public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Alex Williamson <alex.williamson@redhat.com>
To: Joerg Roedel <joro@8bytes.org>
Cc: iommu@lists.linux-foundation.org,
	David Woodhouse <dwmw2@infradead.org>,
	linux-kernel@vger.kernel.org, jroedel@suse.de
Subject: Re: [PATCH 01/26] iommu/vt-d: Keep track of per-iommu domain ids
Date: Thu, 06 Aug 2015 13:19:52 -0600	[thread overview]
Message-ID: <1438888792.4023.118.camel@redhat.com> (raw)
In-Reply-To: <1438787937-7340-2-git-send-email-joro@8bytes.org>

On Wed, 2015-08-05 at 17:18 +0200, Joerg Roedel wrote:
> From: Joerg Roedel <jroedel@suse.de>
> 
> Instead of searching in the domain array for already
> allocated domain ids, keep track of them explicitly.
> 
> Signed-off-by: Joerg Roedel <jroedel@suse.de>
> ---
>  drivers/iommu/intel-iommu.c | 51 +++++++++++++++++++++++----------------------
>  1 file changed, 26 insertions(+), 25 deletions(-)
> 
> diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
> index 0649b94..c3c5167 100644
> --- a/drivers/iommu/intel-iommu.c
> +++ b/drivers/iommu/intel-iommu.c
> @@ -378,6 +378,9 @@ struct dmar_domain {
>  	DECLARE_BITMAP(iommu_bmp, DMAR_UNITS_SUPPORTED);
>  					/* bitmap of iommus this domain uses*/
>  
> +	u16	iommu_did[DMAR_UNITS_SUPPORTED];
> +					/* Domain ids per IOMMU */

Maybe a spec reference here to justify u16 that the 2.3 vt-d spec only
supports up to 16bit domain IDs.

> +
>  	struct list_head devices;	/* all devices' list */
>  	struct iova_domain iovad;	/* iova's that belong to this domain */
>  
> @@ -1543,11 +1546,13 @@ static int iommu_init_domains(struct intel_iommu *iommu)
>  	}
>  
>  	/*
> -	 * if Caching mode is set, then invalid translations are tagged
> -	 * with domainid 0. Hence we need to pre-allocate it.
> +	 * If Caching mode is set, then invalid translations are tagged
> +	 * with domain-id 0, hence we need to pre-allocate it. We also
> +	 * use domain-id 0 as a marker for non-allocated domain-id, so
> +	 * make sure it is not used for a real domain.
>  	 */
> -	if (cap_caching_mode(iommu->cap))
> -		set_bit(0, iommu->domain_ids);
> +	set_bit(0, iommu->domain_ids);
> +
>  	return 0;
>  }
>  
> @@ -1560,9 +1565,10 @@ static void disable_dmar_iommu(struct intel_iommu *iommu)
>  		for_each_set_bit(i, iommu->domain_ids, cap_ndoms(iommu->cap)) {
>  			/*
>  			 * Domain id 0 is reserved for invalid translation
> -			 * if hardware supports caching mode.
> +			 * if hardware supports caching mode and used as
> +			 * a non-allocated marker.
>  			 */
> -			if (cap_caching_mode(iommu->cap) && i == 0)
> +			if (i == 0)
>  				continue;
>  
>  			domain = iommu->domains[i];
> @@ -1624,6 +1630,7 @@ static int __iommu_attach_domain(struct dmar_domain *domain,
>  	if (num < ndomains) {
>  		set_bit(num, iommu->domain_ids);
>  		iommu->domains[num] = domain;
> +		domain->iommu_did[iommu->seq_id] = num;
>  	} else {
>  		num = -ENOSPC;
>  	}
> @@ -1650,12 +1657,10 @@ static int iommu_attach_vm_domain(struct dmar_domain *domain,
>  				  struct intel_iommu *iommu)
>  {
>  	int num;
> -	unsigned long ndomains;
>  
> -	ndomains = cap_ndoms(iommu->cap);
> -	for_each_set_bit(num, iommu->domain_ids, ndomains)
> -		if (iommu->domains[num] == domain)
> -			return num;
> +	num = domain->iommu_did[iommu->seq_id];
> +	if (num)
> +		return num;
>  
>  	return __iommu_attach_domain(domain, iommu);
>  }
> @@ -1664,22 +1669,17 @@ static void iommu_detach_domain(struct dmar_domain *domain,
>  				struct intel_iommu *iommu)
>  {
>  	unsigned long flags;
> -	int num, ndomains;
> +	int num;
>  
>  	spin_lock_irqsave(&iommu->lock, flags);
> -	if (domain_type_is_vm_or_si(domain)) {
> -		ndomains = cap_ndoms(iommu->cap);
> -		for_each_set_bit(num, iommu->domain_ids, ndomains) {
> -			if (iommu->domains[num] == domain) {
> -				clear_bit(num, iommu->domain_ids);
> -				iommu->domains[num] = NULL;
> -				break;
> -			}
> -		}
> -	} else {
> -		clear_bit(domain->id, iommu->domain_ids);
> -		iommu->domains[domain->id] = NULL;
> -	}
> +
> +	num = domain->iommu_did[iommu->seq_id];
> +
> +	WARN_ON(num == 0);

Return -ENODEV and let the caller decide if that's an error?  Then we
could safely call this over all all iommus.

