From patchwork Tue Oct 13 10:21:05 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Vignesh Raghavendra X-Patchwork-Id: 7382771 Return-Path: X-Original-To: patchwork-linux-spi@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork1.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.29.136]) by patchwork1.web.kernel.org (Postfix) with ESMTP id 1E6689F1D5 for ; Tue, 13 Oct 2015 10:21:30 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 1780E207CC for ; Tue, 13 Oct 2015 10:21:29 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id E2F07207C2 for ; Tue, 13 Oct 2015 10:21:27 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752854AbbJMKVP (ORCPT ); Tue, 13 Oct 2015 06:21:15 -0400 Received: from arroyo.ext.ti.com ([192.94.94.40]:53213 "EHLO arroyo.ext.ti.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752514AbbJMKVO (ORCPT ); Tue, 13 Oct 2015 06:21:14 -0400 Received: from dflxv15.itg.ti.com ([128.247.5.124]) by arroyo.ext.ti.com (8.13.7/8.13.7) with ESMTP id t9DALBdr017820; Tue, 13 Oct 2015 05:21:11 -0500 Received: from DLEE71.ent.ti.com (dlee71.ent.ti.com [157.170.170.114]) by dflxv15.itg.ti.com (8.14.3/8.13.8) with ESMTP id t9DALBhI020270; Tue, 13 Oct 2015 05:21:11 -0500 Received: from dlep32.itg.ti.com (157.170.170.100) by DLEE71.ent.ti.com (157.170.170.114) with Microsoft SMTP Server id 14.3.224.2; Tue, 13 Oct 2015 05:21:11 -0500 Received: from uda0132425.apr.dhcp.ti.com (ileax41-snat.itg.ti.com [10.172.224.153]) by dlep32.itg.ti.com (8.14.3/8.13.8) with ESMTP id t9DAL8gp005308; Tue, 13 Oct 2015 05:21:08 -0500 From: Vignesh R To: Mark Brown CC: , , , Vignesh R Subject: [PATCH] spi: spi-ti-qspi: switch to polling mode for better r/w performance Date: Tue, 13 Oct 2015 15:51:05 +0530 Message-ID: <1444731665-11359-1-git-send-email-vigneshr@ti.com> X-Mailer: git-send-email 2.6.1 MIME-Version: 1.0 Sender: linux-spi-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-spi@vger.kernel.org X-Spam-Status: No, score=-6.9 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_HI, T_RP_MATCHES_RCVD, UNPARSEABLE_RELAY autolearn=unavailable version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on mail.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP Currently word completion interrupt is fired for transfer of every word(8bit to 128bit in size). This adds a lot of overhead, and decreases r/w throughput. It hardly takes 3us(@48MHz) for 128bit r/w to complete, hence its better to poll on word complete bit to be set in QSPI_SPI_STATUS_REG instead of using interrupts. This increases the throughput by 30% in both read and write case. So, switch to polling mode instead of interrupts to determine completion of word transfer. Signed-off-by: Vignesh R --- Tested on DRA74 Rev G EVM. drivers/spi/spi-ti-qspi.c | 74 +++++++++++++---------------------------------- 1 file changed, 20 insertions(+), 54 deletions(-) diff --git a/drivers/spi/spi-ti-qspi.c b/drivers/spi/spi-ti-qspi.c index 81b84858cfee..69c1a95b0615 100644 --- a/drivers/spi/spi-ti-qspi.c +++ b/drivers/spi/spi-ti-qspi.c @@ -39,8 +39,6 @@ struct ti_qspi_regs { }; struct ti_qspi { - struct completion transfer_complete; - /* list synchronization */ struct mutex list_lock; @@ -62,10 +60,6 @@ struct ti_qspi { #define QSPI_PID (0x0) #define QSPI_SYSCONFIG (0x10) -#define QSPI_INTR_STATUS_RAW_SET (0x20) -#define QSPI_INTR_STATUS_ENABLED_CLEAR (0x24) -#define QSPI_INTR_ENABLE_SET_REG (0x28) -#define QSPI_INTR_ENABLE_CLEAR_REG (0x2c) #define QSPI_SPI_CLOCK_CNTRL_REG (0x40) #define QSPI_SPI_DC_REG (0x44) #define QSPI_SPI_CMD_REG (0x48) @@ -97,7 +91,6 @@ struct ti_qspi { #define QSPI_RD_DUAL (3 << 