Merge tag 'spi-fix-v6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi

Pull spi fixes from Mark Brown:
 "Several fixes that came in since the merge window, the major one being
  a fix for the spi-mux driver which was broken by the performance
  optimisations due to it peering inside the core's data structures more
  than it should"

* tag 'spi-fix-v6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
  spi: spi: Fix queue hang if previous transfer failed
  spi: mux: Fix mux interaction with fast path optimisations
  spi: cadence-quadspi: Disable irqs during indirect reads
  spi: bitbang: Fix lsb-first Rx

drivers/spi/spi-bitbang-txrx.h

@@ -116,6 +116,7 @@ bitbang_txrx_le_cpha0(struct spi_device *spi,
 {
 	/* if (cpol == 0) this is SPI_MODE_0; else this is SPI_MODE_2 */
 
+	u8 rxbit = bits - 1;
 	u32 oldbit = !(word & 1);
 	/* clock starts at inactive polarity */
 	for (; likely(bits); bits--) {
@@ -135,7 +136,7 @@ bitbang_txrx_le_cpha0(struct spi_device *spi,
 		/* sample LSB (from slave) on leading edge */
 		word >>= 1;
 		if ((flags & SPI_MASTER_NO_RX) == 0)
-			word |= getmiso(spi) << (bits - 1);
+			word |= getmiso(spi) << rxbit;
 		setsck(spi, cpol);
 	}
 	return word;
@@ -148,6 +149,7 @@ bitbang_txrx_le_cpha1(struct spi_device *spi,
 {
 	/* if (cpol == 0) this is SPI_MODE_1; else this is SPI_MODE_3 */
 
+	u8 rxbit = bits - 1;
 	u32 oldbit = !(word & 1);
 	/* clock starts at inactive polarity */
 	for (; likely(bits); bits--) {
@@ -168,7 +170,7 @@ bitbang_txrx_le_cpha1(struct spi_device *spi,
 		/* sample LSB (from slave) on trailing edge */
 		word >>= 1;
 		if ((flags & SPI_MASTER_NO_RX) == 0)
-			word |= getmiso(spi) << (bits - 1);
+			word |= getmiso(spi) << rxbit;
 	}
 	return word;
 }

drivers/spi/spi-cadence-quadspi.c

@@ -39,6 +39,7 @@
 #define CQSPI_DISABLE_DAC_MODE		BIT(1)
 #define CQSPI_SUPPORT_EXTERNAL_DMA	BIT(2)
 #define CQSPI_NO_SUPPORT_WR_COMPLETION	BIT(3)
+#define CQSPI_SLOW_SRAM		BIT(4)
 
 /* Capabilities */
 #define CQSPI_SUPPORTS_OCTAL		BIT(0)
@@ -87,6 +88,7 @@ struct cqspi_st {
 	bool			use_dma_read;
 	u32			pd_dev_id;
 	bool			wr_completion;
+	bool			slow_sram;
 };
 
 struct cqspi_driver_platdata {
@@ -333,7 +335,10 @@ static irqreturn_t cqspi_irq_handler(int this_irq, void *dev)
 		}
 	}
 
+	else if (!cqspi->slow_sram)
 		irq_status &= CQSPI_IRQ_MASK_RD | CQSPI_IRQ_MASK_WR;
+	else
+		irq_status &= CQSPI_REG_IRQ_WATERMARK | CQSPI_IRQ_MASK_WR;
 
 	if (irq_status)
 		complete(&cqspi->transfer_complete);
@@ -673,7 +678,18 @@ static int cqspi_indirect_read_execute(struct cqspi_flash_pdata *f_pdata,
 	/* Clear all interrupts. */
 	writel(CQSPI_IRQ_STATUS_MASK, reg_base + CQSPI_REG_IRQSTATUS);
 
+	/*
+	 * On SoCFPGA platform reading the SRAM is slow due to
+	 * hardware limitation and causing read interrupt storm to CPU,
+	 * so enabling only watermark interrupt to disable all read
+	 * interrupts later as we want to run "bytes to read" loop with
+	 * all the read interrupts disabled for max performance.
+	 */
+
+	if (!cqspi->slow_sram)
 		writel(CQSPI_IRQ_MASK_RD, reg_base + CQSPI_REG_IRQMASK);
+	else
+		writel(CQSPI_REG_IRQ_WATERMARK, reg_base + CQSPI_REG_IRQMASK);
 
 	reinit_completion(&cqspi->transfer_complete);
 	writel(CQSPI_REG_INDIRECTRD_START_MASK,
@@ -684,6 +700,13 @@ static int cqspi_indirect_read_execute(struct cqspi_flash_pdata *f_pdata,
 						 msecs_to_jiffies(CQSPI_READ_TIMEOUT_MS)))
 			ret = -ETIMEDOUT;
 
+		/*
+		 * Disable all read interrupts until
+		 * we are out of "bytes to read"
+		 */
+		if (cqspi->slow_sram)
+			writel(0x0, reg_base + CQSPI_REG_IRQMASK);
+
 		bytes_to_read = cqspi_get_rd_sram_level(cqspi);
 
 		if (ret && bytes_to_read == 0) {
@@ -715,8 +738,11 @@ static int cqspi_indirect_read_execute(struct cqspi_flash_pdata *f_pdata,
 			bytes_to_read = cqspi_get_rd_sram_level(cqspi);
 		}
 
-		if (remaining > 0)
+		if (remaining > 0) {
 			reinit_completion(&cqspi->transfer_complete);
+			if (cqspi->slow_sram)
+				writel(CQSPI_REG_IRQ_WATERMARK, reg_base + CQSPI_REG_IRQMASK);
+		}
 	}
 
 	/* Check indirect done status */
@@ -1667,6 +1693,8 @@ static int cqspi_probe(struct platform_device *pdev)
 			cqspi->use_dma_read = true;
 		if (ddata->quirks & CQSPI_NO_SUPPORT_WR_COMPLETION)
 			cqspi->wr_completion = false;
+		if (ddata->quirks & CQSPI_SLOW_SRAM)
+			cqspi->slow_sram = true;
 
 		if (of_device_is_compatible(pdev->dev.of_node,
 					    "xlnx,versal-ospi-1.0"))
@@ -1779,7 +1807,9 @@ static const struct cqspi_driver_platdata intel_lgm_qspi = {
 };
 
 static const struct cqspi_driver_platdata socfpga_qspi = {
-	.quirks = CQSPI_DISABLE_DAC_MODE | CQSPI_NO_SUPPORT_WR_COMPLETION,
+	.quirks = CQSPI_DISABLE_DAC_MODE
+			| CQSPI_NO_SUPPORT_WR_COMPLETION
+			| CQSPI_SLOW_SRAM,
 };
 
 static const struct cqspi_driver_platdata versal_ospi = {

drivers/spi/spi-mux.c

@@ -161,6 +161,7 @@ static int spi_mux_probe(struct spi_device *spi)
 	ctlr->num_chipselect = mux_control_states(priv->mux);
 	ctlr->bus_num = -1;
 	ctlr->dev.of_node = spi->dev.of_node;
+	ctlr->must_async = true;
 
 	ret = devm_spi_register_controller(&spi->dev, ctlr);
 	if (ret)

drivers/spi/spi.c

@@ -1727,7 +1727,6 @@ static void __spi_pump_messages(struct spi_controller *ctlr, bool in_kthread)
 	spin_unlock_irqrestore(&ctlr->queue_lock, flags);
 
 	ret = __spi_pump_transfer_message(ctlr, msg, was_busy);
-	if (!ret)
 	kthread_queue_work(ctlr->kworker, &ctlr->pump_messages);
 
 	ctlr->cur_msg = NULL;
@@ -4033,7 +4032,7 @@ static int __spi_sync(struct spi_device *spi, struct spi_message *message)
 	 * guard against reentrancy from a different context. The io_mutex
 	 * will catch those cases.
 	 */
-	if (READ_ONCE(ctlr->queue_empty)) {
+	if (READ_ONCE(ctlr->queue_empty) && !ctlr->must_async) {
 		message->actual_length = 0;
 		message->status = -EINPROGRESS;
 

include/linux/spi/spi.h

@@ -469,6 +469,7 @@ extern struct spi_device *spi_new_ancillary_device(struct spi_device *spi, u8 ch
  *	SPI_TRANS_FAIL_NO_START.
  * @queue_empty: signal green light for opportunistically skipping the queue
  *	for spi_sync transfers.
+ * @must_async: disable all fast paths in the core
  *
  * Each SPI controller can communicate with one or more @spi_device
  * children.  These make a small bus, sharing MOSI, MISO and SCK signals
@@ -690,6 +691,7 @@ struct spi_controller {
 
 	/* Flag for enabling opportunistic skipping of the queue in spi_sync */
 	bool			queue_empty;
+	bool			must_async;
 };
 
 static inline void *spi_controller_get_devdata(struct spi_controller *ctlr)