linux-kernel - Re: [PATCH V2] mmc: sdhci-msm: Disable CQE during SDHC reset

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <0e737f52-767f-05d4-829b-4f76c084062c@codeaurora.org>
Date:   Wed, 4 Mar 2020 22:20:52 +0530
From:   Veerabhadrarao Badiganti <vbadigan@...eaurora.org>
To:     Adrian Hunter <adrian.hunter@...el.com>, ulf.hansson@...aro.org
Cc:     linux-mmc@...r.kernel.org, linux-kernel@...r.kernel.org,
        linux-arm-msm@...r.kernel.org, Andy Gross <agross@...nel.org>,
        Bjorn Andersson <bjorn.andersson@...aro.org>
Subject: Re: [PATCH V2] mmc: sdhci-msm: Disable CQE during SDHC reset


On 3/4/2020 7:40 PM, Adrian Hunter wrote:
> On 4/03/20 3:10 pm, Veerabhadrarao Badiganti wrote:
>> Hi Adrian
>>
>> On 3/4/2020 5:58 PM, Adrian Hunter wrote:
>>> On 4/03/20 1:54 pm, Veerabhadrarao Badiganti wrote:
>>>> When SDHC gets reset (E.g. in suspend path), CQE also gets reset
>>>> and goes to disable state. But s/w state still points it as CQE
>>>> is in enabled state. Since s/w and h/w states goes out of sync,
>>>> it results in s/w request timeout for subsequent CQE requests.
>>>>
>>>> To synchronize CQE s/w and h/w state during SDHC reset,
>>>> explicitly disable CQE after reset.
>>> Shouldn't you be calling cqhci_suspend() / cqhci_resume() in the suspend and
>>> resume paths?
>> This issue is seen during mmc runtime suspend.  I can add it
>> sdhci_msm_runtime_suspend
>>
>> but sdhci_msm runtime delay is aggressive, its 50ms. It may get invoked very
>> frequently.
>>
>> So Im of the opinion that disabling CQE very often from platform runtime
>> suspend is overkill.
> It doesn't look like sdhci-msm calls any sdhci.c pm ops, so how does SDHC
> get reset?

With MMC_CAP_AGGRESSIVE_PM flag enabled, it getting called from 
mmc_runtime_suspend()

Below is the call stack()

    sdhci_reset
   sdhci_do_reset
   sdhci_init
   sdhci_set_ios
   mmc_set_initial_state
   mmc_power_off
  _mmc_suspend
   mmc_runtime_suspend

>>>> Signed-off-by: Veerabhadrarao Badiganti <vbadigan@...eaurora.org>
>>>> ---
>>>> Changes since V1:
>>>>      - Disable CQE only when SDHC undergoes s/w reset for all.
>>>> ---
>>>>    drivers/mmc/host/sdhci-msm.c | 9 ++++++++-
>>>>    1 file changed, 8 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/drivers/mmc/host/sdhci-msm.c b/drivers/mmc/host/sdhci-msm.c
>>>> index 53b79ee..75929d3 100644
>>>> --- a/drivers/mmc/host/sdhci-msm.c
>>>> +++ b/drivers/mmc/host/sdhci-msm.c
>>>> @@ -1823,6 +1823,13 @@ static void sdhci_msm_set_regulator_caps(struct
>>>> sdhci_msm_host *msm_host)
>>>>        pr_debug("%s: supported caps: 0x%08x\n", mmc_hostname(mmc), caps);
>>>>    }
>>>>    +static void sdhci_msm_reset(struct sdhci_host *host, u8 mask)
>>>> +{
>>>> +    sdhci_reset(host, mask);
>>>> +    if ((host->mmc->caps2 & MMC_CAP2_CQE) && (mask & SDHCI_RESET_ALL))
>>>> +        cqhci_suspend(host->mmc);
>>>> +}
>>>> +
>>>>    static const struct sdhci_msm_variant_ops mci_var_ops = {
>>>>        .msm_readl_relaxed = sdhci_msm_mci_variant_readl_relaxed,
>>>>        .msm_writel_relaxed = sdhci_msm_mci_variant_writel_relaxed,
>>>> @@ -1861,7 +1868,7 @@ static void sdhci_msm_set_regulator_caps(struct
>>>> sdhci_msm_host *msm_host)
>>>>    MODULE_DEVICE_TABLE(of, sdhci_msm_dt_match);
>>>>      static const struct sdhci_ops sdhci_msm_ops = {
>>>> -    .reset = sdhci_reset,
>>>> +    .reset = sdhci_msm_reset,
>>>>        .set_clock = sdhci_msm_set_clock,
>>>>        .get_min_clock = sdhci_msm_get_min_clock,
>>>>        .get_max_clock = sdhci_msm_get_max_clock,
>>>>