Received: by 2002:ac0:a582:0:0:0:0:0 with SMTP id m2-v6csp99991imm; Mon, 1 Oct 2018 07:17:14 -0700 (PDT) X-Google-Smtp-Source: ACcGV60ez6wY6E/oY6+6lPnSgnlL/DOKwI546M4LzUpK+OtDkuN2HWGuEWIUiZ38cZowy3XJ9TFV X-Received: by 2002:a17:902:700a:: with SMTP id y10-v6mr26131plk.328.1538403434525; Mon, 01 Oct 2018 07:17:14 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1538403434; cv=none; d=google.com; s=arc-20160816; b=GPfq2w2HBQD5qwph34CjSw9V+YiLkG+y2LGBmejnIX4+6FdvG4XlMvas+Z/AbNvnjY qabpWgHq1DlxQKHTYVn1QpsxRejQp8ChJ6bVsrLh2+IzN7pL7LpJtb5Wrj4h+3Y3dMvV EMgEL9Lu8lzVtpMKDjXWERsrB4+G2RdkX1OO5/83bDFbzgtTn9T6VbpicteJFZYve3Y3 vDhJS1pc/Ag90mML4k9t4Z3NrPzsUjgHLgkrGNtet6650cttPtetGYAYPVQY9f4I6i9+ 6eFdfeVAMhIVe4/b0AGUZLxe0YyABVSs7itbRCIK5aoTanjM6kYqr33X2mmSJRjmzrvd dRew== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:sender:content-language :content-transfer-encoding:in-reply-to:mime-version:user-agent:date :message-id:from:references:cc:to:subject:dmarc-filter :dkim-signature:dkim-signature; bh=Ay8yQIwbAB0/Kc4im5elpRf7ssTs++qBLgjBm68f+Uo=; b=Wi+HuWo6eMt09MtrR+R1mEBq3l5lUCdrBPJD41XT8zrtP40wtdutwPBJjfk6y6PLCL In6WApAYC+23wvJV82ja3ixjgW1hc57wIEfxWjSIlJxr6qajO3znv6oDFNX+CdA1RVPR 4RN0/GWZmycyGysquDSXwD9HRjx5QS+RRrmL8sGcILcA2uJdj+SPFjch9UP+/BY99bX6 PdJ2ozFkk+mHLFR/zAGg40Orb3nqZIG7R4/0HU8VKXIfVhLKGhMm1qqM6NVDHXAAO92i rNG/jzHAysAGzO+H84Qvq/2HJxiyGXGfp/HoCoDmeUvX1CEPxa6+nsZxwKjoh2geSgU0 mAKA== ARC-Authentication-Results: i=1; mx.google.com; dkim=pass header.i=@codeaurora.org header.s=default header.b=GyTnkLAk; dkim=pass header.i=@codeaurora.org header.s=default header.b=h1oYBtyj; spf=pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org Return-Path: Received: from vger.kernel.org (vger.kernel.org. [209.132.180.67]) by mx.google.com with ESMTP id u64-v6si12343624pgu.533.2018.10.01.07.16.58; Mon, 01 Oct 2018 07:17:14 -0700 (PDT) Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67; Authentication-Results: mx.google.com; dkim=pass header.i=@codeaurora.org header.s=default header.b=GyTnkLAk; dkim=pass header.i=@codeaurora.org header.s=default header.b=h1oYBtyj; spf=pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729496AbeJAUys (ORCPT + 99 others); Mon, 1 Oct 2018 16:54:48 -0400 Received: from smtp.codeaurora.org ([198.145.29.96]:47494 "EHLO smtp.codeaurora.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729295AbeJAUyr (ORCPT ); Mon, 1 Oct 2018 16:54:47 -0400 Received: by smtp.codeaurora.org (Postfix, from userid 1000) id 6248860C5F; Mon, 1 Oct 2018 14:16:44 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=codeaurora.org; s=default; t=1538403405; bh=aLyfS2DEeZSqcn/dOCderwvj2qW+txEBj2VgR95h9W4=; h=Subject:To:Cc:References:From:Date:In-Reply-To:From; b=GyTnkLAkTqaDpmgSgMY7a02kg+AOrnsVtP9wP7pemCObg1/zqkBsnRyc6vPYqDEdX CqoXN+Ci6ka+of6RmR9Ab86vvloUs86Zs6Mix1+N2Qc9zXHngsdk7gTDY3IOEmobtc auFnK4wXoiSZ0sM7rEtNggW/RBfDCMQVSr5fDY/c= X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on pdx-caf-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-2.7 required=2.0 tests=ALL_TRUSTED,BAYES_00, DKIM_INVALID,DKIM_SIGNED autolearn=no autolearn_force=no version=3.4.0 Received: from [10.206.24.108] (blr-c-bdr-fw-01_globalnat_allzones-outside.qualcomm.com [103.229.19.19]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) (Authenticated sender: sayalil@smtp.codeaurora.org) by smtp.codeaurora.org (Postfix) with ESMTPSA id 7093560B7A; Mon, 1 Oct 2018 14:16:31 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=codeaurora.org; s=default; t=1538403397; bh=aLyfS2DEeZSqcn/dOCderwvj2qW+txEBj2VgR95h9W4=; h=Subject:To:Cc:References:From:Date:In-Reply-To:From; b=h1oYBtyj8li5YshHQqyq/KcR/9IdxV1L7G+HXITL1FAwaeksuoa3E7aVYQn+FxCHw GsD9hRo2fErj833+utWCh2kCeAF+8QLplDmV8OmNzJUlMLN+iXxdPZG90mA4pWgXCy 1OiM0Xl8uw0CXgpbjEZNKRYXy66nU/XxF+jenD7o= DMARC-Filter: OpenDMARC Filter v1.3.2 smtp.codeaurora.org 7093560B7A Authentication-Results: pdx-caf-mail.web.codeaurora.org; dmarc=none (p=none dis=none) header.from=codeaurora.org Authentication-Results: pdx-caf-mail.web.codeaurora.org; spf=none smtp.mailfrom=sayalil@codeaurora.org Subject: Re: [PATCH RFC 2/7] mmc: core: devfreq: Add devfreq based clock scaling support To: Vijay Viswanath , adrian.hunter@intel.com, ulf.hansson@linaro.org, robh+dt@kernel.org, mark.rutland@arm.com Cc: linux-mmc@vger.kernel.org, linux-kernel@vger.kernel.org, shawn.lin@rock-chips.com, linux-arm-msm@vger.kernel.org, georgi.djakov@linaro.org, devicetree@vger.kernel.org, asutoshd@codeaurora.org, stummala@codeaurora.org, venkatg@codeaurora.org, bjorn.andersson@linaro.org, riteshh@codeaurora.org, vbadigan@codeaurora.org, Talel Shenhar References: <1531475583-7050-1-git-send-email-sayalil@codeaurora.org> <1531475583-7050-3-git-send-email-sayalil@codeaurora.org> <36492d8e-9cc7-0462-5366-3612a4214136@codeaurora.org> From: Sayali Lokhande Message-ID: <7664c62a-bfc2-81e1-0593-f7073008de38@codeaurora.org> Date: Mon, 1 Oct 2018 19:46:28 +0530 User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:52.0) Gecko/20100101 Thunderbird/52.9.1 MIME-Version: 1.0 In-Reply-To: <36492d8e-9cc7-0462-5366-3612a4214136@codeaurora.org> Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit Content-Language: en-US Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 7/23/2018 3:31 PM, Vijay Viswanath wrote: > Hi Sayali, > > On 7/13/2018 3:22 PM, Sayali Lokhande wrote: >> This change adds the use of devfreq to MMC. >> Both eMMC and SD card will use it. >> For some workloads, such as video playback, it isn't >> necessary for these cards to run at high speed. >> Running at lower frequency, for example 52MHz, in such >> cases can still meet the deadlines for data transfers. >> Scaling down the clock frequency dynamically has power >> savings not only because the bus is running at lower frequency >> but also has an advantage of scaling down the system core >> voltage, if supported. >> Provide an ondemand clock scaling support similar to the >> cpufreq ondemand governor having two thresholds, >> up_threshold and down_threshold to decide whether to >> increase the frequency or scale it down respectively. >> The sampling interval is in the order of milliseconds. >> If sampling interval is too low, frequent switching of >> frequencies can lead to high power consumption and if >> sampling interval is too high, the clock scaling logic >> would take long time to realize that the underlying >> hardware (controller and card) is busy and scale up >> the clocks. >> >> Signed-off-by: Talel Shenhar >> Signed-off-by: Sayali Lokhande >> --- >>   .../devicetree/bindings/mmc/sdhci-msm.txt          |  10 + >>   drivers/mmc/core/core.c                            | 560 >> +++++++++++++++++++++ >>   drivers/mmc/core/core.h                            |   7 + >>   drivers/mmc/core/debugfs.c                         |  46 ++ >>   drivers/mmc/core/host.c                            |   8 + >>   drivers/mmc/core/mmc.c                             | 200 +++++++- >>   drivers/mmc/core/sd.c                              |  72 ++- >>   drivers/mmc/host/sdhci-msm.c                       |  37 ++ >>   drivers/mmc/host/sdhci-pltfm.c                     |  11 + >>   drivers/mmc/host/sdhci.c                           |  27 + >>   drivers/mmc/host/sdhci.h                           |   8 + >>   