NetBSD/sys/dev/pci/ahd_pci.c

1186 lines
32 KiB
C
Raw Normal View History

/* $NetBSD: ahd_pci.c,v 1.25 2006/11/16 01:33:08 christos Exp $ */
/*
* Product specific probe and attach routines for:
* aic7901 and aic7902 SCSI controllers
*
* Copyright (c) 1994-2001 Justin T. Gibbs.
* Copyright (c) 2000-2002 Adaptec Inc.
* 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.
* 2. Redistributions in binary form must reproduce at minimum a disclaimer
* substantially similar to the "NO WARRANTY" disclaimer below
* ("Disclaimer") and any redistribution must be conditioned upon
* including a substantially similar Disclaimer requirement for further
* binary redistribution.
* 3. Neither the names of the above-listed copyright holders nor the names
* of any contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* Alternatively, this software may be distributed under the terms of the
* GNU General Public License ("GPL") version 2 as published by the Free
* Software Foundation.
*
* NO WARRANTY
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* HOLDERS OR CONTRIBUTORS BE LIABLE FOR 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 DAMAGES.
*
* Id: //depot/aic7xxx/aic7xxx/aic79xx_pci.c#80 $
*
* $FreeBSD: src/sys/dev/aic7xxx/aic79xx_pci.c,v 1.16 2003/06/28 04:39:49 gibbs Exp $
*/
/*
* Ported from FreeBSD by Pascal Renauld, Network Storage Solutions, Inc. - April 2003
*/
2003-07-14 19:47:00 +04:00
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: ahd_pci.c,v 1.25 2006/11/16 01:33:08 christos Exp $");
2003-07-14 19:47:00 +04:00
#define AHD_PCI_IOADDR PCI_MAPREG_START /* I/O Address */
#define AHD_PCI_MEMADDR (PCI_MAPREG_START + 4) /* Mem I/O Address */
#include <dev/ic/aic79xx_osm.h>
#include <dev/ic/aic79xx_inline.h>
static inline uint64_t
ahd_compose_id(u_int device, u_int vendor, u_int subdevice, u_int subvendor)
{
uint64_t id;
id = subvendor
| (subdevice << 16)
| ((uint64_t)vendor << 32)
| ((uint64_t)device << 48);
return (id);
}
#define ID_ALL_MASK 0xFFFFFFFFFFFFFFFFull
#define ID_ALL_IROC_MASK 0xFF7FFFFFFFFFFFFFull
#define ID_DEV_VENDOR_MASK 0xFFFFFFFF00000000ull
#define ID_9005_GENERIC_MASK 0xFFF0FFFF00000000ull
#define ID_9005_GENERIC_IROC_MASK 0xFF70FFFF00000000ull
#define ID_AIC7901 0x800F9005FFFF9005ull
#define ID_AHA_29320A 0x8000900500609005ull
#define ID_AHA_29320ALP 0x8017900500449005ull
#define ID_AIC7901A 0x801E9005FFFF9005ull
#define ID_AHA_29320LP 0x8014900500449005ull
#define ID_AIC7902 0x801F9005FFFF9005ull
#define ID_AIC7902_B 0x801D9005FFFF9005ull
#define ID_AHA_39320 0x8010900500409005ull
#define ID_AHA_29320 0x8012900500429005ull
#define ID_AHA_29320B 0x8013900500439005ull
#define ID_AHA_39320_B 0x8015900500409005ull
#define ID_AHA_39320A 0x8016900500409005ull
#define ID_AHA_39320D 0x8011900500419005ull
#define ID_AHA_39320D_B 0x801C900500419005ull
#define ID_AHA_39320_B_DELL 0x8015900501681028ull
#define ID_AHA_39320D_HP 0x8011900500AC0E11ull
#define ID_AHA_39320D_B_HP 0x801C900500AC0E11ull
#define ID_AIC7902_PCI_REV_A4 0x3
#define ID_AIC7902_PCI_REV_B0 0x10
#define SUBID_HP 0x0E11
#define DEVID_9005_HOSTRAID(id) ((id) & 0x80)
#define DEVID_9005_TYPE(id) ((id) & 0xF)
#define DEVID_9005_TYPE_HBA 0x0 /* Standard Card */
#define DEVID_9005_TYPE_HBA_2EXT 0x1 /* 2 External Ports */
#define DEVID_9005_TYPE_MB 0xF /* On Motherboard */
#define DEVID_9005_MFUNC(id) ((id) & 0x10)
#define DEVID_9005_PACKETIZED(id) ((id) & 0x8000)
#define SUBID_9005_TYPE(id) ((id) & 0xF)
#define SUBID_9005_TYPE_HBA 0x0 /* Standard Card */
#define SUBID_9005_TYPE_MB 0xF /* On Motherboard */
#define SUBID_9005_AUTOTERM(id) (((id) & 0x10) == 0)
#define SUBID_9005_LEGACYCONN_FUNC(id) ((id) & 0x20)
#define SUBID_9005_SEEPTYPE(id) ((id) & 0x0C0) >> 6)
#define SUBID_9005_SEEPTYPE_NONE 0x0
#define SUBID_9005_SEEPTYPE_4K 0x1
static ahd_device_setup_t ahd_aic7901_setup;
static ahd_device_setup_t ahd_aic7901A_setup;
static ahd_device_setup_t ahd_aic7902_setup;
static ahd_device_setup_t ahd_aic790X_setup;
2005-06-28 04:28:41 +04:00
static struct ahd_pci_identity ahd_pci_ident_table [] =
{
/* aic7901 based controllers */
{
ID_AHA_29320A,
ID_ALL_MASK,
"Adaptec 29320A Ultra320 SCSI adapter",
ahd_aic7901_setup
},
{
ID_AHA_29320ALP,
ID_ALL_MASK,
"Adaptec 29320ALP Ultra320 SCSI adapter",
ahd_aic7901_setup
},
/* aic7901A based controllers */
{
ID_AHA_29320LP,
ID_ALL_MASK,
"Adaptec 29320LP Ultra320 SCSI adapter",
ahd_aic7901A_setup
},
2005-02-27 03:26:58 +03:00
/* aic7902 based controllers */
{
ID_AHA_39320,
ID_ALL_MASK,
"Adaptec 39320 Ultra320 SCSI adapter",
ahd_aic7902_setup
},
{
ID_AHA_39320_B,
ID_ALL_MASK,
"Adaptec 39320 Ultra320 SCSI adapter",
ahd_aic7902_setup
},
{
ID_AHA_39320_B_DELL,
ID_ALL_IROC_MASK,
"Adaptec (Dell OEM) 39320 Ultra320 SCSI adapter",
ahd_aic7902_setup
},
{
ID_AHA_39320A,
ID_ALL_MASK,
"Adaptec 39320A Ultra320 SCSI adapter",
ahd_aic7902_setup
},
{
ID_AHA_39320D,
ID_ALL_MASK,
"Adaptec 39320D Ultra320 SCSI adapter",
ahd_aic7902_setup
},
{
ID_AHA_39320D_HP,
ID_ALL_MASK,
"Adaptec (HP OEM) 39320D Ultra320 SCSI adapter",
ahd_aic7902_setup
},
{
ID_AHA_39320D_B,
ID_ALL_MASK,
"Adaptec 39320D Ultra320 SCSI adapter",
ahd_aic7902_setup
},
{
ID_AHA_39320D_B_HP,
ID_ALL_MASK,
"Adaptec (HP OEM) 39320D Ultra320 SCSI adapter",
ahd_aic7902_setup
},
/* Generic chip probes for devices we don't know 'exactly' */
{
ID_AIC7901 & ID_9005_GENERIC_MASK,
ID_9005_GENERIC_MASK,
"Adaptec AIC7901 Ultra320 SCSI adapter",
ahd_aic7901_setup
},
{
ID_AIC7901A & ID_DEV_VENDOR_MASK,
ID_DEV_VENDOR_MASK,
"Adaptec AIC7901A Ultra320 SCSI adapter",
ahd_aic7901A_setup
},
{
ID_AIC7902 & ID_9005_GENERIC_MASK,
ID_9005_GENERIC_MASK,
"Adaptec AIC7902 Ultra320 SCSI adapter",
ahd_aic7902_setup
}
};
2005-06-28 04:28:41 +04:00
static const u_int ahd_num_pci_devs = NUM_ELEMENTS(ahd_pci_ident_table);
#define DEVCONFIG 0x40
#define PCIXINITPAT 0x0000E000ul
#define PCIXINIT_PCI33_66 0x0000E000ul
#define PCIXINIT_PCIX50_66 0x0000C000ul
#define PCIXINIT_PCIX66_100 0x0000A000ul
#define PCIXINIT_PCIX100_133 0x00008000ul
#define PCI_BUS_MODES_INDEX(devconfig) \
(((devconfig) & PCIXINITPAT) >> 13)
static const char *pci_bus_modes[] =
{
"PCI bus mode unknown",
"PCI bus mode unknown",
"PCI bus mode unknown",
"PCI bus mode unknown",
"PCI-X 101-133 MHz",
"PCI-X 67-100 MHz",
"PCI-X 50-66 MHz",
"PCI 33 or 66 MHz"
};
#define TESTMODE 0x00000800ul
#define IRDY_RST 0x00000200ul
#define FRAME_RST 0x00000100ul
#define PCI64BIT 0x00000080ul
#define MRDCEN 0x00000040ul
#define ENDIANSEL 0x00000020ul
#define MIXQWENDIANEN 0x00000008ul
#define DACEN 0x00000004ul
#define STPWLEVEL 0x00000002ul
#define QWENDIANSEL 0x00000001ul
#define DEVCONFIG1 0x44
#define PREQDIS 0x01
#define LATTIME 0x0000ff00ul
static int ahd_check_extport(struct ahd_softc *ahd);
static void ahd_configure_termination(struct ahd_softc *ahd,
u_int adapter_control);
static void ahd_pci_split_intr(struct ahd_softc *ahd, u_int intstat);
2005-06-28 04:28:41 +04:00
static int ahd_pci_test_register_access(struct ahd_softc *);
static int ahd_pci_intr(struct ahd_softc *);
static const struct ahd_pci_identity *
ahd_find_pci_device(pcireg_t id, pcireg_t subid)
{
u_int64_t full_id;
const struct ahd_pci_identity *entry;
u_int i;
full_id = ahd_compose_id(PCI_PRODUCT(id), PCI_VENDOR(id),
PCI_PRODUCT(subid), PCI_VENDOR(subid));
for (i = 0; i < ahd_num_pci_devs; i++) {
entry = &ahd_pci_ident_table[i];
if (entry->full_id == (full_id & entry->id_mask))
return (entry);
}
return (NULL);
}
2005-06-28 04:28:41 +04:00
static int
ahd_pci_probe(struct device *parent, struct cfdata *match,
void *aux)
{
struct pci_attach_args *pa = aux;
const struct ahd_pci_identity *entry;
pcireg_t subid;
subid = pci_conf_read(pa->pa_pc, pa->pa_tag, PCI_SUBSYS_ID_REG);
entry = ahd_find_pci_device(pa->pa_id, subid);
return entry != NULL ? 1 : 0;
}
2005-06-28 04:28:41 +04:00
static void
ahd_pci_attach(struct device *parent, struct device *self, void *aux)
{
struct pci_attach_args *pa = aux;
struct ahd_softc *ahd = (void *)self;
const struct ahd_pci_identity *entry;
uint32_t devconfig;
pcireg_t command;
int error;
pcireg_t subid;
2005-02-27 03:26:58 +03:00
uint16_t subvendor;
pcireg_t reg;
int ioh_valid, ioh2_valid, memh_valid;
pcireg_t memtype;
pci_intr_handle_t ih;
const char *intrstr;
struct ahd_pci_busdata *bd;
ahd_set_name(ahd, ahd->sc_dev.dv_xname);
ahd->parent_dmat = pa->pa_dmat;
2005-02-27 03:26:58 +03:00
command = pci_conf_read(pa->pa_pc, pa->pa_tag, PCI_COMMAND_STATUS_REG);
subid = pci_conf_read(pa->pa_pc, pa->pa_tag, PCI_SUBSYS_ID_REG);
entry = ahd_find_pci_device(pa->pa_id, subid);
if (entry == NULL)
return;
/* Keep information about the PCI bus */
bd = malloc(sizeof (struct ahd_pci_busdata), M_DEVBUF, M_NOWAIT);
if (bd == NULL) {
2003-11-19 00:40:13 +03:00
