linux-mmc.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Scott Branden <sbranden@broadcom.com>
To: Stefan Wahren <info@lategoodbye.de>,
	Stephen Warren <swarren@wwwdotorg.org>,
	Ulf Hansson <ulf.hansson@linaro.org>,
	Russell King <rmk+kernel@arm.linux.org.uk>,
	Peter Griffin <peter.griffin@linaro.org>,
	Chris Ball <chris@printf.net>, Piotr Krol <pietrushnic@gmail.com>
Cc: Ray Jui <rjui@broadcom.com>,
	linux-mmc@vger.kernel.org, linux-kernel@vger.kernel.org,
	bcm-kernel-feedback-list@broadcom.com,
	linux-rpi-kernel@lists.infradead.org,
	Joe Perches <joe@perches.com>
Subject: Re: [PATCHv2 4/5] mmc: shdci-bcm2835: add verify for 32-bit back-to-back workaround
Date: Tue, 22 Dec 2015 11:23:37 -0800	[thread overview]
Message-ID: <5679A339.9000302@broadcom.com> (raw)
In-Reply-To: <5679726A.1070300@lategoodbye.de>

Hi Stefan,

On 15-12-22 07:55 AM, Stefan Wahren wrote:
> Hi Scott,
>
> Am 07.11.2014 um 19:31 schrieb Scott Branden:
>> On 14-11-05 09:01 PM, Stephen Warren wrote:
>>> On 11/05/2014 12:00 AM, Scott Branden wrote:
>>>> On 14-11-04 08:59 PM, Stephen Warren wrote:
>>>>> On 10/30/2014 12:36 AM, Scott Branden wrote:
>>>>>> Add a verify option to driver to print out an error message if a
>>>>>> potential back to back write could cause a clock domain issue.
>>>>>
>>>>>> index f8c450a..11af27f 100644
>>>>>
>>>>>> +#ifdef CONFIG_MMC_SDHCI_BCM2835_VERIFY_WORKAROUND
>>>>>> +    struct sdhci_pltfm_host *pltfm_host = sdhci_priv(host);
>>>>>> +    struct bcm2835_sdhci_host *bcm2835_host = pltfm_host->priv;
>>>>>> +
>>>>>> +    if (bcm2835_host->previous_reg == reg) {
>>>>>> +        if ((reg != SDHCI_HOST_CONTROL)
>>>>>> +            && (reg != SDHCI_CLOCK_CONTROL)) {
>>>>>
>>>>> The comment in patch 3 says the problem doesn't apply to the data
>>>>> register. Why does this check for these two registers rather than data?
>>>> This Verify workaround patch still a work in progress.  I'm still
>>>> getting more info from the silicon designers on the back-to-back
>>>> register writes that are affect.  The spew of 0x20 or 0x28 or 0x2c
>>>> register writes are all ok locations that don't need to be worked
>>>> around.  This patch needs to be corrected with the proper register rules
>>>> still.
>> Thanks for testing.  Yes, I have work to do on the verify patch above
>> still.
>
> do you still have plans to submit a V3 of this patch series?
No, I do not have plans to submit a V3 of this patch series.

I submitted this patch as RPI has a similar controller to the SoCs I am 
familiar with as well as needing similar work arounds   You can take 
over the patchset.  Or, try and get the sdhci-iproc.c driver going on 
RPI.  The sdhci-iproc is the production driver we use on a variety of 
SoCs and support and test this driver.
>
> I attached an improved version of this patch which avoids a possible
> endless loop caused by the dev_err call. So only the first occurence
> of a specific register will be logged.
OK, but is this really necessary?  If veryify workaround ever prints 
anything then the driver workarounds aren't doing what it is supposed to 
anyway?
>
> Regards
> Stefan
>
>
> -------------------8<-------------------------------------------
>
> diff --git a/drivers/mmc/host/Kconfig b/drivers/mmc/host/Kconfig
> index 1526b8a..7b0990f 100644
> --- a/drivers/mmc/host/Kconfig
> +++ b/drivers/mmc/host/Kconfig
> @@ -306,6 +306,15 @@ config MMC_SDHCI_BCM2835
>
>   	  If unsure, say N.
>
> +config MMC_SDHCI_BCM2835_VERIFY_WORKAROUND
> +	bool "Verify BCM2835 workaround does not do back to back writes"
> +	depends on MMC_SDHCI_BCM2835
> +	default y
> +	help
> +	  This enables code that verifies the bcm2835 workaround.
> +	  The verification code checks that back to back writes to the same
> +	  register do not occur.
> +
>   config MMC_SDHCI_F_SDH30
>   	tristate "SDHCI support for Fujitsu Semiconductor F_SDH30"
>   	depends on MMC_SDHCI_PLTFM
> diff --git a/drivers/mmc/host/sdhci-bcm2835.c
> b/drivers/mmc/host/sdhci-bcm2835.c
> index 01ce193d..c1c70df 100644
> --- a/drivers/mmc/host/sdhci-bcm2835.c
> +++ b/drivers/mmc/host/sdhci-bcm2835.c
> @@ -20,15 +20,27 @@
>    */
>
>   #include <linux/delay.h>
> +#include <linux/hashtable.h>
>   #include <linux/module.h>
>   #include <linux/mmc/host.h>
> +#include <linux/slab.h>
>   #include "sdhci-pltfm.h"
>
>   struct bcm2835_sdhci_host {
>   	u32 shadow_cmd;
>   	u32 shadow_blk;
> +	int previous_reg;
>   };
>
> +struct reg_hash {
> +	struct hlist_node node;
> +	int reg;
> +};
> +
> +#define BCM2835_REG_HT_BITS 4
> +
> +static DEFINE_HASHTABLE(bcm2835_used_regs, BCM2835_REG_HT_BITS);
> +
>   #define REG_OFFSET_IN_BITS(reg) ((reg) << 3 & 0x18)
>
>   static inline u32 bcm2835_sdhci_readl(struct sdhci_host *host, int reg)
> @@ -56,8 +68,37 @@ static u8 bcm2835_sdhci_readb(struct sdhci_host
> *host, int reg)
>   }
>
>   static inline void bcm2835_sdhci_writel(struct sdhci_host *host,
> +					u32 val, int reg)
> +{
> +	writel(val, host->ioaddr + reg);
> +}
> +
> +static inline void bcm2835_sdhci_writel_verify(struct sdhci_host *host,
>   						u32 val, int reg)
>   {
> +	struct sdhci_pltfm_host *pltfm_host = sdhci_priv(host);
> +	struct bcm2835_sdhci_host *bcm2835_host = pltfm_host->priv;
> +	struct reg_hash *rh;
> +	struct hlist_head *head;
> +
> +	head = &bcm2835_used_regs[hash_min(reg, BCM2835_REG_HT_BITS)];
> +
> +	if (bcm2835_host->previous_reg == reg) {
> +		if ((reg != SDHCI_HOST_CONTROL) &&
> +		    (reg != SDHCI_CLOCK_CONTROL) &&
> +		    (hlist_empty(head))) {
> +			rh = kzalloc(sizeof(*rh), GFP_KERNEL);
> +			if (WARN_ON(!rh))
> +				return;
> +
> +			rh->reg = reg;
> +			hash_add(bcm2835_used_regs, &rh->node, rh->reg);
> +			dev_err(mmc_dev(host->mmc), "back-to-back write to 0x%x\n",
> +				reg);
> +		}
> +	}
> +	bcm2835_host->previous_reg = reg;
> +
>   	writel(val, host->ioaddr + reg);
>   }
>
> @@ -131,7 +172,11 @@ static const struct sdhci_ops bcm2835_sdhci_ops = {
>   	.read_l = bcm2835_sdhci_readl,
>   	.read_w = bcm2835_sdhci_readw,
>   	.read_b = bcm2835_sdhci_readb,
> +#ifdef CONFIG_MMC_SDHCI_BCM2835_VERIFY_WORKAROUND
> +	.write_l = bcm2835_sdhci_writel_verify,
> +#else
>   	.write_l = bcm2835_sdhci_writel,
> +#endif
>   	.write_w = bcm2835_sdhci_writew,
>   	.write_b = bcm2835_sdhci_writeb,
>   	.set_clock = sdhci_set_clock,
>
>
>
>


  reply	other threads:[~2015-12-22 19:24 UTC|newest]

