bes2600: recover wedged firmware via mmc_hw_reset on link break

When the LMAC active monitor detects 'link break between lmac and host'
(the hw_buf_used==pending watchdog in bes2600_bh_lmac_active_monitor),
bes2600_chrdev_wifi_force_close(hw_priv, true) is invoked to tear the
device down and prepare for a fresh probe. On the wifi_force_close_work
side this calls bes2600_chrdev_do_system_close() which dispatches
sbus_ops->power_switch(0).

On PineTab2 (RK3566 + BES2600WM over SDIO) this recovery path is a
no-op:

  * bes2600_sdio_power_down() writes a SYSTEM_CLOSE host-int message,
    clears MMC_CAP_NONREMOVABLE, and schedules sdio_scan_work, which is
    the literal one-line stub bes_warn("...this function does
    nothing\n").
  * bes2600_sdio_on() (the eventual power_switch(1) counterpart)
    toggles pdata->powerup, which is NULL on PineTab2 because the
    wifi-reset GPIO is owned by sdio_pwrseq, not the bes2600 device
    tree node (see arch/arm64/boot/dts/rockchip/rk3566-pinetab2.dtsi:
    'The reset pin is claimed by sdio_mmcseq, It is better to move it
    to U-Boot so the OS can use it.').

Net result: the chip is never reset. The function drivers are not
removed (the SDIO core has no signal that the card is gone), the
firmware stays wedged, and a subsequent rmmod bes2600 leaves the SDIO
function in a half-torn-down state. modprobe bes2600 then fails with
'probe with driver bes2600_wlan failed with error -123' (-ENOMEDIUM)
on both functions (:1 wifi, :2 BT-companion) until a full system
reboot.

Observed on PineTab2 (linux-pinetab2 6.19.10-danctnix1-1) after ~150
minutes of background-scan rejects (wsm_generic_confirm 0x0007,
[SCAN] Scan failed (-22)) accumulating until the LMAC stopped
acknowledging TX buffers (hw_buf_used:24 pending:24). Reproducible
under sustained scan pressure.

Add a sbus operation bus_reset() that the recovery path can call when
power_switch() has no effective chip-reset signal of its own. Provide
an SDIO implementation that calls mmc_hw_reset(self->func->card),
which on a multi-function SDIO card (PineTab2 binds func 1 for WLAN
and func 2 for the BT-companion path) takes the remove-and-rescan
path: mmc_sdio_hw_reset() marks the card removed and schedules
mmc_rescan, which tears down the bound function drivers and re-detects
the card on the next sweep, in turn reinvoking bes2600_sdio_probe().
With a single function probed it instead invokes mmc_power_cycle()
directly, which on PineTab2 toggles the wifi-reset GPIO via
sdio_pwrseq.

Add bes2600_chrdev_do_bus_reset() as the chrdev-side helper. It
invokes the bus op and then waits on probe_done_wq for the SDIO
remove() callback to clear sbus_priv, mirroring the wait pattern
already used by bes2600_chrdev_do_system_close() so that a subsequent
bes2600_switch_wifi(true) sees a clean state and can wait on the
fresh probe.

Wire it into bes2600_chrdev_wifi_force_close_work(): when halt_dev is
set (the hard-exception path used by both
bes2600_bh_lmac_active_monitor and bes2600_bh_mcu_active_monitor) and
the underlying bus implements bus_reset, take the new recovery path;
otherwise fall back to the legacy power_switch(0) sequence so this
patch is a no-op on USB or any other future bus that does not provide
bus_reset.

mmc_hw_reset() is exported by the MMC core and is the canonical
recovery primitive; calling it without holding the SDIO host claim is
correct because the multi-func remove-and-rescan path acquires the
host claim via the mmc workqueue, and the single-func mmc_power_cycle
path does not require the host claim.

No DT change is required: this works against the existing PineTab2
DTS, where the wifi-reset GPIO and the optional sdio_pwrkey GPIO (on
v2.0 boards) are both already configured as MMC pwrseq resets.