16) #define QSPI_RD_QUAD (7 << 16) #define QSPI_INVAL (4 << 16) -#define QSPI_WC_CMD_INT_EN (1 << 14) #define QSPI_FLEN(n) ((n - 1) << 0) #define QSPI_WLEN_MAX_BITS 128 #define QSPI_WLEN_MAX_BYTES 16 @@ -106,10 +99,6 @@ struct ti_qspi { #define BUSY 0x01 #define WC 0x02 -/* INTERRUPT REGISTER */ -#define QSPI_WC_INT_EN (1 << 1) -#define QSPI_WC_INT_DISABLE (1 << 1) - /* Device Control */ #define QSPI_DD(m, n) (m << (3 + n * 8)) #define QSPI_CKPHA(n) (1 << (2 + n * 8)) @@ -217,6 +206,24 @@ static inline u32 qspi_is_busy(struct ti_qspi *qspi) return stat & BUSY; } +static inline int ti_qspi_poll_wc(struct ti_qspi *qspi) +{ + u32 stat; + unsigned long timeout = jiffies + QSPI_COMPLETION_TIMEOUT; + + do { + stat = ti_qspi_read(qspi, QSPI_SPI_STATUS_REG); + if (stat & WC) + return 0; + cpu_relax(); + } while (time_after(timeout, jiffies)); + + stat = ti_qspi_read(qspi, QSPI_SPI_STATUS_REG); + if (stat & WC) + return 0; + return -ETIMEDOUT; +} + static int qspi_write_msg(struct ti_qspi *qspi, struct spi_transfer *t) { int wlen, count, xfer_len; @@ -275,8 +282,7 @@ static int qspi_write_msg(struct ti_qspi *qspi, struct spi_transfer *t) } ti_qspi_write(qspi, cmd, QSPI_SPI_CMD_REG); - if (!wait_for_completion_timeout(&qspi->transfer_complete, - QSPI_COMPLETION_TIMEOUT)) { + if (ti_qspi_poll_wc(qspi)) { dev_err(qspi->dev, "write timed out\n"); return -ETIMEDOUT; } @@ -315,8 +321,7 @@ static int qspi_read_msg(struct ti_qspi *qspi, struct spi_transfer *t) return -EBUSY; ti_qspi_write(qspi, cmd, QSPI_SPI_CMD_REG); - if (!wait_for_completion_timeout(&qspi->transfer_complete, - QSPI_COMPLETION_TIMEOUT)) { + if (ti_qspi_poll_wc(qspi)) { dev_err(qspi->dev, "read timed out\n"); return -ETIMEDOUT; } @@ -388,9 +393,7 @@ static int ti_qspi_start_transfer_one(struct spi_master *master, qspi->cmd = 0; qspi->cmd |= QSPI_EN_CS(spi->chip_select); qspi->cmd |= QSPI_FLEN(frame_length); - qspi->cmd |= QSPI_WC_CMD_INT_EN; - ti_qspi_write(qspi, QSPI_WC_INT_EN, QSPI_INTR_ENABLE_SET_REG); ti_qspi_write(qspi, qspi->dc, QSPI_SPI_DC_REG); mutex_lock(&qspi->list_lock); @@ -417,31 +420,6 @@ static int ti_qspi_start_transfer_one(struct spi_master *master, return status; } -static irqreturn_t ti_qspi_isr(int irq, void *dev_id) -{ - struct ti_qspi *qspi = dev_id; - u16 int_stat; - u32 stat; - - irqreturn_t ret = IRQ_HANDLED; - - int_stat = ti_qspi_read(qspi, QSPI_INTR_STATUS_ENABLED_CLEAR); - stat = ti_qspi_read(qspi, QSPI_SPI_STATUS_REG); - - if (!int_stat) { - dev_dbg(qspi->dev, "No IRQ triggered\n"); - ret = IRQ_NONE; - goto out; - } - - ti_qspi_write(qspi, QSPI_WC_INT_DISABLE, - QSPI_INTR_STATUS_ENABLED_CLEAR); - if (stat & WC) - complete(&qspi->transfer_complete); -out: - return ret; -} - static int ti_qspi_runtime_resume(struct device *dev) { struct ti_qspi *qspi; @@ -550,22 +528,12 @@ static int ti_qspi_probe(struct platform_device *pdev) } } - ret = devm_request_irq(&pdev->dev, irq, ti_qspi_isr, 0, - dev_name(&pdev->dev), qspi); - if (ret < 0) { - dev_err(&pdev->dev, "Failed to register ISR for IRQ %d\n", - irq); - goto free_master; - } - qspi->fclk = devm_clk_get(&pdev->dev, "fck"); if (IS_ERR(qspi->fclk)) { ret = PTR_ERR(qspi->fclk); dev_err(&pdev->dev, "could not get clk: %d\n", ret); } - init_completion(&qspi->transfer_complete); - pm_runtime_use_autosuspend(&pdev->dev); pm_runtime_set_autosuspend_delay(&pdev->dev, QSPI_AUTOSUSPEND_TIMEOUT); pm_runtime_enable(&pdev->dev); @@ -595,8 +563,6 @@ static int ti_qspi_remove(struct platform_device *pdev) return ret; } - ti_qspi_write(qspi, QSPI_WC_INT_DISABLE, QSPI_INTR_ENABLE_CLEAR_REG); - pm_runtime_put(qspi->dev); pm_runtime_disable(&pdev->dev);