Files
android_kernel_samsung_sm86…/hif/src/hif_main.c
SACHIN AHUJA 147467b9f5 qcacmn: Add the SOCID for WCN6750_V2 chip
SOC ID is missing for WCN6750_V2 and as a result correct
string for chip is not sent to CNE. This leads to the
issue in Wifi calls.

Add the SOC ID for WCN6750_V2.

CRs-Fixed: 3552566
Change-Id: I80d19742f6ffa1c5b03f1f2576c6e4c0c4a1c1db
2023-07-13 19:23:57 -07:00

3348 lines
82 KiB
C

/*
* Copyright (c) 2015-2021 The Linux Foundation. All rights reserved.
* Copyright (c) 2021-2023 Qualcomm Innovation Center, Inc. All rights reserved.
*
* Permission to use, copy, modify, and/or distribute this software for
* any purpose with or without fee is hereby granted, provided that the
* above copyright notice and this permission notice appear in all
* copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL
* WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE
* AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
* DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
* PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
* TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
* PERFORMANCE OF THIS SOFTWARE.
*/
#include "targcfg.h"
#include "qdf_lock.h"
#include "qdf_status.h"
#include "qdf_status.h"
#include <qdf_atomic.h> /* qdf_atomic_read */
#include <targaddrs.h>
#include "hif_io32.h"
#include <hif.h>
#include <target_type.h>
#include "regtable.h"
#define ATH_MODULE_NAME hif
#include <a_debug.h>
#include "hif_main.h"
#include "hif_hw_version.h"
#if (defined(HIF_PCI) || defined(HIF_SNOC) || defined(HIF_AHB) || \
defined(HIF_IPCI))
#include "ce_tasklet.h"
#include "ce_api.h"
#endif
#include "qdf_trace.h"
#include "qdf_status.h"
#include "hif_debug.h"
#include "mp_dev.h"
#if defined(QCA_WIFI_QCA8074) || defined(QCA_WIFI_QCA6018) || \
defined(QCA_WIFI_QCA5018) || defined(QCA_WIFI_QCA9574) || \
defined(QCA_WIFI_QCA5332)
#include "hal_api.h"
#endif
#include "hif_napi.h"
#include "hif_unit_test_suspend_i.h"
#include "qdf_module.h"
#ifdef HIF_CE_LOG_INFO
#include <qdf_notifier.h>
#include <qdf_hang_event_notifier.h>
#endif
#include <linux/cpumask.h>
#include <pld_common.h>
#include "ce_internal.h"
#include <qdf_tracepoint.h>
void hif_dump(struct hif_opaque_softc *hif_ctx, uint8_t cmd_id, bool start)
{
hif_trigger_dump(hif_ctx, cmd_id, start);
}
/**
* hif_get_target_id(): hif_get_target_id
* @scn: scn
*
* Return the virtual memory base address to the caller
*
* @scn: hif_softc
*
* Return: A_target_id_t
*/
A_target_id_t hif_get_target_id(struct hif_softc *scn)
{
return scn->mem;
}
/**
* hif_get_targetdef(): hif_get_targetdef
* @hif_ctx: hif context
*
* Return: void *
*/
void *hif_get_targetdef(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
return scn->targetdef;
}
#ifdef FORCE_WAKE
#ifndef QCA_WIFI_WCN6450
void hif_srng_init_phase(struct hif_opaque_softc *hif_ctx,
bool init_phase)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (ce_srng_based(scn))
hal_set_init_phase(scn->hal_soc, init_phase);
}
#else
void hif_srng_init_phase(struct hif_opaque_softc *hif_ctx,
bool init_phase)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
hal_set_init_phase(scn->hal_soc, init_phase);
}
#endif
#endif /* FORCE_WAKE */
#ifdef HIF_IPCI
void hif_shutdown_notifier_cb(void *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
scn->recovery = true;
}
#endif
/**
* hif_vote_link_down(): unvote for link up
* @hif_ctx: hif context
*
* Call hif_vote_link_down to release a previous request made using
* hif_vote_link_up. A hif_vote_link_down call should only be made
* after a corresponding hif_vote_link_up, otherwise you could be
* negating a vote from another source. When no votes are present
* hif will not guarantee the linkstate after hif_bus_suspend.
*
* SYNCHRONIZE WITH hif_vote_link_up by only calling in MC thread
* and initialization deinitialization sequencences.
*
* Return: n/a
*/
void hif_vote_link_down(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
QDF_BUG(scn);
if (scn->linkstate_vote == 0)
QDF_DEBUG_PANIC("linkstate_vote(%d) has already been 0",
scn->linkstate_vote);
scn->linkstate_vote--;
hif_info("Down_linkstate_vote %d", scn->linkstate_vote);
if (scn->linkstate_vote == 0)
hif_bus_prevent_linkdown(scn, false);
}
/**
* hif_vote_link_up(): vote to prevent bus from suspending
* @hif_ctx: hif context
*
* Makes hif guarantee that fw can message the host normally
* during suspend.
*
* SYNCHRONIZE WITH hif_vote_link_up by only calling in MC thread
* and initialization deinitialization sequencences.
*
* Return: n/a
*/
void hif_vote_link_up(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
QDF_BUG(scn);
scn->linkstate_vote++;
hif_info("Up_linkstate_vote %d", scn->linkstate_vote);
if (scn->linkstate_vote == 1)
hif_bus_prevent_linkdown(scn, true);
}
/**
* hif_can_suspend_link(): query if hif is permitted to suspend the link
* @hif_ctx: hif context
*
* Hif will ensure that the link won't be suspended if the upperlayers
* don't want it to.
*
* SYNCHRONIZATION: MC thread is stopped before bus suspend thus
* we don't need extra locking to ensure votes dont change while
* we are in the process of suspending or resuming.
*
* Return: false if hif will guarantee link up during suspend.
*/
bool hif_can_suspend_link(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
QDF_BUG(scn);
return scn->linkstate_vote == 0;
}
/**
* hif_hia_item_address(): hif_hia_item_address
* @target_type: target_type
* @item_offset: item_offset
*
* Return: n/a
*/
uint32_t hif_hia_item_address(uint32_t target_type, uint32_t item_offset)
{
switch (target_type) {
case TARGET_TYPE_AR6002:
return AR6002_HOST_INTEREST_ADDRESS + item_offset;
case TARGET_TYPE_AR6003:
return AR6003_HOST_INTEREST_ADDRESS + item_offset;
case TARGET_TYPE_AR6004:
return AR6004_HOST_INTEREST_ADDRESS + item_offset;
case TARGET_TYPE_AR6006:
return AR6006_HOST_INTEREST_ADDRESS + item_offset;
case TARGET_TYPE_AR9888:
return AR9888_HOST_INTEREST_ADDRESS + item_offset;
case TARGET_TYPE_AR6320:
case TARGET_TYPE_AR6320V2:
return AR6320_HOST_INTEREST_ADDRESS + item_offset;
case TARGET_TYPE_ADRASTEA:
/* ADRASTEA doesn't have a host interest address */
ASSERT(0);
return 0;
case TARGET_TYPE_AR900B:
return AR900B_HOST_INTEREST_ADDRESS + item_offset;
case TARGET_TYPE_QCA9984:
return QCA9984_HOST_INTEREST_ADDRESS + item_offset;
case TARGET_TYPE_QCA9888:
return QCA9888_HOST_INTEREST_ADDRESS + item_offset;
default:
ASSERT(0);
return 0;
}
}
/**
* hif_max_num_receives_reached() - check max receive is reached
* @scn: HIF Context
* @count: unsigned int.
*
* Output check status as bool
*
* Return: bool
*/
bool hif_max_num_receives_reached(struct hif_softc *scn, unsigned int count)
{
if (QDF_IS_EPPING_ENABLED(hif_get_conparam(scn)))
return count > 120;
else
return count > MAX_NUM_OF_RECEIVES;
}
/**
* init_buffer_count() - initial buffer count
* @maxSize: qdf_size_t
*
* routine to modify the initial buffer count to be allocated on an os
* platform basis. Platform owner will need to modify this as needed
*
* Return: qdf_size_t
*/
qdf_size_t init_buffer_count(qdf_size_t maxSize)
{
return maxSize;
}
/**
* hif_save_htc_htt_config_endpoint() - save htt_tx_endpoint
* @hif_ctx: hif context
* @htc_htt_tx_endpoint: htt_tx_endpoint
*
* Return: void
*/
void hif_save_htc_htt_config_endpoint(struct hif_opaque_softc *hif_ctx,
int htc_htt_tx_endpoint)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (!scn) {
hif_err("scn or scn->hif_sc is NULL!");
return;
}
scn->htc_htt_tx_endpoint = htc_htt_tx_endpoint;
}
qdf_export_symbol(hif_save_htc_htt_config_endpoint);
static const struct qwlan_hw qwlan_hw_list[] = {
{
.id = AR6320_REV1_VERSION,
.subid = 0,
.name = "QCA6174_REV1",
},
{
.id = AR6320_REV1_1_VERSION,
.subid = 0x1,
.name = "QCA6174_REV1_1",
},
{
.id = AR6320_REV1_3_VERSION,
.subid = 0x2,
.name = "QCA6174_REV1_3",
},
{
.id = AR6320_REV2_1_VERSION,
.subid = 0x4,
.name = "QCA6174_REV2_1",
},
{
.id = AR6320_REV2_1_VERSION,
.subid = 0x5,
.name = "QCA6174_REV2_2",
},
{
.id = AR6320_REV3_VERSION,
.subid = 0x6,
.name = "QCA6174_REV2.3",
},
{
.id = AR6320_REV3_VERSION,
.subid = 0x8,
.name = "QCA6174_REV3",
},
{
.id = AR6320_REV3_VERSION,
.subid = 0x9,
.name = "QCA6174_REV3_1",
},
{
.id = AR6320_REV3_2_VERSION,
.subid = 0xA,
.name = "AR6320_REV3_2_VERSION",
},
{
.id = QCA6390_V1,
.subid = 0x0,
.name = "QCA6390_V1",
},
{
.id = QCA6490_V1,
.subid = 0x0,
.name = "QCA6490_V1",
},
{
.id = WCN3990_v1,
.subid = 0x0,
.name = "WCN3990_V1",
},
{
.id = WCN3990_v2,
.subid = 0x0,
.name = "WCN3990_V2",
},
{
.id = WCN3990_v2_1,
.subid = 0x0,
.name = "WCN3990_V2.1",
},
{
.id = WCN3998,
.subid = 0x0,
.name = "WCN3998",
},
{
.id = QCA9379_REV1_VERSION,
.subid = 0xC,
.name = "QCA9379_REV1",
},
{
.id = QCA9379_REV1_VERSION,
.subid = 0xD,
.name = "QCA9379_REV1_1",
},
{
.id = MANGO_V1,
.subid = 0xF,
.name = "MANGO_V1",
},
{
.id = PEACH_V1,
.subid = 0,
.name = "PEACH_V1",
},
{
.id = KIWI_V1,
.subid = 0,
.name = "KIWI_V1",
},
{
.id = KIWI_V2,
.subid = 0,
.name = "KIWI_V2",
},
{
.id = WCN6750_V1,
.subid = 0,
.name = "WCN6750_V1",
},
{
.id = WCN6750_V2,
.subid = 0,
.name = "WCN6750_V2",
},
{
.id = WCN6450_V1,
.subid = 0,
.name = "WCN6450_V1",
},
{
.id = QCA6490_v2_1,
.subid = 0,
.name = "QCA6490",
},
{
.id = QCA6490_v2,
.subid = 0,
.name = "QCA6490",
},
{
.id = WCN3990_TALOS,
.subid = 0,
.name = "WCN3990",
},
{
.id = WCN3990_MOOREA,
.subid = 0,
.name = "WCN3990",
},
{
.id = WCN3990_SAIPAN,
.subid = 0,
.name = "WCN3990",
},
{
.id = WCN3990_RENNELL,
.subid = 0,
.name = "WCN3990",
},
{
.id = WCN3990_BITRA,
.subid = 0,
.name = "WCN3990",
},
{
.id = WCN3990_DIVAR,
.subid = 0,
.name = "WCN3990",
},
{
.id = WCN3990_ATHERTON,
.subid = 0,
.name = "WCN3990",
},
{
.id = WCN3990_STRAIT,
.subid = 0,
.name = "WCN3990",
},
{
.id = WCN3990_NETRANI,
.subid = 0,
.name = "WCN3990",
},
{
.id = WCN3990_CLARENCE,
.subid = 0,
.name = "WCN3990",
}
};
/**
* hif_get_hw_name(): get a human readable name for the hardware
* @info: Target Info
*
* Return: human readable name for the underlying wifi hardware.
*/
static const char *hif_get_hw_name(struct hif_target_info *info)
{
int i;
hif_debug("target version = %d, target revision = %d",
info->target_version,
info->target_revision);
if (info->hw_name)
return info->hw_name;
for (i = 0; i < ARRAY_SIZE(qwlan_hw_list); i++) {
if (info->target_version == qwlan_hw_list[i].id &&
info->target_revision == qwlan_hw_list[i].subid) {
return qwlan_hw_list[i].name;
}
}
info->hw_name = qdf_mem_malloc(64);
if (!info->hw_name)
return "Unknown Device (nomem)";
i = qdf_snprint(info->hw_name, 64, "HW_VERSION=%x.",
info->target_version);
if (i < 0)
return "Unknown Device (snprintf failure)";
else
return info->hw_name;
}
/**
* hif_get_hw_info(): hif_get_hw_info
* @scn: scn
* @version: version
* @revision: revision
* @target_name: target name
*
* Return: n/a
*/
void hif_get_hw_info(struct hif_opaque_softc *scn, u32 *version, u32 *revision,
const char **target_name)
{
struct hif_target_info *info = hif_get_target_info_handle(scn);
struct hif_softc *sc = HIF_GET_SOFTC(scn);
if (sc->bus_type == QDF_BUS_TYPE_USB)
hif_usb_get_hw_info(sc);
*version = info->target_version;
*revision = info->target_revision;
*target_name = hif_get_hw_name(info);
}
/**
* hif_get_dev_ba(): API to get device base address.
* @hif_handle: hif handle
*
* Return: device base address
*/
void *hif_get_dev_ba(struct hif_opaque_softc *hif_handle)
{
struct hif_softc *scn = (struct hif_softc *)hif_handle;
return scn->mem;
}
qdf_export_symbol(hif_get_dev_ba);
/**
* hif_get_dev_ba_ce(): API to get device ce base address.
* @hif_handle: hif handle
*
* Return: dev mem base address for CE
*/
void *hif_get_dev_ba_ce(struct hif_opaque_softc *hif_handle)
{
struct hif_softc *scn = (struct hif_softc *)hif_handle;
return scn->mem_ce;
}
qdf_export_symbol(hif_get_dev_ba_ce);
/**
* hif_get_dev_ba_pmm(): API to get device pmm base address.
* @hif_handle: scn
*
* Return: dev mem base address for PMM
*/
void *hif_get_dev_ba_pmm(struct hif_opaque_softc *hif_handle)
{
struct hif_softc *scn = (struct hif_softc *)hif_handle;
return scn->mem_pmm_base;
}
qdf_export_symbol(hif_get_dev_ba_pmm);
uint32_t hif_get_soc_version(struct hif_opaque_softc *hif_handle)
{
struct hif_softc *scn = (struct hif_softc *)hif_handle;
return scn->target_info.soc_version;
}
qdf_export_symbol(hif_get_soc_version);
/**
* hif_get_dev_ba_cmem(): API to get device ce base address.
* @hif_handle: hif handle
*
* Return: dev mem base address for CMEM
*/
void *hif_get_dev_ba_cmem(struct hif_opaque_softc *hif_handle)
{
struct hif_softc *scn = (struct hif_softc *)hif_handle;
return scn->mem_cmem;
}
qdf_export_symbol(hif_get_dev_ba_cmem);
#ifdef FEATURE_RUNTIME_PM
void hif_runtime_prevent_linkdown(struct hif_softc *scn, bool is_get)
{
if (is_get)
qdf_runtime_pm_prevent_suspend(&scn->prevent_linkdown_lock);
else
qdf_runtime_pm_allow_suspend(&scn->prevent_linkdown_lock);
}
static inline
void hif_rtpm_lock_init(struct hif_softc *scn)
{
qdf_runtime_lock_init(&scn->prevent_linkdown_lock);
}
static inline
void hif_rtpm_lock_deinit(struct hif_softc *scn)
{
qdf_runtime_lock_deinit(&scn->prevent_linkdown_lock);
}
#else
static inline
void hif_rtpm_lock_init(struct hif_softc *scn)
{
}
static inline
void hif_rtpm_lock_deinit(struct hif_softc *scn)
{
}
#endif
#ifdef WLAN_CE_INTERRUPT_THRESHOLD_CONFIG
/**
* hif_get_interrupt_threshold_cfg_from_psoc() - Retrieve ini cfg from psoc
* @scn: hif context
* @psoc: psoc objmgr handle
*
* Return: None
*/
static inline
void hif_get_interrupt_threshold_cfg_from_psoc(struct hif_softc *scn,
struct wlan_objmgr_psoc *psoc)
{
if (psoc) {
scn->ini_cfg.ce_status_ring_timer_threshold =
cfg_get(psoc,
CFG_CE_STATUS_RING_TIMER_THRESHOLD);
scn->ini_cfg.ce_status_ring_batch_count_threshold =
cfg_get(psoc,
CFG_CE_STATUS_RING_BATCH_COUNT_THRESHOLD);
}
}
#else
static inline
void hif_get_interrupt_threshold_cfg_from_psoc(struct hif_softc *scn,
struct wlan_objmgr_psoc *psoc)
{
}
#endif /* WLAN_CE_INTERRUPT_THRESHOLD_CONFIG */
/**
* hif_get_cfg_from_psoc() - Retrieve ini cfg from psoc
* @scn: hif context
* @psoc: psoc objmgr handle
*
* Return: None
*/
static inline
void hif_get_cfg_from_psoc(struct hif_softc *scn,
struct wlan_objmgr_psoc *psoc)
{
if (psoc) {
scn->ini_cfg.disable_wake_irq =
cfg_get(psoc, CFG_DISABLE_WAKE_IRQ);
/**
* Wake IRQ can't share the same IRQ with the copy engines
* In one MSI mode, we don't know whether wake IRQ is triggered
* or not in wake IRQ handler. known issue CR 2055359
* If you want to support Wake IRQ. Please allocate at least
* 2 MSI vector. The first is for wake IRQ while the others
* share the second vector
*/
if (pld_is_one_msi(scn->qdf_dev->dev)) {
hif_debug("Disable wake IRQ once it is one MSI mode");
scn->ini_cfg.disable_wake_irq = true;
}
hif_get_interrupt_threshold_cfg_from_psoc(scn, psoc);
}
}
#if defined(HIF_CE_LOG_INFO) || defined(HIF_BUS_LOG_INFO)
/**
* hif_recovery_notifier_cb - Recovery notifier callback to log
* hang event data
* @block: notifier block
* @state: state
* @data: notifier data
*
* Return: status
*/
static
int hif_recovery_notifier_cb(struct notifier_block *block, unsigned long state,
void *data)
{
struct qdf_notifer_data *notif_data = data;
qdf_notif_block *notif_block;
struct hif_softc *hif_handle;
bool bus_id_invalid;
if (!data || !block)
return -EINVAL;
notif_block = qdf_container_of(block, qdf_notif_block, notif_block);
hif_handle = notif_block->priv_data;
if (!hif_handle)
return -EINVAL;
bus_id_invalid = hif_log_bus_info(hif_handle, notif_data->hang_data,
&notif_data->offset);
if (bus_id_invalid)
return NOTIFY_STOP_MASK;
hif_log_ce_info(hif_handle, notif_data->hang_data,
&notif_data->offset);
return 0;
}
/**
* hif_register_recovery_notifier - Register hif recovery notifier
* @hif_handle: hif handle
*
* Return: status
*/
static
QDF_STATUS hif_register_recovery_notifier(struct hif_softc *hif_handle)
{
qdf_notif_block *hif_notifier;
if (!hif_handle)
return QDF_STATUS_E_FAILURE;
hif_notifier = &hif_handle->hif_recovery_notifier;
hif_notifier->notif_block.notifier_call = hif_recovery_notifier_cb;
hif_notifier->priv_data = hif_handle;
return qdf_hang_event_register_notifier(hif_notifier);
}
/**
* hif_unregister_recovery_notifier - Un-register hif recovery notifier
* @hif_handle: hif handle
*
* Return: status
*/
static
QDF_STATUS hif_unregister_recovery_notifier(struct hif_softc *hif_handle)
{
qdf_notif_block *hif_notifier = &hif_handle->hif_recovery_notifier;
return qdf_hang_event_unregister_notifier(hif_notifier);
}
#else
static inline
QDF_STATUS hif_register_recovery_notifier(struct hif_softc *hif_handle)
{
return QDF_STATUS_SUCCESS;
}
static inline
QDF_STATUS hif_unregister_recovery_notifier(struct hif_softc *hif_handle)
{
return QDF_STATUS_SUCCESS;
}
#endif
#ifdef HIF_CPU_PERF_AFFINE_MASK
/**
* __hif_cpu_hotplug_notify() - CPU hotplug event handler
* @context: HIF context
* @cpu: CPU Id of the CPU generating the event
* @cpu_up: true if the CPU is online
*
* Return: None
*/
static void __hif_cpu_hotplug_notify(void *context,
uint32_t cpu, bool cpu_up)
{
struct hif_softc *scn = context;
if (!scn)
return;
if (hif_is_driver_unloading(scn) || hif_is_recovery_in_progress(scn))
return;
if (cpu_up) {
hif_config_irq_set_perf_affinity_hint(GET_HIF_OPAQUE_HDL(scn));
hif_debug("Setting affinity for online CPU: %d", cpu);
} else {
hif_debug("Skip setting affinity for offline CPU: %d", cpu);
}
}
/**
* hif_cpu_hotplug_notify - cpu core up/down notification
* handler
* @context: HIF context
* @cpu: CPU generating the event
* @cpu_up: true if the CPU is online
*
* Return: None
*/
static void hif_cpu_hotplug_notify(void *context, uint32_t cpu, bool cpu_up)
{
struct qdf_op_sync *op_sync;
if (qdf_op_protect(&op_sync))
return;
__hif_cpu_hotplug_notify(context, cpu, cpu_up);
qdf_op_unprotect(op_sync);
}
static void hif_cpu_online_cb(void *context, uint32_t cpu)
{
hif_cpu_hotplug_notify(context, cpu, true);
}
static void hif_cpu_before_offline_cb(void *context, uint32_t cpu)
{
hif_cpu_hotplug_notify(context, cpu, false);
}
static void hif_cpuhp_register(struct hif_softc *scn)
{
if (!scn) {
hif_info_high("cannot register hotplug notifiers");
return;
}
qdf_cpuhp_register(&scn->cpuhp_event_handle,
scn,
hif_cpu_online_cb,
hif_cpu_before_offline_cb);
}
static void hif_cpuhp_unregister(struct hif_softc *scn)
{
if (!scn) {
hif_info_high("cannot unregister hotplug notifiers");
return;
}
qdf_cpuhp_unregister(&scn->cpuhp_event_handle);
}
#else
static void hif_cpuhp_register(struct hif_softc *scn)
{
}
static void hif_cpuhp_unregister(struct hif_softc *scn)
{
}
#endif /* ifdef HIF_CPU_PERF_AFFINE_MASK */
#ifdef HIF_DETECTION_LATENCY_ENABLE
/*
* Bitmask to control enablement of latency detection for the tasklets,
* bit-X represents for tasklet of WLAN_CE_X.
*/
#ifndef DETECTION_LATENCY_TASKLET_MASK
#define DETECTION_LATENCY_TASKLET_MASK (BIT(2) | BIT(7))
#endif
static inline int
__hif_tasklet_latency(struct hif_softc *scn, bool from_timer, int idx)
{
qdf_time_t sched_time =
scn->latency_detect.tasklet_info[idx].sched_time;
qdf_time_t exec_time =
scn->latency_detect.tasklet_info[idx].exec_time;
qdf_time_t curr_time = qdf_system_ticks();
uint32_t threshold = scn->latency_detect.threshold;
qdf_time_t expect_exec_time =
sched_time + qdf_system_msecs_to_ticks(threshold);
/* 2 kinds of check here.
* from_timer==true: check if tasklet stall
* from_timer==false: check tasklet execute comes late
*/
if (from_timer ?
(qdf_system_time_after(sched_time, exec_time) &&
qdf_system_time_after(curr_time, expect_exec_time)) :
qdf_system_time_after(exec_time, expect_exec_time)) {
hif_err("tasklet[%d] latency detected: from_timer %d, curr_time %lu, sched_time %lu, exec_time %lu, threshold %ums, timeout %ums, cpu_id %d, called: %ps",
idx, from_timer, curr_time, sched_time,
exec_time, threshold,
scn->latency_detect.timeout,
qdf_get_cpu(), (void *)_RET_IP_);
qdf_trigger_self_recovery(NULL,
QDF_TASKLET_CREDIT_LATENCY_DETECT);
return -ETIMEDOUT;
}
return 0;
}
/**
* hif_tasklet_latency_detect_enabled() - check whether latency detect
* is enabled for the tasklet which is specified by idx
* @scn: HIF opaque context
* @idx: CE id
*
* Return: true if latency detect is enabled for the specified tasklet,
* false otherwise.
