Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Kamal Dasu <kdasu.kdev@gmail.com>
To:     Kamal Dasu <kdasu.kdev@gmail.com>,
        bcm-kernel-feedback-list@broadcom.com,
        Mark Brown <broonie@kernel.org>
Cc:     linux-spi@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: [Patch v2 4/9] spi: bcm-qspi: Drive MSPI peripheral SSb pin on cs_change
Date:   Sun, 19 Apr 2020 15:23:33 -0400
Message-Id: <20200419192339.32023-5-kdasu.kdev@gmail.com>
In-Reply-To: <20200419192339.32023-1-kdasu.kdev@gmail.com>
References: <20200419192339.32023-1-kdasu.kdev@gmail.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

As per the spi core implementation for MSPI devices when the transfer is
the last one in the message, the chip may stay selected until the next
transfer. On multi-device SPI busses with nothing blocking messages going
to other devices, this is just a performance hint; starting a message to
another device deselects this one. But in other cases, this can be used
to ensure correctness. Some devices need protocol transactions to be built
from a series of spi_message submissions, where the content of one message
is determined by the results of previous messages and where the whole
transaction ends when the chipselect goes intactive.

On CS change after completing the last serial transfer, the MSPI driver
drives SSb pin CDRAM register correctly according comments in core spi.h
as shown below:

case 1) EOM =1, cs_change =0: SSb inactive
case 2) EOM =1, cs_change =1: SSb active
case 3) EOM =0, cs_change =0: SSb active
case 4) EOM =0, cs_change =1: SSb inactive

Signed-off-by: Kamal Dasu <kdasu.kdev@gmail.com>
---
 drivers/spi/spi-bcm-qspi.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

diff --git a/drivers/spi/spi-bcm-qspi.c b/drivers/spi/spi-bcm-qspi.c
index d901dcb10d06..c48c399dce53 100644
--- a/drivers/spi/spi-bcm-qspi.c
+++ b/drivers/spi/spi-bcm-qspi.c
@@ -615,19 +615,15 @@ static int update_qspi_trans_byte_count(struct bcm_qspi *qspi,
 		if (qt->trans->cs_change &&
 		    (flags & TRANS_STATUS_BREAK_CS_CHANGE))
 			ret |= TRANS_STATUS_BREAK_CS_CHANGE;
-		if (ret)
-			goto done;
 
-		dev_dbg(&qspi->pdev->dev, "advance msg exit\n");
 		if (bcm_qspi_mspi_transfer_is_last(qspi, qt))
-			ret = TRANS_STATUS_BREAK_EOM;
+			ret |= TRANS_STATUS_BREAK_EOM;
 		else
-			ret = TRANS_STATUS_BREAK_NO_BYTES;
+			ret |= TRANS_STATUS_BREAK_NO_BYTES;
 
 		qt->trans = NULL;
 	}
 
-done:
 	dev_dbg(&qspi->pdev->dev, "trans %p len %d byte %d ret %x\n",
 		qt->trans, qt->trans ? qt->trans->len : 0, qt->byte, ret);
 	return ret;
@@ -774,7 +770,16 @@ static int write_to_hw(struct bcm_qspi *qspi, struct spi_device *spi)
 	bcm_qspi_write(qspi, MSPI, MSPI_NEWQP, 0);
 	bcm_qspi_write(qspi, MSPI, MSPI_ENDQP, slot - 1);
 
-	if (tstatus & TRANS_STATUS_BREAK_DESELECT) {
+	/*
+	 *  case 1) EOM =1, cs_change =0: SSb inactive
+	 *  case 2) EOM =1, cs_change =1: SSb stay active
+	 *  case 3) EOM =0, cs_change =0: SSb stay active
+	 *  case 4) EOM =0, cs_change =1: SSb inactive
+	 */
+	if (((tstatus & TRANS_STATUS_BREAK_DESELECT)
+	     == TRANS_STATUS_BREAK_CS_CHANGE) ||
+	    ((tstatus & TRANS_STATUS_BREAK_DESELECT)
+	     == TRANS_STATUS_BREAK_EOM)) {
 		mspi_cdram = read_cdram_slot(qspi, slot - 1) &
 			~MSPI_CDRAM_CONT_BIT;
 		write_cdram_slot(qspi, slot - 1, mspi_cdram);
-- 
2.17.1