Thread overview: 53+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <Scott Branden <sbranden@broadcom.com>
2014-10-15  2:01 ` [PATCH 0/1] sdhci-bcm2835: added quirk and removed udelay in write ops Scott Branden
2014-10-15  2:01   ` [PATCH 1/1] mmc: " Scott Branden
2014-10-17  2:50     ` Stephen Warren
2014-10-15 16:43 ` Scott Branden
2014-10-18  2:37   ` Stephen Warren
2014-10-18  6:40     ` Scott Branden
2014-10-30  6:36 ` [PATCHv2 0/5] " Scott Branden
2014-10-30  6:36   ` [PATCHv2 1/5] mmc: sdhci-bcm2835: group read and write functions to improve readability Scott Branden
2014-10-30  6:36   ` [PATCHv2 2/5] mmc: sdhci-bcm2835: make shift calculations consistent Scott Branden
2014-11-05  4:48     ` Stephen Warren
2014-11-05  5:19       ` Scott Branden
2014-10-30  6:36   ` [PATCHv2 3/5] mmc: shdci-bcm2835: add efficient back-to-back write workaround Scott Branden
2014-11-05  4:57     ` Stephen Warren
2014-11-05  6:55       ` Scott Branden
2014-11-06  4:48         ` Stephen Warren
2014-11-07 18:29           ` Scott Branden
2014-10-30  6:36   ` [PATCHv2 4/5] mmc: shdci-bcm2835: add verify for 32-bit back-to-back workaround Scott Branden
2014-11-05  4:44     ` Stephen Warren
2014-11-05  5:26       ` Scott Branden
2014-11-05  4:59     ` Stephen Warren
2014-11-05  7:00       ` Scott Branden
2014-11-06  5:01         ` Stephen Warren
2014-11-07 18:31           ` Scott Branden
2015-12-22 15:55             ` Stefan Wahren
2015-12-22 19:23               ` Scott Branden [this message]
2015-12-22 20:13                 ` Stefan Wahren
2014-10-30  6:36   ` [PATCHv2 5/5] mmc: sdhci-bcm2835: add sdhci quirk SDHCI_QUIRK_MULTIBLOCK_READ_ACMD12 Scott Branden
2014-11-05  5:00     ` Stephen Warren
2014-11-05  7:02       ` Scott Branden
2014-11-06  4:50         ` Stephen Warren
2014-11-07 18:30           ` Scott Branden
2014-12-05  0:11 ` [PATCH] mmc: sdhci: add quirk for ACMD23 broken Scott Branden
2014-12-05  0:16 ` [PATCH v2] " Scott Branden
2014-12-17 18:36   ` Scott Branden
2014-12-17 19:48   ` Chris Ball
2014-12-17 20:42     ` Scott Branden
2015-02-10  0:06 ` [PATCH 0/4] Add support for IPROC SDHCI controller Scott Branden
2015-02-10  0:06   ` [PATCH 1/4] mmc: sdhci: add quirk for ACMD23 broken Scott Branden
2015-02-10  0:06   ` [PATCH 2/4] mmc: sdhci: do not set AUTO_CMD12 for multi-block CMD53 Scott Branden
2015-02-10  0:06   ` [PATCH 3/4] mmc: sdhci-iproc: add IPROC SDHCI driver Scott Branden
2015-02-10  0:06   ` [PATCH 4/4] mmc: sdhci-iproc: add device tree bindings Scott Branden
2015-03-02 23:50     ` Florian Fainelli
2015-03-04 23:14       ` Scott Branden
     [not found]   ` <1423526791-29453-1-git-send-email-sbranden-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
2015-02-26 17:28     ` [PATCH 0/4] Add support for IPROC SDHCI controller Scott Branden
2015-03-05 15:59 ` [PATCH RESEND " Scott Branden
2015-03-05 15:59   ` [PATCH 1/4] mmc: sdhci: add quirk for ACMD23 broken Scott Branden
2015-03-05 15:59   ` [PATCH 2/4] mmc: sdhci: do not set AUTO_CMD12 for multi-block CMD53 Scott Branden
2015-03-05 15:59   ` [PATCH 3/4] mmc: sdhci-iproc: add IPROC SDHCI driver Scott Branden
2015-03-05 15:59   ` [PATCH 4/4] mmc: sdhci-iproc: add device tree bindings Scott Branden
2015-03-05 16:16   ` [PATCH RESEND 0/4] Add support for IPROC SDHCI controller Ulf Hansson
2015-03-05 19:57     ` Florian Fainelli
2015-03-10 18:35 ` [PATCH] mmc: sdhci: fix card presence logic in sdhci_request function Scott Branden
2015-03-13 10:14   ` Ulf Hansson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=5679A339.9000302@broadcom.com \
    --to=sbranden@broadcom.com \
    --cc=bcm-kernel-feedback-list@broadcom.com \
    --cc=chris@printf.net \
    --cc=info@lategoodbye.de \
    --cc=joe@perches.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mmc@vger.kernel.org \
    --cc=linux-rpi-kernel@lists.infradead.org \
    --cc=peter.griffin@linaro.org \
    --cc=pietrushnic@gmail.com \
    --cc=rjui@broadcom.com \
    --cc=rmk+kernel@arm.linux.org.uk \
    --cc=swarren@wwwdotorg.org \
    --cc=ulf.hansson@linaro.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).