#define SEC_SC_HALTED 0x4 /* per-context only */
#define SEC_SPC_FREEZE 0x8 /* per-HFI only */
-#define VL15CTXT 1
#define MIN_KERNEL_KCTXTS 2
+#define FIRST_KERNEL_KCTXT 1
#define NUM_MAP_REGS 32
/* Bit offset into the GUID which carries HFI id information */
[C_RX_TID_FLGMS] = RXE32_DEV_CNTR_ELEM(RxTidFLGMs,
RCV_TID_FLOW_GEN_MISMATCH_CNT,
CNTR_NORMAL),
-[C_RX_CTX_RHQS] = RXE32_DEV_CNTR_ELEM(RxCtxRHQS, RCV_CONTEXT_RHQ_STALL,
- CNTR_NORMAL),
[C_RX_CTX_EGRS] = RXE32_DEV_CNTR_ELEM(RxCtxEgrS, RCV_CONTEXT_EGR_STALL,
CNTR_NORMAL),
[C_RCV_TID_FLSMS] = RXE32_DEV_CNTR_ELEM(RxTidFLSMs,
/* ======================================================================== */
-/* return true if this is chip revision revision a0 */
-int is_a0(struct hfi1_devdata *dd)
-{
- return ((dd->revision >> CCE_REVISION_CHIP_REV_MINOR_SHIFT)
- & CCE_REVISION_CHIP_REV_MINOR_MASK) == 0;
-}
-
/* return true if this is chip revision revision a */
int is_ax(struct hfi1_devdata *dd)
{
u8 chip_rev_minor =
dd->revision >> CCE_REVISION_CHIP_REV_MINOR_SHIFT
& CCE_REVISION_CHIP_REV_MINOR_MASK;
- return !!(chip_rev_minor & 0x10);
+ return (chip_rev_minor & 0xF0) == 0x10;
}
/*
dd_dev_info(dd, "CCE Error: %s\n",
cce_err_status_string(buf, sizeof(buf), reg));
- if ((reg & CCE_ERR_STATUS_CCE_CLI2_ASYNC_FIFO_PARITY_ERR_SMASK)
- && is_a0(dd)
- && (dd->icode != ICODE_FUNCTIONAL_SIMULATOR)) {
+ if ((reg & CCE_ERR_STATUS_CCE_CLI2_ASYNC_FIFO_PARITY_ERR_SMASK) &&
+ is_ax(dd) && (dd->icode != ICODE_FUNCTIONAL_SIMULATOR)) {
/* this error requires a manual drop into SPC freeze mode */
/* then a fix up */
start_freeze_handling(dd->pport, FREEZE_SELF);
* Freeze mode recovery is disabled for the errors
* in RXE_FREEZE_ABORT_MASK
*/
- if (is_a0(dd) && (reg & RXE_FREEZE_ABORT_MASK))
+ if (is_ax(dd) && (reg & RXE_FREEZE_ABORT_MASK))
flags = FREEZE_ABORT;
start_freeze_handling(dd->pport, flags);
if (reg & ALL_TXE_EGRESS_FREEZE_ERR)
start_freeze_handling(dd->pport, 0);
- if (is_a0(dd) && (reg &
+ if (is_ax(dd) && (reg &
SEND_EGRESS_ERR_STATUS_TX_CREDIT_RETURN_VL_ERR_SMASK)
&& (dd->icode != ICODE_FUNCTIONAL_SIMULATOR))
start_freeze_handling(dd->pport, 0);
/* else this is _p */
version = emulator_rev(dd);
- if (!is_a0(dd))
+ if (!is_ax(dd))
version = 0x2d; /* all B0 use 0x2d or higher settings */
if (version <= 0x12) {
write_csr(dd, CCE_CTRL, CCE_CTRL_SPC_UNFREEZE_SMASK);
wait_for_freeze_status(dd, 0);
- if (is_a0(dd)) {
+ if (is_ax(dd)) {
write_csr(dd, CCE_CTRL, CCE_CTRL_SPC_FREEZE_SMASK);
wait_for_freeze_status(dd, 1);
write_csr(dd, CCE_CTRL, CCE_CTRL_SPC_UNFREEZE_SMASK);
* REPLAY_BUF_MBE_SMASK
* FLIT_INPUT_BUF_MBE_SMASK
*/
- if (is_a0(dd)) { /* fixed in B0 */
+ if (is_ax(dd)) { /* fixed in B0 */
reg = read_csr(dd, DC_LCB_CFG_LINK_KILL_EN);
reg |= DC_LCB_CFG_LINK_KILL_EN_REPLAY_BUF_MBE_SMASK
| DC_LCB_CFG_LINK_KILL_EN_FLIT_INPUT_BUF_MBE_SMASK;
*/
void apply_link_downgrade_policy(struct hfi1_pportdata *ppd, int refresh_widths)
{
- int skip = 1;
int do_bounce = 0;
- u16 lwde = ppd->link_width_downgrade_enabled;
+ int tries;
+ u16 lwde;
u16 tx, rx;
+ /* use the hls lock to avoid a race with actual link up */
+ tries = 0;
+retry:
mutex_lock(&ppd->hls_lock);
/* only apply if the link is up */
- if (ppd->host_link_state & HLS_UP)
- skip = 0;
- mutex_unlock(&ppd->hls_lock);
- if (skip)
- return;
+ if (!(ppd->host_link_state & HLS_UP)) {
+ /* still going up..wait and retry */
+ if (ppd->host_link_state & HLS_GOING_UP) {
+ if (++tries < 1000) {
+ mutex_unlock(&ppd->hls_lock);
+ usleep_range(100, 120); /* arbitrary */
+ goto retry;
+ }
+ dd_dev_err(ppd->dd,
+ "%s: giving up waiting for link state change\n",
+ __func__);
+ }
+ goto done;
+ }
+
+ lwde = ppd->link_width_downgrade_enabled;
if (refresh_widths) {
get_link_widths(ppd->dd, &tx, &rx);
do_bounce = 1;
}
+done:
+ mutex_unlock(&ppd->hls_lock);
+
if (do_bounce) {
set_link_down_reason(ppd, OPA_LINKDOWN_REASON_WIDTH_POLICY, 0,
OPA_LINKDOWN_REASON_WIDTH_POLICY);
*/
static int write_lcb_via_8051(struct hfi1_devdata *dd, u32 addr, u64 data)
{
+ u32 regno;
+ int ret;
- if (acquire_lcb_access(dd, 0) == 0) {
- write_csr(dd, addr, data);
- release_lcb_access(dd, 0);
- return 0;
+ if (dd->icode == ICODE_FUNCTIONAL_SIMULATOR ||
+ (dd->dc8051_ver < dc8051_ver(0, 20))) {
+ if (acquire_lcb_access(dd, 0) == 0) {
+ write_csr(dd, addr, data);
+ release_lcb_access(dd, 0);
+ return 0;
+ }
+ return -EBUSY;
}
- return -EBUSY;
+
+ /* register is an index of LCB registers: (offset - base) / 8 */
+ regno = (addr - DC_LCB_CFG_RUN) >> 3;
+ ret = do_8051_command(dd, HCMD_WRITE_LCB_CSR, regno, &data);
+ if (ret != HCMD_SUCCESS)
+ return -EBUSY;
+ return 0;
}
/*
* waiting for a command.
*/
+ /*
+ * When writing a LCB CSR, out_data contains the full value to
+ * to be written, while in_data contains the relative LCB
+ * address in 7:0. Do the work here, rather than the caller,
+ * of distrubting the write data to where it needs to go:
+ *
+ * Write data
+ * 39:00 -> in_data[47:8]
+ * 47:40 -> DC8051_CFG_EXT_DEV_0.RETURN_CODE
+ * 63:48 -> DC8051_CFG_EXT_DEV_0.RSP_DATA
+ */
+ if (type == HCMD_WRITE_LCB_CSR) {
+ in_data |= ((*out_data) & 0xffffffffffull) << 8;
+ reg = ((((*out_data) >> 40) & 0xff) <<
+ DC_DC8051_CFG_EXT_DEV_0_RETURN_CODE_SHIFT)
+ | ((((*out_data) >> 48) & 0xffff) <<
+ DC_DC8051_CFG_EXT_DEV_0_RSP_DATA_SHIFT);
+ write_csr(dd, DC_DC8051_CFG_EXT_DEV_0, reg);
+ }
+
/*
* Do two writes: the first to stabilize the type and req_data, the
* second to activate.