Signed-off-by: Markus Fritsche <fritsche.markus@gmail.com>

bes2600/bes2600_sdio.c

@@ -16,6 +16,7 @@
 #include <linux/mmc/host.h>
 #include <linux/mmc/sdio_func.h>
 #include <linux/mmc/card.h>
+#include <linux/mmc/core.h>
 #include <linux/mmc/sdio.h>
 #include <linux/spinlock.h>
 #include <net/mac80211.h>
@@ -1777,6 +1778,33 @@ static void bes2600_sdio_halt_device(struct sbus_priv *self)
 	sdio_work_debug(self);
 }
 
+/*
+ * Trigger an SDIO bus reset via mmc_hw_reset().
+ *
+ * With multiple SDIO functions probed (PineTab2 binds func 1 for WLAN and
+ * func 2 for the BT-companion path) mmc_sdio_hw_reset() takes the
+ * remove-and-rescan path: it marks the card removed and schedules
+ * mmc_rescan, which tears down the bound function drivers and re-detects
+ * the card on the next sweep, in turn reinvoking bes2600_sdio_probe().
+ *
+ * With a single function probed it instead invokes mmc_power_cycle()
+ * directly, which on PineTab2 toggles the wifi-reset GPIO via sdio_pwrseq.
+ *
+ * In both cases the chip ends up in a freshly reset state, which is the
+ * goal of the recovery path.
+ *
+ * mmc_hw_reset() must be called without holding the SDIO host claim --
+ * the multi-func remove-and-rescan path acquires the host claim via the
+ * mmc workqueue.
+ */
+static int bes2600_sdio_bus_reset(struct sbus_priv *self)
+{
+	if (!self || !self->func || !self->func->card)
+		return -EINVAL;
+
+	return mmc_hw_reset(self->func->card);
+}
+
 static bool bes2600_sdio_wakeup_source(struct sbus_priv *self)
 {
 	struct bes2600_platform_data_sdio *pdata = bes2600_get_platform_data();
@@ -1815,6 +1843,7 @@ static struct sbus_ops bes2600_sdio_sbus_ops = {
 	.gpio_sleep		= bes2600_gpio_allow_mcu_sleep,
 	.halt_device		= bes2600_sdio_halt_device,
 	.wakeup_source		= bes2600_sdio_wakeup_source,
+	.bus_reset		= bes2600_sdio_bus_reset,
 };
 
 static void bes2600_sdio_en_lp_cb(struct bes2600_common *hw_priv)

bes2600/bes_chardev.c

@@ -442,6 +442,48 @@ int bes2600_chrdev_do_system_close(const struct sbus_ops *sbus_ops, struct sbus_
 	return ret;
 }
 
+/*
+ * Hard-reset the bus and wait for the bus core to remove the chip.
+ *
+ * Used by the firmware-wedge recovery path on platforms where the normal
+ * power_switch(0) sequence has no effective chip-reset signal. The bus
+ * implementation triggers an asynchronous re-detect; this helper waits for
+ * the resulting remove() callback to clear bes2600_cdev.sbus_priv so that a
+ * subsequent bes2600_switch_wifi(true) sees a clean state and can wait on
+ * the fresh probe.
+ */
+int bes2600_chrdev_do_bus_reset(const struct sbus_ops *sbus_ops, struct sbus_priv *priv)
+{
+	int ret;
+	long status;
+
+	if (!sbus_ops || !priv)
+		return -EINVAL;
+
+	if (!sbus_ops->bus_reset)
+		return -EOPNOTSUPP;
+
+	bes_info("trigger bus reset to recover wedged firmware.\n");
+
+	ret = sbus_ops->bus_reset(priv);
+	if (ret) {
+		bes_err("bus_reset failed: %d\n", ret);
+		return ret;
+	}
+
+	/*
+	 * The bus reset is asynchronous: the bus core schedules a rescan
+	 * which removes the bound function drivers and then re-detects the
+	 * chip. Wait for the remove callback to clear sbus_priv. Do not
+	 * dereference 'priv' after this point -- it may already be freed.
+	 */
+	status = wait_event_timeout(bes2600_cdev.probe_done_wq,
+				    !bes2600_cdev.sbus_priv, HZ * 3);
+	WARN_ON(status <= 0);
+
+	return 0;
+}
+
 bool bes2600_chrdev_is_wifi_opened(void)
 {
 	bool wifi_opened = false;
@@ -540,8 +582,21 @@ static void bes2600_chrdev_wifi_force_close_work(struct work_struct *work)
 		/* unregister wifi */
 		bes2600_switch_wifi(0);
 