aprint_error("%s: unable to allocate bus-specific data\n", ahd_name(ahd));
return;
}
memset(bd, 0, sizeof(struct ahd_pci_busdata));
bd->pc = pa->pa_pc;
bd->tag = pa->pa_tag;
bd->func = pa->pa_function;
bd->dev = pa->pa_device;
ahd->bus_data = bd;
ahd->description = entry->name;
ahd->seep_config = malloc(sizeof(*ahd->seep_config),
M_DEVBUF, M_NOWAIT);
if (ahd->seep_config == NULL) {
2003-11-19 00:40:13 +03:00
aprint_error("%s: cannot malloc seep_config!\n", ahd_name(ahd));
return;
}
memset(ahd->seep_config, 0, sizeof(*ahd->seep_config));
LIST_INIT(&ahd->pending_scbs);
ahd_timer_init(&ahd->reset_timer);
ahd_timer_init(&ahd->stat_timer);
ahd->flags = AHD_SPCHK_ENB_A|AHD_RESET_BUS_A|AHD_TERM_ENB_A
| AHD_EXTENDED_TRANS_A|AHD_STPWLEVEL_A;
ahd->int_coalescing_timer = AHD_INT_COALESCING_TIMER_DEFAULT;
ahd->int_coalescing_maxcmds = AHD_INT_COALESCING_MAXCMDS_DEFAULT;
ahd->int_coalescing_mincmds = AHD_INT_COALESCING_MINCMDS_DEFAULT;
ahd->int_coalescing_threshold = AHD_INT_COALESCING_THRESHOLD_DEFAULT;
ahd->int_coalescing_stop_threshold = AHD_INT_COALESCING_STOP_THRESHOLD_DEFAULT;
if (ahd_platform_alloc(ahd, NULL) != 0) {
ahd_free(ahd);
return;
}
/*
* Record if this is an HP board.
*/
subvendor = PCI_VENDOR(subid);
if (subvendor == SUBID_HP)
ahd->flags |= AHD_HP_BOARD;
error = entry->setup(ahd, pa);
if (error != 0)
return;
devconfig = pci_conf_read(pa->pa_pc, pa->pa_tag, DEVCONFIG);
if ((devconfig & PCIXINITPAT) == PCIXINIT_PCI33_66) {
ahd->chip |= AHD_PCI;
/* Disable PCIX workarounds when running in PCI mode. */
ahd->bugs &= ~AHD_PCIX_BUG_MASK;
} else {
ahd->chip |= AHD_PCIX;
}
ahd->bus_description = pci_bus_modes[PCI_BUS_MODES_INDEX(devconfig)];
memh_valid = ioh_valid = ioh2_valid = 0;
if (!pci_get_capability(pa->pa_pc, pa->pa_tag, PCI_CAP_PCIX,
&bd->pcix_off, NULL)) {
if (ahd->chip & AHD_PCIX)
2003-11-19 00:40:13 +03:00
aprint_error("%s: warning: can't find PCI-X capability\n",
ahd->sc_dev.dv_xname);
ahd->chip &= ~AHD_PCIX;
ahd->chip |= AHD_PCI;
ahd->bugs &= ~AHD_PCIX_BUG_MASK;
}
/*
* Map PCI Registers
*/
if ((ahd->bugs & AHD_PCIX_MMAPIO_BUG) == 0) {
memtype = pci_mapreg_type(pa->pa_pc, pa->pa_tag,
AHD_PCI_MEMADDR);
switch (memtype) {
case PCI_MAPREG_TYPE_MEM | PCI_MAPREG_MEM_TYPE_32BIT:
case PCI_MAPREG_TYPE_MEM | PCI_MAPREG_MEM_TYPE_64BIT:
memh_valid = (pci_mapreg_map(pa, AHD_PCI_MEMADDR,
2005-02-27 03:26:58 +03:00
memtype, 0, &ahd->tags[0],
&ahd->bshs[0],
NULL, NULL) == 0);
if (memh_valid) {
ahd->tags[1] = ahd->tags[0];
bus_space_subregion(ahd->tags[0], ahd->bshs[0],
/*offset*/0x100,
/*size*/0x100,
&ahd->bshs[1]);
if (ahd_pci_test_register_access(ahd) != 0)
memh_valid = 0;
}
break;
2005-02-27 03:26:58 +03:00
default:
memh_valid = 0;
2003-11-19 00:40:13 +03:00
aprint_error("%s: unknown memory type: 0x%x\n",
ahd_name(ahd), memtype);
break;
}
if (memh_valid) {
command &= ~PCI_COMMAND_IO_ENABLE;
pci_conf_write(pa->pa_pc, pa->pa_tag,
PCI_COMMAND_STATUS_REG, command);
}
#ifdef AHD_DEBUG
printf("%s: doing memory mapping tag0 0x%x, tag1 0x%x, "
"shs0 0x%lx, shs1 0x%lx\n",
ahd_name(ahd), ahd->tags[0], ahd->tags[1],
ahd->bshs[0], ahd->bshs[1]);
#endif
}
if (command & PCI_COMMAND_IO_ENABLE) {
2005-02-27 03:26:58 +03:00
/* First BAR */
ioh_valid = (pci_mapreg_map(pa, AHD_PCI_IOADDR,
PCI_MAPREG_TYPE_IO, 0,
&ahd->tags[0], &ahd->bshs[0],
NULL, NULL) == 0);
2005-02-27 03:26:58 +03:00
/* 2nd BAR */
ioh2_valid = (pci_mapreg_map(pa, AHD_PCI_IOADDR1,
PCI_MAPREG_TYPE_IO, 0,
&ahd->tags[1], &ahd->bshs[1],
NULL, NULL) == 0);
if (ioh_valid && ioh2_valid) {
KASSERT(memh_valid == 0);
command &= ~PCI_COMMAND_MEM_ENABLE;
pci_conf_write(pa->pa_pc, pa->pa_tag,
PCI_COMMAND_STATUS_REG, command);
2005-02-27 03:26:58 +03:00
}
#ifdef AHD_DEBUG
printf("%s: doing io mapping tag0 0x%x, tag1 0x%x, "
"shs0 0x%lx, shs1 0x%lx\n", ahd_name(ahd), ahd->tags[0],
ahd->tags[1], ahd->bshs[0], ahd->bshs[1]);
#endif
}
if (memh_valid == 0 && (ioh_valid == 0 || ioh2_valid == 0)) {
2003-11-19 00:40:13 +03:00
aprint_error("%s: unable to map registers\n", ahd_name(ahd));
return;
}
2003-11-19 00:40:13 +03:00
aprint_normal("\n");
aprint_naive("\n");
/* power up chip */
if ((error = pci_activate(pa->pa_pc, pa->pa_tag, ahd,
pci_activate_null)) && error != EOPNOTSUPP) {
aprint_error("%s: cannot activate %d\n", ahd->sc_dev.dv_xname,
error);
return;
}
/*
* Should we bother disabling 39Bit addressing
* based on installed memory?