*/
use_all_mask = 0;
if ((be16_to_cpu(new_bc->overall_shared_limit) <
- be16_to_cpu(cur_bc.overall_shared_limit))
- || (is_a0(dd) && any_shared_limit_changing)) {
+ be16_to_cpu(cur_bc.overall_shared_limit)) ||
+ (is_ax(dd) && any_shared_limit_changing)) {
set_global_shared(dd, 0);
cur_bc.overall_shared_limit = 0;
use_all_mask = 1;
*/
static int disable_data_vls(struct hfi1_devdata *dd)
{
- if (is_a0(dd))
+ if (is_ax(dd))
return 1;
pio_send_control(dd, PSC_DATA_VL_DISABLE);
*/
int open_fill_data_vls(struct hfi1_devdata *dd)
{
- if (is_a0(dd))
+ if (is_ax(dd))
return 1;
pio_send_control(dd, PSC_DATA_VL_ENABLE);
& RCV_TID_CTRL_TID_BASE_INDEX_MASK)
<< RCV_TID_CTRL_TID_BASE_INDEX_SHIFT);
write_kctxt_csr(dd, ctxt, RCV_TID_CTRL, reg);
- if (ctxt == VL15CTXT)
- write_csr(dd, RCV_VL15, VL15CTXT);
+ if (ctxt == HFI1_CTRL_CTXT)
+ write_csr(dd, RCV_VL15, HFI1_CTRL_CTXT);
}
if (op & HFI1_RCVCTRL_CTXT_DIS) {
write_csr(dd, RCV_VL15, 0);
+ /*
+ * When receive context is being disabled turn on tail
+ * update with a dummy tail address and then disable
+ * receive context.
+ */
+ if (dd->rcvhdrtail_dummy_physaddr) {
+ write_kctxt_csr(dd, ctxt, RCV_HDR_TAIL_ADDR,
+ dd->rcvhdrtail_dummy_physaddr);
+ rcvctrl |= RCV_CTXT_CTRL_TAIL_UPD_SMASK;
+ }
+
rcvctrl &= ~RCV_CTXT_CTRL_ENABLE_SMASK;
}
if (op & HFI1_RCVCTRL_INTRAVAIL_ENB)
if (op & (HFI1_RCVCTRL_TAILUPD_DIS | HFI1_RCVCTRL_CTXT_DIS))
/*
* If the context has been disabled and the Tail Update has
- * been cleared, clear the RCV_HDR_TAIL_ADDR CSR so
- * it doesn't contain an address that is invalid.
+ * been cleared, set the RCV_HDR_TAIL_ADDR CSR to dummy address
+ * so it doesn't contain an address that is invalid.
*/
- write_kctxt_csr(dd, ctxt, RCV_HDR_TAIL_ADDR, 0);
+ write_kctxt_csr(dd, ctxt, RCV_HDR_TAIL_ADDR,
+ dd->rcvhdrtail_dummy_physaddr);
}
u32 hfi1_read_cntrs(struct hfi1_devdata *dd, loff_t pos, char **namep,
/* turn off interrupts */
if (dd->num_msix_entries) {
/* MSI-X */
- hfi1_nomsix(dd);
+ pci_disable_msix(dd->pcidev);
} else {
/* INTx */
disable_intx(dd->pcidev);
msix_intr);
}
-static void remap_receive_available_interrupt(struct hfi1_devdata *dd,
- int rx, int msix_intr)
-{
- remap_intr(dd, IS_RCVAVAIL_START + rx, msix_intr);
-}
-
static int request_intx_irq(struct hfi1_devdata *dd)
{
int ret;
- snprintf(dd->intx_name, sizeof(dd->intx_name), DRIVER_NAME"_%d",
- dd->unit);
+ snprintf(dd->intx_name, sizeof(dd->intx_name), DRIVER_NAME "_%d",
+ dd->unit);
ret = request_irq(dd->pcidev->irq, general_interrupt,
IRQF_SHARED, dd->intx_name, dd);
if (ret)
int first_general, last_general;
int first_sdma, last_sdma;
int first_rx, last_rx;
- int first_cpu, restart_cpu, curr_cpu;
+ int first_cpu, curr_cpu;
int rcv_cpu, sdma_cpu;
int i, ret = 0, possible;
int ht;
topology_sibling_cpumask(cpumask_first(local_mask)));
for (i = possible/ht; i < possible; i++)
cpumask_clear_cpu(i, def);
- /* reset possible */
- possible = cpumask_weight(def);
/* def now has full cores on chosen node*/
first_cpu = cpumask_first(def);
if (nr_cpu_ids >= first_cpu)
first_cpu++;
- restart_cpu = first_cpu;
- curr_cpu = restart_cpu;
+ curr_cpu = first_cpu;
- for (i = first_cpu; i < dd->n_krcv_queues + first_cpu; i++) {
+ /* One context is reserved as control context */
+ for (i = first_cpu; i < dd->n_krcv_queues + first_cpu - 1; i++) {
cpumask_clear_cpu(curr_cpu, def);
