv.tv_sec;
e->same_count = 1;
e->event_data = *evt;
e->application = 0;
}
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&elock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
int gdt_read_event(int handle, gdt_evt_str *estr)
{
gdt_evt_str *e;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
int eindex;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_DPRINTF(GDT_D_MISC, ("gdt_read_event(%d)\n", handle));
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_lock(&elock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
if (handle == -1)
eindex = eoldidx;
else
eindex = handle;
estr->event_source = 0;
if (eindex >= GDT_MAX_EVENTS) {
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&elock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return eindex;
}
e = &ebuffer[eindex];
if (e->event_source != 0) {
if (eindex != elastidx) {
if (++eindex == GDT_MAX_EVENTS)
eindex = 0;
} else {
eindex = -1;
}
memcpy(estr, e, sizeof(gdt_evt_str));
}
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&elock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
return eindex;
}
void gdt_readapp_event(u_int8_t application, gdt_evt_str *estr)
{
gdt_evt_str *e;
int found = FALSE;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
int eindex;
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
GDT_DPRINTF(GDT_D_MISC, ("gdt_readapp_event(%d)\n", application));
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_lock(&elock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
eindex = eoldidx;
for (;;) {
e = &ebuffer[eindex];
if (e->event_source == 0)
break;
if ((e->application & application) == 0) {
e->application |= application;
found = TRUE;
break;
}
if (eindex == elastidx)
break;
if (++eindex == GDT_MAX_EVENTS)
eindex = 0;
}
if (found)
memcpy(estr, e, sizeof(gdt_evt_str));
else
estr->event_source = 0;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&elock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
void gdt_clear_events()
{
GDT_DPRINTF(GDT_D_MISC, ("gdt_clear_events\n"));
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_lock(&elock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
eoldidx = elastidx = 0;
ebuffer[0].event_source = 0;
Lock iir(4) and mark it MPSAFE. - Add a per-device mutex to the softc and use it for bus_dma tags, CAM SIMs, callouts, and interrupt handler. - Switch from timeout(9) to callout(9). - Add a separate global mutex to protect the global event buffer ring. - Return completed index from iir_intr_locked() and remove the global gdt_wait_* variables. - Remove global list of gdt softcs and replace its use with devclass_get_device(). - Use si_drv1 to store softc pointer in the SDEV_PER_HBA case instead of minor numbers. - Do math on osreldate instead of dubious char math on osrelease[] that didn't work on 10.0+. - Use bus_*() instead of bus_space_*(). - Use device_printf() instead of printf() with a unit number. Tested by: no one
2014-11-13 22:00:18 +00:00
mtx_unlock(&elock);
Add the 'iir' driver, for the Intel Integrated RAID controllers and prior ICP Vortex models. This driver was developed by Achim Leubner of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel. Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com> MFC after: 2 weeks
2002-01-20 08:51:08 +00:00
}
Reference in New Issue Copy Permalink