*/
static inline bool
hif_tasklet_latency_detect_enabled(struct hif_softc *scn, int idx)
{
if (QDF_GLOBAL_MISSION_MODE != hif_get_conparam(scn))
return false;
if (!scn->latency_detect.enable_detection)
return false;
if (idx < 0 || idx >= HIF_TASKLET_IN_MONITOR ||
!qdf_test_bit(idx, scn->latency_detect.tasklet_bmap))
return false;
return true;
}
void hif_tasklet_latency_record_exec(struct hif_softc *scn, int idx)
{
if (!hif_tasklet_latency_detect_enabled(scn, idx))
return;
/*
* hif_set_enable_detection(true) might come between
* hif_tasklet_latency_record_sched() and
* hif_tasklet_latency_record_exec() during wlan startup, then the
* sched_time is 0 but exec_time is not, and hit the timeout case in
* __hif_tasklet_latency().
* To avoid such issue, skip exec_time recording if sched_time has not
* been recorded.
*/
if (!scn->latency_detect.tasklet_info[idx].sched_time)
return;
scn->latency_detect.tasklet_info[idx].exec_time = qdf_system_ticks();
__hif_tasklet_latency(scn, false, idx);
}
void hif_tasklet_latency_record_sched(struct hif_softc *scn, int idx)
{
if (!hif_tasklet_latency_detect_enabled(scn, idx))
return;
scn->latency_detect.tasklet_info[idx].sched_cpuid = qdf_get_cpu();
scn->latency_detect.tasklet_info[idx].sched_time = qdf_system_ticks();
}
static inline void hif_credit_latency(struct hif_softc *scn, bool from_timer)
{
qdf_time_t credit_request_time =
scn->latency_detect.credit_request_time;
qdf_time_t credit_report_time = scn->latency_detect.credit_report_time;
qdf_time_t curr_jiffies = qdf_system_ticks();
uint32_t threshold = scn->latency_detect.threshold;
int cpu_id = qdf_get_cpu();
/* 2 kinds of check here.
* from_timer==true: check if credit report stall
* from_timer==false: check credit report comes late
*/
if ((from_timer ?
qdf_system_time_after(credit_request_time, credit_report_time) :
qdf_system_time_after(credit_report_time, credit_request_time)) &&
qdf_system_time_after(curr_jiffies,
credit_request_time +
qdf_system_msecs_to_ticks(threshold))) {
hif_err("credit report latency: from timer %d, curr_jiffies %lu, credit_request_time %lu, credit_report_time %lu, threshold %ums, timeout %ums, cpu_id %d, called: %ps",
from_timer, curr_jiffies, credit_request_time,
credit_report_time, threshold,
scn->latency_detect.timeout,
cpu_id, (void *)_RET_IP_);
goto latency;
}
return;
latency:
qdf_trigger_self_recovery(NULL, QDF_TASKLET_CREDIT_LATENCY_DETECT);
}
static inline void hif_tasklet_latency(struct hif_softc *scn, bool from_timer)
{
int i, ret;
for (i = 0; i < HIF_TASKLET_IN_MONITOR; i++) {
if (!qdf_test_bit(i, scn->latency_detect.tasklet_bmap))
continue;
ret = __hif_tasklet_latency(scn, from_timer, i);
if (ret)
return;
}
}
/**
* hif_check_detection_latency(): to check if latency for tasklet/credit
*
* @scn: hif context
* @from_timer: if called from timer handler
* @bitmap_type: indicate if check tasklet or credit
*
* Return: none
*/
void hif_check_detection_latency(struct hif_softc *scn,
bool from_timer,
uint32_t bitmap_type)
{
if (QDF_GLOBAL_MISSION_MODE != hif_get_conparam(scn))
return;
if (!scn->latency_detect.enable_detection)
return;
if (bitmap_type & BIT(HIF_DETECT_TASKLET))
hif_tasklet_latency(scn, from_timer);
if (bitmap_type & BIT(HIF_DETECT_CREDIT))
hif_credit_latency(scn, from_timer);
}
static void hif_latency_detect_timeout_handler(void *arg)
{
struct hif_softc *scn = (struct hif_softc *)arg;
int next_cpu, i;
qdf_cpu_mask cpu_mask = {0};
struct hif_latency_detect *detect = &scn->latency_detect;
hif_check_detection_latency(scn, true,
BIT(HIF_DETECT_TASKLET) |
BIT(HIF_DETECT_CREDIT));
/* it need to make sure timer start on a different cpu,
* so it can detect the tasklet schedule stall, but there
* is still chance that, after timer has been started, then
* irq/tasklet happens on the same cpu, then tasklet will
* execute before softirq timer, if this tasklet stall, the
* timer can't detect it, we can accept this as a limitation,
* if tasklet stall, anyway other place will detect it, just
* a little later.
*/
qdf_cpumask_copy(&cpu_mask, (const qdf_cpu_mask *)cpu_active_mask);
for (i = 0; i < HIF_TASKLET_IN_MONITOR; i++) {
if (!qdf_test_bit(i, detect->tasklet_bmap))
continue;
qdf_cpumask_clear_cpu(detect->tasklet_info[i].sched_cpuid,
&cpu_mask);
}
next_cpu = cpumask_first(&cpu_mask);
if (qdf_unlikely(next_cpu >= nr_cpu_ids)) {
hif_debug("start timer on local");
/* it doesn't found a available cpu, start on local cpu*/
qdf_timer_mod(&detect->timer, detect->timeout);
} else {
qdf_timer_start_on(&detect->timer, detect->timeout, next_cpu);
}
}
static void hif_latency_detect_timer_init(struct hif_softc *scn)
{
scn->latency_detect.timeout =
DETECTION_TIMER_TIMEOUT;
scn->latency_detect.threshold =
DETECTION_LATENCY_THRESHOLD;
hif_info("timer timeout %u, latency threshold %u",
scn->latency_detect.timeout,
scn->latency_detect.threshold);
scn->latency_detect.is_timer_started = false;
qdf_timer_init(NULL,
&scn->latency_detect.timer,
&hif_latency_detect_timeout_handler,
scn,
QDF_TIMER_TYPE_SW_SPIN);
}
static void hif_latency_detect_timer_deinit(struct hif_softc *scn)
{
hif_info("deinit timer");
qdf_timer_free(&scn->latency_detect.timer);
}
static void hif_latency_detect_init(struct hif_softc *scn)
{
uint32_t tasklet_mask;
int i;
if (QDF_GLOBAL_MISSION_MODE != hif_get_conparam(scn))
return;
tasklet_mask = DETECTION_LATENCY_TASKLET_MASK;
hif_info("tasklet mask is 0x%x", tasklet_mask);
for (i = 0; i < HIF_TASKLET_IN_MONITOR; i++) {
if (BIT(i) & tasklet_mask)
qdf_set_bit(i, scn->latency_detect.tasklet_bmap);
}
hif_latency_detect_timer_init(scn);
}
static void hif_latency_detect_deinit(struct hif_softc *scn)
{
int i;
if (QDF_GLOBAL_MISSION_MODE != hif_get_conparam(scn))
return;
hif_latency_detect_timer_deinit(scn);
for (i = 0; i < HIF_TASKLET_IN_MONITOR; i++)
qdf_clear_bit(i, scn->latency_detect.tasklet_bmap);
}
void hif_latency_detect_timer_start(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (QDF_GLOBAL_MISSION_MODE != hif_get_conparam(scn))
return;
hif_debug_rl("start timer");
if (scn->latency_detect.is_timer_started) {
hif_info("timer has been started");
return;
}
qdf_timer_start(&scn->latency_detect.timer,
scn->latency_detect.timeout);
scn->latency_detect.is_timer_started = true;
}
void hif_latency_detect_timer_stop(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (QDF_GLOBAL_MISSION_MODE != hif_get_conparam(scn))
return;
hif_debug_rl("stop timer");
qdf_timer_sync_cancel(&scn->latency_detect.timer);
scn->latency_detect.is_timer_started = false;
}
void hif_latency_detect_credit_record_time(
enum hif_credit_exchange_type type,
struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (!scn) {
hif_err("Could not do runtime put, scn is null");
return;
}
if (QDF_GLOBAL_MISSION_MODE != hif_get_conparam(scn))
return;
if (HIF_REQUEST_CREDIT == type)
scn->latency_detect.credit_request_time = qdf_system_ticks();
else if (HIF_PROCESS_CREDIT_REPORT == type)
scn->latency_detect.credit_report_time = qdf_system_ticks();
hif_check_detection_latency(scn, false, BIT(HIF_DETECT_CREDIT));
}
void hif_set_enable_detection(struct hif_opaque_softc *hif_ctx, bool value)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (!scn) {
hif_err("Could not do runtime put, scn is null");
return;
}
if (QDF_GLOBAL_MISSION_MODE != hif_get_conparam(scn))
return;
scn->latency_detect.enable_detection = value;
}
#else
static inline void hif_latency_detect_init(struct hif_softc *scn)
{}
static inline void hif_latency_detect_deinit(struct hif_softc *scn)
{}
#endif
#ifdef WLAN_FEATURE_AFFINITY_MGR
#define AFFINITY_THRESHOLD 5000000
static inline void
hif_affinity_mgr_init(struct hif_softc *scn, struct wlan_objmgr_psoc *psoc)
{
unsigned int cpus;
qdf_cpu_mask allowed_mask = {0};
scn->affinity_mgr_supported =
(cfg_get(psoc, CFG_IRQ_AFFINE_AUDIO_USE_CASE) &&
qdf_walt_get_cpus_taken_supported());
hif_info("Affinity Manager supported: %d", scn->affinity_mgr_supported);
if (!scn->affinity_mgr_supported)
return;
scn->time_threshold = AFFINITY_THRESHOLD;
qdf_for_each_possible_cpu(cpus)
if (qdf_topology_physical_package_id(cpus) ==
CPU_CLUSTER_TYPE_LITTLE)
qdf_cpumask_set_cpu(cpus, &allowed_mask);
qdf_cpumask_copy(&scn->allowed_mask, &allowed_mask);
}
#else
static inline void
hif_affinity_mgr_init(struct hif_softc *scn, struct wlan_objmgr_psoc *psoc)
{
}
#endif
#ifdef FEATURE_DIRECT_LINK
/**
* hif_init_direct_link_rcv_pipe_num(): Initialize the direct link receive
* pipe number
* @scn: hif context
*
* Return: None
*/
static inline
void hif_init_direct_link_rcv_pipe_num(struct hif_softc *scn)
{
scn->dl_recv_pipe_num = INVALID_PIPE_NO;
}
#else
static inline
void hif_init_direct_link_rcv_pipe_num(struct hif_softc *scn)
{
}
#endif
struct hif_opaque_softc *hif_open(qdf_device_t qdf_ctx,
uint32_t mode,
enum qdf_bus_type bus_type,
struct hif_driver_state_callbacks *cbk,
struct wlan_objmgr_psoc *psoc)
{
struct hif_softc *scn;
QDF_STATUS status = QDF_STATUS_SUCCESS;
int bus_context_size = hif_bus_get_context_size(bus_type);
if (bus_context_size == 0) {
hif_err("context size 0 not allowed");
return NULL;
}
scn = (struct hif_softc *)qdf_mem_malloc(bus_context_size);
if (!scn)
return GET_HIF_OPAQUE_HDL(scn);
scn->qdf_dev = qdf_ctx;
scn->hif_con_param = mode;
qdf_atomic_init(&scn->active_tasklet_cnt);
qdf_atomic_init(&scn->active_grp_tasklet_cnt);
qdf_atomic_init(&scn->link_suspended);
qdf_atomic_init(&scn->tasklet_from_intr);
hif_system_pm_set_state_on(GET_HIF_OPAQUE_HDL(scn));
qdf_mem_copy(&scn->callbacks, cbk,
sizeof(struct hif_driver_state_callbacks));
scn->bus_type = bus_type;
hif_allow_ep_vote_access(GET_HIF_OPAQUE_HDL(scn));
hif_get_cfg_from_psoc(scn, psoc);
hif_set_event_hist_mask(GET_HIF_OPAQUE_HDL(scn));
status = hif_bus_open(scn, bus_type);
if (status != QDF_STATUS_SUCCESS) {
hif_err("hif_bus_open error = %d, bus_type = %d",
status, bus_type);
qdf_mem_free(scn);
scn = NULL;
goto out;
}
hif_rtpm_lock_init(scn);
hif_cpuhp_register(scn);
hif_latency_detect_init(scn);
hif_affinity_mgr_init(scn, psoc);
hif_init_direct_link_rcv_pipe_num(scn);
out:
return GET_HIF_OPAQUE_HDL(scn);
}
#ifdef ADRASTEA_RRI_ON_DDR
/**
* hif_uninit_rri_on_ddr(): free consistent memory allocated for rri
* @scn: hif context
*
* Return: none
*/
void hif_uninit_rri_on_ddr(struct hif_softc *scn)
{
if (scn->vaddr_rri_on_ddr)
qdf_mem_free_consistent(scn->qdf_dev, scn->qdf_dev->dev,
RRI_ON_DDR_MEM_SIZE,
scn->vaddr_rri_on_ddr,
scn->paddr_rri_on_ddr, 0);
scn->vaddr_rri_on_ddr = NULL;
}
#endif
/**
* hif_close(): hif_close
* @hif_ctx: hif_ctx
*
* Return: n/a
*/
void hif_close(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (!scn) {
hif_err("hif_opaque_softc is NULL");
return;
}
hif_latency_detect_deinit(scn);
if (scn->athdiag_procfs_inited) {
athdiag_procfs_remove();
scn->athdiag_procfs_inited = false;
}
if (scn->target_info.hw_name) {
char *hw_name = scn->target_info.hw_name;
scn->target_info.hw_name = "ErrUnloading";
qdf_mem_free(hw_name);
}
hif_uninit_rri_on_ddr(scn);
hif_cleanup_static_buf_to_target(scn);
hif_cpuhp_unregister(scn);
hif_rtpm_lock_deinit(scn);
hif_bus_close(scn);
qdf_mem_free(scn);
}
/**
* hif_get_num_active_grp_tasklets() - get the number of active
* datapath group tasklets pending to be completed.