*/
if (sizeof(bus_addr_t) > 4)
ahd->flags |= AHD_39BIT_ADDRESSING;
/*
* If we need to support high memory, enable dual
* address cycles. This bit must be set to enable
* high address bit generation even if we are on a
* 64bit bus (PCI64BIT set in devconfig).
*/
if ((ahd->flags & (AHD_39BIT_ADDRESSING|AHD_64BIT_ADDRESSING)) != 0) {
uint32_t dvconfig;
2003-11-19 00:40:13 +03:00
aprint_normal("%s: Enabling 39Bit Addressing\n", ahd_name(ahd));
dvconfig = pci_conf_read(pa->pa_pc, pa->pa_tag, DEVCONFIG);
dvconfig |= DACEN;
pci_conf_write(pa->pa_pc, pa->pa_tag, DEVCONFIG, dvconfig);
}
2005-02-27 03:26:58 +03:00
/* Ensure busmastering is enabled */
reg = pci_conf_read(pa->pa_pc, pa->pa_tag, PCI_COMMAND_STATUS_REG);
pci_conf_write(pa->pa_pc, pa->pa_tag, PCI_COMMAND_STATUS_REG,
reg | PCI_COMMAND_MASTER_ENABLE);
ahd_softc_init(ahd);
2005-02-27 03:26:58 +03:00
/*
* Map the interrupt routines
*/
ahd->bus_intr = ahd_pci_intr;
error = ahd_reset(ahd, /*reinit*/FALSE);
if (error != 0) {
ahd_free(ahd);
return;
}
if (pci_intr_map(pa, &ih)) {
2003-11-19 00:40:13 +03:00
aprint_error("%s: couldn't map interrupt\n", ahd_name(ahd));
ahd_free(ahd);
return;
}
intrstr = pci_intr_string(pa->pa_pc, ih);
ahd->ih = pci_intr_establish(pa->pa_pc, ih, IPL_BIO, ahd_intr, ahd);
if (ahd->ih == NULL) {
2003-11-19 00:40:13 +03:00
aprint_error("%s: couldn't establish interrupt",
ahd_name(ahd));
if (intrstr != NULL)
2003-11-19 00:40:13 +03:00
aprint_error(" at %s", intrstr);
aprint_error("\n");
ahd_free(ahd);
return;
}
if (intrstr != NULL)
2003-11-19 00:40:13 +03:00
aprint_normal("%s: interrupting at %s\n", ahd_name(ahd),
intrstr);
/* Get the size of the cache */
ahd->pci_cachesize = pci_conf_read(pa->pa_pc, pa->pa_tag, PCI_BHLC_REG);
ahd->pci_cachesize *= 4;
ahd_set_modes(ahd, AHD_MODE_SCSI, AHD_MODE_SCSI);
/* See if we have a SEEPROM and perform auto-term */
error = ahd_check_extport(ahd);
if (error != 0)
return;
/* Core initialization */
error = ahd_init(ahd);
if (error != 0)
return;
/*
* Link this softc in with all other ahd instances.
*/
ahd_attach(ahd);
}
2005-06-28 04:28:41 +04:00
CFATTACH_DECL(ahd_pci, sizeof(struct ahd_softc),
ahd_pci_probe, ahd_pci_attach, NULL, NULL);
/*
* Perform some simple tests that should catch situations where
* our registers are invalidly mapped.
*/
2005-06-28 04:28:41 +04:00
static int
ahd_pci_test_register_access(struct ahd_softc *ahd)
{
uint32_t cmd;
struct ahd_pci_busdata *bd = ahd->bus_data;
u_int targpcistat;
uint32_t pci_status1;
int error;
uint8_t hcntrl;
error = EIO;
/*
* Enable PCI error interrupt status, but suppress NMIs
* generated by SERR raised due to target aborts.
*/
cmd = pci_conf_read(bd->pc, bd->tag, PCI_COMMAND_STATUS_REG);
pci_conf_write(bd->pc, bd->tag, PCI_COMMAND_STATUS_REG,
cmd & ~PCI_COMMAND_SERR_ENABLE);
/*
* First a simple test to see if any
* registers can be read. Reading
* HCNTRL has no side effects and has
* at least one bit that is guaranteed to
* be zero so it is a good register to
* use for this test.