cpumask_set_cpu(curr_cpu, rcv);
- if (curr_cpu >= possible)
- curr_cpu = restart_cpu;
- else
- curr_cpu++;
+ curr_cpu = cpumask_next(curr_cpu, def);
+ if (curr_cpu >= nr_cpu_ids)
+ break;
}
/* def mask has non-rcv, rcv has recv mask */
rcv_cpu = cpumask_first(rcv);
handler = general_interrupt;
arg = dd;
snprintf(me->name, sizeof(me->name),
- DRIVER_NAME"_%d", dd->unit);
+ DRIVER_NAME "_%d", dd->unit);
err_info = "general";
} else if (first_sdma <= i && i < last_sdma) {
idx = i - first_sdma;
handler = sdma_interrupt;
arg = sde;
snprintf(me->name, sizeof(me->name),
- DRIVER_NAME"_%d sdma%d", dd->unit, idx);
+ DRIVER_NAME "_%d sdma%d", dd->unit, idx);
err_info = "sdma";
remap_sdma_interrupts(dd, idx, i);
} else if (first_rx <= i && i < last_rx) {
thread = receive_context_thread;
arg = rcd;
snprintf(me->name, sizeof(me->name),
- DRIVER_NAME"_%d kctxt%d", dd->unit, idx);
+ DRIVER_NAME "_%d kctxt%d", dd->unit, idx);
err_info = "receive context";
- remap_receive_available_interrupt(dd, idx, i);
+ remap_intr(dd, IS_RCVAVAIL_START + idx, i);
} else {
/* not in our expected range - complain, then
ignore it */
if (handler == sdma_interrupt) {
dd_dev_info(dd, "sdma engine %d cpu %d\n",
sde->this_idx, sdma_cpu);
+ sde->cpu = sdma_cpu;
cpumask_set_cpu(sdma_cpu, dd->msix_entries[i].mask);
sdma_cpu = cpumask_next(sdma_cpu, def);
if (sdma_cpu >= nr_cpu_ids)
sdma_cpu = cpumask_first(def);
} else if (handler == receive_context_interrupt) {
- dd_dev_info(dd, "rcv ctxt %d cpu %d\n",
- rcd->ctxt, rcv_cpu);
- cpumask_set_cpu(rcv_cpu, dd->msix_entries[i].mask);
- rcv_cpu = cpumask_next(rcv_cpu, rcv);
- if (rcv_cpu >= nr_cpu_ids)
- rcv_cpu = cpumask_first(rcv);
+ dd_dev_info(dd, "rcv ctxt %d cpu %d\n", rcd->ctxt,
+ (rcd->ctxt == HFI1_CTRL_CTXT) ?
+ cpumask_first(def) : rcv_cpu);
+ if (rcd->ctxt == HFI1_CTRL_CTXT) {
+ /* map to first default */
+ cpumask_set_cpu(cpumask_first(def),
+ dd->msix_entries[i].mask);
+ } else {
+ cpumask_set_cpu(rcv_cpu,
+ dd->msix_entries[i].mask);
+ rcv_cpu = cpumask_next(rcv_cpu, rcv);
+ if (rcv_cpu >= nr_cpu_ids)
+ rcv_cpu = cpumask_first(rcv);
+ }
} else {
/* otherwise first def */
dd_dev_info(dd, "%s cpu %d\n",
/*
* Kernel contexts: (to be fixed later):
* - min or 2 or 1 context/numa
- * - Context 0 - default/errors
- * - Context 1 - VL15
+ * - Context 0 - control context (VL15/multicast/error)
+ * - Context 1 - default context
*/
if (n_krcvqs)
- num_kernel_contexts = n_krcvqs + MIN_KERNEL_KCTXTS;
+ /*
+ * Don't count context 0 in n_krcvqs since
+ * is isn't used for normal verbs traffic.
+ *
+ * krcvqs will reflect number of kernel
+ * receive contexts above 0.
+ */
+ num_kernel_contexts = n_krcvqs + MIN_KERNEL_KCTXTS - 1;
else
num_kernel_contexts = num_online_nodes();
num_kernel_contexts =
/* We might want to retain this state across FLR if we ever use it */
write_csr(dd, ASIC_CFG_DRV_STR, 0);
- write_csr(dd, ASIC_CFG_THERM_POLL_EN, 0);
+ /* ASIC_CFG_THERM_POLL_EN leave alone */
/* ASIC_STS_THERM read-only */
/* ASIC_CFG_RESET leave alone */
/* restore command and BARs */
restore_pci_variables(dd);
- if (is_a0(dd)) {
+ if (is_ax(dd)) {
dd_dev_info(dd, "Resetting CSRs with FLR\n");
hfi1_pcie_flr(dd);
restore_pci_variables(dd);
write_csr(dd, CCE_DC_CTRL, 0);
/* Set the LED off */
- if (is_a0(dd))
+ if (is_ax(dd))
setextled(dd, 0);
/*
* Clear the QSFP reset.