-		/* power down device if wifi is only opened */
-		if (bes2600_chrdev_check_system_close()) {
+		/*
+		 * Hard exception with a bus_reset implementation: tear the
+		 * bus down via mmc_hw_reset() (or equivalent) so the next
+		 * bringup probes a freshly reset chip. On PineTab2 this is
+		 * the only effective recovery path -- the existing
+		 * power_switch(0)/(1) sequence has no chip-reset signal of
+		 * its own (sdio_pwrseq owns wifi_reset).
+		 *
+		 * Soft close, or hard close on a board without bus_reset:
+		 * fall back to the legacy power_switch(0) sequence.
+		 */
+		if (bes2600_cdev.halt_dev && bes2600_cdev.sbus_ops->bus_reset) {
+			bes2600_chrdev_do_bus_reset(bes2600_cdev.sbus_ops,
+						    bes2600_cdev.sbus_priv);
+		} else if (bes2600_chrdev_check_system_close()) {
 			bes2600_chrdev_do_system_close(bes2600_cdev.sbus_ops,
 						bes2600_cdev.sbus_priv);
 		}

bes2600/bes_chardev.h

@@ -60,6 +60,7 @@ struct sbus_priv *bes2600_chrdev_get_sbus_priv_data(void);
 /* used to control device power down */
 int bes2600_chrdev_check_system_close(void);
 int bes2600_chrdev_do_system_close(const struct sbus_ops *sbus_ops, struct sbus_priv *priv);
+int bes2600_chrdev_do_bus_reset(const struct sbus_ops *sbus_ops, struct sbus_priv *priv);
 void bes2600_chrdev_wakeup_bt(void);
 void bes2600_chrdev_wifi_force_close(struct bes2600_common *hw_priv, bool halt_dev);
 void bes2600_chrdev_usb_remove(struct bes2600_common *hw_priv);

bes2600/sbus.h

@@ -75,6 +75,14 @@ struct sbus_ops {
 	void (*halt_device)(struct sbus_priv *self);
 	bool (*wakeup_source)(struct sbus_priv *self);
 	int (*reboot)(struct sbus_priv *self);
+	/*
+	 * Force the host bus to re-detect and re-probe the chip. Called
+	 * from the firmware-wedge recovery path when power_switch() has no
+	 * effective chip-reset signal of its own (e.g. PineTab2, where the
+	 * wifi-reset GPIO is owned by sdio_pwrseq, not the bes2600 node).
+	 * Returns 0 on success or a negative errno.
+	 */
+	int (*bus_reset)(struct sbus_priv *self);
 };
 
 void bes2600_irq_handler(struct bes2600_common *priv);

bes2600/scan.c

@@ -14,11 +14,50 @@
 #include "scan.h"
 #include "sta.h"
 #include "pm.h"
+#include "epta_coex.h"
 #include "epta_request.h"
 #include "bes_pwr.h"
 
+/*
+ * After this many consecutive WSM scan rejections from firmware, stop
+ * issuing new scans for BES2600_SCAN_BACKOFF_JIFFIES and let the state
+ * that's rejecting them (coex window, firmware-internal busy) clear.
+ */
+#define BES2600_SCAN_REJECT_THRESHOLD	3
+#define BES2600_SCAN_BACKOFF_JIFFIES	(10 * HZ)
+
 static void bes2600_scan_restart_delayed(struct bes2600_vif *priv);
 