> +
> +	clear_bit(num, iommu->domain_ids);
> +	iommu->domains[num]		 = NULL;
> +
>  	spin_unlock_irqrestore(&iommu->lock, flags);
>  }
>  
> @@ -1708,6 +1708,7 @@ static int domain_detach_iommu(struct dmar_domain *domain,
>  	if (test_and_clear_bit(iommu->seq_id, domain->iommu_bmp)) {
>  		count = --domain->iommu_count;
>  		domain_update_iommu_cap(domain);
> +		domain->iommu_did[iommu->seq_id] = 0;
>  	}
>  	spin_unlock_irqrestore(&domain->iommu_lock, flags);
>  




  reply	other threads:[~2015-08-06 19:19 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-08-05 15:18 [PATCH 00/26] iommu/vt-d: Clean up device<->domain attachment Joerg Roedel
2015-08-05 15:18 ` [PATCH 01/26] iommu/vt-d: Keep track of per-iommu domain ids Joerg Roedel
2015-08-06 19:19   ` Alex Williamson [this message]
2015-08-07 11:07     ` Joerg Roedel
2015-08-05 15:18 ` [PATCH 02/26] iommu/vt-d: Add access functions for iommu->domains Joerg Roedel
2015-08-05 15:18 ` [PATCH 03/26] iommu/vt-d: Split up iommu->domains array Joerg Roedel
2015-08-06 19:20   ` Alex Williamson
2015-08-07 11:11     ` Joerg Roedel
2015-08-05 15:18 ` [PATCH 04/26] iommu/vt-d: Get rid of iommu_attach_vm_domain() Joerg Roedel
2015-08-05 15:18 ` [PATCH 05/26] iommu/vt-d: Calculate translation in domain_context_mapping_one Joerg Roedel
2015-08-06 19:20   ` Alex Williamson
2015-08-07 11:13     ` Joerg Roedel
2015-08-05 15:18 ` [PATCH 06/26] iommu/vt-d: Simplify domain_context_mapping_one Joerg Roedel
2015-08-05 15:18 ` [PATCH 07/26] iommu/vt-d: Pass dmar_domain directly into iommu_flush_iotlb_psi Joerg Roedel
2015-08-05 15:18 ` [PATCH 08/26] iommu/vt-d: Don't pre-allocate domain ids for si_domain Joerg Roedel
2015-08-05 15:18 ` [PATCH 09/26] iommu/vt-d: Kill dmar_domain->id Joerg Roedel
2015-08-05 15:18 ` [PATCH 10/26] iommu/vt-d: Replace iommu_bmp with a refcount Joerg Roedel
2015-08-06 19:19   ` Alex Williamson
2015-08-07 11:17     ` Joerg Roedel
2015-08-05 15:18 ` [PATCH 11/26] iommu/vt-d: Simplify io/tlb flushing in intel_iommu_unmap Joerg Roedel
2015-08-05 15:18 ` [PATCH 12/26] iommu/vt-d: Simplify domain_remove_one_dev_info() Joerg Roedel
2015-08-05 15:18 ` [PATCH 13/26] iommu/vt-d: Simplify domain_remove_dev_info() Joerg Roedel
2015-08-05 15:18 ` [PATCH 14/26] iommu/vt-d: Move context-mapping into dmar_insert_dev_info Joerg Roedel
2015-08-05 15:18 ` [PATCH 15/26] iommu/vt-d: Rename dmar_insert_dev_info() Joerg Roedel
2015-08-05 15:18 ` [PATCH 16/26] iommu/vt-d: Rename domain_remove_one_dev_info() Joerg Roedel
2015-08-05 15:18 ` [PATCH 17/26] iommu/vt-d: Rename iommu_detach_dependent_devices() Joerg Roedel
2015-08-05 15:18 ` [PATCH 18/26] iommu/vt-d: Pass an iommu pointer to domain_init() Joerg Roedel
2015-08-05 15:18 ` [PATCH 19/26] iommu/vt-d: Establish domain<->iommu link in dmar_insert_one_dev_info Joerg Roedel
2015-08-05 15:18 ` [PATCH 20/26] iommu/vt-d: Unify domain->iommu attach/detachment Joerg Roedel
2015-08-05 15:18 ` [PATCH 21/26] iommu/vt-d: Only call domain_remove_one_dev_info to detach old domain Joerg Roedel
2015-08-05 15:18 ` [PATCH 22/26] iommu/vt-d: Get rid of domain->iommu_lock Joerg Roedel
2015-08-05 15:18 ` [PATCH 23/26] iommu/vt-d: Remove dmar_global_lock from device_notifier Joerg Roedel
2015-08-05 15:18 ` [PATCH 24/26] iommu/vt-d: Pass device_domain_info to __dmar_remove_one_dev_info Joerg Roedel
2015-08-05 15:18 ` [PATCH 25/26] iommu/vt-d: Only insert alias dev_info if there is an alias Joerg Roedel
2015-08-05 15:18 ` [PATCH 26/26] iommu/vt-d: Avoid duplicate device_domain_info structures Joerg Roedel
2015-08-06 19:19 ` [PATCH 00/26] iommu/vt-d: Clean up device<->domain attachment Alex Williamson
2015-08-07 11:22   ` Joerg Roedel

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=1438888792.4023.118.camel@redhat.com \
    --to=alex.williamson@redhat.com \
    --cc=dwmw2@infradead.org \
    --cc=iommu@lists.linux-foundation.org \
    --cc=joro@8bytes.org \
    --cc=jroedel@suse.de \
    --cc=linux-kernel@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox