[RFC] mmc: cqhci: commit descriptors before setting the doorbell

[RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Faiz Abbas]

Add a write memory barrier to make sure that descriptors are actually
written to memory before ringing the doorbell.

Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
---

This patch fixes a very infrequent ADMA error (1 out of 100 times) that
I have been seeing after enabling command queuing for J721e.
Also looking at memory-barriers.txt and this commit[1],
it looks like we should be doing this before any descriptor write
followed by a doorbell ring operation. It'll be nice if someone with more
expertise in memory barriers can comment.

[1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
    doorbell")

 drivers/mmc/host/cqhci.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/mmc/host/cqhci.c b/drivers/mmc/host/cqhci.c
index f7bdae5354c3..5047f7343ffc 100644
--- a/drivers/mmc/host/cqhci.c
+++ b/drivers/mmc/host/cqhci.c
@@ -611,7 +611,8 @@ static int cqhci_request(struct mmc_host *mmc, struct mmc_request *mrq)
 	cq_host->slot[tag].flags = 0;
 
 	cq_host->qcnt += 1;
-
+	/* Make sure descriptors are ready before ringing the doorbell */
+	wmb();
 	cqhci_writel(cq_host, 1 << tag, CQHCI_TDBR);
 	if (!(cqhci_readl(cq_host, CQHCI_TDBR) & (1 << tag)))
 		pr_debug("%s: cqhci: doorbell not set for tag %d\n",
-- 
2.19.2

[RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Faiz Abbas]

Add a write memory barrier to make sure that descriptors are actually
written to memory before ringing the doorbell.

Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
---

This patch fixes a very infrequent ADMA error (1 out of 100 times) that
I have been seeing after enabling command queuing for J721e.
Also looking at memory-barriers.txt and this commit[1],
it looks like we should be doing this before any descriptor write
followed by a doorbell ring operation. It'll be nice if someone with more
expertise in memory barriers can comment.

[1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
    doorbell")

 drivers/mmc/host/cqhci.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/mmc/host/cqhci.c b/drivers/mmc/host/cqhci.c
index f7bdae5354c3..5047f7343ffc 100644
--- a/drivers/mmc/host/cqhci.c
+++ b/drivers/mmc/host/cqhci.c
@@ -611,7 +611,8 @@ static int cqhci_request(struct mmc_host *mmc, struct mmc_request *mrq)
 	cq_host->slot[tag].flags = 0;
 
 	cq_host->qcnt += 1;
-
+	/* Make sure descriptors are ready before ringing the doorbell */
+	wmb();
 	cqhci_writel(cq_host, 1 << tag, CQHCI_TDBR);
 	if (!(cqhci_readl(cq_host, CQHCI_TDBR) & (1 << tag)))
 		pr_debug("%s: cqhci: doorbell not set for tag %d\n",
-- 
2.19.2

Re: [RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Faiz Abbas]

Hi,

On 15/10/19 12:08 AM, Faiz Abbas wrote:
> Add a write memory barrier to make sure that descriptors are actually
> written to memory before ringing the doorbell.
> 
> Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
> ---
> 
> This patch fixes a very infrequent ADMA error (1 out of 100 times) that
> I have been seeing after enabling command queuing for J721e.
> Also looking at memory-barriers.txt and this commit[1],
> it looks like we should be doing this before any descriptor write
> followed by a doorbell ring operation. It'll be nice if someone with more
> expertise in memory barriers can comment.
> 
> [1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
>     doorbell")

So I see that cqhci_readl/writel() use readl/writel_relaxed() which
seems to be causing this issue. Should I just fix this by converting
those to readl/writel with memory barriers instead?

Thanks,
Faiz

Re: [RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Faiz Abbas]

Hi,

On 15/10/19 12:08 AM, Faiz Abbas wrote:
> Add a write memory barrier to make sure that descriptors are actually
> written to memory before ringing the doorbell.
> 
> Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
> ---
> 
> This patch fixes a very infrequent ADMA error (1 out of 100 times) that
> I have been seeing after enabling command queuing for J721e.
> Also looking at memory-barriers.txt and this commit[1],
> it looks like we should be doing this before any descriptor write
> followed by a doorbell ring operation. It'll be nice if someone with more
> expertise in memory barriers can comment.
> 
> [1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
>     doorbell")

So I see that cqhci_readl/writel() use readl/writel_relaxed() which
seems to be causing this issue. Should I just fix this by converting
those to readl/writel with memory barriers instead?

Thanks,
Faiz

Re: [RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Adrian Hunter]

On 15/10/19 10:55 AM, Faiz Abbas wrote:
> Hi,
> 
> On 15/10/19 12:08 AM, Faiz Abbas wrote:
>> Add a write memory barrier to make sure that descriptors are actually
>> written to memory before ringing the doorbell.
>>
>> Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
>> ---
>>
>> This patch fixes a very infrequent ADMA error (1 out of 100 times) that
>> I have been seeing after enabling command queuing for J721e.
>> Also looking at memory-barriers.txt and this commit[1],
>> it looks like we should be doing this before any descriptor write
>> followed by a doorbell ring operation. It'll be nice if someone with more
>> expertise in memory barriers can comment.
>>
>> [1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
>>     doorbell")
> 
> So I see that cqhci_readl/writel() use readl/writel_relaxed() which
> seems to be causing this issue. Should I just fix this by converting
> those to readl/writel with memory barriers instead?

Perhaps we could do both changes i.e. add wmb() and convert to non-relaxed
readl/writel

Re: [RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Faiz Abbas]

Adrian,

On 15/10/19 7:15 PM, Adrian Hunter wrote:
> On 15/10/19 10:55 AM, Faiz Abbas wrote:
>> Hi,
>>
>> On 15/10/19 12:08 AM, Faiz Abbas wrote:
>>> Add a write memory barrier to make sure that descriptors are actually
>>> written to memory before ringing the doorbell.
>>>
>>> Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
>>> ---
>>>
>>> This patch fixes a very infrequent ADMA error (1 out of 100 times) that
>>> I have been seeing after enabling command queuing for J721e.
>>> Also looking at memory-barriers.txt and this commit[1],
>>> it looks like we should be doing this before any descriptor write
>>> followed by a doorbell ring operation. It'll be nice if someone with more
>>> expertise in memory barriers can comment.
>>>
>>> [1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
>>>     doorbell")
>>
>> So I see that cqhci_readl/writel() use readl/writel_relaxed() which
>> seems to be causing this issue. Should I just fix this by converting
>> those to readl/writel with memory barriers instead?
> 
> Perhaps we could do both changes i.e. add wmb() and convert to non-relaxed
> readl/writel
> 

readl is implemented as  readl_relaxed(); __rmb();
and
writel is implemented as wmb(); writel_relaxed();

I think another wmb() before writel will be redundant.

Maybe this patch is good enough in itself.

Thanks,
Faiz

Re: [RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Faiz Abbas]

Adrian,

On 15/10/19 7:15 PM, Adrian Hunter wrote:
> On 15/10/19 10:55 AM, Faiz Abbas wrote:
>> Hi,
>>
>> On 15/10/19 12:08 AM, Faiz Abbas wrote:
>>> Add a write memory barrier to make sure that descriptors are actually
>>> written to memory before ringing the doorbell.
>>>
>>> Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
>>> ---
>>>
>>> This patch fixes a very infrequent ADMA error (1 out of 100 times) that
>>> I have been seeing after enabling command queuing for J721e.
>>> Also looking at memory-barriers.txt and this commit[1],
>>> it looks like we should be doing this before any descriptor write
>>> followed by a doorbell ring operation. It'll be nice if someone with more
>>> expertise in memory barriers can comment.
>>>
>>> [1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
>>>     doorbell")
>>
>> So I see that cqhci_readl/writel() use readl/writel_relaxed() which
>> seems to be causing this issue. Should I just fix this by converting
>> those to readl/writel with memory barriers instead?
> 
> Perhaps we could do both changes i.e. add wmb() and convert to non-relaxed
> readl/writel
> 

readl is implemented as  readl_relaxed(); __rmb();
and
writel is implemented as wmb(); writel_relaxed();

I think another wmb() before writel will be redundant.

Maybe this patch is good enough in itself.

Thanks,
Faiz

Re: [RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Faiz Abbas]

Adrian,

On 16/10/19 5:46 PM, Faiz Abbas wrote:
> Adrian,
> 
> On 15/10/19 7:15 PM, Adrian Hunter wrote:
>> On 15/10/19 10:55 AM, Faiz Abbas wrote:
>>> Hi,
>>>
>>> On 15/10/19 12:08 AM, Faiz Abbas wrote:
>>>> Add a write memory barrier to make sure that descriptors are actually
>>>> written to memory before ringing the doorbell.
>>>>
>>>> Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
>>>> ---
>>>>
>>>> This patch fixes a very infrequent ADMA error (1 out of 100 times) that
>>>> I have been seeing after enabling command queuing for J721e.
>>>> Also looking at memory-barriers.txt and this commit[1],
>>>> it looks like we should be doing this before any descriptor write
>>>> followed by a doorbell ring operation. It'll be nice if someone with more
>>>> expertise in memory barriers can comment.
>>>>
>>>> [1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
>>>>     doorbell")
>>>
>>> So I see that cqhci_readl/writel() use readl/writel_relaxed() which
>>> seems to be causing this issue. Should I just fix this by converting
>>> those to readl/writel with memory barriers instead?
>>
>> Perhaps we could do both changes i.e. add wmb() and convert to non-relaxed
>> readl/writel
>>
> 
> readl is implemented as  readl_relaxed(); __rmb();
> and
> writel is implemented as wmb(); writel_relaxed();
> 
> I think another wmb() before writel will be redundant.
> 
> Maybe this patch is good enough in itself.
> 

Do you agree?

Thanks,
Faiz

Re: [RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Faiz Abbas]

Adrian,

On 16/10/19 5:46 PM, Faiz Abbas wrote:
> Adrian,
> 
> On 15/10/19 7:15 PM, Adrian Hunter wrote:
>> On 15/10/19 10:55 AM, Faiz Abbas wrote:
>>> Hi,
>>>
>>> On 15/10/19 12:08 AM, Faiz Abbas wrote:
>>>> Add a write memory barrier to make sure that descriptors are actually
>>>> written to memory before ringing the doorbell.
>>>>
>>>> Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
>>>> ---
>>>>
>>>> This patch fixes a very infrequent ADMA error (1 out of 100 times) that
>>>> I have been seeing after enabling command queuing for J721e.
>>>> Also looking at memory-barriers.txt and this commit[1],
>>>> it looks like we should be doing this before any descriptor write
>>>> followed by a doorbell ring operation. It'll be nice if someone with more
>>>> expertise in memory barriers can comment.
>>>>
>>>> [1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
>>>>     doorbell")
>>>
>>> So I see that cqhci_readl/writel() use readl/writel_relaxed() which
>>> seems to be causing this issue. Should I just fix this by converting
>>> those to readl/writel with memory barriers instead?
>>
>> Perhaps we could do both changes i.e. add wmb() and convert to non-relaxed
>> readl/writel
>>
> 
> readl is implemented as  readl_relaxed(); __rmb();
> and
> writel is implemented as wmb(); writel_relaxed();
> 
> I think another wmb() before writel will be redundant.
> 
> Maybe this patch is good enough in itself.
> 

Do you agree?

Thanks,
Faiz

Re: [RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Adrian Hunter]

On 18/10/19 10:22 AM, Faiz Abbas wrote:
> Adrian,
> 
> On 16/10/19 5:46 PM, Faiz Abbas wrote:
>> Adrian,
>>
>> On 15/10/19 7:15 PM, Adrian Hunter wrote:
>>> On 15/10/19 10:55 AM, Faiz Abbas wrote:
>>>> Hi,
>>>>
>>>> On 15/10/19 12:08 AM, Faiz Abbas wrote:
>>>>> Add a write memory barrier to make sure that descriptors are actually
>>>>> written to memory before ringing the doorbell.
>>>>>
>>>>> Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
>>>>> ---
>>>>>
>>>>> This patch fixes a very infrequent ADMA error (1 out of 100 times) that
>>>>> I have been seeing after enabling command queuing for J721e.
>>>>> Also looking at memory-barriers.txt and this commit[1],
>>>>> it looks like we should be doing this before any descriptor write
>>>>> followed by a doorbell ring operation. It'll be nice if someone with more
>>>>> expertise in memory barriers can comment.
>>>>>
>>>>> [1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
>>>>>     doorbell")
>>>>
>>>> So I see that cqhci_readl/writel() use readl/writel_relaxed() which
>>>> seems to be causing this issue. Should I just fix this by converting
>>>> those to readl/writel with memory barriers instead?
>>>
>>> Perhaps we could do both changes i.e. add wmb() and convert to non-relaxed
>>> readl/writel
>>>
>>
>> readl is implemented as  readl_relaxed(); __rmb();
>> and
>> writel is implemented as wmb(); writel_relaxed();
>>
>> I think another wmb() before writel will be redundant.
>>
>> Maybe this patch is good enough in itself.
>>
> 
> Do you agree?

Sure.

Acked-by: Adrian Hunter <adrian.hunter@intel.com>

Re: [RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Ulf Hansson]

On Mon, 14 Oct 2019 at 20:37, Faiz Abbas <faiz_abbas@ti.com> wrote:
>
> Add a write memory barrier to make sure that descriptors are actually
> written to memory before ringing the doorbell.
>
> Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>

Applied for fixes and by adding a stable tag, thanks!

BTW, do you have a valid commit that it fixes?

Kind regards
Uffe


> ---
>
> This patch fixes a very infrequent ADMA error (1 out of 100 times) that
> I have been seeing after enabling command queuing for J721e.
> Also looking at memory-barriers.txt and this commit[1],
> it looks like we should be doing this before any descriptor write
> followed by a doorbell ring operation. It'll be nice if someone with more
> expertise in memory barriers can comment.
>
> [1] ad1a1b9cd67a ("scsi: ufs: commit descriptors before setting the
>     doorbell")
>
>  drivers/mmc/host/cqhci.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mmc/host/cqhci.c b/drivers/mmc/host/cqhci.c
> index f7bdae5354c3..5047f7343ffc 100644
> --- a/drivers/mmc/host/cqhci.c
> +++ b/drivers/mmc/host/cqhci.c
> @@ -611,7 +611,8 @@ static int cqhci_request(struct mmc_host *mmc, struct mmc_request *mrq)
>         cq_host->slot[tag].flags = 0;
>
>         cq_host->qcnt += 1;
> -
> +       /* Make sure descriptors are ready before ringing the doorbell */
> +       wmb();
>         cqhci_writel(cq_host, 1 << tag, CQHCI_TDBR);
>         if (!(cqhci_readl(cq_host, CQHCI_TDBR) & (1 << tag)))
>                 pr_debug("%s: cqhci: doorbell not set for tag %d\n",
> --
> 2.19.2
>

Re: [RFC] mmc: cqhci: commit descriptors before setting the doorbell

[Faiz Abbas]

Hi Uffe,

On 18/10/19 4:28 PM, Ulf Hansson wrote:
> On Mon, 14 Oct 2019 at 20:37, Faiz Abbas <faiz_abbas@ti.com> wrote:
>>
>> Add a write memory barrier to make sure that descriptors are actually
>> written to memory before ringing the doorbell.
>>
>> Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
> 
> Applied for fixes and by adding a stable tag, thanks!
> 
> BTW, do you have a valid commit that it fixes?
> 
You can add:

Fixes: a4080225f51d ("mmc: cqhci: support for command queue enabled host")

Thanks,
Faiz