IOMMU Archive on lore.kernel.org
 help / color / Atom feed
From: "Sironi, Filippo via iommu" <iommu@lists.linux-foundation.org>
To: Joerg Roedel <joro@8bytes.org>
Cc: "iommu@lists.linux-foundation.org"
	<iommu@lists.linux-foundation.org>,
	"jroedel@suse.de" <jroedel@suse.de>
Subject: Re: [PATCH 2/6] iommu/amd: Remove amd_iommu_devtable_lock
Date: Wed, 25 Sep 2019 15:52:49 +0000
Message-ID: <1EDB77EE-1289-4BCC-A8FF-E1D11D161D04@amazon.de> (raw)
In-Reply-To: <ABED92CE-D4E0-4DC0-BA2E-E7C66B6833A6@amazon.de>



> On 25. Sep 2019, at 08:50, Sironi, Filippo <sironi@amazon.de> wrote:
> 
> 
> 
>> On 25. Sep 2019, at 06:22, Joerg Roedel <joro@8bytes.org> wrote:
>> 
>> From: Joerg Roedel <jroedel@suse.de>
>> 
>> The lock is not necessary because the device table does not
>> contain shared state that needs protection. Locking is only
>> needed on an individual entry basis, and that needs to
>> happen on the iommu_dev_data level.
>> 
>> Fixes: 92d420ec028d ("iommu/amd: Relax locking in dma_ops path")
>> Signed-off-by: Joerg Roedel <jroedel@suse.de>
>> ---
>> drivers/iommu/amd_iommu.c | 23 ++++++-----------------
>> 1 file changed, 6 insertions(+), 17 deletions(-)
>> 
>> diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c
>> index 042854bbc5bc..37a9c04fc728 100644
>> --- a/drivers/iommu/amd_iommu.c
>> +++ b/drivers/iommu/amd_iommu.c
>> @@ -70,7 +70,6 @@
>> */
>> #define AMD_IOMMU_PGSIZES	((~0xFFFUL) & ~(2ULL << 38))
>> 
>> -static DEFINE_SPINLOCK(amd_iommu_devtable_lock);
>> static DEFINE_SPINLOCK(pd_bitmap_lock);
>> 
>> /* List of all available dev_data structures */
>> @@ -2080,10 +2079,11 @@ static void do_detach(struct iommu_dev_data *dev_data)
>> static int __attach_device(struct iommu_dev_data *dev_data,
>> 			   struct protection_domain *domain)
>> {
>> +	unsigned long flags;
>> 	int ret;
>> 
>> 	/* lock domain */
>> -	spin_lock(&domain->lock);
>> +	spin_lock_irqsave(&domain->lock, flags);
>> 
>> 	ret = -EBUSY;
>> 	if (dev_data->domain != NULL)
>> @@ -2097,7 +2097,7 @@ static int __attach_device(struct iommu_dev_data *dev_data,
>> out_unlock:
>> 
>> 	/* ready */
>> -	spin_unlock(&domain->lock);
>> +	spin_unlock_irqrestore(&domain->lock, flags);
>> 
>> 	return ret;
>> }
>> @@ -2181,7 +2181,6 @@ static int attach_device(struct device *dev,
>> {
>> 	struct pci_dev *pdev;
>> 	struct iommu_dev_data *dev_data;
>> -	unsigned long flags;
>> 	int ret;
>> 
>> 	dev_data = get_dev_data(dev);
>> @@ -2209,9 +2208,7 @@ static int attach_device(struct device *dev,
>> 	}
>> 
>> skip_ats_check:
>> -	spin_lock_irqsave(&amd_iommu_devtable_lock, flags);
>> 	ret = __attach_device(dev_data, domain);
>> -	spin_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
>> 
>> 	/*
>> 	 * We might boot into a crash-kernel here. The crashed kernel
>> @@ -2231,14 +2228,15 @@ static int attach_device(struct device *dev,
>> static void __detach_device(struct iommu_dev_data *dev_data)
>> {
>> 	struct protection_domain *domain;
>> +	unsigned long flags;
>> 
>> 	domain = dev_data->domain;
>> 
>> -	spin_lock(&domain->lock);
>> +	spin_lock_irqsave(&domain->lock, flags);
>> 
>> 	do_detach(dev_data);
>> 
>> -	spin_unlock(&domain->lock);
>> +	spin_unlock_irqrestore(&domain->lock, flags);
>> }
>> 
>> /*
>> @@ -2248,7 +2246,6 @@ static void detach_device(struct device *dev)
>> {
>> 	struct protection_domain *domain;
>> 	struct iommu_dev_data *dev_data;
>> -	unsigned long flags;
>> 
>> 	dev_data = get_dev_data(dev);
>> 	domain   = dev_data->domain;
>> @@ -2262,10 +2259,7 @@ static void detach_device(struct device *dev)
>> 	if (WARN_ON(!dev_data->domain))
>> 		return;
>> 
>> -	/* lock device table */
>> -	spin_lock_irqsave(&amd_iommu_devtable_lock, flags);
>> 	__detach_device(dev_data);
>> -	spin_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
>> 
>> 	if (!dev_is_pci(dev))
>> 		return;
>> @@ -2910,9 +2904,6 @@ int __init amd_iommu_init_dma_ops(void)
>> static void cleanup_domain(struct protection_domain *domain)
>> {
>> 	struct iommu_dev_data *entry;
>> -	unsigned long flags;
>> -
>> -	spin_lock_irqsave(&amd_iommu_devtable_lock, flags);
>> 
>> 	while (!list_empty(&domain->dev_list)) {
>> 		entry = list_first_entry(&domain->dev_list,
>> @@ -2920,8 +2911,6 @@ static void cleanup_domain(struct protection_domain *domain)
>> 		BUG_ON(!entry->domain);
>> 		__detach_device(entry);
>> 	}
>> -
>> -	spin_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
>> }
> 
> I'm guessing that it is free not to take the domain lock in cleanup_domain
> since this is called when free-ing the domain.  Right?

My guess was wrong but I see that you've fixed this in patch 3/6.

>> static void protection_domain_free(struct protection_domain *domain)
>> -- 
>> 2.17.1

Reviewed-by: Filippo Sironi <sironi@amazon.de>



Amazon Development Center Germany GmbH
Krausenstr. 38
10117 Berlin
Geschaeftsfuehrung: Christian Schlaeger, Ralf Herbrich
Eingetragen am Amtsgericht Charlottenburg unter HRB 149173 B
Sitz: Berlin
Ust-ID: DE 289 237 879



_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu

  reply index

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-09-25 13:22 [PATCH 0/6] iommu/amd: Locking Fixes Joerg Roedel
2019-09-25 13:22 ` [PATCH 1/6] iommu/amd: Remove domain->updated Joerg Roedel
2019-09-25 15:45   ` Sironi, Filippo via iommu
2019-09-26  6:18   ` Jerry Snitselaar
2019-09-25 13:22 ` [PATCH 2/6] iommu/amd: Remove amd_iommu_devtable_lock Joerg Roedel
2019-09-25 15:50   ` Sironi, Filippo via iommu
2019-09-25 15:52     ` Sironi, Filippo via iommu [this message]
2019-09-26  6:24   ` Jerry Snitselaar
2019-09-25 13:22 ` [PATCH 3/6] iommu/amd: Take domain->lock for complete attach/detach path Joerg Roedel
2019-09-25 15:53   ` Sironi, Filippo via iommu
2019-09-26  6:34   ` Jerry Snitselaar
2019-09-25 13:22 ` [PATCH 4/6] iommu/amd: Check for busy devices earlier in attach_device() Joerg Roedel
2019-09-25 15:55   ` Sironi, Filippo via iommu
2019-09-26  6:37   ` Jerry Snitselaar
2019-09-25 13:22 ` [PATCH 5/6] iommu/amd: Lock dev_data in attach/detach code paths Joerg Roedel
2019-09-25 15:56   ` Sironi, Filippo via iommu
2019-09-26  6:41   ` Jerry Snitselaar
2019-09-25 13:23 ` [PATCH 6/6] iommu/amd: Lock code paths traversing protection_domain->dev_list Joerg Roedel
2019-09-25 15:58   ` Sironi, Filippo via iommu
2019-09-26  6:48   ` Jerry Snitselaar
2019-09-26  0:25 ` [PATCH 0/6] iommu/amd: Locking Fixes Jerry Snitselaar
2019-09-26  5:46   ` Joerg Roedel

Reply instructions:

You may reply publically 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=1EDB77EE-1289-4BCC-A8FF-E1D11D161D04@amazon.de \
    --to=iommu@lists.linux-foundation.org \
    --cc=joro@8bytes.org \
    --cc=jroedel@suse.de \
    --cc=sironi@amazon.de \
    /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

IOMMU Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-iommu/0 linux-iommu/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-iommu linux-iommu/ https://lore.kernel.org/linux-iommu \
		iommu@lists.linux-foundation.org
	public-inbox-index linux-iommu

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.linux-foundation.lists.iommu


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git