spi/fsl-espi: Don't spin forever on SPIE_RXCNT

Nobuteru Hayashi · broonie · commit a12ddd60ed0a · 2016-03-28T19:41:18.000+01:00
Infinite loop on SPIE_RXCNT occurred.

while (SPIE_RXCNT(events) &lt; min(4, mspi-&gt;len)) {
        cpu_relax();
        events = mpc8xxx_spi_read_reg(&amp;reg_base-&gt;event);
}

We met a soft lockup at fsl_espi_cpu_irq() because of this.

Fix it by using spin_event_timeout() so that fsl_espi_cpu_irq()
can break loop with timeouts dmesg.

Signed-off-by: Nobuteru Hayashi &lt;hayashi.nbb@ncos.nec.co.jp&gt;
Signed-off-by: Mark Brown &lt;broonie@kernel.org&gt;
diff --git a/drivers/spi/spi-fsl-espi.c b/drivers/spi/spi-fsl-espi.c
@@ -539,11 +539,18 @@ void fsl_espi_cpu_irq(struct mpc8xxx_spi *mspi, u32 events)
 	if (events & SPIE_NE) {
 		u32 rx_data, tmp;
 		u8 rx_data_8;
+		int ret;
 
 		/* Spin until RX is done */
-		while (SPIE_RXCNT(events) < min(4, mspi->len)) {
-			cpu_relax();
-			events = mpc8xxx_spi_read_reg(&reg_base->event);
+		if (SPIE_RXCNT(events) < min(4, mspi->len)) {
+			ret = spin_event_timeout(
+				!(SPIE_RXCNT(events =
+				mpc8xxx_spi_read_reg(&reg_base->event)) <
+						min(4, mspi->len)),
+						10000, 0); /* 10 msec */
+			if (!ret)
+				dev_err(mspi->dev,
+					 "tired waiting for SPIE_RXCNT\n");
 		}
 
 		if (mspi->len >= 4) {