* @scn: HIF context
*
* Returns: the number of datapath group tasklets which are active
*/
static inline int hif_get_num_active_grp_tasklets(struct hif_softc *scn)
{
return qdf_atomic_read(&scn->active_grp_tasklet_cnt);
}
#if (defined(QCA_WIFI_QCA8074) || defined(QCA_WIFI_QCA6018) || \
defined(QCA_WIFI_QCA6290) || defined(QCA_WIFI_QCA6390) || \
defined(QCA_WIFI_QCN9000) || defined(QCA_WIFI_QCA6490) || \
defined(QCA_WIFI_QCA6750) || defined(QCA_WIFI_QCA5018) || \
defined(QCA_WIFI_KIWI) || defined(QCA_WIFI_QCN9224) || \
defined(QCA_WIFI_QCN6432) || \
defined(QCA_WIFI_QCA9574)) || defined(QCA_WIFI_QCA5332)
/**
* hif_get_num_pending_work() - get the number of entries in
* the workqueue pending to be completed.
* @scn: HIF context
*
* Returns: the number of tasklets which are active
*/
static inline int hif_get_num_pending_work(struct hif_softc *scn)
{
return hal_get_reg_write_pending_work(scn->hal_soc);
}
#elif defined(FEATURE_HIF_DELAYED_REG_WRITE)
static inline int hif_get_num_pending_work(struct hif_softc *scn)
{
return qdf_atomic_read(&scn->active_work_cnt);
}
#else
static inline int hif_get_num_pending_work(struct hif_softc *scn)
{
return 0;
}
#endif
QDF_STATUS hif_try_complete_tasks(struct hif_softc *scn)
{
uint32_t task_drain_wait_cnt = 0;
int tasklet = 0, grp_tasklet = 0, work = 0;
while ((tasklet = hif_get_num_active_tasklets(scn)) ||
(grp_tasklet = hif_get_num_active_grp_tasklets(scn)) ||
(work = hif_get_num_pending_work(scn))) {
if (++task_drain_wait_cnt > HIF_TASK_DRAIN_WAIT_CNT) {
hif_err("pending tasklets %d grp tasklets %d work %d",
tasklet, grp_tasklet, work);
QDF_DEBUG_PANIC("Complete tasks takes more than %u ms: tasklets %d grp tasklets %d work %d",
HIF_TASK_DRAIN_WAIT_CNT * 10,
tasklet, grp_tasklet, work);
return QDF_STATUS_E_FAULT;
}
hif_info("waiting for tasklets %d grp tasklets %d work %d",
tasklet, grp_tasklet, work);
msleep(10);
}
return QDF_STATUS_SUCCESS;
}
#ifdef HIF_HAL_REG_ACCESS_SUPPORT
void hif_reg_window_write(struct hif_softc *scn, uint32_t offset,
uint32_t value)
{
hal_write32_mb(scn->hal_soc, offset, value);
}
uint32_t hif_reg_window_read(struct hif_softc *scn, uint32_t offset)
{
return hal_read32_mb(scn->hal_soc, offset);
}
#endif
#if defined(HIF_IPCI) && defined(FEATURE_HAL_DELAYED_REG_WRITE)
QDF_STATUS hif_try_prevent_ep_vote_access(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
uint32_t work_drain_wait_cnt = 0;
uint32_t wait_cnt = 0;
int work = 0;
qdf_atomic_set(&scn->dp_ep_vote_access,
HIF_EP_VOTE_ACCESS_DISABLE);
qdf_atomic_set(&scn->ep_vote_access,
HIF_EP_VOTE_ACCESS_DISABLE);
while ((work = hif_get_num_pending_work(scn))) {
if (++work_drain_wait_cnt > HIF_WORK_DRAIN_WAIT_CNT) {
qdf_atomic_set(&scn->dp_ep_vote_access,
HIF_EP_VOTE_ACCESS_ENABLE);
qdf_atomic_set(&scn->ep_vote_access,
HIF_EP_VOTE_ACCESS_ENABLE);
hif_err("timeout wait for pending work %d ", work);
return QDF_STATUS_E_FAULT;
}
qdf_sleep(10);
}
if (pld_is_pci_ep_awake(scn->qdf_dev->dev) == -ENOTSUPP)
return QDF_STATUS_SUCCESS;
while (pld_is_pci_ep_awake(scn->qdf_dev->dev)) {
if (++wait_cnt > HIF_EP_WAKE_RESET_WAIT_CNT) {
hif_err("Release EP vote is not proceed by Fw");
return QDF_STATUS_E_FAULT;
}
qdf_sleep(5);
}
return QDF_STATUS_SUCCESS;
}
void hif_set_ep_intermediate_vote_access(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
uint8_t vote_access;
vote_access = qdf_atomic_read(&scn->ep_vote_access);
if (vote_access != HIF_EP_VOTE_ACCESS_DISABLE)
hif_info("EP vote changed from:%u to intermediate state",
vote_access);
if (QDF_IS_STATUS_ERROR(hif_try_prevent_ep_vote_access(hif_ctx)))
QDF_BUG(0);
qdf_atomic_set(&scn->ep_vote_access,
HIF_EP_VOTE_INTERMEDIATE_ACCESS);
}
void hif_allow_ep_vote_access(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
qdf_atomic_set(&scn->dp_ep_vote_access,
HIF_EP_VOTE_ACCESS_ENABLE);
qdf_atomic_set(&scn->ep_vote_access,
HIF_EP_VOTE_ACCESS_ENABLE);
}
void hif_set_ep_vote_access(struct hif_opaque_softc *hif_ctx,
uint8_t type, uint8_t access)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (type == HIF_EP_VOTE_DP_ACCESS)
qdf_atomic_set(&scn->dp_ep_vote_access, access);
else
qdf_atomic_set(&scn->ep_vote_access, access);
}
uint8_t hif_get_ep_vote_access(struct hif_opaque_softc *hif_ctx,
uint8_t type)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (type == HIF_EP_VOTE_DP_ACCESS)
return qdf_atomic_read(&scn->dp_ep_vote_access);
else
return qdf_atomic_read(&scn->ep_vote_access);
}
#endif
#ifdef FEATURE_HIF_DELAYED_REG_WRITE
#ifdef MEMORY_DEBUG
#define HIF_REG_WRITE_QUEUE_LEN 128
#else
#define HIF_REG_WRITE_QUEUE_LEN 32
#endif
/**
* hif_print_reg_write_stats() - Print hif delayed reg write stats
* @hif_ctx: hif opaque handle
*
* Return: None
*/
void hif_print_reg_write_stats(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
struct CE_state *ce_state;
uint32_t *hist;
int i;
hist = scn->wstats.sched_delay;
hif_debug("wstats: enq %u deq %u coal %u direct %u q_depth %u max_q %u sched-delay hist %u %u %u %u",
qdf_atomic_read(&scn->wstats.enqueues),
scn->wstats.dequeues,
qdf_atomic_read(&scn->wstats.coalesces),
qdf_atomic_read(&scn->wstats.direct),
qdf_atomic_read(&scn->wstats.q_depth),
scn->wstats.max_q_depth,
hist[HIF_REG_WRITE_SCHED_DELAY_SUB_100us],
hist[HIF_REG_WRITE_SCHED_DELAY_SUB_1000us],
hist[HIF_REG_WRITE_SCHED_DELAY_SUB_5000us],
hist[HIF_REG_WRITE_SCHED_DELAY_GT_5000us]);
for (i = 0; i < scn->ce_count; i++) {
ce_state = scn->ce_id_to_state[i];
if (!ce_state)
continue;
hif_debug("ce%d: enq %u deq %u coal %u direct %u",
i, ce_state->wstats.enqueues,
ce_state->wstats.dequeues,
ce_state->wstats.coalesces,
ce_state->wstats.direct);
}
}
/**
* hif_is_reg_write_tput_level_high() - throughput level for delayed reg writes
* @scn: hif_softc pointer
*
* Return: true if throughput is high, else false.
*/
static inline bool hif_is_reg_write_tput_level_high(struct hif_softc *scn)
{
int bw_level = hif_get_bandwidth_level(GET_HIF_OPAQUE_HDL(scn));
return (bw_level >= PLD_BUS_WIDTH_MEDIUM) ? true : false;
}
/**
* hif_reg_write_fill_sched_delay_hist() - fill reg write delay histogram
* @scn: hif_softc pointer
* @delay_us: delay in us
*
* Return: None
*/
static inline void hif_reg_write_fill_sched_delay_hist(struct hif_softc *scn,
uint64_t delay_us)
{
uint32_t *hist;
hist = scn->wstats.sched_delay;
if (delay_us < 100)
hist[HIF_REG_WRITE_SCHED_DELAY_SUB_100us]++;
else if (delay_us < 1000)
hist[HIF_REG_WRITE_SCHED_DELAY_SUB_1000us]++;
else if (delay_us < 5000)
hist[HIF_REG_WRITE_SCHED_DELAY_SUB_5000us]++;
else
hist[HIF_REG_WRITE_SCHED_DELAY_GT_5000us]++;
}
/**
* hif_process_reg_write_q_elem() - process a register write queue element
* @scn: hif_softc pointer
* @q_elem: pointer to hal register write queue element
*
* Return: The value which was written to the address
*/
static int32_t
hif_process_reg_write_q_elem(struct hif_softc *scn,
struct hif_reg_write_q_elem *q_elem)
{
struct CE_state *ce_state = q_elem->ce_state;
uint32_t write_val = -1;
qdf_spin_lock_bh(&ce_state->ce_index_lock);
ce_state->reg_write_in_progress = false;
ce_state->wstats.dequeues++;
if (ce_state->src_ring) {
q_elem->dequeue_val = ce_state->src_ring->write_index;
hal_write32_mb(scn->hal_soc, ce_state->ce_wrt_idx_offset,
ce_state->src_ring->write_index);
write_val = ce_state->src_ring->write_index;
} else if (ce_state->dest_ring) {
q_elem->dequeue_val = ce_state->dest_ring->write_index;
hal_write32_mb(scn->hal_soc, ce_state->ce_wrt_idx_offset,
ce_state->dest_ring->write_index);
write_val = ce_state->dest_ring->write_index;
} else {
hif_debug("invalid reg write received");
qdf_assert(0);
}
q_elem->valid = 0;
ce_state->last_dequeue_time = q_elem->dequeue_time;
qdf_spin_unlock_bh(&ce_state->ce_index_lock);
return write_val;
}
/**
* hif_reg_write_work() - Worker to process delayed writes
* @arg: hif_softc pointer
*
* Return: None
*/
static void hif_reg_write_work(void *arg)
{
struct hif_softc *scn = arg;
struct hif_reg_write_q_elem *q_elem;
uint32_t offset;
uint64_t delta_us;
int32_t q_depth, write_val;
uint32_t num_processed = 0;
int32_t ring_id;
q_elem = &scn->reg_write_queue[scn->read_idx];
q_elem->work_scheduled_time = qdf_get_log_timestamp();
q_elem->cpu_id = qdf_get_cpu();
/* Make sure q_elem consistent in the memory for multi-cores */
qdf_rmb();
if (!q_elem->valid)
return;
q_depth = qdf_atomic_read(&scn->wstats.q_depth);
if (q_depth > scn->wstats.max_q_depth)
scn->wstats.max_q_depth = q_depth;
if (hif_prevent_link_low_power_states(GET_HIF_OPAQUE_HDL(scn))) {
scn->wstats.prevent_l1_fails++;
return;
}
while (true) {
qdf_rmb();
if (!q_elem->valid)
break;
q_elem->dequeue_time = qdf_get_log_timestamp();
ring_id = q_elem->ce_state->id;
offset = q_elem->offset;
delta_us = qdf_log_timestamp_to_usecs(q_elem->dequeue_time -
q_elem->enqueue_time);
hif_reg_write_fill_sched_delay_hist(scn, delta_us);
scn->wstats.dequeues++;
qdf_atomic_dec(&scn->wstats.q_depth);
write_val = hif_process_reg_write_q_elem(scn, q_elem);
hif_debug("read_idx %u ce_id %d offset 0x%x dequeue_val %d",
scn->read_idx, ring_id, offset, write_val);
qdf_trace_dp_del_reg_write(ring_id, q_elem->enqueue_val,
q_elem->dequeue_val,
q_elem->enqueue_time,
q_elem->dequeue_time);
num_processed++;
scn->read_idx = (scn->read_idx + 1) &
(HIF_REG_WRITE_QUEUE_LEN - 1);
q_elem = &scn->reg_write_queue[scn->read_idx];
}
hif_allow_link_low_power_states(GET_HIF_OPAQUE_HDL(scn));
/*
* Decrement active_work_cnt by the number of elements dequeued after
* hif_allow_link_low_power_states.
