All of lore.kernel.org
 help / color / mirror / Atom feed
From: Khiem Nguyen <khiem.nguyen.xt@renesas.com>
To: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: khiem.nguyen.xt@renesas.com, iommu@lists.linux-foundation.org,
	linux-sh@vger.kernel.org, linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH v2 01/10] iommu/ipmmu-vmsa: Refactor micro-TLB lookup
Date: Mon, 14 Jul 2014 18:15:27 +0900	[thread overview]
Message-ID: <53C39FAF.2060306@renesas.com> (raw)
In-Reply-To: <1509851.5UQ4JhkPb5@avalon>

Hi Laurent,

On 7/14/2014 6:01 PM, Laurent Pinchart wrote:
> On Monday 14 July 2014 09:19:23 Khiem Nguyen wrote:
>> On 7/10/2014 7:37 PM, Laurent Pinchart wrote:
>>> On Thursday 10 July 2014 09:03:26 Khiem Nguyen wrote:
>>>> On 5/15/2014 7:40 PM, Laurent Pinchart wrote:
>>>>> Cache the micro-TLB number in archdata allocated in the .add_device
>>>>> handler instead of looking it up when the deviced is attached and
>>>>> detached. This simplifies the .attach_dev and .detach_dev operations and
>>>>> prepares for DT support.
>>>>
>>>> [snip]
>>>>
>>>>> Signed-off-by: Laurent Pinchart
>>>>> <laurent.pinchart+renesas@ideasonboard.com>
>>>>
>>>> [snip]
>>>>
>>>>> +static int ipmmu_find_utlb(struct ipmmu_vmsa_device *mmu, struct device
>>>>> *dev)
>>>>> +{
>>>>> +	const struct ipmmu_vmsa_master *master = mmu->pdata->masters;
>>>>> +	const char *devname = dev_name(dev);
>>>>> +	unsigned int i;
>>>>> +
>>>>> +	for (i = 0; i < mmu->pdata->num_masters; ++i, ++master) {
>>>>> +		if (strcmp(master->name, devname) == 0)
>>>>> +			return master->utlb;
>>>>> +	}
>>>>> +
>>>>> +	return -1;
>>>>> +}
>>>>
>>>> [snip]
>>>>
>>>>>  static int ipmmu_add_device(struct device *dev)
>>>>
>>>> [snip]
>>>>
>>>>>  	list_for_each_entry(mmu, &ipmmu_devices, list) {
>>>>> -		master = ipmmu_find_master(mmu, dev);
>>>>> -		if (master) {
>>>>> +		utlb = ipmmu_find_utlb(mmu, dev);
>>>>> +		if (utlb >= 0) {
>>>>>  			/*
>>>>> -			 * TODO Take a reference to the master to protect
>>>>> +			 * TODO Take a reference to the MMU to protect
>>>>>  			 * against device removal.
>>>>>  			 */
>>>>>  			break;
>>>>
>>>> [snip]
>>>>
>>>>> +	archdata->mmu = mmu;
>>>>> +	archdata->utlb = utlb;
>>>>
>>>> [snip]
>>>>
>>>> I have one question for ipmmu_add_device().
>>>>
>>>> In my understanding, your code will find utlb for device
>>>> base on device name.
>>>> For any device, it will /only/ return utlb number of first match.
>>>>
>>>> How about the case that a device name connected with more than 1 utlb ?
>>>> e.g DU device (rcar-du-r8a7790) in Lager
>>>>
>>>> Was that case already covered in your code ?
>>>
>>> For the DU case, the R8A7790 contains /two DU devices/, each connected to
>>> a single utlb. The IPMMU driver will thus work fine in that case.
>>
>> As my understanding, in board-lager-reference.c,
>> DU devices are registered with 1 device name "rcar-du-r8a7790".
>>
>> I also added some logs in ipmmu_add_device() to observe
>> the mapping between device name and utlb number.
>> And I saw that only one utlb is mapped with DU.
>>
>> Do I miss anything here ?
> 
> No, you're right, my bad. This is definitely a limitation of the driver at the 
> moment.

OK. Thanks for your confirmation.

> 
>>> I agree that this is a problem in general though, other devices (such as
>>> the DMAC) are connected to more than one utlb. This is currently not
>>> supported by the driver.
>>
>> Thanks for your confirmation.
>> I guess the driver will be improved in near future.
> 
> Yes, it will.
> 
I'm happy to wait for your update. :)

-- 
Best regards,
KHIEM Nguyen

WARNING: multiple messages have this Message-ID (diff)
From: Khiem Nguyen <khiem.nguyen.xt@renesas.com>
To: linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH v2 01/10] iommu/ipmmu-vmsa: Refactor micro-TLB lookup
Date: Mon, 14 Jul 2014 09:15:27 +0000	[thread overview]
Message-ID: <53C39FAF.2060306@renesas.com> (raw)
In-Reply-To: <1509851.5UQ4JhkPb5@avalon>

Hi Laurent,

On 7/14/2014 6:01 PM, Laurent Pinchart wrote:
> On Monday 14 July 2014 09:19:23 Khiem Nguyen wrote:
>> On 7/10/2014 7:37 PM, Laurent Pinchart wrote:
>>> On Thursday 10 July 2014 09:03:26 Khiem Nguyen wrote:
>>>> On 5/15/2014 7:40 PM, Laurent Pinchart wrote:
>>>>> Cache the micro-TLB number in archdata allocated in the .add_device
>>>>> handler instead of looking it up when the deviced is attached and
>>>>> detached. This simplifies the .attach_dev and .detach_dev operations and
>>>>> prepares for DT support.
>>>>
>>>> [snip]
>>>>
>>>>> Signed-off-by: Laurent Pinchart
>>>>> <laurent.pinchart+renesas@ideasonboard.com>
>>>>
>>>> [snip]
>>>>
>>>>> +static int ipmmu_find_utlb(struct ipmmu_vmsa_device *mmu, struct device
>>>>> *dev)
>>>>> +{
>>>>> +	const struct ipmmu_vmsa_master *master = mmu->pdata->masters;
>>>>> +	const char *devname = dev_name(dev);
>>>>> +	unsigned int i;
>>>>> +
>>>>> +	for (i = 0; i < mmu->pdata->num_masters; ++i, ++master) {
>>>>> +		if (strcmp(master->name, devname) = 0)
>>>>> +			return master->utlb;
>>>>> +	}
>>>>> +
>>>>> +	return -1;
>>>>> +}
>>>>
>>>> [snip]
>>>>
>>>>>  static int ipmmu_add_device(struct device *dev)
>>>>
>>>> [snip]
>>>>
>>>>>  	list_for_each_entry(mmu, &ipmmu_devices, list) {
>>>>> -		master = ipmmu_find_master(mmu, dev);
>>>>> -		if (master) {
>>>>> +		utlb = ipmmu_find_utlb(mmu, dev);
>>>>> +		if (utlb >= 0) {
>>>>>  			/*
>>>>> -			 * TODO Take a reference to the master to protect
>>>>> +			 * TODO Take a reference to the MMU to protect
>>>>>  			 * against device removal.
>>>>>  			 */
>>>>>  			break;
>>>>
>>>> [snip]
>>>>
>>>>> +	archdata->mmu = mmu;
>>>>> +	archdata->utlb = utlb;
>>>>
>>>> [snip]
>>>>
>>>> I have one question for ipmmu_add_device().
>>>>
>>>> In my understanding, your code will find utlb for device
>>>> base on device name.
>>>> For any device, it will /only/ return utlb number of first match.
>>>>
>>>> How about the case that a device name connected with more than 1 utlb ?
>>>> e.g DU device (rcar-du-r8a7790) in Lager
>>>>
>>>> Was that case already covered in your code ?
>>>
>>> For the DU case, the R8A7790 contains /two DU devices/, each connected to
>>> a single utlb. The IPMMU driver will thus work fine in that case.
>>
>> As my understanding, in board-lager-reference.c,
>> DU devices are registered with 1 device name "rcar-du-r8a7790".
>>
>> I also added some logs in ipmmu_add_device() to observe
>> the mapping between device name and utlb number.
>> And I saw that only one utlb is mapped with DU.
>>
>> Do I miss anything here ?
> 
> No, you're right, my bad. This is definitely a limitation of the driver at the 
> moment.

OK. Thanks for your confirmation.

> 
>>> I agree that this is a problem in general though, other devices (such as
>>> the DMAC) are connected to more than one utlb. This is currently not
>>> supported by the driver.
>>
>> Thanks for your confirmation.
>> I guess the driver will be improved in near future.
> 
> Yes, it will.
> 
I'm happy to wait for your update. :)

-- 
Best regards,
KHIEM Nguyen

WARNING: multiple messages have this Message-ID (diff)
From: khiem.nguyen.xt@renesas.com (Khiem Nguyen)
To: linux-arm-kernel@lists.infradead.org
Subject: [PATCH v2 01/10] iommu/ipmmu-vmsa: Refactor micro-TLB lookup
Date: Mon, 14 Jul 2014 18:15:27 +0900	[thread overview]
Message-ID: <53C39FAF.2060306@renesas.com> (raw)
In-Reply-To: <1509851.5UQ4JhkPb5@avalon>

Hi Laurent,

On 7/14/2014 6:01 PM, Laurent Pinchart wrote:
> On Monday 14 July 2014 09:19:23 Khiem Nguyen wrote:
>> On 7/10/2014 7:37 PM, Laurent Pinchart wrote:
>>> On Thursday 10 July 2014 09:03:26 Khiem Nguyen wrote:
>>>> On 5/15/2014 7:40 PM, Laurent Pinchart wrote:
>>>>> Cache the micro-TLB number in archdata allocated in the .add_device
>>>>> handler instead of looking it up when the deviced is attached and
>>>>> detached. This simplifies the .attach_dev and .detach_dev operations and
>>>>> prepares for DT support.
>>>>
>>>> [snip]
>>>>
>>>>> Signed-off-by: Laurent Pinchart
>>>>> <laurent.pinchart+renesas@ideasonboard.com>
>>>>
>>>> [snip]
>>>>
>>>>> +static int ipmmu_find_utlb(struct ipmmu_vmsa_device *mmu, struct device
>>>>> *dev)
>>>>> +{
>>>>> +	const struct ipmmu_vmsa_master *master = mmu->pdata->masters;
>>>>> +	const char *devname = dev_name(dev);
>>>>> +	unsigned int i;
>>>>> +
>>>>> +	for (i = 0; i < mmu->pdata->num_masters; ++i, ++master) {
>>>>> +		if (strcmp(master->name, devname) == 0)
>>>>> +			return master->utlb;
>>>>> +	}
>>>>> +
>>>>> +	return -1;
>>>>> +}
>>>>
>>>> [snip]
>>>>
>>>>>  static int ipmmu_add_device(struct device *dev)
>>>>
>>>> [snip]
>>>>
>>>>>  	list_for_each_entry(mmu, &ipmmu_devices, list) {
>>>>> -		master = ipmmu_find_master(mmu, dev);
>>>>> -		if (master) {
>>>>> +		utlb = ipmmu_find_utlb(mmu, dev);
>>>>> +		if (utlb >= 0) {
>>>>>  			/*
>>>>> -			 * TODO Take a reference to the master to protect
>>>>> +			 * TODO Take a reference to the MMU to protect
>>>>>  			 * against device removal.
>>>>>  			 */
>>>>>  			break;
>>>>
>>>> [snip]
>>>>
>>>>> +	archdata->mmu = mmu;
>>>>> +	archdata->utlb = utlb;
>>>>
>>>> [snip]
>>>>
>>>> I have one question for ipmmu_add_device().
>>>>
>>>> In my understanding, your code will find utlb for device
>>>> base on device name.
>>>> For any device, it will /only/ return utlb number of first match.
>>>>
>>>> How about the case that a device name connected with more than 1 utlb ?
>>>> e.g DU device (rcar-du-r8a7790) in Lager
>>>>
>>>> Was that case already covered in your code ?
>>>
>>> For the DU case, the R8A7790 contains /two DU devices/, each connected to
>>> a single utlb. The IPMMU driver will thus work fine in that case.
>>
>> As my understanding, in board-lager-reference.c,
>> DU devices are registered with 1 device name "rcar-du-r8a7790".
>>
>> I also added some logs in ipmmu_add_device() to observe
>> the mapping between device name and utlb number.
>> And I saw that only one utlb is mapped with DU.
>>
>> Do I miss anything here ?
> 
> No, you're right, my bad. This is definitely a limitation of the driver at the 
> moment.

OK. Thanks for your confirmation.

> 
>>> I agree that this is a problem in general though, other devices (such as
>>> the DMAC) are connected to more than one utlb. This is currently not
>>> supported by the driver.
>>
>> Thanks for your confirmation.
>> I guess the driver will be improved in near future.
> 
> Yes, it will.
> 
I'm happy to wait for your update. :)

-- 
Best regards,
KHIEM Nguyen

  reply	other threads:[~2014-07-14  9:15 UTC|newest]

Thread overview: 69+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-05-15 10:40 [PATCH v2 00/10] Renesas ipmmu-vmsa: Miscellaneous cleanups and fixes Laurent Pinchart
2014-05-15 10:40 ` Laurent Pinchart
2014-05-15 10:40 ` Laurent Pinchart
2014-05-15 10:40 ` [PATCH v2 01/10] iommu/ipmmu-vmsa: Refactor micro-TLB lookup Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
     [not found]   ` <1400150451-13469-2-git-send-email-laurent.pinchart+renesas-ryLnwIuWjnjg/C1BVhZhaw@public.gmane.org>
2014-07-10  0:03     ` Khiem Nguyen
2014-07-10  0:03       ` Khiem Nguyen
2014-07-10  0:03       ` Khiem Nguyen
2014-07-10 10:37       ` Laurent Pinchart
2014-07-10 10:37         ` Laurent Pinchart
2014-07-10 10:37         ` Laurent Pinchart
2014-07-14  0:19         ` Khiem Nguyen
2014-07-14  0:19           ` Khiem Nguyen
2014-07-14  0:19           ` Khiem Nguyen
     [not found]           ` <53C3220B.30506-zM6kxYcvzFBBDgjK7y7TUQ@public.gmane.org>
2014-07-14  9:01             ` Laurent Pinchart
2014-07-14  9:01               ` Laurent Pinchart
2014-07-14  9:01               ` Laurent Pinchart
2014-07-14  9:15               ` Khiem Nguyen [this message]
2014-07-14  9:15                 ` Khiem Nguyen
2014-07-14  9:15                 ` Khiem Nguyen
2014-05-15 10:40 ` [PATCH v2 02/10] iommu/ipmmu-vmsa: Cleanup failures of ARM mapping creation or attachment Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
     [not found]   ` <1400150451-13469-3-git-send-email-laurent.pinchart+renesas-ryLnwIuWjnjg/C1BVhZhaw@public.gmane.org>
2014-05-26  9:23     ` Joerg Roedel
2014-05-26  9:23       ` Joerg Roedel
2014-05-26  9:23       ` Joerg Roedel
     [not found]       ` <20140526092310.GA11299-zLv9SwRftAIdnm+yROfE0A@public.gmane.org>
2014-05-26 10:08         ` Laurent Pinchart
2014-05-26 10:08           ` Laurent Pinchart
2014-05-26 10:08           ` Laurent Pinchart
2014-05-26 11:31           ` Joerg Roedel
2014-05-26 11:31             ` Joerg Roedel
2014-05-26 11:31             ` Joerg Roedel
2014-05-26 11:35             ` Laurent Pinchart
2014-05-26 11:35               ` Laurent Pinchart
2014-05-26 11:35               ` Laurent Pinchart
2014-06-16 15:11           ` Joerg Roedel
2014-06-16 15:11             ` Joerg Roedel
2014-06-16 15:11             ` Joerg Roedel
2014-06-17 23:20             ` Laurent Pinchart
2014-06-17 23:20               ` Laurent Pinchart
2014-06-17 23:20               ` Laurent Pinchart
     [not found] ` <1400150451-13469-1-git-send-email-laurent.pinchart+renesas-ryLnwIuWjnjg/C1BVhZhaw@public.gmane.org>
2014-05-15 10:40   ` [PATCH v2 03/10] iommu/ipmmu-vmsa: Fix the supported page sizes Laurent Pinchart
2014-05-15 10:40     ` Laurent Pinchart
2014-05-15 10:40     ` Laurent Pinchart
2014-05-15 10:40   ` [PATCH v2 04/10] iommu/ipmmu-vmsa: Define driver-specific page directory sizes Laurent Pinchart
2014-05-15 10:40     ` Laurent Pinchart
2014-05-15 10:40     ` Laurent Pinchart
2014-05-15 10:40   ` [PATCH v2 05/10] iommu/ipmmu-vmsa: Set the PTE contiguous hint bit when possible Laurent Pinchart
2014-05-15 10:40     ` Laurent Pinchart
2014-05-15 10:40     ` Laurent Pinchart
2014-05-21 16:11   ` [PATCH v2 00/10] Renesas ipmmu-vmsa: Miscellaneous cleanups and fixes Laurent Pinchart
2014-05-21 16:11     ` Laurent Pinchart
2014-05-21 16:11     ` Laurent Pinchart
2014-05-15 10:40 ` [PATCH v2 06/10] iommu/ipmmu-vmsa: PMD is never folded, PUD always is Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40 ` [PATCH v2 07/10] iommu/ipmmu-vmsa: Rewrite page table management Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40 ` [PATCH v2 08/10] iommu/ipmmu-vmsa: Support 2MB mappings Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40 ` [PATCH v2 09/10] iommu/ipmmu-vmsa: Remove stage 2 PTE bits definitions Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40 ` [PATCH v2 10/10] iommu/ipmmu-vmsa: Support clearing mappings Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart
2014-05-15 10:40   ` Laurent Pinchart

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=53C39FAF.2060306@renesas.com \
    --to=khiem.nguyen.xt@renesas.com \
    --cc=iommu@lists.linux-foundation.org \
    --cc=laurent.pinchart@ideasonboard.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-sh@vger.kernel.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.