- * A0 leaves the out lines floating on power on, then on an FLR
- * enforces a 0 on all out pins. The driver does not touch
+ * An FLR enforces a 0 on all out pins. The driver does not touch
* ASIC_QSFPn_OUT otherwise. This leaves RESET_N low and
- * anything plugged constantly in reset, if it pays attention
+ * anything plugged constantly in reset, if it pays attention
* to RESET_N.
- * A prime example of this is SiPh. For now, set all pins high.
+ * Prime examples of this are optical cables. Set all pins high.
* I2CCLK and I2CDAT will change per direction, and INT_N and
* MODPRS_N are input only and their value is ignored.
*/
- if (is_a0(dd)) {
- write_csr(dd, ASIC_QSFP1_OUT, 0x1f);
- write_csr(dd, ASIC_QSFP2_OUT, 0x1f);
- }
+ write_csr(dd, ASIC_QSFP1_OUT, 0x1f);
+ write_csr(dd, ASIC_QSFP2_OUT, 0x1f);
}
static void init_early_variables(struct hfi1_devdata *dd)
/* assign link credit variables */
dd->vau = CM_VAU;
dd->link_credits = CM_GLOBAL_CREDITS;
- if (is_a0(dd))
+ if (is_ax(dd))
dd->link_credits--;
dd->vcu = cu_to_vcu(hfi1_cu);
/* enough room for 8 MAD packets plus header - 17K */
u64 ctxt = first_ctxt;
for (i = 0; i < 256;) {
- if (ctxt == VL15CTXT) {
- ctxt++;
- if (ctxt > last_ctxt)
- ctxt = first_ctxt;
- continue;
- }
reg |= ctxt << (8 * (i % 8));
i++;
ctxt++;
unsigned qpns_per_vl, ctxt, i, qpn, n = 1, m;
u64 *rsmmap;
u64 reg;
- u8 rxcontext = is_a0(dd) ? 0 : 0xff; /* 0 is default if a0 ver. */
+ u8 rxcontext = is_ax(dd) ? 0 : 0xff; /* 0 is default if a0 ver. */
/* validate */
if (dd->n_krcv_queues <= MIN_KERNEL_KCTXTS ||
/* Enable RSM */
add_rcvctrl(dd, RCV_CTRL_RCV_RSM_ENABLE_SMASK);
kfree(rsmmap);
- /* map everything else (non-VL15) to context 0 */
- init_qpmap_table(
- dd,
- 0,
- 0);
+ /* map everything else to first context */
+ init_qpmap_table(dd, FIRST_KERNEL_KCTXT, MIN_KERNEL_KCTXTS - 1);
dd->qos_shift = n + 1;
return;
bail:
dd->qos_shift = 1;
- init_qpmap_table(
- dd,
- dd->n_krcv_queues > MIN_KERNEL_KCTXTS ? MIN_KERNEL_KCTXTS : 0,
- dd->n_krcv_queues - 1);
+ init_qpmap_table(dd, FIRST_KERNEL_KCTXT, dd->n_krcv_queues - 1);
}
static void init_rxe(struct hfi1_devdata *dd)
* Enable send-side J_KEY integrity check, unless this is A0 h/w
* (due to A0 erratum).
*/
- if (!is_a0(dd)) {
+ if (!is_ax(dd)) {
reg = read_kctxt_csr(dd, sctxt, SEND_CTXT_CHECK_ENABLE);
reg |= SEND_CTXT_CHECK_ENABLE_CHECK_JOB_KEY_SMASK;
write_kctxt_csr(dd, sctxt, SEND_CTXT_CHECK_ENABLE, reg);
* This check would not have been enabled for A0 h/w, see
* set_ctxt_jkey().
*/
- if (!is_a0(dd)) {
+ if (!is_ax(dd)) {
reg = read_kctxt_csr(dd, sctxt, SEND_CTXT_CHECK_ENABLE);
reg &= ~SEND_CTXT_CHECK_ENABLE_CHECK_JOB_KEY_SMASK;
write_kctxt_csr(dd, sctxt, SEND_CTXT_CHECK_ENABLE, reg);
acquire_hw_mutex(dd);
dd_dev_info(dd, "Initializing thermal sensor\n");
-
+ /* Disable polling of thermal readings */
+ write_csr(dd, ASIC_CFG_THERM_POLL_EN, 0x0);
+ msleep(100);
/* Thermal Sensor Initialization */
/* Step 1: Reset the Thermal SBus Receiver */
ret = sbus_request_slow(dd, SBUS_THERMAL, 0x0,