* This makes sure that hif_try_complete_tasks will wait till we make
* the bus access in hif_allow_link_low_power_states. This will avoid
* race condition between delayed register worker and bus suspend
* (system suspend or runtime suspend).
*
* The following decrement should be done at the end!
*/
qdf_atomic_sub(num_processed, &scn->active_work_cnt);
}
/**
* hif_delayed_reg_write_deinit() - De-Initialize delayed reg write processing
* @scn: hif_softc pointer
*
* De-initialize main data structures to process register writes in a delayed
* workqueue.
*
* Return: None
*/
static void hif_delayed_reg_write_deinit(struct hif_softc *scn)
{
qdf_flush_work(&scn->reg_write_work);
qdf_disable_work(&scn->reg_write_work);
qdf_flush_workqueue(0, scn->reg_write_wq);
qdf_destroy_workqueue(0, scn->reg_write_wq);
qdf_mem_free(scn->reg_write_queue);
}
/**
* hif_delayed_reg_write_init() - Initialization function for delayed reg writes
* @scn: hif_softc pointer
*
* Initialize main data structures to process register writes in a delayed
* workqueue.
*/
static QDF_STATUS hif_delayed_reg_write_init(struct hif_softc *scn)
{
qdf_atomic_init(&scn->active_work_cnt);
scn->reg_write_wq =
qdf_alloc_high_prior_ordered_workqueue("hif_register_write_wq");
qdf_create_work(0, &scn->reg_write_work, hif_reg_write_work, scn);
scn->reg_write_queue = qdf_mem_malloc(HIF_REG_WRITE_QUEUE_LEN *
sizeof(*scn->reg_write_queue));
if (!scn->reg_write_queue) {
hif_err("unable to allocate memory for delayed reg write");
QDF_BUG(0);
return QDF_STATUS_E_NOMEM;
}
/* Initial value of indices */
scn->read_idx = 0;
qdf_atomic_set(&scn->write_idx, -1);
return QDF_STATUS_SUCCESS;
}
static void hif_reg_write_enqueue(struct hif_softc *scn,
struct CE_state *ce_state,
uint32_t value)
{
struct hif_reg_write_q_elem *q_elem;
uint32_t write_idx;
if (ce_state->reg_write_in_progress) {
hif_debug("Already in progress ce_id %d offset 0x%x value %u",
ce_state->id, ce_state->ce_wrt_idx_offset, value);
qdf_atomic_inc(&scn->wstats.coalesces);
ce_state->wstats.coalesces++;
return;
}
write_idx = qdf_atomic_inc_return(&scn->write_idx);
write_idx = write_idx & (HIF_REG_WRITE_QUEUE_LEN - 1);
q_elem = &scn->reg_write_queue[write_idx];
if (q_elem->valid) {
hif_err("queue full");
QDF_BUG(0);
return;
}
qdf_atomic_inc(&scn->wstats.enqueues);
ce_state->wstats.enqueues++;
qdf_atomic_inc(&scn->wstats.q_depth);
q_elem->ce_state = ce_state;
q_elem->offset = ce_state->ce_wrt_idx_offset;
q_elem->enqueue_val = value;
q_elem->enqueue_time = qdf_get_log_timestamp();
/*
* Before the valid flag is set to true, all the other
* fields in the q_elem needs to be updated in memory.
* Else there is a chance that the dequeuing worker thread
* might read stale entries and process incorrect srng.
*/
qdf_wmb();
q_elem->valid = true;
/*
* After all other fields in the q_elem has been updated
* in memory successfully, the valid flag needs to be updated
* in memory in time too.
* Else there is a chance that the dequeuing worker thread
* might read stale valid flag and the work will be bypassed
* for this round. And if there is no other work scheduled
* later, this hal register writing won't be updated any more.
*/
qdf_wmb();
ce_state->reg_write_in_progress = true;
qdf_atomic_inc(&scn->active_work_cnt);
hif_debug("write_idx %u ce_id %d offset 0x%x value %u",
write_idx, ce_state->id, ce_state->ce_wrt_idx_offset, value);
qdf_queue_work(scn->qdf_dev, scn->reg_write_wq,
&scn->reg_write_work);
}
void hif_delayed_reg_write(struct hif_softc *scn, uint32_t ctrl_addr,
uint32_t val)
{
struct CE_state *ce_state;
int ce_id = COPY_ENGINE_ID(ctrl_addr);
ce_state = scn->ce_id_to_state[ce_id];
if (!ce_state->htt_tx_data && !ce_state->htt_rx_data) {
hif_reg_write_enqueue(scn, ce_state, val);
return;
}
if (hif_is_reg_write_tput_level_high(scn) ||
(PLD_MHI_STATE_L0 == pld_get_mhi_state(scn->qdf_dev->dev))) {
hal_write32_mb(scn->hal_soc, ce_state->ce_wrt_idx_offset, val);
qdf_atomic_inc(&scn->wstats.direct);
ce_state->wstats.direct++;
} else {
hif_reg_write_enqueue(scn, ce_state, val);
}
}
#else
static inline QDF_STATUS hif_delayed_reg_write_init(struct hif_softc *scn)
{
return QDF_STATUS_SUCCESS;
}
static inline void hif_delayed_reg_write_deinit(struct hif_softc *scn)
{
}
#endif
#if defined(QCA_WIFI_WCN6450)
static QDF_STATUS hif_hal_attach(struct hif_softc *scn)
{
scn->hal_soc = hal_attach(hif_softc_to_hif_opaque_softc(scn),
scn->qdf_dev);
if (!scn->hal_soc)
return QDF_STATUS_E_FAILURE;
return QDF_STATUS_SUCCESS;
}
static QDF_STATUS hif_hal_detach(struct hif_softc *scn)
{
hal_detach(scn->hal_soc);
scn->hal_soc = NULL;
return QDF_STATUS_SUCCESS;
}
#elif (defined(QCA_WIFI_QCA8074) || defined(QCA_WIFI_QCA6018) || \
defined(QCA_WIFI_QCA6290) || defined(QCA_WIFI_QCA6390) || \
defined(QCA_WIFI_QCN9000) || defined(QCA_WIFI_QCA6490) || \
defined(QCA_WIFI_QCA6750) || defined(QCA_WIFI_QCA5018) || \
defined(QCA_WIFI_KIWI) || defined(QCA_WIFI_QCN9224) || \
defined(QCA_WIFI_QCA9574)) || defined(QCA_WIFI_QCA5332)
static QDF_STATUS hif_hal_attach(struct hif_softc *scn)
{
if (ce_srng_based(scn)) {
scn->hal_soc = hal_attach(
hif_softc_to_hif_opaque_softc(scn),
scn->qdf_dev);
if (!scn->hal_soc)
return QDF_STATUS_E_FAILURE;
}
return QDF_STATUS_SUCCESS;
}
static QDF_STATUS hif_hal_detach(struct hif_softc *scn)
{
if (ce_srng_based(scn)) {
hal_detach(scn->hal_soc);
scn->hal_soc = NULL;
}
return QDF_STATUS_SUCCESS;
}
#else
static QDF_STATUS hif_hal_attach(struct hif_softc *scn)
{
return QDF_STATUS_SUCCESS;
}
static QDF_STATUS hif_hal_detach(struct hif_softc *scn)
{
return QDF_STATUS_SUCCESS;
}
#endif
int hif_init_dma_mask(struct device *dev, enum qdf_bus_type bus_type)
{
int ret;
switch (bus_type) {
case QDF_BUS_TYPE_IPCI:
ret = qdf_set_dma_coherent_mask(dev,
DMA_COHERENT_MASK_DEFAULT);
if (ret) {
hif_err("Failed to set dma mask error = %d", ret);
return ret;
}
break;
default:
/* Follow the existing sequence for other targets */
break;
}
return 0;
}
/**
* hif_enable(): hif_enable
* @hif_ctx: hif_ctx
* @dev: dev
* @bdev: bus dev
* @bid: bus ID
* @bus_type: bus type
* @type: enable type
*
* Return: QDF_STATUS
*/
QDF_STATUS hif_enable(struct hif_opaque_softc *hif_ctx, struct device *dev,
void *bdev,
const struct hif_bus_id *bid,
enum qdf_bus_type bus_type,
enum hif_enable_type type)
{
QDF_STATUS status;
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (!scn) {
hif_err("hif_ctx = NULL");
return QDF_STATUS_E_NULL_VALUE;
}
status = hif_enable_bus(scn, dev, bdev, bid, type);
if (status != QDF_STATUS_SUCCESS) {
hif_err("hif_enable_bus error = %d", status);
return status;
}
status = hif_hal_attach(scn);
if (status != QDF_STATUS_SUCCESS) {
hif_err("hal attach failed");
goto disable_bus;
}
if (hif_delayed_reg_write_init(scn) != QDF_STATUS_SUCCESS) {
hif_err("unable to initialize delayed reg write");
goto hal_detach;
}
if (hif_bus_configure(scn)) {
hif_err("Target probe failed");
status = QDF_STATUS_E_FAILURE;
goto hal_detach;
}
hif_ut_suspend_init(scn);
hif_register_recovery_notifier(scn);
hif_latency_detect_timer_start(hif_ctx);
/*
* Flag to avoid potential unallocated memory access from MSI
* interrupt handler which could get scheduled as soon as MSI
* is enabled, i.e to take care of the race due to the order
* in where MSI is enabled before the memory, that will be
* in interrupt handlers, is allocated.
*/
scn->hif_init_done = true;
hif_debug("OK");
return QDF_STATUS_SUCCESS;
hal_detach:
hif_hal_detach(scn);
disable_bus:
hif_disable_bus(scn);
return status;
}
void hif_disable(struct hif_opaque_softc *hif_ctx, enum hif_disable_type type)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (!scn)
return;
hif_delayed_reg_write_deinit(scn);
hif_set_enable_detection(hif_ctx, false);
hif_latency_detect_timer_stop(hif_ctx);
hif_unregister_recovery_notifier(scn);
hif_nointrs(scn);
if (scn->hif_init_done == false)
hif_shutdown_device(hif_ctx);
else
hif_stop(hif_ctx);
hif_hal_detach(scn);
hif_disable_bus(scn);
hif_wlan_disable(scn);
scn->notice_send = false;
hif_debug("X");
}
#ifdef CE_TASKLET_DEBUG_ENABLE
void hif_enable_ce_latency_stats(struct hif_opaque_softc *hif_ctx, uint8_t val)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (!scn)
return;
scn->ce_latency_stats = val;
}
#endif
void hif_display_stats(struct hif_opaque_softc *hif_ctx)
{
hif_display_bus_stats(hif_ctx);
}
qdf_export_symbol(hif_display_stats);
void hif_clear_stats(struct hif_opaque_softc *hif_ctx)
{
hif_clear_bus_stats(hif_ctx);
}
/**
* hif_crash_shutdown_dump_bus_register() - dump bus registers
* @hif_ctx: hif_ctx
*
* Return: n/a
*/
#if defined(TARGET_RAMDUMP_AFTER_KERNEL_PANIC) && defined(WLAN_FEATURE_BMI)
static void hif_crash_shutdown_dump_bus_register(void *hif_ctx)
{
struct hif_opaque_softc *scn = hif_ctx;
if (hif_check_soc_status(scn))
return;
if (hif_dump_registers(scn))
hif_err("Failed to dump bus registers!");
}
/**
* hif_crash_shutdown(): hif_crash_shutdown
*
* This function is called by the platform driver to dump CE registers
*
* @hif_ctx: hif_ctx
*
* Return: n/a
*/
void hif_crash_shutdown(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
if (!hif_ctx)
return;
if (scn->bus_type == QDF_BUS_TYPE_SNOC) {
hif_warn("RAM dump disabled for bustype %d", scn->bus_type);
return;
}
if (TARGET_STATUS_RESET == scn->target_status) {
hif_warn("Target is already asserted, ignore!");
return;
}
if (hif_is_load_or_unload_in_progress(scn)) {
hif_err("Load/unload is in progress, ignore!");
return;
}
hif_crash_shutdown_dump_bus_register(hif_ctx);
hif_set_target_status(hif_ctx, TARGET_STATUS_RESET);
if (ol_copy_ramdump(hif_ctx))
goto out;
hif_info("RAM dump collecting completed!");
out:
return;
}
#else
void hif_crash_shutdown(struct hif_opaque_softc *hif_ctx)
{
hif_debug("Collecting target RAM dump disabled");
}
#endif /* TARGET_RAMDUMP_AFTER_KERNEL_PANIC */
#ifdef QCA_WIFI_3_0
/**
* hif_check_fw_reg(): hif_check_fw_reg
* @scn: scn
*
* Return: int
*/
int hif_check_fw_reg(struct hif_opaque_softc *scn)
{
return 0;
}
#endif
/**
* hif_read_phy_mem_base(): hif_read_phy_mem_base
* @scn: scn
* @phy_mem_base: physical mem base
*
* Return: n/a
*/
void hif_read_phy_mem_base(struct hif_softc *scn, qdf_dma_addr_t *phy_mem_base)
{
*phy_mem_base = scn->mem_pa;
}
qdf_export_symbol(hif_read_phy_mem_base);
/**
* hif_get_device_type(): hif_get_device_type
* @device_id: device_id
* @revision_id: revision_id
* @hif_type: returned hif_type
* @target_type: returned target_type
*
* Return: int
*/
int hif_get_device_type(uint32_t device_id,
uint32_t revision_id,
uint32_t *hif_type, uint32_t *target_type)
{
int ret = 0;
switch (device_id) {
case ADRASTEA_DEVICE_ID_P2_E12:
*hif_type = HIF_TYPE_ADRASTEA;
*target_type = TARGET_TYPE_ADRASTEA;
break;
case AR9888_DEVICE_ID:
*hif_type = HIF_TYPE_AR9888;
*target_type = TARGET_TYPE_AR9888;
break;
case AR6320_DEVICE_ID:
switch (revision_id) {
case AR6320_FW_1_1:
case AR6320_FW_1_3:
*hif_type = HIF_TYPE_AR6320;
*target_type = TARGET_TYPE_AR6320;
break;
case AR6320_FW_2_0:
case AR6320_FW_3_0:
case AR6320_FW_3_2:
*hif_type = HIF_TYPE_AR6320V2;
*target_type = TARGET_TYPE_AR6320V2;
break;
default:
hif_err("dev_id = 0x%x, rev_id = 0x%x",
device_id, revision_id);
ret = -ENODEV;
goto end;
}
break;
case AR9887_DEVICE_ID:
*hif_type = HIF_TYPE_AR9888;
*target_type = TARGET_TYPE_AR9888;
hif_info(" *********** AR9887 **************");
break;
case QCA9984_DEVICE_ID:
*hif_type = HIF_TYPE_QCA9984;
*target_type = TARGET_TYPE_QCA9984;
hif_info(" *********** QCA9984 *************");
break;
case QCA9888_DEVICE_ID:
*hif_type = HIF_TYPE_QCA9888;
*target_type = TARGET_TYPE_QCA9888;
hif_info(" *********** QCA9888 *************");
break;
case AR900B_DEVICE_ID:
*hif_type = HIF_TYPE_AR900B;
*target_type = TARGET_TYPE_AR900B;
hif_info(" *********** AR900B *************");
break;
case QCA8074_DEVICE_ID:
*hif_type = HIF_TYPE_QCA8074;
*target_type = TARGET_TYPE_QCA8074;
hif_info(" *********** QCA8074 *************");
break;
case QCA6290_EMULATION_DEVICE_ID:
case QCA6290_DEVICE_ID:
*hif_type = HIF_TYPE_QCA6290;
*target_type = TARGET_TYPE_QCA6290;
hif_info(" *********** QCA6290EMU *************");
break;
case QCN9000_DEVICE_ID:
*hif_type = HIF_TYPE_QCN9000;
*target_type = TARGET_TYPE_QCN9000;
hif_info(" *********** QCN9000 *************");
break;
case QCN9224_DEVICE_ID:
*hif_type = HIF_TYPE_QCN9224;
*target_type = TARGET_TYPE_QCN9224;
hif_info(" *********** QCN9224 *************");
break;
case QCN6122_DEVICE_ID:
*hif_type = HIF_TYPE_QCN6122;
*target_type = TARGET_TYPE_QCN6122;
hif_info(" *********** QCN6122 *************");
break;
case QCN9160_DEVICE_ID:
*hif_type = HIF_TYPE_QCN9160;
*target_type = TARGET_TYPE_QCN9160;
hif_info(" *********** QCN9160 *************");
break;
case QCN6432_DEVICE_ID:
*hif_type = HIF_TYPE_QCN6432;
*target_type = TARGET_TYPE_QCN6432;
hif_info(" *********** QCN6432 *************");
break;
case QCN7605_DEVICE_ID:
case QCN7605_COMPOSITE:
case QCN7605_STANDALONE:
case QCN7605_STANDALONE_V2:
case QCN7605_COMPOSITE_V2:
*hif_type = HIF_TYPE_QCN7605;
*target_type = TARGET_TYPE_QCN7605;
hif_info(" *********** QCN7605 *************");
break;
case QCA6390_DEVICE_ID:
case QCA6390_EMULATION_DEVICE_ID:
*hif_type = HIF_TYPE_QCA6390;
*target_type = TARGET_TYPE_QCA6390;
hif_info(" *********** QCA6390 *************");
break;
case QCA6490_DEVICE_ID:
case QCA6490_EMULATION_DEVICE_ID:
*hif_type = HIF_TYPE_QCA6490;
*target_type = TARGET_TYPE_QCA6490;
hif_info(" *********** QCA6490 *************");
break;
case QCA6750_DEVICE_ID:
case QCA6750_EMULATION_DEVICE_ID:
*hif_type = HIF_TYPE_QCA6750;
*target_type = TARGET_TYPE_QCA6750;
hif_info(" *********** QCA6750 *************");
break;
case KIWI_DEVICE_ID:
*hif_type = HIF_TYPE_KIWI;
*target_type = TARGET_TYPE_KIWI;
hif_info(" *********** KIWI *************");
break;
case MANGO_DEVICE_ID:
*hif_type = HIF_TYPE_MANGO;
*target_type = TARGET_TYPE_MANGO;
hif_info(" *********** MANGO *************");
break;
case PEACH_DEVICE_ID:
*hif_type = HIF_TYPE_PEACH;
*target_type = TARGET_TYPE_PEACH;
hif_info(" *********** PEACH *************");
break;
case QCA8074V2_DEVICE_ID:
*hif_type = HIF_TYPE_QCA8074V2;
*target_type = TARGET_TYPE_QCA8074V2;
hif_info(" *********** QCA8074V2 *************");
break;
case QCA6018_DEVICE_ID:
case RUMIM2M_DEVICE_ID_NODE0:
case RUMIM2M_DEVICE_ID_NODE1:
case RUMIM2M_DEVICE_ID_NODE2:
case RUMIM2M_DEVICE_ID_NODE3:
case RUMIM2M_DEVICE_ID_NODE4:
case RUMIM2M_DEVICE_ID_NODE5:
*hif_type = HIF_TYPE_QCA6018;
*target_type = TARGET_TYPE_QCA6018;
hif_info(" *********** QCA6018 *************");
break;
case QCA5018_DEVICE_ID:
*hif_type = HIF_TYPE_QCA5018;
*target_type = TARGET_TYPE_QCA5018;
hif_info(" *********** qca5018 *************");
break;
case QCA5332_DEVICE_ID:
*hif_type = HIF_TYPE_QCA5332;
*target_type = TARGET_TYPE_QCA5332;
hif_info(" *********** QCA5332 *************");
break;
case QCA9574_DEVICE_ID:
*hif_type = HIF_TYPE_QCA9574;
*target_type = TARGET_TYPE_QCA9574;
hif_info(" *********** QCA9574 *************");
break;
case WCN6450_DEVICE_ID:
*hif_type = HIF_TYPE_WCN6450;
*target_type = TARGET_TYPE_WCN6450;
hif_info(" *********** WCN6450 *************");
break;
default:
hif_err("Unsupported device ID = 0x%x!", device_id);
ret = -ENODEV;
break;
}
if (*target_type == TARGET_TYPE_UNKNOWN) {
hif_err("Unsupported target_type!");
ret = -ENODEV;
}
end:
return ret;
}
/**
* hif_get_bus_type() - return the bus type
* @hif_hdl: HIF Context
*
* Return: enum qdf_bus_type
*/
enum qdf_bus_type hif_get_bus_type(struct hif_opaque_softc *hif_hdl)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_hdl);
return scn->bus_type;
}
/*
* Target info and ini parameters are global to the driver
* Hence these structures are exposed to all the modules in
* the driver and they don't need to maintains multiple copies
* of the same info, instead get the handle from hif and
* modify them in hif
*/
/**
* hif_get_ini_handle() - API to get hif_config_param handle
* @hif_ctx: HIF Context
*
* Return: pointer to hif_config_info
*/
struct hif_config_info *hif_get_ini_handle(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *sc = HIF_GET_SOFTC(hif_ctx);
return &sc->hif_config;
}
/**
* hif_get_target_info_handle() - API to get hif_target_info handle
* @hif_ctx: HIF context
*
* Return: Pointer to hif_target_info
*/
struct hif_target_info *hif_get_target_info_handle(
struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *sc = HIF_GET_SOFTC(hif_ctx);
return &sc->target_info;
}
qdf_export_symbol(hif_get_target_info_handle);
#ifdef RECEIVE_OFFLOAD
void hif_offld_flush_cb_register(struct hif_opaque_softc *scn,
void (offld_flush_handler)(void *))
{
if (hif_napi_enabled(scn, -1))
hif_napi_rx_offld_flush_cb_register(scn, offld_flush_handler);
else
hif_err("NAPI not enabled");
}
qdf_export_symbol(hif_offld_flush_cb_register);
void hif_offld_flush_cb_deregister(struct hif_opaque_softc *scn)
{
if (hif_napi_enabled(scn, -1))
hif_napi_rx_offld_flush_cb_deregister(scn);
else
hif_err("NAPI not enabled");
}
qdf_export_symbol(hif_offld_flush_cb_deregister);
int hif_get_rx_ctx_id(int ctx_id, struct hif_opaque_softc *hif_hdl)
{
if (hif_napi_enabled(hif_hdl, -1))
return NAPI_PIPE2ID(ctx_id);
else
return ctx_id;
}
#else /* RECEIVE_OFFLOAD */
int hif_get_rx_ctx_id(int ctx_id, struct hif_opaque_softc *hif_hdl)
{
return 0;
}
qdf_export_symbol(hif_get_rx_ctx_id);
#endif /* RECEIVE_OFFLOAD */
#if defined(FEATURE_LRO)
/**
* hif_get_lro_info - Returns LRO instance for instance ID
* @ctx_id: LRO instance ID
* @hif_hdl: HIF Context
*
* Return: Pointer to LRO instance.