*/
hcntrl = ahd_inb(ahd, HCNTRL);
if (hcntrl == 0xFF)
goto fail;
/*
* Next create a situation where write combining
* or read prefetching could be initiated by the
* CPU or host bridge. Our device does not support
* either, so look for data corruption and/or flaged
* PCI errors. First pause without causing another
* chip reset.
*/
hcntrl &= ~CHIPRST;
ahd_outb(ahd, HCNTRL, hcntrl|PAUSE);
while (ahd_is_paused(ahd) == 0)
;
/* Clear any PCI errors that occurred before our driver attached. */
ahd_set_modes(ahd, AHD_MODE_CFG, AHD_MODE_CFG);
targpcistat = ahd_inb(ahd, TARGPCISTAT);
ahd_outb(ahd, TARGPCISTAT, targpcistat);
pci_status1 = pci_conf_read(bd->pc, bd->tag, PCI_COMMAND_STATUS_REG);
pci_conf_write(bd->pc, bd->tag, PCI_COMMAND_STATUS_REG, pci_status1);
ahd_set_modes(ahd, AHD_MODE_SCSI, AHD_MODE_SCSI);
ahd_outb(ahd, CLRINT, CLRPCIINT);
ahd_outb(ahd, SEQCTL0, PERRORDIS);
ahd_outl(ahd, SRAM_BASE, 0x5aa555aa);
if (ahd_inl(ahd, SRAM_BASE) != 0x5aa555aa)
goto fail;
if ((ahd_inb(ahd, INTSTAT) & PCIINT) != 0) {
u_int trgpcistat;
ahd_set_modes(ahd, AHD_MODE_CFG, AHD_MODE_CFG);
trgpcistat = ahd_inb(ahd, TARGPCISTAT);
if ((trgpcistat & STA) != 0)
goto fail;
}
error = 0;
fail:
if ((ahd_inb(ahd, INTSTAT) & PCIINT) != 0) {
ahd_set_modes(ahd, AHD_MODE_CFG, AHD_MODE_CFG);
targpcistat = ahd_inb(ahd, TARGPCISTAT);
/* Silently clear any latched errors. */
ahd_outb(ahd, TARGPCISTAT, targpcistat);
pci_status1 = pci_conf_read(bd->pc, bd->tag,
PCI_COMMAND_STATUS_REG);
pci_conf_write(bd->pc, bd->tag, PCI_COMMAND_STATUS_REG,
pci_status1);
ahd_outb(ahd, CLRINT, CLRPCIINT);
}
ahd_outb(ahd, SEQCTL0, PERRORDIS|FAILDIS);
pci_conf_write(bd->pc, bd->tag, PCI_COMMAND_STATUS_REG, cmd);
return (error);
}
/*
* Check the external port logic for a serial eeprom
* and termination/cable detection contrls.
*/
static int
ahd_check_extport(struct ahd_softc *ahd)
{
Apply the following change checked in 2003/05/04 00:20:07 by gibbs to the FreeBSD ahd driver: Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-08-29 04:09:59 +04:00
struct vpd_config vpd;
struct seeprom_config *sc;
u_int adapter_control;
int have_seeprom;
int error;
sc = ahd->seep_config;
have_seeprom = ahd_acquire_seeprom(ahd);
if (have_seeprom) {
u_int start_addr;
Apply the following change checked in 2003/05/04 00:20:07 by gibbs to the FreeBSD ahd driver: Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-08-29 04:09:59 +04:00
/*
* Fetch VPD for this function and parse it.
*/
#ifdef AHD_DEBUG
printf("%s: Reading VPD from SEEPROM...",
ahd_name(ahd));
#endif
/* Address is always in units of 16bit words */
start_addr = ((2 * sizeof(*sc))
+ (sizeof(vpd) * (ahd->channel - 'A'))) / 2;
2005-02-27 03:26:58 +03:00
Apply the following change checked in 2003/05/04 00:20:07 by gibbs to the FreeBSD ahd driver: Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-08-29 04:09:59 +04:00
error = ahd_read_seeprom(ahd, (uint16_t *)&vpd,
start_addr, sizeof(vpd)/2,
/*bytestream*/TRUE);
if (error == 0)
error = ahd_parse_vpddata(ahd, &vpd);
#ifdef AHD_DEBUG
printf("%s: VPD parsing %s\n",
ahd_name(ahd),
error == 0 ? "successful" : "failed");
#endif
#ifdef AHD_DEBUG
printf("%s: Reading SEEPROM...", ahd_name(ahd));
#endif
/* Address is always in units of 16bit words */
start_addr = (sizeof(*sc) / 2) * (ahd->channel - 'A');
error = ahd_read_seeprom(ahd, (uint16_t *)sc,
Apply the following change checked in 2003/05/04 00:20:07 by gibbs to the FreeBSD ahd driver: Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-08-29 04:09:59 +04:00
start_addr, sizeof(*sc)/2,
/*bytestream*/FALSE);
if (error != 0) {
#ifdef AHD_DEBUG
printf("Unable to read SEEPROM\n");
#endif
have_seeprom = 0;
} else {
have_seeprom = ahd_verify_cksum(sc);
#ifdef AHD_DEBUG
if (have_seeprom == 0)
printf ("checksum error\n");
else
printf ("done.\n");
#endif
}
ahd_release_seeprom(ahd);
}
if (!have_seeprom) {
u_int nvram_scb;
/*
* Pull scratch ram settings and treat them as
* if they are the contents of an seeprom if
* the 'ADPT', 'BIOS', or 'ASPI' signature is found
* in SCB 0xFF. We manually compose the data as 16bit
* values to avoid endian issues.
