From mboxrd@z Thu Jan 1 00:00:00 1970 From: Veerabhadrarao Badiganti Subject: Re: [PATCH v5 1/5] mmc: core: Extend mmc_of_parse() to parse CQE bindings Date: Thu, 7 May 2020 22:03:14 +0530 Message-ID: References: <1588031768-23677-1-git-send-email-chun-hung.wu@mediatek.com> <1588031768-23677-2-git-send-email-chun-hung.wu@mediatek.com> <9bc2454f-0b42-e256-7927-2564b56f369f@codeaurora.org> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit Return-path: In-Reply-To: Content-Language: en-US Sender: linux-arm-msm-owner@vger.kernel.org To: Ulf Hansson Cc: Chun-Hung Wu , =?UTF-8?B?TWljaGHFgiBNaXJvc8WCYXc=?= , Jonathan Hunter , Al Cooper , Adrian Hunter , Florian Fainelli , BCM Kernel Feedback , Andy Gross , Bjorn Andersson , Michal Simek , Thierry Reding , Chaotian Jing , Rob Herring , Mark Rutland , Matthias Brugger , Linus Walleij , Pavel Machek , Kate Stewart List-Id: linux-tegra@vger.kernel.org On 5/6/2020 10:06 PM, Ulf Hansson wrote: > On Wed, 6 May 2020 at 15:01, Veerabhadrarao Badiganti > wrote: >> >> On 4/28/2020 5:26 AM, Chun-Hung Wu wrote: >>> Parse CQE bindings "supports-cqe" and "disable-cqe-dcmd" >>> in mmc_of_parse(). >>> >>> Signed-off-by: Chun-Hung Wu >>> --- >>> drivers/mmc/core/host.c | 5 +++++ >>> 1 file changed, 5 insertions(+) >>> >>> diff --git a/drivers/mmc/core/host.c b/drivers/mmc/core/host.c >>> index c876872..47521c6 100644 >>> --- a/drivers/mmc/core/host.c >>> +++ b/drivers/mmc/core/host.c >>> @@ -302,6 +302,11 @@ int mmc_of_parse(struct mmc_host *host) >>> host->caps2 |= MMC_CAP2_NO_SD; >>> if (device_property_read_bool(dev, "no-mmc")) >>> host->caps2 |= MMC_CAP2_NO_MMC; >>> + if (device_property_read_bool(dev, "supports-cqe")) >>> + host->caps2 |= MMC_CAP2_CQE; >> This change is breaking emmc driver on qcom platforms where this dt >> property is defined. >> >> [ 1.543453] cqhci_deactivate+0xc/0x38 >> [ 1.545627] sdhci_msm_reset+0x40/0x58 >> [ 1.549447] sdhci_do_reset+0x48/0x7c >> [ 1.553180] __sdhci_read_caps+0x7c/0x214 >> [ 1.556913] sdhci_setup_host+0x58/0xce8 >> [ 1.560905] sdhci_msm_probe+0x588/0x8a4 >> [ 1.564900] platform_drv_probe+0x4c/0xb0 >> >> So, we cant have this flag defined before sdhci_setup_host(). >> >> I will have to clear this cap and re-enable it in our initialization. > Thanks for reporting! I have dropped all the four patches from > Chun-Hung, so we can figure out how to fix this. > > Please help to review the next version of the series. Thanks Ulf. Hi Chun-Hung, On qcom controller CQE also gets reset when SDHC is reset. So we have to explicitly disable CQE by invoking  cqhci_deactivate() during sdhc reset SDHC gets reset in sdhci_setup_host() even before cqe is initialized. With MMC_CAP2_CQE_DCMD cap set even before sdhci_set_host(), we are getting null pointer access with cqhci_deactivate(). If CQE getting reset with SDHC reset is generic (applicable to other controllers) then you have revisit your logic. If its not the case then only qcom driver would get affected. I see you are updating sdhci-msm.c file as-well. How about including below change besides your change? @@ -1658,6 +1658,8 @@ static int sdhci_msm_cqe_add_host(struct sdhci_host *host,         if (host->caps & SDHCI_CAN_64BIT)                 host->alloc_desc_sz = 16; +       /* Clear the CQE cap during setup host */ +       msm_host->mmc->caps2 &= ~MMC_CAP2_CQE; +         ret = sdhci_setup_host(host); >>> + if (!device_property_read_bool(dev, "disable-cqe-dcmd")) { >>> + host->caps2 |= MMC_CAP2_CQE_DCMD; >>> + } >>> >>> /* Must be after "non-removable" check */ >>> if (device_property_read_u32(dev, "fixed-emmc-driver-type", &drv_type) == 0) { > Kind regards > Uffe