*/
void *hif_get_lro_info(int ctx_id, struct hif_opaque_softc *hif_hdl)
{
void *data;
if (hif_napi_enabled(hif_hdl, -1))
data = hif_napi_get_lro_info(hif_hdl, ctx_id);
else
data = hif_ce_get_lro_ctx(hif_hdl, ctx_id);
return data;
}
#endif
/**
* hif_get_target_status - API to get target status
* @hif_ctx: HIF Context
*
* Return: enum hif_target_status
*/
enum hif_target_status hif_get_target_status(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
return scn->target_status;
}
qdf_export_symbol(hif_get_target_status);
/**
* hif_set_target_status() - API to set target status
* @hif_ctx: HIF Context
* @status: Target Status
*
* Return: void
*/
void hif_set_target_status(struct hif_opaque_softc *hif_ctx, enum
hif_target_status status)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
scn->target_status = status;
}
/**
* hif_init_ini_config() - API to initialize HIF configuration parameters
* @hif_ctx: HIF Context
* @cfg: HIF Configuration
*
* Return: void
*/
void hif_init_ini_config(struct hif_opaque_softc *hif_ctx,
struct hif_config_info *cfg)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
qdf_mem_copy(&scn->hif_config, cfg, sizeof(struct hif_config_info));
}
/**
* hif_get_conparam() - API to get driver mode in HIF
* @scn: HIF Context
*
* Return: driver mode of operation
*/
uint32_t hif_get_conparam(struct hif_softc *scn)
{
if (!scn)
return 0;
return scn->hif_con_param;
}
/**
* hif_get_callbacks_handle() - API to get callbacks Handle
* @scn: HIF Context
*
* Return: pointer to HIF Callbacks
*/
struct hif_driver_state_callbacks *hif_get_callbacks_handle(
struct hif_softc *scn)
{
return &scn->callbacks;
}
/**
* hif_is_driver_unloading() - API to query upper layers if driver is unloading
* @scn: HIF Context
*
* Return: True/False
*/
bool hif_is_driver_unloading(struct hif_softc *scn)
{
struct hif_driver_state_callbacks *cbk = hif_get_callbacks_handle(scn);
if (cbk && cbk->is_driver_unloading)
return cbk->is_driver_unloading(cbk->context);
return false;
}
/**
* hif_is_load_or_unload_in_progress() - API to query upper layers if
* load/unload in progress
* @scn: HIF Context
*
* Return: True/False
*/
bool hif_is_load_or_unload_in_progress(struct hif_softc *scn)
{
struct hif_driver_state_callbacks *cbk = hif_get_callbacks_handle(scn);
if (cbk && cbk->is_load_unload_in_progress)
return cbk->is_load_unload_in_progress(cbk->context);
return false;
}
/**
* hif_is_recovery_in_progress() - API to query upper layers if recovery in
* progress
* @scn: HIF Context
*
* Return: True/False
*/
bool hif_is_recovery_in_progress(struct hif_softc *scn)
{
struct hif_driver_state_callbacks *cbk = hif_get_callbacks_handle(scn);
if (cbk && cbk->is_recovery_in_progress)
return cbk->is_recovery_in_progress(cbk->context);
return false;
}
#if defined(HIF_PCI) || defined(HIF_SNOC) || defined(HIF_AHB) || \
defined(HIF_IPCI)
/**
* hif_update_pipe_callback() - API to register pipe specific callbacks
* @osc: Opaque softc
* @pipeid: pipe id
* @callbacks: callbacks to register
*
* Return: void
*/
void hif_update_pipe_callback(struct hif_opaque_softc *osc,
u_int8_t pipeid,
struct hif_msg_callbacks *callbacks)
{
struct hif_softc *scn = HIF_GET_SOFTC(osc);
struct HIF_CE_state *hif_state = HIF_GET_CE_STATE(scn);
struct HIF_CE_pipe_info *pipe_info;
QDF_BUG(pipeid < CE_COUNT_MAX);
hif_debug("pipeid: %d", pipeid);
pipe_info = &hif_state->pipe_info[pipeid];
qdf_mem_copy(&pipe_info->pipe_callbacks,
callbacks, sizeof(pipe_info->pipe_callbacks));
}
qdf_export_symbol(hif_update_pipe_callback);
/**
* hif_is_target_ready() - API to query if target is in ready state
* progress
* @scn: HIF Context
*
* Return: True/False
*/
bool hif_is_target_ready(struct hif_softc *scn)
{
struct hif_driver_state_callbacks *cbk = hif_get_callbacks_handle(scn);
if (cbk && cbk->is_target_ready)
return cbk->is_target_ready(cbk->context);
/*
* if callback is not registered then there is no way to determine
* if target is ready. In-such case return true to indicate that
* target is ready.
*/
return true;
}
qdf_export_symbol(hif_is_target_ready);
int hif_get_bandwidth_level(struct hif_opaque_softc *hif_handle)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_handle);
struct hif_driver_state_callbacks *cbk = hif_get_callbacks_handle(scn);
if (cbk && cbk->get_bandwidth_level)
return cbk->get_bandwidth_level(cbk->context);
return 0;
}
qdf_export_symbol(hif_get_bandwidth_level);
#ifdef DP_MEM_PRE_ALLOC
void *hif_mem_alloc_consistent_unaligned(struct hif_softc *scn,
qdf_size_t size,
qdf_dma_addr_t *paddr,
uint32_t ring_type,
uint8_t *is_mem_prealloc)
{
void *vaddr = NULL;
struct hif_driver_state_callbacks *cbk =
hif_get_callbacks_handle(scn);
*is_mem_prealloc = false;
if (cbk && cbk->prealloc_get_consistent_mem_unaligned) {
vaddr = cbk->prealloc_get_consistent_mem_unaligned(size,
paddr,
ring_type);
if (vaddr) {
*is_mem_prealloc = true;
goto end;
}
}
vaddr = qdf_mem_alloc_consistent(scn->qdf_dev,
scn->qdf_dev->dev,
size,
paddr);
end:
dp_info("%s va_unaligned %pK pa_unaligned %pK size %d ring_type %d",
*is_mem_prealloc ? "pre-alloc" : "dynamic-alloc", vaddr,
(void *)*paddr, (int)size, ring_type);
return vaddr;
}
void hif_mem_free_consistent_unaligned(struct hif_softc *scn,
qdf_size_t size,
void *vaddr,
qdf_dma_addr_t paddr,
qdf_dma_context_t memctx,
uint8_t is_mem_prealloc)
{
struct hif_driver_state_callbacks *cbk =
hif_get_callbacks_handle(scn);
if (is_mem_prealloc) {
if (cbk && cbk->prealloc_put_consistent_mem_unaligned) {
cbk->prealloc_put_consistent_mem_unaligned(vaddr);
} else {
dp_warn("dp_prealloc_put_consistent_unligned NULL");
QDF_BUG(0);
}
} else {
qdf_mem_free_consistent(scn->qdf_dev, scn->qdf_dev->dev,
size, vaddr, paddr, memctx);
}
}
void hif_prealloc_get_multi_pages(struct hif_softc *scn, uint32_t desc_type,
qdf_size_t elem_size, uint16_t elem_num,
struct qdf_mem_multi_page_t *pages,
bool cacheable)
{
struct hif_driver_state_callbacks *cbk =
hif_get_callbacks_handle(scn);
if (cbk && cbk->prealloc_get_multi_pages)
cbk->prealloc_get_multi_pages(desc_type, elem_size, elem_num,
pages, cacheable);
if (!pages->num_pages)
qdf_mem_multi_pages_alloc(scn->qdf_dev, pages,
elem_size, elem_num, 0, cacheable);
}
void hif_prealloc_put_multi_pages(struct hif_softc *scn, uint32_t desc_type,
struct qdf_mem_multi_page_t *pages,
bool cacheable)
{
struct hif_driver_state_callbacks *cbk =
hif_get_callbacks_handle(scn);
if (cbk && cbk->prealloc_put_multi_pages &&
pages->is_mem_prealloc)
cbk->prealloc_put_multi_pages(desc_type, pages);
if (!pages->is_mem_prealloc)
qdf_mem_multi_pages_free(scn->qdf_dev, pages, 0,
cacheable);
}
#endif
/**
* hif_batch_send() - API to access hif specific function
* ce_batch_send.
* @osc: HIF Context
* @msdu: list of msdus to be sent
* @transfer_id: transfer id
* @len: downloaded length
* @sendhead:
*
* Return: list of msds not sent
*/
qdf_nbuf_t hif_batch_send(struct hif_opaque_softc *osc, qdf_nbuf_t msdu,
uint32_t transfer_id, u_int32_t len, uint32_t sendhead)
{
void *ce_tx_hdl = hif_get_ce_handle(osc, CE_HTT_TX_CE);
if (!ce_tx_hdl)
return NULL;
return ce_batch_send((struct CE_handle *)ce_tx_hdl, msdu, transfer_id,
len, sendhead);
}
qdf_export_symbol(hif_batch_send);
/**
* hif_update_tx_ring() - API to access hif specific function
* ce_update_tx_ring.
* @osc: HIF Context
* @num_htt_cmpls: number of htt compl received.
*
* Return: void
*/
void hif_update_tx_ring(struct hif_opaque_softc *osc, u_int32_t num_htt_cmpls)
{
void *ce_tx_hdl = hif_get_ce_handle(osc, CE_HTT_TX_CE);
ce_update_tx_ring(ce_tx_hdl, num_htt_cmpls);
}
qdf_export_symbol(hif_update_tx_ring);
/**
* hif_send_single() - API to access hif specific function
* ce_send_single.
* @osc: HIF Context
* @msdu : msdu to be sent
* @transfer_id: transfer id
* @len : downloaded length
*
* Return: msdu sent status
*/
QDF_STATUS hif_send_single(struct hif_opaque_softc *osc, qdf_nbuf_t msdu,
uint32_t transfer_id, u_int32_t len)
{
void *ce_tx_hdl = hif_get_ce_handle(osc, CE_HTT_TX_CE);
if (!ce_tx_hdl)
return QDF_STATUS_E_NULL_VALUE;
return ce_send_single((struct CE_handle *)ce_tx_hdl, msdu, transfer_id,
len);
}
qdf_export_symbol(hif_send_single);
#endif
/**
* hif_reg_write() - API to access hif specific function
* hif_write32_mb.
* @hif_ctx : HIF Context
* @offset : offset on which value has to be written
* @value : value to be written
*
* Return: None
*/
void hif_reg_write(struct hif_opaque_softc *hif_ctx, uint32_t offset,
uint32_t value)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
hif_write32_mb(scn, scn->mem + offset, value);
}
qdf_export_symbol(hif_reg_write);
/**
* hif_reg_read() - API to access hif specific function
* hif_read32_mb.
* @hif_ctx : HIF Context
* @offset : offset from which value has to be read
*
* Return: Read value
*/
uint32_t hif_reg_read(struct hif_opaque_softc *hif_ctx, uint32_t offset)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
return hif_read32_mb(scn, scn->mem + offset);
}
qdf_export_symbol(hif_reg_read);
/**
* hif_ramdump_handler(): generic ramdump handler
* @scn: struct hif_opaque_softc
*
* Return: None
*/
void hif_ramdump_handler(struct hif_opaque_softc *scn)
{
if (hif_get_bus_type(scn) == QDF_BUS_TYPE_USB)
hif_usb_ramdump_handler(scn);
}
hif_pm_wake_irq_type hif_pm_get_wake_irq_type(struct hif_opaque_softc *hif_ctx)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
return scn->wake_irq_type;
}
irqreturn_t hif_wake_interrupt_handler(int irq, void *context)
{
struct hif_softc *scn = context;
hif_info("wake interrupt received on irq %d", irq);
hif_rtpm_set_monitor_wake_intr(0);
hif_rtpm_request_resume();
if (scn->initial_wakeup_cb)
scn->initial_wakeup_cb(scn->initial_wakeup_priv);
if (hif_is_ut_suspended(scn))
hif_ut_fw_resume(scn);
qdf_pm_system_wakeup();
return IRQ_HANDLED;
}
void hif_set_initial_wakeup_cb(struct hif_opaque_softc *hif_ctx,
void (*callback)(void *),
void *priv)
{
struct hif_softc *scn = HIF_GET_SOFTC(hif_ctx);
scn->initial_wakeup_cb = callback;
scn->initial_wakeup_priv = priv;
}
void hif_set_ce_service_max_yield_time(struct hif_opaque_softc *hif,
uint32_t ce_service_max_yield_time)
{
struct hif_softc *hif_ctx = HIF_GET_SOFTC(hif);
hif_ctx->ce_service_max_yield_time =
ce_service_max_yield_time * 1000;
}
unsigned long long
hif_get_ce_service_max_yield_time(struct hif_opaque_softc *hif)
{
struct hif_softc *hif_ctx = HIF_GET_SOFTC(hif);
return hif_ctx->ce_service_max_yield_time;
}
void hif_set_ce_service_max_rx_ind_flush(struct hif_opaque_softc *hif,
uint8_t ce_service_max_rx_ind_flush)
{
struct hif_softc *hif_ctx = HIF_GET_SOFTC(hif);
if (ce_service_max_rx_ind_flush == 0 ||
ce_service_max_rx_ind_flush > MSG_FLUSH_NUM)
hif_ctx->ce_service_max_rx_ind_flush = MSG_FLUSH_NUM;
else
hif_ctx->ce_service_max_rx_ind_flush =
ce_service_max_rx_ind_flush;
}
#ifdef SYSTEM_PM_CHECK
void __hif_system_pm_set_state(struct hif_opaque_softc *hif,
enum hif_system_pm_state state)
{
struct hif_softc *hif_ctx = HIF_GET_SOFTC(hif);
qdf_atomic_set(&hif_ctx->sys_pm_state, state);
}
int32_t hif_system_pm_get_state(struct hif_opaque_softc *hif)
{
struct hif_softc *hif_ctx = HIF_GET_SOFTC(hif);
return qdf_atomic_read(&hif_ctx->sys_pm_state);
}
int hif_system_pm_state_check(struct hif_opaque_softc *hif)
{
struct hif_softc *hif_ctx = HIF_GET_SOFTC(hif);
int32_t sys_pm_state;
if (!hif_ctx) {
hif_err("hif context is null");
return -EFAULT;
}
sys_pm_state = qdf_atomic_read(&hif_ctx->sys_pm_state);
if (sys_pm_state == HIF_SYSTEM_PM_STATE_BUS_SUSPENDING ||
sys_pm_state == HIF_SYSTEM_PM_STATE_BUS_SUSPENDED) {
hif_info("Triggering system wakeup");
qdf_pm_system_wakeup();
return -EAGAIN;
}
return 0;
}
#endif
#ifdef WLAN_FEATURE_AFFINITY_MGR
/*
* hif_audio_cpu_affinity_allowed() - Check if audio cpu affinity allowed
*
* @scn: hif handle
* @cfg: hif affinity manager configuration for IRQ
* @audio_taken_cpu: Current CPUs which are taken by audio.
* @current_time: Current system time.
*
* This API checks for 2 conditions
* 1) Last audio taken mask and current taken mask are different
* 2) Last time when IRQ was affined away due to audio taken CPUs is
* more than time threshold (5 Seconds in current case).
* If both condition satisfies then only return true.
*
* Return: bool: true if it is allowed to affine away audio taken cpus.
*/
static inline bool
hif_audio_cpu_affinity_allowed(struct hif_softc *scn,
struct hif_cpu_affinity *cfg,
qdf_cpu_mask audio_taken_cpu,
uint64_t current_time)
{
if (!qdf_cpumask_equal(&audio_taken_cpu, &cfg->walt_taken_mask) &&
(qdf_log_timestamp_to_usecs(current_time -
cfg->last_affined_away)
< scn->time_threshold))
return false;
return true;
}
/*
* hif_affinity_mgr_check_update_mask() - Check if cpu mask need to be updated
*
* @scn: hif handle
* @cfg: hif affinity manager configuration for IRQ
* @audio_taken_cpu: Current CPUs which are taken by audio.
* @cpu_mask: CPU mask which need to be updated.
* @current_time: Current system time.
*
* This API checks if Pro audio use case is running and if cpu_mask need
* to be updated
*
* Return: QDF_STATUS
*/
static inline QDF_STATUS
hif_affinity_mgr_check_update_mask(struct hif_softc *scn,
struct hif_cpu_affinity *cfg,
qdf_cpu_mask audio_taken_cpu,
qdf_cpu_mask *cpu_mask,
uint64_t current_time)
{
qdf_cpu_mask allowed_mask;
/*
* Case 1: audio_taken_mask is empty
* Check if passed cpu_mask and wlan_requested_mask is same or not.
* If both mask are different copy wlan_requested_mask(IRQ affinity
* mask requested by WLAN) to cpu_mask.
*
* Case 2: audio_taken_mask is not empty
* 1. Only allow update if last time when IRQ was affined away due to
* audio taken CPUs is more than 5 seconds or update is requested
* by WLAN
* 2. Only allow silver cores to be affined away.
* 3. Check if any allowed CPUs for audio use case is set in cpu_mask.
* i. If any CPU mask is set, mask out that CPU from the cpu_mask
* ii. If after masking out audio taken cpu(Silver cores) cpu_mask
* is empty, set mask to all cpu except cpus taken by audio.
* Example:
*| Audio mask | mask allowed | cpu_mask | WLAN req mask | new cpu_mask|
*| 0x00 | 0x00 | 0x0C | 0x0C | 0x0C |
*| 0x00 | 0x00 | 0x03 | 0x03 | 0x03 |
*| 0x00 | 0x00 | 0xFC | 0x03 | 0x03 |
*| 0x00 | 0x00 | 0x03 | 0x0C | 0x0C |
*| 0x0F | 0x03 | 0x0C | 0x0C | 0x0C |
*| 0x0F | 0x03 | 0x03 | 0x03 | 0xFC |
*| 0x03 | 0x03 | 0x0C | 0x0C | 0x0C |
*| 0x03 | 0x03 | 0x03 | 0x03 | 0xFC |
*| 0x03 | 0x03 | 0xFC | 0x03 | 0xFC |
*| 0xF0 | 0x00 | 0x0C | 0x0C | 0x0C |
*| 0xF0 | 0x00 | 0x03 | 0x03 | 0x03 |
*/
/* Check if audio taken mask is empty*/
if (qdf_likely(qdf_cpumask_empty(&audio_taken_cpu))) {
/* If CPU mask requested by WLAN for the IRQ and
* cpu_mask passed CPU mask set for IRQ is different
* Copy requested mask into cpu_mask and return
*/
if (qdf_unlikely(!qdf_cpumask_equal(cpu_mask,
&cfg->wlan_requested_mask))) {
qdf_cpumask_copy(cpu_mask, &cfg->wlan_requested_mask);
return QDF_STATUS_SUCCESS;
}
return QDF_STATUS_E_ALREADY;
}
if (!(hif_audio_cpu_affinity_allowed(scn, cfg, audio_taken_cpu,
current_time) ||
cfg->update_requested))
return QDF_STATUS_E_AGAIN;
/* Only allow Silver cores to be affine away */
qdf_cpumask_and(&allowed_mask, &scn->allowed_mask, &audio_taken_cpu);
if (qdf_cpumask_intersects(cpu_mask, &allowed_mask)) {
/* If any of taken CPU(Silver cores) mask is set in cpu_mask,
* mask out the audio taken CPUs from the cpu_mask.
*/
qdf_cpumask_andnot(cpu_mask, &cfg->wlan_requested_mask,
&allowed_mask);
/* If cpu_mask is empty set it to all CPUs
* except taken by audio(Silver cores)
*/
if (qdf_unlikely(qdf_cpumask_empty(cpu_mask)))
qdf_cpumask_complement(cpu_mask, &allowed_mask);
return QDF_STATUS_SUCCESS;
}
return QDF_STATUS_E_ALREADY;
}
static inline QDF_STATUS
hif_check_and_affine_irq(struct hif_softc *scn, struct hif_cpu_affinity *cfg,
qdf_cpu_mask audio_taken_cpu, qdf_cpu_mask cpu_mask,
uint64_t current_time)
{
QDF_STATUS status;
status = hif_affinity_mgr_check_update_mask(scn, cfg,
audio_taken_cpu,
&cpu_mask,
current_time);
/* Set IRQ affinity if CPU mask was updated */
if (QDF_IS_STATUS_SUCCESS(status)) {
status = hif_irq_set_affinity_hint(cfg->irq,
&cpu_mask);
if (QDF_IS_STATUS_SUCCESS(status)) {
/* Store audio taken CPU mask */
qdf_cpumask_copy(&cfg->walt_taken_mask,
&audio_taken_cpu);
/* Store CPU mask which was set for IRQ*/
qdf_cpumask_copy(&cfg->current_irq_mask,
&cpu_mask);
/* Set time when IRQ affinity was updated */
cfg->last_updated = current_time;
if (hif_audio_cpu_affinity_allowed(scn, cfg,
audio_taken_cpu,
current_time))
/* If CPU mask was updated due to CPU
* taken by audio, update
* last_affined_away time
*/
cfg->last_affined_away = current_time;
}
}
return status;
}
void hif_affinity_mgr_affine_irq(struct hif_softc *scn)
{
bool audio_affinity_allowed = false;
int i, j, ce_id;
uint64_t current_time;
char cpu_str[10];
QDF_STATUS status;
qdf_cpu_mask cpu_mask, audio_taken_cpu;
struct HIF_CE_state *hif_state;
struct hif_exec_context *hif_ext_group;
struct CE_attr *host_ce_conf;
struct HIF_CE_state *ce_sc;
struct hif_cpu_affinity *cfg;
if (!scn->affinity_mgr_supported)
return;
current_time = hif_get_log_timestamp();
/* Get CPU mask for audio taken CPUs */
audio_taken_cpu = qdf_walt_get_cpus_taken();
ce_sc = HIF_GET_CE_STATE(scn);
host_ce_conf = ce_sc->host_ce_config;
for (ce_id = 0; ce_id < scn->ce_count; ce_id++) {
if (host_ce_conf[ce_id].flags & CE_ATTR_DISABLE_INTR)
continue;
cfg = &scn->ce_irq_cpu_mask[ce_id];
qdf_cpumask_copy(&cpu_mask, &cfg->current_irq_mask);
status =
hif_check_and_affine_irq(scn, cfg, audio_taken_cpu,
cpu_mask, current_time);
if (QDF_IS_STATUS_SUCCESS(status))
audio_affinity_allowed = true;
}
hif_state = HIF_GET_CE_STATE(scn);
for (i = 0; i < hif_state->hif_num_extgroup; i++) {
hif_ext_group = hif_state->hif_ext_group[i];
for (j = 0; j < hif_ext_group->numirq; j++) {
cfg = &scn->irq_cpu_mask[hif_ext_group->grp_id][j];
qdf_cpumask_copy(&cpu_mask, &cfg->current_irq_mask);
status =
hif_check_and_affine_irq(scn, cfg, audio_taken_cpu,
cpu_mask, current_time);
if (QDF_IS_STATUS_SUCCESS(status)) {
qdf_atomic_set(&hif_ext_group->force_napi_complete, -1);
audio_affinity_allowed = true;
}
}
}
if (audio_affinity_allowed) {
qdf_thread_cpumap_print_to_pagebuf(false, cpu_str,
&audio_taken_cpu);
hif_info("Audio taken CPU mask: %s", cpu_str);
}
}
static inline QDF_STATUS
hif_affinity_mgr_set_irq_affinity(struct hif_softc *scn, uint32_t irq,
struct hif_cpu_affinity *cfg,
qdf_cpu_mask *cpu_mask)
{
uint64_t current_time;
char cpu_str[10];
QDF_STATUS status, mask_updated;
qdf_cpu_mask audio_taken_cpu = qdf_walt_get_cpus_taken();
current_time = hif_get_log_timestamp();
qdf_cpumask_copy(&cfg->wlan_requested_mask, cpu_mask);
cfg->update_requested = true;
mask_updated = hif_affinity_mgr_check_update_mask(scn, cfg,
audio_taken_cpu,
cpu_mask,
current_time);
status = hif_irq_set_affinity_hint(irq, cpu_mask);
if (QDF_IS_STATUS_SUCCESS(status)) {
qdf_cpumask_copy(&cfg->walt_taken_mask, &audio_taken_cpu);
qdf_cpumask_copy(&cfg->current_irq_mask, cpu_mask);
if (QDF_IS_STATUS_SUCCESS(mask_updated)) {
cfg->last_updated = current_time;
if (hif_audio_cpu_affinity_allowed(scn, cfg,
audio_taken_cpu,
current_time)) {
cfg->last_affined_away = current_time;
qdf_thread_cpumap_print_to_pagebuf(false,
cpu_str,
&audio_taken_cpu);
hif_info_rl("Audio taken CPU mask: %s",
cpu_str);
}
}
}
cfg->update_requested = false;
return status;
}
QDF_STATUS
hif_affinity_mgr_set_qrg_irq_affinity(struct hif_softc *scn, uint32_t irq,
uint32_t grp_id, uint32_t irq_index,
qdf_cpu_mask *cpu_mask)
{
struct hif_cpu_affinity *cfg;
if (!scn->affinity_mgr_supported)
return hif_irq_set_affinity_hint(irq, cpu_mask);
cfg = &scn->irq_cpu_mask[grp_id][irq_index];
return hif_affinity_mgr_set_irq_affinity(scn, irq, cfg, cpu_mask);
}
QDF_STATUS
hif_affinity_mgr_set_ce_irq_affinity(struct hif_softc *scn, uint32_t irq,
uint32_t ce_id, qdf_cpu_mask *cpu_mask)
{
struct hif_cpu_affinity *cfg;
if (!scn->affinity_mgr_supported)
return hif_irq_set_affinity_hint(irq, cpu_mask);
cfg = &scn->ce_irq_cpu_mask[ce_id];
return hif_affinity_mgr_set_irq_affinity(scn, irq, cfg, cpu_mask);
}
void
hif_affinity_mgr_init_ce_irq(struct hif_softc *scn, int id, int irq)
{
unsigned int cpus;
qdf_cpu_mask cpu_mask = {0};
struct hif_cpu_affinity *cfg = NULL;
if (!scn->affinity_mgr_supported)
return;
/* Set CPU Mask to Silver core */
qdf_for_each_possible_cpu(cpus)
if (qdf_topology_physical_package_id(cpus) ==
CPU_CLUSTER_TYPE_LITTLE)
qdf_cpumask_set_cpu(cpus, &cpu_mask);
cfg = &scn->ce_irq_cpu_mask[id];
qdf_cpumask_copy(&cfg->current_irq_mask, &cpu_mask);
qdf_cpumask_copy(&cfg->wlan_requested_mask, &cpu_mask);
cfg->irq = irq;
cfg->last_updated = 0;
cfg->last_affined_away = 0;
cfg->update_requested = false;
}
void
hif_affinity_mgr_init_grp_irq(struct hif_softc *scn, int grp_id,
int irq_num, int irq)
{
unsigned int cpus;
qdf_cpu_mask cpu_mask = {0};
struct hif_cpu_affinity *cfg = NULL;
if (!scn->affinity_mgr_supported)
return;
/* Set CPU Mask to Silver core */
qdf_for_each_possible_cpu(cpus)
if (qdf_topology_physical_package_id(cpus) ==
CPU_CLUSTER_TYPE_LITTLE)
qdf_cpumask_set_cpu(cpus, &cpu_mask);
cfg = &scn->irq_cpu_mask[grp_id][irq_num];
qdf_cpumask_copy(&cfg->current_irq_mask, &cpu_mask);
qdf_cpumask_copy(&cfg->wlan_requested_mask, &cpu_mask);
cfg->irq = irq;
cfg->last_updated = 0;
cfg->last_affined_away = 0;
cfg->update_requested = false;
}
#endif