*/
ahd_set_scbptr(ahd, 0xFF);
nvram_scb = ahd_inb_scbram(ahd, SCB_BASE + NVRAM_SCB_OFFSET);
if (nvram_scb != 0xFF
&& ((ahd_inb_scbram(ahd, SCB_BASE + 0) == 'A'
&& ahd_inb_scbram(ahd, SCB_BASE + 1) == 'D'
&& ahd_inb_scbram(ahd, SCB_BASE + 2) == 'P'
&& ahd_inb_scbram(ahd, SCB_BASE + 3) == 'T')
|| (ahd_inb_scbram(ahd, SCB_BASE + 0) == 'B'
&& ahd_inb_scbram(ahd, SCB_BASE + 1) == 'I'
&& ahd_inb_scbram(ahd, SCB_BASE + 2) == 'O'
&& ahd_inb_scbram(ahd, SCB_BASE + 3) == 'S')
|| (ahd_inb_scbram(ahd, SCB_BASE + 0) == 'A'
&& ahd_inb_scbram(ahd, SCB_BASE + 1) == 'S'
&& ahd_inb_scbram(ahd, SCB_BASE + 2) == 'P'
&& ahd_inb_scbram(ahd, SCB_BASE + 3) == 'I'))) {
uint16_t *sc_data;
int i;
ahd_set_scbptr(ahd, nvram_scb);
sc_data = (uint16_t *)sc;
for (i = 0; i < 64; i += 2)
*sc_data++ = ahd_inw_scbram(ahd, SCB_BASE+i);
have_seeprom = ahd_verify_cksum(sc);
if (have_seeprom)
ahd->flags |= AHD_SCB_CONFIG_USED;
}
}
#ifdef AHD_DEBUG
if ((have_seeprom != 0) && (ahd_debug & AHD_DUMP_SEEPROM) != 0) {
Apply the following change checked in 2003/05/04 00:20:07 by gibbs to the FreeBSD ahd driver: Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-08-29 04:09:59 +04:00
uint16_t *sc_data;
int i;
printf("%s: Seeprom Contents:", ahd_name(ahd));
Apply the following change checked in 2003/05/04 00:20:07 by gibbs to the FreeBSD ahd driver: Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-08-29 04:09:59 +04:00
sc_data = (uint16_t *)sc;
for (i = 0; i < (sizeof(*sc)); i += 2)
Apply the following change checked in 2003/05/04 00:20:07 by gibbs to the FreeBSD ahd driver: Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-08-29 04:09:59 +04:00
printf("\n\t0x%.4x", sc_data[i]);
printf("\n");
}
#endif
if (!have_seeprom) {
2003-11-19 00:40:13 +03:00
aprint_error("%s: No SEEPROM available.\n", ahd_name(ahd));
ahd->flags |= AHD_USEDEFAULTS;
error = ahd_default_config(ahd);
adapter_control = CFAUTOTERM|CFSEAUTOTERM;
free(ahd->seep_config, M_DEVBUF);
ahd->seep_config = NULL;
} else {
error = ahd_parse_cfgdata(ahd, sc);
adapter_control = sc->adapter_control;
}
if (error != 0)
return (error);
ahd_configure_termination(ahd, adapter_control);
return (0);
}
static void
ahd_configure_termination(struct ahd_softc *ahd, u_int adapter_control)
{
int error;
u_int sxfrctl1;
uint8_t termctl;
uint32_t devconfig;
struct ahd_pci_busdata *bd = ahd->bus_data;
devconfig = pci_conf_read(bd->pc, bd->tag, DEVCONFIG);
devconfig &= ~STPWLEVEL;
if ((ahd->flags & AHD_STPWLEVEL_A) != 0)
devconfig |= STPWLEVEL;
#ifdef AHD_DEBUG
printf("%s: STPWLEVEL is %s\n",
ahd_name(ahd), (devconfig & STPWLEVEL) ? "on" : "off");
#endif
pci_conf_write(bd->pc, bd->tag, DEVCONFIG, devconfig);
2005-02-27 03:26:58 +03:00
/* Make sure current sensing is off. */
if ((ahd->flags & AHD_CURRENT_SENSING) != 0) {
(void)ahd_write_flexport(ahd, FLXADDR_ROMSTAT_CURSENSECTL, 0);
}
/*
* Read to sense. Write to set.
*/
error = ahd_read_flexport(ahd, FLXADDR_TERMCTL, &termctl);
if ((adapter_control & CFAUTOTERM) == 0) {
if (bootverbose)
printf("%s: Manual Primary Termination\n",
ahd_name(ahd));
termctl &= ~(FLX_TERMCTL_ENPRILOW|FLX_TERMCTL_ENPRIHIGH);
if ((adapter_control & CFSTERM) != 0)
termctl |= FLX_TERMCTL_ENPRILOW;
if ((adapter_control & CFWSTERM) != 0)
termctl |= FLX_TERMCTL_ENPRIHIGH;
} else if (error != 0) {
if (bootverbose)
printf("%s: Primary Auto-Term Sensing failed! "
"Using Defaults.\n", ahd_name(ahd));
termctl = FLX_TERMCTL_ENPRILOW|FLX_TERMCTL_ENPRIHIGH;
}
if ((adapter_control & CFSEAUTOTERM) == 0) {
if (bootverbose)
printf("%s: Manual Secondary Termination\n",
ahd_name(ahd));
termctl &= ~(FLX_TERMCTL_ENSECLOW|FLX_TERMCTL_ENSECHIGH);
if ((adapter_control & CFSELOWTERM) != 0)
termctl |= FLX_TERMCTL_ENSECLOW;
if ((adapter_control & CFSEHIGHTERM) != 0)
termctl |= FLX_TERMCTL_ENSECHIGH;
} else if (error != 0) {
if (bootverbose)
printf("%s: Secondary Auto-Term Sensing failed! "
"Using Defaults.\n", ahd_name(ahd));
termctl |= FLX_TERMCTL_ENSECLOW|FLX_TERMCTL_ENSECHIGH;
}
/*
* Now set the termination based on what we found.
*/
sxfrctl1 = ahd_inb(ahd, SXFRCTL1) & ~STPWEN;
if ((termctl & FLX_TERMCTL_ENPRILOW) != 0) {
ahd->flags |= AHD_TERM_ENB_A;
sxfrctl1 |= STPWEN;
}
/* Must set the latch once in order to be effective. */
ahd_outb(ahd, SXFRCTL1, sxfrctl1|STPWEN);
ahd_outb(ahd, SXFRCTL1, sxfrctl1);
error = ahd_write_flexport(ahd, FLXADDR_TERMCTL, termctl);
if (error != 0) {
2003-11-19 00:40:13 +03:00
aprint_error("%s: Unable to set termination settings!\n",
ahd_name(ahd));
} else {
if (bootverbose) {
printf("%s: Primary High byte termination %sabled\n",
ahd_name(ahd),
(termctl & FLX_TERMCTL_ENPRIHIGH) ? "En" : "Dis");
printf("%s: Primary Low byte termination %sabled\n",
ahd_name(ahd),
(termctl & FLX_TERMCTL_ENPRILOW) ? "En" : "Dis");
printf("%s: Secondary High byte termination %sabled\n",
ahd_name(ahd),
(termctl & FLX_TERMCTL_ENSECHIGH) ? "En" : "Dis");
printf("%s: Secondary Low byte termination %sabled\n",
ahd_name(ahd),
(termctl & FLX_TERMCTL_ENSECLOW) ? "En" : "Dis");
}
}
return;
}
#define DPE 0x80
#define SSE 0x40
#define RMA 0x20
#define RTA 0x10
#define STA 0x08
#define DPR 0x01
static const char *split_status_source[] =
{
"DFF0",
"DFF1",
"OVLY",
"CMC",
};
static const char *pci_status_source[] =
{
"DFF0",
"DFF1",
"SG",
"CMC",
"OVLY",
"NONE",
"MSI",
"TARG"
};
static const char *split_status_strings[] =
{
"%s: Received split response in %s.\n",
"%s: Received split completion error message in %s\n",
"%s: Receive overrun in %s\n",
"%s: Count not complete in %s\n",
"%s: Split completion data bucket in %s\n",
"%s: Split completion address error in %s\n",
"%s: Split completion byte count error in %s\n",
"%s: Signaled Target-abort to early terminate a split in %s\n"
};
static const char *pci_status_strings[] =
{
"%s: Data Parity Error has been reported via PERR# in %s\n",
"%s: Target initial wait state error in %s\n",
"%s: Split completion read data parity error in %s\n",
"%s: Split completion address attribute parity error in %s\n",
"%s: Received a Target Abort in %s\n",
"%s: Received a Master Abort in %s\n",
"%s: Signal System Error Detected in %s\n",
"%s: Address or Write Phase Parity Error Detected in %s.\n"
};
2005-06-28 04:28:41 +04:00
static int
ahd_pci_intr(struct ahd_softc *ahd)
{
uint8_t pci_status[8];
ahd_mode_state saved_modes;
u_int pci_status1;
u_int intstat;
u_int i;
u_int reg;
struct ahd_pci_busdata *bd = ahd->bus_data;
2005-02-27 03:26:58 +03:00
intstat = ahd_inb(ahd, INTSTAT);
if ((intstat & SPLTINT) != 0)
ahd_pci_split_intr(ahd, intstat);
if ((intstat & PCIINT) == 0)
return 0;
printf("%s: PCI error Interrupt\n", ahd_name(ahd));
saved_modes = ahd_save_modes(ahd);
ahd_dump_card_state(ahd);
ahd_set_modes(ahd, AHD_MODE_CFG, AHD_MODE_CFG);
for (i = 0, reg = DF0PCISTAT; i < 8; i++, reg++) {
if (i == 5)
continue;
pci_status[i] = ahd_inb(ahd, reg);
/* Clear latched errors. So our interrupt deasserts. */
ahd_outb(ahd, reg, pci_status[i]);
}
for (i = 0; i < 8; i++) {
u_int bit;
2005-02-27 03:26:58 +03:00
if (i == 5)
continue;
for (bit = 0; bit < 8; bit++) {
if ((pci_status[i] & (0x1 << bit)) != 0) {
static const char *s;
s = pci_status_strings[bit];
if (i == 7/*TARG*/ && bit == 3)
s = "%s: Signaled Target Abort\n";
printf(s, ahd_name(ahd), pci_status_source[i]);
}
2005-02-27 03:26:58 +03:00
}
}
pci_status1 = pci_conf_read(bd->pc, bd->tag, PCI_COMMAND_STATUS_REG);
pci_conf_write(bd->pc, bd->tag, PCI_COMMAND_STATUS_REG , pci_status1);
ahd_restore_modes(ahd, saved_modes);
ahd_outb(ahd, CLRINT, CLRPCIINT);
ahd_unpause(ahd);
return 1;
}
static void
ahd_pci_split_intr(struct ahd_softc *ahd, u_int intstat)
{
uint8_t split_status[4];
uint8_t split_status1[4];
uint8_t sg_split_status[2];
uint8_t sg_split_status1[2];
ahd_mode_state saved_modes;
u_int i;
pcireg_t pcix_status;
struct ahd_pci_busdata *bd = ahd->bus_data;
/*
* Check for splits in all modes. Modes 0 and 1
* additionally have SG engine splits to look at.
*/
pcix_status = pci_conf_read(bd->pc, bd->tag,
bd->pcix_off + PCI_PCIX_STATUS);
printf("%s: PCI Split Interrupt - PCI-X status = 0x%x\n",
ahd_name(ahd), pcix_status);
saved_modes = ahd_save_modes(ahd);
for (i = 0; i < 4; i++) {
ahd_set_modes(ahd, i, i);
split_status[i] = ahd_inb(ahd, DCHSPLTSTAT0);
split_status1[i] = ahd_inb(ahd, DCHSPLTSTAT1);
/* Clear latched errors. So our interrupt deasserts. */
ahd_outb(ahd, DCHSPLTSTAT0, split_status[i]);
ahd_outb(ahd, DCHSPLTSTAT1, split_status1[i]);
Apply the following change checked in 2003/05/04 00:20:07 by gibbs to the FreeBSD ahd driver: Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-08-29 04:09:59 +04:00
if (i > 1)
continue;
sg_split_status[i] = ahd_inb(ahd, SGSPLTSTAT0);
sg_split_status1[i] = ahd_inb(ahd, SGSPLTSTAT1);
/* Clear latched errors. So our interrupt deasserts. */
ahd_outb(ahd, SGSPLTSTAT0, sg_split_status[i]);
ahd_outb(ahd, SGSPLTSTAT1, sg_split_status1[i]);
}
for (i = 0; i < 4; i++) {
u_int bit;
for (bit = 0; bit < 8; bit++) {
if ((split_status[i] & (0x1 << bit)) != 0) {
static const char *s;
s = split_status_strings[bit];
printf(s, ahd_name(ahd),
split_status_source[i]);
}
Apply the following change checked in 2003/05/04 00:20:07 by gibbs to the FreeBSD ahd driver: Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-08-29 04:09:59 +04:00
if (i > 0)
continue;
if ((sg_split_status[i] & (0x1 << bit)) != 0) {
static const char *s;
s = split_status_strings[bit];
printf(s, ahd_name(ahd), "SG");
}
}
}
/*
* Clear PCI-X status bits.
*/
pci_conf_write(bd->pc, bd->tag, bd->pcix_off + PCI_PCIX_STATUS,
pcix_status);
ahd_outb(ahd, CLRINT, CLRSPLTINT);
ahd_restore_modes(ahd, saved_modes);
}
static int
ahd_aic7901_setup(struct ahd_softc *ahd, struct pci_attach_args *pa)
{
ahd->chip = AHD_AIC7901;
ahd->features = AHD_AIC7901_FE;
return (ahd_aic790X_setup(ahd, pa));
}
static int
ahd_aic7901A_setup(struct ahd_softc *ahd, struct pci_attach_args *pa)
{
ahd->chip = AHD_AIC7901A;
ahd->features = AHD_AIC7901A_FE;
return (ahd_aic790X_setup(ahd, pa));
}
static int
ahd_aic7902_setup(struct ahd_softc *ahd, struct pci_attach_args *pa)
{
ahd->chip = AHD_AIC7902;
ahd->features = AHD_AIC7902_FE;
return (ahd_aic790X_setup(ahd, pa));
}
static int
ahd_aic790X_setup(struct ahd_softc *ahd, struct pci_attach_args *pa)
{
u_int rev;
rev = PCI_REVISION(pa->pa_class);
#ifdef AHD_DEBUG
printf("\n%s: aic7902 chip revision 0x%x\n", ahd_name(ahd), rev);
#endif
if (rev < ID_AIC7902_PCI_REV_A4) {
2003-11-19 00:40:13 +03:00
aprint_error("%s: Unable to attach to unsupported chip revision %d\n",
ahd_name(ahd), rev);
pci_conf_write(pa->pa_pc, pa->pa_tag, PCI_COMMAND_STATUS_REG, 0);
return (ENXIO);
}
ahd->channel = (pa->pa_function == 1) ? 'B' : 'A';
if (rev < ID_AIC7902_PCI_REV_B0) {
/*
* Enable A series workarounds.
*/
ahd->bugs |= AHD_SENT_SCB_UPDATE_BUG|AHD_ABORT_LQI_BUG
| AHD_PKT_BITBUCKET_BUG|AHD_LONG_SETIMO_BUG
| AHD_NLQICRC_DELAYED_BUG|AHD_SCSIRST_BUG
| AHD_LQO_ATNO_BUG|AHD_AUTOFLUSH_BUG
| AHD_CLRLQO_AUTOCLR_BUG|AHD_PCIX_MMAPIO_BUG
| AHD_PCIX_CHIPRST_BUG|AHD_PCIX_SCBRAM_RD_BUG
| AHD_PKTIZED_STATUS_BUG|AHD_PKT_LUN_BUG
| AHD_MDFF_WSCBPTR_BUG|AHD_REG_SLOW_SETTLE_BUG
| AHD_SET_MODE_BUG|AHD_BUSFREEREV_BUG
Apply the following change checked in 2003/05/04 00:20:07 by gibbs to the FreeBSD ahd driver: Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-08-29 04:09:59 +04:00
| AHD_NONPACKFIFO_BUG|AHD_PACED_NEGTABLE_BUG
| AHD_FAINT_LED_BUG;
/*
* IO Cell parameter setup.
*/
AHD_SET_PRECOMP(ahd, AHD_PRECOMP_CUTBACK_29);
if ((ahd->flags & AHD_HP_BOARD) == 0)
AHD_SET_SLEWRATE(ahd, AHD_SLEWRATE_DEF_REVA);
} else {
u_int devconfig1;
ahd->features |= AHD_RTI|AHD_NEW_IOCELL_OPTS
| AHD_NEW_DFCNTRL_OPTS;
ahd->bugs |= AHD_LQOOVERRUN_BUG|AHD_EARLY_REQ_BUG;
/*
* Some issues have been resolved in the 7901B.
*/
if ((ahd->features & AHD_MULTI_FUNC) != 0)
ahd->bugs |= AHD_INTCOLLISION_BUG|AHD_ABORT_LQI_BUG;
/*
* IO Cell parameter setup.
*/
AHD_SET_PRECOMP(ahd, AHD_PRECOMP_CUTBACK_29);
AHD_SET_SLEWRATE(ahd, AHD_SLEWRATE_DEF_REVB);
AHD_SET_AMPLITUDE(ahd, AHD_AMPLITUDE_DEF);
/*
* Set the PREQDIS bit for H2B which disables some workaround
* that doesn't work on regular PCI busses.
* XXX - Find out exactly what this does from the hardware
* folks!
*/
devconfig1 = pci_conf_read(pa->pa_pc, pa->pa_tag, DEVCONFIG1);
pci_conf_write(pa->pa_pc, pa->pa_tag, DEVCONFIG1, devconfig1|PREQDIS);
devconfig1 = pci_conf_read(pa->pa_pc, pa->pa_tag, DEVCONFIG1);
}
return (0);
}