include/linux/mmc/card.h                           |   5 + >>   include/linux/mmc/host.h                           |  70 +++ >>   13 files changed, 1059 insertions(+), 2 deletions(-) >> >> diff --git a/Documentation/devicetree/bindings/mmc/sdhci-msm.txt >> b/Documentation/devicetree/bindings/mmc/sdhci-msm.txt >> index 502b3b8..bd8470a 100644 >> --- a/Documentation/devicetree/bindings/mmc/sdhci-msm.txt >> +++ b/Documentation/devicetree/bindings/mmc/sdhci-msm.txt >> @@ -26,6 +26,15 @@ Required properties: >>       "cal"    - reference clock for RCLK delay calibration (optional) >>       "sleep"    - sleep clock for RCLK delay calibration (optional) >>   +Optional Properties: >> +- qcom,devfreq,freq-table - specifies supported frequencies for >> clock scaling. >> +                    Clock scaling logic shall toggle between these >> frequencies based >> +                    on card load. In case the defined frequencies >> are over or below >> +                    the supported card frequencies, they will be >> overridden >> +                    during card init. In case this entry is not >> supplied, >> +                    the driver will construct one based on the card >> +                    supported max and min frequencies. >> +                    The frequencies must be ordered from lowest to >> highest. >>   Example: >>         sdhc_1: sdhci@f9824900 { >> @@ -43,6 +52,7 @@ Example: >>             clocks = <&gcc GCC_SDCC1_APPS_CLK>, <&gcc >> GCC_SDCC1_AHB_CLK>; >>           clock-names = "core", "iface"; >> +        qcom,devfreq,freq-table = <52000000 200000000>; >>       }; >>         sdhc_2: sdhci@f98a4900 { >> diff --git a/drivers/mmc/core/core.c b/drivers/mmc/core/core.c >> index 281826d..0eaee42 100644 >> --- a/drivers/mmc/core/core.c >> +++ b/drivers/mmc/core/core.c >> @@ -14,6 +14,7 @@ >>   #include >>   #include >>   #include >> +#include >>   #include >>   #include >>   #include >> @@ -112,6 +113,556 @@ static inline void >> mmc_should_fail_request(struct mmc_host *host, >>     #endif /* CONFIG_FAIL_MMC_REQUEST */ >>   +static bool mmc_is_data_request(struct mmc_request *mmc_request) >> +{ >> +    switch (mmc_request->cmd->opcode) { >> +    case MMC_READ_SINGLE_BLOCK: >> +    case MMC_READ_MULTIPLE_BLOCK: >> +    case MMC_WRITE_BLOCK: >> +    case MMC_WRITE_MULTIPLE_BLOCK: >> +        return true; >> +    default: >> +        return false; >> +    } >> +} >> + >> +static void mmc_clk_scaling_start_busy(struct mmc_host *host, bool >> lock_needed) >> +{ >> +    struct mmc_devfeq_clk_scaling *clk_scaling = &host->clk_scaling; >> + >> +    if (!clk_scaling->enable) >> +        return; >> + >> +    if (lock_needed) >> +        spin_lock_bh(&clk_scaling->lock); >> + >> +    clk_scaling->start_busy = ktime_get(); >> +    clk_scaling->is_busy_started = true; >> + >> +    if (lock_needed) >> +        spin_unlock_bh(&clk_scaling->lock); >> +} >> + >> +static void mmc_clk_scaling_stop_busy(struct mmc_host *host, bool >> lock_needed) >> +{ >> +    struct mmc_devfeq_clk_scaling *clk_scaling = &host->clk_scaling; >> + >> +    if (!clk_scaling->enable) >> +        return; >> + >> +    if (lock_needed) >> +        spin_lock_bh(&clk_scaling->lock); >> + >> +    if (!clk_scaling->is_busy_started) { >> +        WARN_ON(1); >> +        goto out; >> +    } >> + >> +    clk_scaling->total_busy_time_us += >> +        ktime_to_us(ktime_sub(ktime_get(), >> +            clk_scaling->start_busy)); >> +    pr_debug("%s: accumulated busy time is %lu usec\n", >> +        mmc_hostname(host), clk_scaling->total_busy_time_us); >> +    clk_scaling->is_busy_started = false; >> + >> +out: >> +    if (lock_needed) >> +        spin_unlock_bh(&clk_scaling->lock); >> +} >> + >> +/** >> + * mmc_can_scale_clk() - Check clock scaling capability >> + * @host: pointer to mmc host structure >> + */ >> +bool mmc_can_scale_clk(struct mmc_host *host) >> +{ >> +    if (!host) { >> +        pr_err("bad host parameter\n"); >> +        WARN_ON(1); >> +        return false; >> +    } >> + >> +    return host->caps2 & MMC_CAP2_CLK_SCALE; >> +} >> +EXPORT_SYMBOL(mmc_can_scale_clk); >> + >> +static int mmc_devfreq_get_dev_status(struct device *dev, >> +        struct devfreq_dev_status *status) >> +{ >> +    struct mmc_host *host = container_of(dev, struct mmc_host, >> class_dev); >> +    struct mmc_devfeq_clk_scaling *clk_scaling; >> + >> +    if (!host) { >> +        pr_err("bad host parameter\n"); >> +        WARN_ON(1); >> +        return -EINVAL; >> +    } >> + >> +    clk_scaling = &host->clk_scaling; >> + >> +    if (!clk_scaling->enable) >> +        return 0; >> + >> +    spin_lock_bh(&clk_scaling->lock); >> + >> +    /* accumulate the busy time of ongoing work */ >> +    memset(status, 0, sizeof(*status)); >> +    if (clk_scaling->is_busy_started) { >> +            mmc_clk_scaling_stop_busy(host, false); >> +            mmc_clk_scaling_start_busy(host, false); >> +    } >> + >> +    status->busy_time = clk_scaling->total_busy_time_us; >> +    status->total_time = ktime_to_us(ktime_sub(ktime_get(), >> +        clk_scaling->measure_interval_start)); >> +    clk_scaling->total_busy_time_us = 0; >> +    status->current_frequency = clk_scaling->curr_freq; >> +    clk_scaling->measure_interval_start = ktime_get(); >> + >> +    pr_debug("%s: status: load = %lu%% - total_time=%lu busy_time = >> %lu, clk=%lu\n", >> +        mmc_hostname(host), >> +        (status->busy_time*100)/status->total_time, >> +        status->total_time, status->busy_time, >> +        status->current_frequency); >> + >> +    spin_unlock_bh(&clk_scaling->lock); >> + >> +    return 0; >> +} >> + >> +static bool mmc_is_valid_state_for_clk_scaling(struct mmc_host *host) >> +{ >> +    struct mmc_card *card = host->card; >> +    u32 status; >> + >> +    /* >> +     * If the current partition type is RPMB, clock switching may not >> +     * work properly as sending tuning command (CMD21) is illegal in >> +     * this mode. >> +     */ >> +    if (!card || (mmc_card_mmc(card) && >> +            (card->part_curr == EXT_CSD_PART_CONFIG_ACC_RPMB || >> +            mmc_card_doing_bkops(card)))) >> +        return false; >> + >> +    if (mmc_send_status(card, &status)) { >> +        pr_err("%s: Get card status fail\n", mmc_hostname(card->host)); >> +        return false; >> +    } >> + >> +    return R1_CURRENT_STATE(status) == R1_STATE_TRAN; >> +} >> + >> +int mmc_clk_update_freq(struct mmc_host *host, >> +        unsigned long freq, enum mmc_load state) >> +{ >> +    int err = 0; >> + >> +    if (!host) { >> +        pr_err("bad host parameter\n"); >> +        WARN_ON(1); >> +        return -EINVAL; >> +    } >> + >> +    /* make sure the card supports the frequency we want */ >> +    if (unlikely(freq > host->card->clk_scaling_highest)) { >> +        freq = host->card->clk_scaling_highest; >> +        pr_warn("%s: %s: frequency was overridden to %lu\n", >> +                mmc_hostname(host), __func__, >> +                host->card->clk_scaling_highest); >> +    } >> + >> +    if (unlikely(freq < host->card->clk_scaling_lowest)) { >> +        freq = host->card->clk_scaling_lowest; >> +        pr_warn("%s: %s: frequency was overridden to %lu\n", >> +            mmc_hostname(host), __func__, >> +            host->card->clk_scaling_lowest); >> +    } >> + >> +    if (freq == host->clk_scaling.curr_freq) >> +        goto out; >> + >> +    if (host->ops->notify_load) { >> +        err = host->ops->notify_load(host, state); >> +        if (err) { >> +            pr_err("%s: %s: fail on notify_load\n", >> +                mmc_hostname(host), __func__); >> +            goto out; >> +        } >> +    } >> + >> +    if (!mmc_is_valid_state_for_clk_scaling(host)) { >> +        pr_debug("%s: invalid state for clock scaling - skipping", >> +            mmc_hostname(host)); >> +        goto invalid_state; >> +    } >> + >> +    err = host->bus_ops->change_bus_speed(host, &freq); >> +    if (!err) >> +        host->clk_scaling.curr_freq = freq; >> +    else >> +        pr_err("%s: %s: failed (%d) at freq=%lu\n", >> +            mmc_hostname(host), __func__, err, freq); >> + >> +invalid_state: >> +    if (err) { >> +        /* restore previous state */ >> +        if (host->ops->notify_load) >> +            if (host->ops->notify_load(host, >> +                host->clk_scaling.state)) >> +                pr_err("%s: %s: fail on notify_load restore\n", >> +                    mmc_hostname(host), __func__); >> +    } >> +out: >> +    return err; >> +} >> +EXPORT_SYMBOL(mmc_clk_update_freq); >> + >> +static int mmc_devfreq_set_target(struct device *dev, >> +                unsigned long *freq, u32 devfreq_flags) >> +{ >> +    struct mmc_host *host = container_of(dev, struct mmc_host, >> class_dev); >> +    struct mmc_devfeq_clk_scaling *clk_scaling; >> +    int err = 0; >> +    int abort; >> +    unsigned long pflags = current->flags; >> + >> +    /* Ensure scaling would happen even in memory pressure >> conditions */ >> +    current->flags |= PF_MEMALLOC; >> + >> +    if (!(host && freq)) { >> +        pr_err("%s: unexpected host/freq parameter\n", __func__); >> +        err = -EINVAL; >> +        goto out; >> +    } >> + >> +    clk_scaling = &host->clk_scaling; >> + >> +    if (!clk_scaling->enable) >> +        goto out; >> + >> +    pr_debug("%s: target freq = %lu (%s)\n", mmc_hostname(host), >> +        *freq, current->comm); >> + >> +    if ((clk_scaling->curr_freq == *freq) || >> +        clk_scaling->skip_clk_scale_freq_update) >> +        goto out; >> + > > Suppose devfreq issued a low_freq request while mmc is handling a > request already issued to card (so mmc host is claimed by issuing > context). So we will set target freq to LOW_FREQ and quit from here. > Now, if devfreq comes again and issues a HIGH_FREQ request (before mmc > thread had a chance to change the freq to LOW in > mmc_deferred_scaling), we will skip this devfreq request (since > current freq is high already). Then, when mmc thread comes, it will > execute deferred scaling to LOW_FREQ. > > How about comparing *freq with clk_scaling->target_freq ? Agree. Will update this in next patchset. > >> +    /* No need to scale the clocks if they are gated */ >> +    if (!host->ios.clock) >> +        goto out; >> + >> +    spin_lock_bh(&clk_scaling->lock); >> +    if (clk_scaling->clk_scaling_in_progress) { >> +        pr_debug("%s: clocks scaling is already in-progress by mmc >> thread\n", >> +            mmc_hostname(host)); >> +        spin_unlock_bh(&clk_scaling->lock); >> +        goto out; >> +    } >> +    clk_scaling->need_freq_change = true; >> +    clk_scaling->target_freq = *freq; >> +    clk_scaling->state = *freq < clk_scaling->curr_freq ? >> +        MMC_LOAD_LOW : MMC_LOAD_HIGH; >> +    spin_unlock_bh(&clk_scaling->lock); >> + >> +    abort = __mmc_claim_host(host, NULL, &clk_scaling->devfreq_abort); >> +    if (abort) >> +        goto out; >> + >> +    /* >> +     * In case we were able to claim host there is no need to >> +     * defer the frequency change. It will be done now >> +     */ >> +    clk_scaling->need_freq_change = false; >> + >> +    err = mmc_clk_update_freq(host, *freq, clk_scaling->state); >> +    if (err && err != -EAGAIN) { >> +        pr_err("%s: clock scale to %lu failed with error %d\n", >> +            mmc_hostname(host), *freq, err); >> +    } else { >> +        pr_debug("%s: clock change to %lu finished successfully >> (%s)\n", >> +            mmc_hostname(host), *freq, current->comm); >> +    } >> + >> +    mmc_release_host(host); >> +out: >> +    current->flags &= ~PF_MEMALLOC; >> +    current->flags |= pflags & PF_MEMALLOC; >> +    return err; >> +} >> + >> +/** >> + * mmc_deferred_scaling() - scale clocks from data path (mmc thread >> context) >> + * @host: pointer to mmc host structure >> + * >> + * This function does clock scaling in case "need_freq_change" flag >> was set >> + * by the clock scaling logic. >> + */ >> +void mmc_deferred_scaling(struct mmc_host *host) >> +{ >> +    unsigned long target_freq; >> +    int err; >> + >> +    if (!host->clk_scaling.enable) >> +        return; >> + >> +    spin_lock_bh(&host->clk_scaling.lock); >> + >> +    if (host->clk_scaling.clk_scaling_in_progress || >> +        !(host->clk_scaling.need_freq_change)) { >> +        spin_unlock_bh(&host->clk_scaling.lock); >> +        return; >> +    } >> + >> + >> +    atomic_inc(&host->clk_scaling.devfreq_abort); >> +    target_freq = host->clk_scaling.target_freq; >> +    host->clk_scaling.clk_scaling_in_progress = true; >> +    host->clk_scaling.need_freq_change = false; >> +    spin_unlock_bh(&host->clk_scaling.lock); >> +    pr_debug("%s: doing deferred frequency change (%lu) (%s)\n", >> +                mmc_hostname(host), >> +                target_freq, current->comm); >> + >> +    err = mmc_clk_update_freq(host, target_freq, >> +        host->clk_scaling.state); >> +    if (err && err != -EAGAIN) { >> +        pr_err("%s: failed on deferred scale clocks (%d)\n", >> +            mmc_hostname(host), err); >> +    } else { >> +        pr_debug("%s: clocks were successfully scaled to %lu (%s)\n", >> +            mmc_hostname(host), >> +            target_freq, current->comm); >> +    } >> +    host->clk_scaling.clk_scaling_in_progress = false; >> +    atomic_dec(&host->clk_scaling.devfreq_abort); >> +} >> +EXPORT_SYMBOL(mmc_deferred_scaling); >> + >> +static int mmc_devfreq_create_freq_table(struct mmc_host *host) >> +{ >> +    int i; >> +    struct mmc_devfeq_clk_scaling *clk_scaling = &host->clk_scaling; >> + >> +    pr_debug("%s: supported: lowest=%lu, highest=%lu\n", >> +        mmc_hostname(host), >> +        host->card->clk_scaling_lowest, >> +        host->card->clk_scaling_highest); >> + >> +    /* >> +     * Create the frequency table and initialize it with default >> values. >> +     * Initialize it with platform specific frequencies if the >> frequency >> +     * table supplied by platform driver is present, otherwise >> initialize >> +     * it with min and max frequencies supported by the card. >> +     */ >> +    if (!clk_scaling->freq_table) { >> +        if (clk_scaling->pltfm_freq_table_sz) >> +            clk_scaling->freq_table_sz = >> +                clk_scaling->pltfm_freq_table_sz; >> +        else >> +            clk_scaling->freq_table_sz = 2; >> + >> +        clk_scaling->freq_table = kzalloc( >> +            (clk_scaling->freq_table_sz * >> +            sizeof(*(clk_scaling->freq_table))), GFP_KERNEL); >> +        if (!clk_scaling->freq_table) >> +            return -ENOMEM; >> + >> +        if (clk_scaling->pltfm_freq_table) { >> +            memcpy(clk_scaling->freq_table, >> +                clk_scaling->pltfm_freq_table, >> +                (clk_scaling->pltfm_freq_table_sz * >> +                sizeof(*(clk_scaling->pltfm_freq_table)))); >> +        } else { >> +            pr_debug("%s: no frequency table defined -  setting >> default\n", >> +                mmc_hostname(host)); >> +            clk_scaling->freq_table[0] = >> +                host->card->clk_scaling_lowest; >> +            clk_scaling->freq_table[1] = >> +                host->card->clk_scaling_highest; >> +            goto out; >> +        } >> +    } >> + >> +    if (host->card->clk_scaling_lowest > >> +        clk_scaling->freq_table[0]) >> +        pr_debug("%s: frequency table undershot possible freq\n", >> +            mmc_hostname(host)); >> + >> +    for (i = 0; i < clk_scaling->freq_table_sz; i++) { >> +        if (clk_scaling->freq_table[i] <= >> +            host->card->clk_scaling_highest) >> +            continue; >> +        clk_scaling->freq_table[i] = >> +            host->card->clk_scaling_highest; >> +        clk_scaling->freq_table_sz = i + 1; >> +        pr_debug("%s: frequency table overshot possible freq (%d)\n", >> +                mmc_hostname(host), clk_scaling->freq_table[i]); >> +        break; >> +    } >> + >> +out: >> +    /** >> +     * devfreq requires unsigned long type freq_table while the >> +     * freq_table in clk_scaling is un32. Here allocates an individual >> +     * memory space for it and release it when exit clock scaling. >> +     */ >> +    clk_scaling->devfreq_profile.freq_table =  kzalloc( >> +            clk_scaling->freq_table_sz * >> + sizeof(*(clk_scaling->devfreq_profile.freq_table)), >> +            GFP_KERNEL); >> +    if (!clk_scaling->devfreq_profile.freq_table) >> +        return -ENOMEM; >> +    clk_scaling->devfreq_profile.max_state = >> clk_scaling->freq_table_sz; >> + >> +    for (i = 0; i < clk_scaling->freq_table_sz; i++) { >> +        clk_scaling->devfreq_profile.freq_table[i] = >> +            clk_scaling->freq_table[i]; >> +        pr_debug("%s: freq[%d] = %u\n", >> +            mmc_hostname(host), i, clk_scaling->freq_table[i]); >> +    } >> + >> +    return 0; >> +} >> + >> +/** >> + * mmc_init_devfreq_clk_scaling() - Initialize clock scaling >> + * @host: pointer to mmc host structure >> + * >> + * Initialize clock scaling for supported hosts. It is assumed that >> the caller >> + * ensure clock is running at maximum possible frequency before >> calling this >> + * function. Shall use struct devfreq_simple_ondemand_data to configure >> + * governor. >> + */ >> +int mmc_init_clk_scaling(struct mmc_host *host) >> +{ >> +    int err; >> + >> +    if (!host || !host->card) { >> +        pr_err("%s: unexpected host/card parameters\n", >> +            __func__); >> +        return -EINVAL; >> +    } >> + >> +    if (!mmc_can_scale_clk(host) || >> +        !host->bus_ops->change_bus_speed) { >> +        pr_debug("%s: clock scaling is not supported\n", >> +            mmc_hostname(host)); >> +        return 0; >> +    } >> + >> +    pr_debug("registering %s dev (%p) to devfreq", >> +        mmc_hostname(host), >> +        mmc_classdev(host)); >> + >> +    if (host->clk_scaling.devfreq) { >> +        pr_err("%s: dev is already registered for dev %p\n", >> +            mmc_hostname(host), >> +            mmc_dev(host)); >> +        return -EPERM; >> +    } >> +    spin_lock_init(&host->clk_scaling.lock); >> +    atomic_set(&host->clk_scaling.devfreq_abort, 0); >> +    host->clk_scaling.curr_freq = host->ios.clock; >> +    host->clk_scaling.clk_scaling_in_progress = false; >> +    host->clk_scaling.need_freq_change = false; >> +    host->clk_scaling.is_busy_started = false; >> + >> +    host->clk_scaling.devfreq_profile.polling_ms = >> +        host->clk_scaling.polling_delay_ms; >> +    host->clk_scaling.devfreq_profile.get_dev_status = >> +        mmc_devfreq_get_dev_status; >> +    host->clk_scaling.devfreq_profile.target = mmc_devfreq_set_target; >> +    host->clk_scaling.devfreq_profile.initial_freq = host->ios.clock; >> + >> +    host->clk_scaling.ondemand_gov_data.simple_scaling = true; >> +    host->clk_scaling.ondemand_gov_data.upthreshold = >> +        host->clk_scaling.upthreshold; >> +    host->clk_scaling.ondemand_gov_data.downdifferential = >> +        host->clk_scaling.upthreshold - >> host->clk_scaling.downthreshold; >> + >> +    err = mmc_devfreq_create_freq_table(host); >> +    if (err) { >> +        pr_err("%s: fail to create devfreq frequency table\n", >> +            mmc_hostname(host)); >> +        return err; >> +    } >> + >> +    pr_debug("%s: adding devfreq with: upthreshold=%u >> downthreshold=%u polling=%u\n", >> +        mmc_hostname(host), >> +        host->clk_scaling.ondemand_gov_data.upthreshold, >> + host->clk_scaling.ondemand_gov_data.downdifferential, >> +        host->clk_scaling.devfreq_profile.polling_ms); >> +    host->clk_scaling.devfreq = devfreq_add_device( >> +        mmc_classdev(host), >> +        &host->clk_scaling.devfreq_profile, >> +        "simple_ondemand", >> +        &host->clk_scaling.ondemand_gov_data); >> +    if (!host->clk_scaling.devfreq) { >> +        pr_err("%s: unable to register with devfreq\n", >> +            mmc_hostname(host)); >> +        return -EPERM; >> +    } >> + >> +    pr_debug("%s: clk scaling is enabled for device %s (%p) with >> devfreq %p (clock = %uHz)\n", >> +        mmc_hostname(host), >> +        dev_name(mmc_classdev(host)), >> +        mmc_classdev(host), >> +        host->clk_scaling.devfreq, >> +        host->ios.clock); >> + >> +    host->clk_scaling.enable = true; >> + >> +    return err; >> +} >> +EXPORT_SYMBOL(mmc_init_clk_scaling); >> + >> +/** >> + * mmc_exit_devfreq_clk_scaling() - Disable clock scaling >> + * @host: pointer to mmc host structure >> + * >> + * Disable clock scaling permanently. >> + */ >> +int mmc_exit_clk_scaling(struct mmc_host *host) >> +{ >> +    int err; >> + >> +    if (!host) { >> +        pr_err("%s: bad host parameter\n", __func__); >> +        WARN_ON(1); >> +        return -EINVAL; >> +    } >> + >> +    if (!mmc_can_scale_clk(host)) >> +        return 0; >> + >> +    if (!host->clk_scaling.devfreq) { >> +        pr_err("%s: %s: no devfreq is assosiated with this device\n", >> +            mmc_hostname(host), __func__); >> +        return -EPERM; >> +    } >> + >> +    err = devfreq_remove_device(host->clk_scaling.devfreq); >> +    if (err) { >> +        pr_err("%s: remove devfreq failed (%d)\n", >> +            mmc_hostname(host), err); >> +        return err; >> +    } >> + >> +    kfree(host->clk_scaling.devfreq_profile.freq_table); >> + >> +    host->clk_scaling.devfreq = NULL; >> +    atomic_set(&host->clk_scaling.devfreq_abort, 1); >> + >> +    kfree(host->clk_scaling.freq_table); >> +    host->clk_scaling.freq_table = NULL; >> + >> +    pr_debug("%s: devfreq was removed\n", mmc_hostname(host)); >> + >> +    return 0; >> +} >> +EXPORT_SYMBOL(mmc_exit_clk_scaling); >> + >>   static inline void mmc_complete_cmd(struct mmc_request *mrq) >>   { >>       if (mrq->cap_cmd_during_tfr && >> !completion_done(&mrq->cmd_completion)) >> @@ -143,6 +694,9 @@ void mmc_request_done(struct mmc_host *host, >> struct mmc_request *mrq) >>       struct mmc_command *cmd = mrq->cmd; >>       int err = cmd->error; >>   +    if (host->clk_scaling.is_busy_started) >> +        mmc_clk_scaling_stop_busy(host, true); >> + >>       /* Flag re-tuning needed on CRC errors */ >>       if ((cmd->opcode != MMC_SEND_TUNING_BLOCK && >>           cmd->opcode != MMC_SEND_TUNING_BLOCK_HS200) && >> @@ -354,6 +908,12 @@ int mmc_start_request(struct mmc_host *host, >> struct mmc_request *mrq) >>           return err; >>         led_trigger_event(host->led, LED_FULL); >> + >> +    if (mmc_is_data_request(mrq)) { >> +        mmc_deferred_scaling(host); >> +        mmc_clk_scaling_start_busy(host, true); >> +    } >> + >>       __mmc_start_request(host, mrq); >>         return 0; >> diff --git a/drivers/mmc/core/core.h b/drivers/mmc/core/core.h >> index 9d8f09a..fc0a9b7 100644 >> --- a/drivers/mmc/core/core.h >> +++ b/drivers/mmc/core/core.h >> @@ -34,6 +34,7 @@ struct mmc_bus_ops { >>       int (*shutdown)(struct mmc_host *); >>       int (*hw_reset)(struct mmc_host *); >>       int (*sw_reset)(struct mmc_host *); >> +    int (*change_bus_speed)(struct mmc_host *, unsigned long *); >>   }; >>     void mmc_attach_bus(struct mmc_host *host, const struct >> mmc_bus_ops *ops); >> @@ -46,6 +47,8 @@ struct device_node *mmc_of_find_child_device(struct >> mmc_host *host, >>     void mmc_set_chip_select(struct mmc_host *host, int mode); >>   void mmc_set_clock(struct mmc_host *host, unsigned int hz); >> +int mmc_clk_update_freq(struct mmc_host *host, >> +        unsigned long freq, enum mmc_load state); >>   void mmc_set_bus_mode(struct mmc_host *host, unsigned int mode); >>   void mmc_set_bus_width(struct mmc_host *host, unsigned int width); >>   u32 mmc_select_voltage(struct mmc_host *host, u32 ocr); >> @@ -91,6 +94,10 @@ static inline void mmc_delay(unsigned int ms) >>   void mmc_add_card_debugfs(struct mmc_card *card); >>   void mmc_remove_card_debugfs(struct mmc_card *card); >>   +extern bool mmc_can_scale_clk(struct mmc_host *host); >> +extern int mmc_init_clk_scaling(struct mmc_host *host); >> +extern int mmc_exit_clk_scaling(struct mmc_host *host); >> + >>   int mmc_execute_tuning(struct mmc_card *card); >>   int mmc_hs200_to_hs400(struct mmc_card *card); >>   int mmc_hs400_to_hs200(struct mmc_card *card); >> diff --git a/drivers/mmc/core/debugfs.c b/drivers/mmc/core/debugfs.c >> index d2275c5..630ca8e 100644 >> --- a/drivers/mmc/core/debugfs.c >> +++ b/drivers/mmc/core/debugfs.c >> @@ -225,6 +225,43 @@ static int mmc_clock_opt_set(void *data, u64 val) >>   DEFINE_SIMPLE_ATTRIBUTE(mmc_clock_fops, mmc_clock_opt_get, >> mmc_clock_opt_set, >>       "%llu\n"); >>   +#include >> + >> +static int mmc_scale_get(void *data, u64 *val) >> +{ >> +    struct mmc_host *host = data; >> + >> +    *val = host->clk_scaling.curr_freq; >> + >> +    return 0; >> +} >> + >> +static int mmc_scale_set(void *data, u64 val) >> +{ >> +    int err = 0; >> +    struct mmc_host *host = data; >> + >> +    mmc_claim_host(host); >> + >> +    /* change frequency from sysfs manually */ >> +    err = mmc_clk_update_freq(host, val, host->clk_scaling.state); >> +    if (err == -EAGAIN) >> +        err = 0; >> +    else if (err) >> +        pr_err("%s: clock scale to %llu failed with error %d\n", >> +            mmc_hostname(host), val, err); >> +    else >> +        pr_debug("%s: clock change to %llu finished successfully >> (%s)\n", >> +            mmc_hostname(host), val, current->comm); >> + >> +    mmc_release_host(host); >> + >> +    return err; >> +} >> + >> +DEFINE_SIMPLE_ATTRIBUTE(mmc_scale_fops, mmc_scale_get, mmc_scale_set, >> +    "%llu\n"); >> + >>   void mmc_add_host_debugfs(struct mmc_host *host) >>   { >>       struct dentry *root; >> @@ -253,6 +290,15 @@ void mmc_add_host_debugfs(struct mmc_host *host) >>               &mmc_clock_fops)) >>           goto err_node; >>   +    if (!debugfs_create_file("scale", 0600, root, host, >> +        &mmc_scale_fops)) >> +        goto err_node; >> + >> +    if (!debugfs_create_bool("skip_clk_scale_freq_update", >> +        0600, root, >> +        &host->clk_scaling.skip_clk_scale_freq_update)) >> +        goto err_node; >> + >>   #ifdef CONFIG_FAIL_MMC_REQUEST >>       if (fail_request) >>           setup_fault_attr(&fail_default_attr, fail_request); >> diff --git a/drivers/mmc/core/host.c b/drivers/mmc/core/host.c >> index abf9e88..1e46aa4 100644 >> --- a/drivers/mmc/core/host.c >> +++ b/drivers/mmc/core/host.c >> @@ -32,6 +32,10 @@ >>   #include "pwrseq.h" >>   #include "sdio_ops.h" >>   +#define MMC_DEVFRQ_DEFAULT_UP_THRESHOLD 35 >> +#define MMC_DEVFRQ_DEFAULT_DOWN_THRESHOLD 5 >> +#define MMC_DEVFRQ_DEFAULT_POLLING_MSEC 100 >> + >>   #define cls_dev_to_mmc_host(d)    container_of(d, struct mmc_host, >> class_dev) >>     static DEFINE_IDA(mmc_host_ida); >> @@ -435,6 +439,10 @@ int mmc_add_host(struct mmc_host *host) >>           return err; >> led_trigger_register_simple(dev_name(&host->class_dev), &host->led); >> +    host->clk_scaling.upthreshold = MMC_DEVFRQ_DEFAULT_UP_THRESHOLD; >> +    host->clk_scaling.downthreshold = >> MMC_DEVFRQ_DEFAULT_DOWN_THRESHOLD; >> +    host->clk_scaling.polling_delay_ms = >> MMC_DEVFRQ_DEFAULT_POLLING_MSEC; >> +    host->clk_scaling.skip_clk_scale_freq_update = false; >>     #ifdef CONFIG_DEBUG_FS >>       mmc_add_host_debugfs(host); >> diff --git a/drivers/mmc/core/mmc.c b/drivers/mmc/core/mmc.c >> index 4466f5d..c8aedf3 100644 >> --- a/drivers/mmc/core/mmc.c >> +++ b/drivers/mmc/core/mmc.c >> @@ -1526,6 +1526,170 @@ static int mmc_hs200_tuning(struct mmc_card >> *card) >>   } >>     /* >> + * Scale down from HS400 to HS in order to allow frequency change. >> + * This is needed for cards that doesn't support changing frequency >> in HS400 >> + */ >> +static int mmc_scale_low(struct mmc_host *host, unsigned long freq) >> +{ >> +    int err = 0; >> + >> +    mmc_set_timing(host, MMC_TIMING_LEGACY); >> +    mmc_set_clock(host, MMC_HIGH_26_MAX_DTR); >> + >> +    err = mmc_select_hs(host->card); >> +    if (err) { >> +        pr_err("%s: %s: scaling low: failed (%d)\n", >> +               mmc_hostname(host), __func__, err); >> +        return err; >> +    } >> + >> +    err = mmc_select_bus_width(host->card); >> +    if (err < 0) { >> +        pr_err("%s: %s: select_bus_width failed(%d)\n", >> +            mmc_hostname(host), __func__, err); >> +        return err; >> +    } >> + >> +    mmc_set_clock(host, freq); >> + >> +    return 0; >> +} >> + >> +/* >> + * Scale UP from HS to HS200/H400 >> + */ >> +static int mmc_scale_high(struct mmc_host *host) >> +{ >> +    int err = 0; >> + >> +    if (mmc_card_ddr52(host->card)) { >> +        mmc_set_timing(host, MMC_TIMING_LEGACY); >> +        mmc_set_clock(host, MMC_HIGH_26_MAX_DTR); >> +    } >> + >> +    if (!host->card->ext_csd.strobe_support) { >> +        if (!(host->card->mmc_avail_type & EXT_CSD_CARD_TYPE_HS200)) { >> +            pr_err("%s: %s: card does not support HS200\n", >> +                mmc_hostname(host), __func__); >> +            WARN_ON(1); >> +            return -EPERM; >> +        } >> + >> +        err = mmc_select_hs200(host->card); >> +        if (err) { >> +            pr_err("%s: %s: selecting HS200 failed (%d)\n", >> +                mmc_hostname(host), __func__, err); >> +            return err; >> +        } >> + >> +        mmc_set_bus_speed(host->card); >> + >> +        err = mmc_hs200_tuning(host->card); >> +        if (err) { >> +            pr_err("%s: %s: hs200 tuning failed (%d)\n", >> +                mmc_hostname(host), __func__, err); >> +            return err; >> +        } >> + >> +        if (!(host->card->mmc_avail_type & EXT_CSD_CARD_TYPE_HS400)) { >> +            pr_debug("%s: card does not support HS400\n", >> +                mmc_hostname(host)); >> +            return 0; >> +        } >> +    } >> + >> +    err = mmc_select_hs400(host->card); >> +    if (err) { >> +        pr_err("%s: %s: select hs400 failed (%d)\n", >> +            mmc_hostname(host), __func__, err); >> +        return err; >> +    } >> + >> +    return err; >> +} >> + >> +static int mmc_set_clock_bus_speed(struct mmc_card *card, unsigned >> long freq) >> +{ >> +    int err = 0; >> + >> +    if (freq == MMC_HS200_MAX_DTR) >> +        err = mmc_scale_high(card->host); >> +    else >> +        err = mmc_scale_low(card->host, freq); >> + >> +    return err; >> +} >> + >> +static inline unsigned long mmc_ddr_freq_accommodation(unsigned long >> freq) >> +{ >> +    if (freq == MMC_HIGH_DDR_MAX_DTR) >> +        return freq; >> + >> +    return freq/2; >> +} >> + >> +/** >> + * mmc_change_bus_speed() - Change MMC card bus frequency at runtime >> + * @host: pointer to mmc host structure >> + * @freq: pointer to desired frequency to be set >> + * >> + * Change the MMC card bus frequency at runtime after the card is >> + * initialized. Callers are expected to make sure of the card's >> + * state (DATA/RCV/TRANSFER) before changing the frequency at runtime. >> + * >> + * If the frequency to change is greater than max. supported by card, >> + * *freq is changed to max. supported by card. If it is less than min. >> + * supported by host, *freq is changed to min. supported by host. >> + * Host is assumed to be calimed while calling this funciton. >> + */ >> +static int mmc_change_bus_speed(struct mmc_host *host, unsigned long >> *freq) >> +{ >> +    int err = 0; >> +    struct mmc_card *card; >> +    unsigned long actual_freq; >> + >> +    card = host->card; >> + >> +    if (!card || !freq) { >> +        err = -EINVAL; >> +        goto out; >> +    } >> +    actual_freq = *freq; >> + >> +    WARN_ON(!host->claimed); >> + >> +    /* >> +     * For scaling up/down HS400 we'll need special handling, >> +     * for other timings we can simply do clock frequency change >> +     */ >> +    if (mmc_card_hs400(card) || >> +        (!mmc_card_hs200(host->card) && *freq == MMC_HS200_MAX_DTR)) { >> +        err = mmc_set_clock_bus_speed(card, *freq); >> +        if (err) { >> +            pr_err("%s: %s: failed (%d)to set bus and clock speed >> (freq=%lu)\n", >> +                mmc_hostname(host), __func__, err, *freq); >> +            goto out; >> +        } >> +    } else if (mmc_card_hs200(host->card)) { >> +        mmc_set_clock(host, *freq); >> +        err = mmc_hs200_tuning(host->card); >> +        if (err) { >> +            pr_warn("%s: %s: tuning execution failed %d\n", >> +                mmc_hostname(card->host), >> +                __func__, err); >> +            mmc_set_clock(host, host->clk_scaling.curr_freq); >> +        } >> +    } else { >> +        if (mmc_card_ddr52(host->card)) >> +            actual_freq = mmc_ddr_freq_accommodation(*freq); >> +        mmc_set_clock(host, actual_freq); >> +    } >> + >> +out: >> +    return err; >> +} >> + >> +/* >>    * Handle the detection and initialisation of a card. >>    * >>    * In the case of a resume, "oldcard" will contain the card >> @@ -1751,6 +1915,16 @@ static int mmc_init_card(struct mmc_host >> *host, u32 ocr, >>           } >>       } >>   +    card->clk_scaling_lowest = host->f_min; >> +    if ((card->mmc_avail_type & EXT_CSD_CARD_TYPE_HS400) || >> +            (card->mmc_avail_type & EXT_CSD_CARD_TYPE_HS200)) >> +        card->clk_scaling_highest = card->ext_csd.hs200_max_dtr; >> +    else if ((card->mmc_avail_type & EXT_CSD_CARD_TYPE_HS) || >> +            (card->mmc_avail_type & EXT_CSD_CARD_TYPE_DDR_52)) >> +        card->clk_scaling_highest = card->ext_csd.hs_max_dtr; >> +    else >> +        card->clk_scaling_highest = card->csd.max_dtr; >> + >>       /* >>        * Choose the power class with selected bus interface >>        */ >> @@ -1942,6 +2116,7 @@ static int mmc_poweroff_notify(struct mmc_card >> *card, unsigned int notify_type) >>    */ >>   static void mmc_remove(struct mmc_host *host) >>   { >> +    mmc_exit_clk_scaling(host); >>       mmc_remove_card(host->card); >>       host->card = NULL; >>   } >> @@ -2064,6 +2239,13 @@ static int mmc_shutdown(struct mmc_host *host) >>       int err = 0; >>         /* >> +     * Exit clock scaling so that it doesn't kick in after >> +     * power off notification is sent >> +     */ >> +    if (host->caps2 & MMC_CAP2_CLK_SCALE) >> +        mmc_exit_clk_scaling(host); >> + >> +    /* >>        * In a specific case for poweroff notify, we need to resume >> the card >>        * before we can shutdown it properly. >>        */ >> @@ -2132,6 +2314,7 @@ static int mmc_can_reset(struct mmc_card *card) >>   static int _mmc_hw_reset(struct mmc_host *host) >>   { >>       struct mmc_card *card = host->card; >> +    int ret; >>         /* >>        * In the case of recovery, we can't expect flushing the cache >> to work >> @@ -2151,7 +2334,15 @@ static int _mmc_hw_reset(struct mmc_host *host) >>           mmc_power_cycle(host, card->ocr); >>           mmc_pwrseq_reset(host); >>       } >> -    return mmc_init_card(host, card->ocr, card); >> + >> +    ret = mmc_init_card(host, card->ocr, card); >> +    if (ret) { >> +        pr_err("%s: %s: mmc_init_card failed (%d)\n", >> +            mmc_hostname(host), __func__, ret); >> +        return ret; >> +    } >> + >> +    return ret; >>   } >>     static const struct mmc_bus_ops mmc_ops = { >> @@ -2164,6 +2355,7 @@ static int _mmc_hw_reset(struct mmc_host *host) >>       .alive = mmc_alive, >>       .shutdown = mmc_shutdown, >>       .hw_reset = _mmc_hw_reset, >> +    .change_bus_speed = mmc_change_bus_speed, >>   }; >>     /* >> @@ -2220,6 +2412,12 @@ int mmc_attach_mmc(struct mmc_host *host) >>           goto remove_card; >>         mmc_claim_host(host); >> +    err = mmc_init_clk_scaling(host); >> +    if (err) { >> +        mmc_release_host(host); >> +        goto remove_card; >> +    } >> + >>       return 0; >>     remove_card: >> diff --git a/drivers/mmc/core/sd.c b/drivers/mmc/core/sd.c >> index d0d9f90..40144c1 100644 >> --- a/drivers/mmc/core/sd.c >> +++ b/drivers/mmc/core/sd.c >> @@ -892,7 +892,10 @@ unsigned mmc_sd_get_max_clock(struct mmc_card >> *card) >>   { >>       unsigned max_dtr = (unsigned int)-1; >>   -    if (mmc_card_hs(card)) { >> +    if (mmc_card_uhs(card)) { >> +        if (max_dtr > card->sw_caps.uhs_max_dtr) >> +            max_dtr = card->sw_caps.uhs_max_dtr; >> +    } else if (mmc_card_hs(card)) { >>           if (max_dtr > card->sw_caps.hs_max_dtr) >>               max_dtr = card->sw_caps.hs_max_dtr; >>       } else if (max_dtr > card->csd.max_dtr) { >> @@ -1059,6 +1062,9 @@ static int mmc_sd_init_card(struct mmc_host >> *host, u32 ocr, >>           } >>       } >>   +    card->clk_scaling_highest = mmc_sd_get_max_clock(card); >> +    card->clk_scaling_lowest = host->f_min; >> + >>       if (host->caps2 & MMC_CAP2_AVOID_3_3V && >>           host->ios.signal_voltage == MMC_SIGNAL_VOLTAGE_330) { >>           pr_err("%s: Host failed to negotiate down from 3.3V\n", >> @@ -1082,6 +1088,7 @@ static int mmc_sd_init_card(struct mmc_host >> *host, u32 ocr, >>    */ >>   static void mmc_sd_remove(struct mmc_host *host) >>   { >> +    mmc_exit_clk_scaling(host); >>       mmc_remove_card(host->card); >>       host->card = NULL; >>   } >> @@ -1228,6 +1235,62 @@ static int mmc_sd_hw_reset(struct mmc_host *host) >>       return mmc_sd_init_card(host, host->card->ocr, host->card); >>   } >>   +/** >> + * mmc_sd_change_bus_speed() - Change SD card bus frequency at runtime >> + * @host: pointer to mmc host structure >> + * @freq: pointer to desired frequency to be set >> + * >> + * Change the SD card bus frequency at runtime after the card is >> + * initialized. Callers are expected to make sure of the card's >> + * state (DATA/RCV/TRANSFER) beforing changing the frequency at >> runtime. >> + * >> + * If the frequency to change is greater than max. supported by card, >> + * *freq is changed to max. supported by card and if it is less than >> min. >> + * supported by host, *freq is changed to min. supported by host. >> + */ >> +static int mmc_sd_change_bus_speed(struct mmc_host *host, unsigned >> long *freq) >> +{ >> +    int err = 0; >> +    struct mmc_card *card; >> + >> +    mmc_claim_host(host); >> +    /* >> +     * Assign card pointer after claiming host to avoid race >> +     * conditions that may arise during removal of the card. >> +     */ >> +    card = host->card; >> + >> +    /* sanity checks */ >> +    if (!card || !freq) { >> +        err = -EINVAL; >> +        goto out; >> +    } >> + >> +    mmc_set_clock(host, (unsigned int) (*freq)); >> + >> +    if (!mmc_host_is_spi(card->host) && mmc_card_uhs(card) >> +            && card->host->ops->execute_tuning) { >> +        /* >> +         * We try to probe host driver for tuning for any >> +         * frequency, it is host driver responsibility to >> +         * perform actual tuning only when required. >> +         */ >> +        err = card->host->ops->execute_tuning(card->host, >> +                MMC_SEND_TUNING_BLOCK); >> + >> +        if (err) { >> +            pr_warn("%s: %s: tuning execution failed %d. Restoring >> to previous clock %lu\n", >> +                   mmc_hostname(card->host), __func__, err, >> +                   host->clk_scaling.curr_freq); >> +            mmc_set_clock(host, host->clk_scaling.curr_freq); >> +        } >> +    } >> + >> +out: >> +    mmc_release_host(host); >> +    return err; >> +} >> + >>   static const struct mmc_bus_ops mmc_sd_ops = { >>       .remove = mmc_sd_remove, >>       .detect = mmc_sd_detect, >> @@ -1238,6 +1301,7 @@ static int mmc_sd_hw_reset(struct mmc_host *host) >>       .alive = mmc_sd_alive, >>       .shutdown = mmc_sd_suspend, >>       .hw_reset = mmc_sd_hw_reset, >> +    .change_bus_speed = mmc_sd_change_bus_speed, >>   }; >>     /* >> @@ -1292,6 +1356,12 @@ int mmc_attach_sd(struct mmc_host *host) >>           goto remove_card; >>         mmc_claim_host(host); >> +    err = mmc_init_clk_scaling(host); >> +    if (err) { >> +        mmc_release_host(host); >> +        goto remove_card; >> +    } >> + >>       return 0; >>     remove_card: >> diff --git a/drivers/mmc/host/sdhci-msm.c b/drivers/mmc/host/sdhci-msm.c >> index b5519a5..e9fe8c6 100644 >> --- a/drivers/mmc/host/sdhci-msm.c >> +++ b/drivers/mmc/host/sdhci-msm.c >> @@ -1705,6 +1705,43 @@ static int sdhci_msm_register_vreg(struct >> sdhci_msm_host *msm_host) >>     MODULE_DEVICE_TABLE(of, sdhci_msm_dt_match); >>   +int sdhci_msm_dt_get_array(struct device *dev, const char *prop_name, >> +                 u32 **out, int *len, u32 size) >> +{ >> +    int ret = 0; >> +    struct device_node *np = dev->of_node; >> +    size_t sz; >> +    u32 *arr = NULL; >> + >> +    if (!of_get_property(np, prop_name, len)) { >> +        ret = -EINVAL; >> +        goto out; >> +    } >> +    sz = *len = *len / sizeof(*arr); >> +    if (sz <= 0 || (size > 0 && (sz > size))) { >> +        dev_err(dev, "%s invalid size\n", prop_name); >> +        ret = -EINVAL; >> +        goto out; >> +    } >> + >> +    arr = devm_kzalloc(dev, sz * sizeof(*arr), GFP_KERNEL); >> +    if (!arr) { >> +        ret = -ENOMEM; >> +        goto out; >> +    } >> + >> +    ret = of_property_read_u32_array(np, prop_name, arr, sz); >> +    if (ret < 0) { >> +        dev_err(dev, "%s failed reading array %d\n", prop_name, ret); >> +        goto out; >> +    } >> +    *out = arr; >> +out: >> +    if (ret) >> +        *len = 0; >> +    return ret; >> +} >> + >>   static const struct sdhci_ops sdhci_msm_ops = { >>       .reset = sdhci_reset, >>       .set_clock = sdhci_msm_set_clock, >> diff --git a/drivers/mmc/host/sdhci-pltfm.c >> b/drivers/mmc/host/sdhci-pltfm.c >> index 02bea61..354fc68 100644 >> --- a/drivers/mmc/host/sdhci-pltfm.c >> +++ b/drivers/mmc/host/sdhci-pltfm.c >> @@ -36,6 +36,9 @@ >>   #endif >>   #include "sdhci-pltfm.h" >>   +int sdhci_msm_dt_get_array(struct device *dev, const char *prop_name, >> +                 u32 **out, int *len, u32 size); >> + >>   unsigned int sdhci_pltfm_clk_get_max_clock(struct sdhci_host *host) >>   { >>       struct sdhci_pltfm_host *pltfm_host = sdhci_priv(host); >> @@ -101,6 +104,14 @@ void sdhci_get_of_property(struct >> platform_device *pdev) >>         of_property_read_u32(np, "clock-frequency", &pltfm_host->clock); >>   +    if (sdhci_msm_dt_get_array(&pdev->dev, "qcom,devfreq,freq-table", >> +            &host->mmc->clk_scaling.pltfm_freq_table, >> + &host->mmc->clk_scaling.pltfm_freq_table_sz, 0)) >> +        pr_debug("no clock scaling frequencies were supplied\n"); >> +    else if (!host->mmc->clk_scaling.pltfm_freq_table || >> +            !host->mmc->clk_scaling.pltfm_freq_table_sz) >> +        pr_err("bad dts clock scaling frequencies\n"); >> + >>       if (of_find_property(np, "keep-power-in-suspend", NULL)) >>           host->mmc->pm_caps |= MMC_PM_KEEP_POWER; >>   diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c >> index 162b9af..f0aafab 100644 >> --- a/drivers/mmc/host/sdhci.c >> +++ b/drivers/mmc/host/sdhci.c >> @@ -2427,6 +2427,32 @@ static void sdhci_card_event(struct mmc_host >> *mmc) >>       spin_unlock_irqrestore(&host->lock, flags); >>   } >>   +static inline void sdhci_update_power_policy(struct sdhci_host *host, >> +        enum sdhci_power_policy policy) >> +{ >> +    host->power_policy = policy; >> +} >> + >> +static int sdhci_notify_load(struct mmc_host *mmc, enum mmc_load state) >> +{ >> +    int err = 0; >> +    struct sdhci_host *host = mmc_priv(mmc); >> + >> +    switch (state) { >> +    case MMC_LOAD_HIGH: >> +        sdhci_update_power_policy(host, SDHCI_PERFORMANCE_MODE); >> +        break; >> +    case MMC_LOAD_LOW: >> +        sdhci_update_power_policy(host, SDHCI_POWER_SAVE_MODE); >> +        break; >> +    default: >> +        err = -EINVAL; >> +        break; >> +    } >> + >> +    return err; >> +} >> + >>   static const struct mmc_host_ops sdhci_ops = { >>       .request    = sdhci_request, >>       .post_req    = sdhci_post_req, >> @@ -2441,6 +2467,7 @@ static void sdhci_card_event(struct mmc_host *mmc) >>       .execute_tuning            = sdhci_execute_tuning, >>       .card_event            = sdhci_card_event, >>       .card_busy    = sdhci_card_busy, >> +    .notify_load = sdhci_notify_load, >>   }; >> /*****************************************************************************\ >> diff --git a/drivers/mmc/host/sdhci.h b/drivers/mmc/host/sdhci.h >> index 3b0c97a..740471f 100644 >> --- a/drivers/mmc/host/sdhci.h >> +++ b/drivers/mmc/host/sdhci.h >> @@ -346,6 +346,12 @@ enum sdhci_cookie { >>       COOKIE_MAPPED,        /* mapped by sdhci_prepare_data() */ >>   }; >>   +enum sdhci_power_policy { >> +    SDHCI_PERFORMANCE_MODE, >> +    SDHCI_POWER_SAVE_MODE, >> +    SDHCI_POWER_POLICY_NUM /* Always keep this one last */ >> +}; >> + >>   struct sdhci_host { >>       /* Data set by hardware interface driver */ >>       const char *hw_name;    /* Hardware bus name */ >> @@ -562,6 +568,8 @@ struct sdhci_host { >>       /* Delay (ms) between tuning commands */ >>       int            tuning_delay; >>   +    enum sdhci_power_policy power_policy; >> + >>       /* Host SDMA buffer boundary. */ >>       u32            sdma_boundary; >>   diff --git a/include/linux/mmc/card.h b/include/linux/mmc/card.h >> index de73778..c713581 100644 >> --- a/include/linux/mmc/card.h >> +++ b/include/linux/mmc/card.h >> @@ -245,6 +245,10 @@ struct mmc_card { >>       struct mmc_host        *host;        /* the host this device >> belongs to */ >>       struct device        dev;        /* the device */ >>       u32            ocr;        /* the current OCR setting */ >> +    unsigned long        clk_scaling_lowest;    /* lowest scaleable*/ >> +                            /* frequency */ >> +    unsigned long        clk_scaling_highest;    /* highest >> scaleable */ >> + >>       unsigned int        rca;        /* relative card address of >> device */ >>       unsigned int        type;        /* card type */ >>   #define MMC_TYPE_MMC        0        /* MMC card */ >> @@ -308,6 +312,7 @@ struct mmc_card { >>       unsigned int    nr_parts; >>         unsigned int        bouncesz;    /* Bounce buffer size */ >> +    unsigned int        part_curr; >>   }; >>     static inline bool mmc_large_sector(struct mmc_card *card) >> diff --git a/include/linux/mmc/host.h b/include/linux/mmc/host.h >> index 64300a4..321ab39 100644 >> --- a/include/linux/mmc/host.h >> +++ b/include/linux/mmc/host.h >> @@ -12,6 +12,7 @@ >>     #include >>   #include >> +#include >>   #include >>     #include >> @@ -82,6 +83,12 @@ struct mmc_ios { >>     struct mmc_host; >>   +/* states to represent load on the host */ >> +enum mmc_load { >> +    MMC_LOAD_HIGH, >> +    MMC_LOAD_LOW, >> +}; >> + >>   struct mmc_host_ops { >>       /* >>        * It is optional for the host to implement pre_req and >> post_req in >> @@ -161,6 +168,7 @@ struct mmc_host_ops { >>        */ >>       int    (*multi_io_quirk)(struct mmc_card *card, >>                     unsigned int direction, int blk_size); >> +    int    (*notify_load)(struct mmc_host *, enum mmc_load); >>   }; >>     struct mmc_cqe_ops { >> @@ -260,9 +268,60 @@ struct mmc_ctx { >>       struct task_struct *task; >>   }; >>   +/** >> + * struct mmc_devfeq_clk_scaling - main context for MMC clock >> scaling logic >> + * >> + * @lock: spinlock to protect statistics >> + * @devfreq: struct that represent mmc-host as a client for devfreq >> + * @devfreq_profile: MMC device profile, mostly polling interval and >> callbacks >> + * @ondemand_gov_data: struct supplied to ondemmand governor >> (thresholds) >> + * @state: load state, can be HIGH or LOW. used to notify >> mmc_host_ops callback >> + * @start_busy: timestamped armed once a data request is started >> + * @measure_interval_start: timestamped armed once a measure >> interval started >> + * @devfreq_abort: flag to sync between different contexts relevant >> to devfreq >> + * @skip_clk_scale_freq_update: flag that enable/disable frequency >> change >> + * @freq_table_sz: table size of frequencies supplied to devfreq >> + * @freq_table: frequencies table supplied to devfreq >> + * @curr_freq: current frequency >> + * @polling_delay_ms: polling interval for status collection used by >> devfreq >> + * @upthreshold: up-threshold supplied to ondemand governor >> + * @downthreshold: down-threshold supplied to ondemand governor >> + * @need_freq_change: flag indicating if a frequency change is required >> + * @clk_scaling_in_progress: flag indicating if there's ongoing >> frequency change >> + * @is_busy_started: flag indicating if a request is handled by the HW >> + * @enable: flag indicating if the clock scaling logic is enabled >> for this host >> + */ >> +struct mmc_devfeq_clk_scaling { >> +    spinlock_t    lock; >> +    struct        devfreq *devfreq; >> +    struct        devfreq_dev_profile devfreq_profile; >> +    struct        devfreq_simple_ondemand_data ondemand_gov_data; >> +    enum mmc_load    state; >> +    ktime_t        start_busy; >> +    ktime_t        measure_interval_start; >> +    atomic_t    devfreq_abort; >> +    bool        skip_clk_scale_freq_update; >> +    int        freq_table_sz; >> +    int        pltfm_freq_table_sz; >> +    u32        *freq_table; >> +    u32        *pltfm_freq_table; >> +    unsigned long    total_busy_time_us; >> +    unsigned long    target_freq; >> +    unsigned long    curr_freq; >> +    unsigned long    polling_delay_ms; >> +    unsigned int    upthreshold; >> +    unsigned int    downthreshold; >> +    bool        need_freq_change; >> +    bool        clk_scaling_in_progress; >> +    bool        is_busy_started; >> +    bool        enable; >> +}; >> + >> + >>   struct mmc_host { >>       struct device        *parent; >>       struct device        class_dev; >> +    struct mmc_devfeq_clk_scaling    clk_scaling; >>       int            index; >>       const struct mmc_host_ops *ops; >>       struct mmc_pwrseq    *pwrseq; >> @@ -360,6 +419,7 @@ struct mmc_host { >>   #define MMC_CAP2_CQE        (1 << 23)    /* Has eMMC command queue >> engine */ >>   #define MMC_CAP2_CQE_DCMD    (1 << 24)    /* CQE can issue a direct >> command */ >>   #define MMC_CAP2_AVOID_3_3V    (1 << 25)    /* Host must negotiate >> down from 3.3V */ >> +#define MMC_CAP2_CLK_SCALE      (1 << 26)       /* Allow dynamic clk >> scaling */ >>         int            fixed_drv_type;    /* fixed driver type for >> non-removable media */ >>   @@ -523,6 +583,16 @@ static inline int >> mmc_regulator_set_vqmmc(struct mmc_host *mmc, >>   u32 mmc_vddrange_to_ocrmask(int vdd_min, int vdd_max); >>   int mmc_regulator_get_supply(struct mmc_host *mmc); >>   +static inline void mmc_host_clear_sdr104(struct mmc_host *host) >> +{ >> +    host->caps &= ~MMC_CAP_UHS_SDR104; >> +} >> + >> +static inline void mmc_host_set_sdr104(struct mmc_host *host) >> +{ >> +    host->caps |= MMC_CAP_UHS_SDR104; >> +} >> + >>   static inline int mmc_card_is_removable(struct mmc_host *host) >>   { >>       return !(host->caps & MMC_CAP_NONREMOVABLE); >> > > Thanks, > Vijay