+/*
+ * Decide whether to skip sending the next WSM scan command without
+ * bothering the firmware. Two triggers:
+ *
+ *  1. BT A2DP is streaming in non-FDD coex mode. The firmware is
+ *     known to reject scan requests during that window; short-
+ *     circuiting here saves a WSM round-trip and avoids the
+ *     wsm_generic_confirm / scan_work warning cascade that follows.
+ *
+ *  2. We already saw >= BES2600_SCAN_REJECT_THRESHOLD consecutive
+ *     rejections on recent scan attempts and the backoff window has
+ *     not yet elapsed. Whatever was rejecting them is likely still
+ *     rejecting them; give it time.
+ *
+ * Returns true if the caller should abandon the scan iteration.
+ */
+static bool bes2600_scan_should_defer(struct bes2600_common *hw_priv)
+{
+#ifdef WIFI_BT_COEXIST_EPTA_ENABLE
+	if (!coex_is_fdd_mode() && coex_is_bt_a2dp())
+		return true;
+#endif
+
+	if (hw_priv->scan.reject_count >= BES2600_SCAN_REJECT_THRESHOLD &&
+	    time_before(jiffies, hw_priv->scan.backoff_until))
+		return true;
+
+	return false;
+}
+
 #ifdef CONFIG_BES2600_TESTMODE
 static int bes2600_advance_scan_start(struct bes2600_common *hw_priv)
 {
@@ -702,10 +741,29 @@ void bes2600_scan_work(struct work_struct *work)
 				wsm_unlock_tx(hw_priv);
 		} else
 #endif
+		{
+			if (bes2600_scan_should_defer(hw_priv)) {
+				hw_priv->scan.status = -EBUSY;
+				hw_priv->scan.reject_count++;
+				hw_priv->scan.backoff_until =
+					jiffies + BES2600_SCAN_BACKOFF_JIFFIES;
+				wiphy_dbg(priv->hw->wiphy,
+					  "[SCAN] deferred (coex/backoff, reject_count=%u)\n",
+					  hw_priv->scan.reject_count);
+				kfree(scan.ch);
+				goto fail;
+			}
 			hw_priv->scan.status = bes2600_scan_start(priv, &scan);
+		}
 		kfree(scan.ch);
-		if (WARN_ON(hw_priv->scan.status))
+		if (hw_priv->scan.status) {
+			hw_priv->scan.reject_count++;
+			hw_priv->scan.backoff_until =
+				jiffies + BES2600_SCAN_BACKOFF_JIFFIES;
+			/* Lower callers already logged the reason at wiphy_warn. */
 			goto fail;
+		}
+		hw_priv->scan.reject_count = 0;
 		hw_priv->scan.curr = it;
 	}
 	up(&hw_priv->conf_lock);

bes2600/scan.h

@@ -42,6 +42,17 @@ struct bes2600_scan {
 	struct delayed_work probe_work;
 	int direct_probe;
 	u8 if_id;
+	/*
+	 * Track consecutive firmware-side WSM scan rejections so we can
+	 * back off briefly instead of re-issuing the same scan on every
+	 * mac80211 background-scan tick. Firmware returns WSM status != 0
+	 * for a handful of transient conditions (BT A2DP active in non-
+	 * FDD coex, firmware-internal busy windows) and keeps rejecting
+	 * until the state clears; retrying at full cadence just floods
+	 * dmesg.
+	 */
+	unsigned int reject_count;
+	unsigned long backoff_until;
 };
 
 int bes2600_hw_scan(struct ieee80211_hw *hw,

bes2600/wsm.c

@@ -134,8 +134,20 @@ static int wsm_generic_confirm(struct bes2600_common *hw_priv,
 				 struct wsm_buf *buf)
 {
 	u32 status = WSM_GET32(buf);
-	if (WARN(status != WSM_STATUS_SUCCESS, "wsm_generic_confirm ret %u", status))
+
+	/*
+	 * A non-SUCCESS status here is a firmware-side policy decision for
+	 * the command whose confirm this is -- commonly WSM status 2 for
+	 * scan (0x0407) rejected because of a coex window or transient
+	 * firmware-busy state. It is not a driver/kernel bug, so avoid the
+	 * WARN()/stack-trace treatment; the caller already emits a
+	 * wiphy_warn identifying the request id and will propagate the
+	 * error to mac80211.
+	 */
+	if (status != WSM_STATUS_SUCCESS) {
+		bes_devel("%s ret %u\n", __func__, status);
 		return -EINVAL;
+	}
 	return 0;
 
 underflow: