linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH -next] PCI: Use GFP_ATOMIC in resource_alignment_store()
@ 2019-08-31 12:49 YueHaibing
  2019-09-02  7:50 ` Christoph Hellwig
  2019-09-03  8:22 ` [PATCH v2 -next] PCI: Don't use GFP_KERNEL for kstrbdup in resource_alignment_store YueHaibing
  0 siblings, 2 replies; 5+ messages in thread
From: YueHaibing @ 2019-08-31 12:49 UTC (permalink / raw)
  To: Bjorn Helgaas, Logan Gunthorpe
  Cc: YueHaibing, linux-pci, linux-kernel, kernel-janitors

When allocating memory, the GFP_KERNEL cannot be used during the
spin_lock period. It may cause scheduling when holding spin_lock.

Fixes: f13755318675 ("PCI: Move pci_[get|set]_resource_alignment_param() into their callers")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
---
 drivers/pci/pci.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 484e35349565..0b5fc6736f3f 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -6148,7 +6148,7 @@ static ssize_t resource_alignment_store(struct bus_type *bus,
 	spin_lock(&resource_alignment_lock);
 
 	kfree(resource_alignment_param);
-	resource_alignment_param = kstrndup(buf, count, GFP_KERNEL);
+	resource_alignment_param = kstrndup(buf, count, GFP_ATOMIC);
 
 	spin_unlock(&resource_alignment_lock);




^ permalink raw reply related	[flat|nested] 5+ messages in thread
* Re: [PATCH -next] PCI: Use GFP_ATOMIC in resource_alignment_store()
  2019-08-31 12:49 [PATCH -next] PCI: Use GFP_ATOMIC in resource_alignment_store() YueHaibing
@ 2019-09-02  7:50 ` Christoph Hellwig
  2019-09-03 15:51   ` Logan Gunthorpe
  2019-09-03  8:22 ` [PATCH v2 -next] PCI: Don't use GFP_KERNEL for kstrbdup in resource_alignment_store YueHaibing
  1 sibling, 1 reply; 5+ messages in thread
From: Christoph Hellwig @ 2019-09-02  7:50 UTC (permalink / raw)
  To: YueHaibing
  Cc: Bjorn Helgaas, Logan Gunthorpe, linux-pci, linux-kernel, kernel-janitors

On Sat, Aug 31, 2019 at 12:49:32PM +0000, YueHaibing wrote:
> When allocating memory, the GFP_KERNEL cannot be used during the
> spin_lock period. It may cause scheduling when holding spin_lock.
> 
> Fixes: f13755318675 ("PCI: Move pci_[get|set]_resource_alignment_param() into their callers")
> Signed-off-by: YueHaibing <yuehaibing@huawei.com>
> ---
>  drivers/pci/pci.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 484e35349565..0b5fc6736f3f 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -6148,7 +6148,7 @@ static ssize_t resource_alignment_store(struct bus_type *bus,
>  	spin_lock(&resource_alignment_lock);
>  
>  	kfree(resource_alignment_param);
> -	resource_alignment_param = kstrndup(buf, count, GFP_KERNEL);
> +	resource_alignment_param = kstrndup(buf, count, GFP_ATOMIC);
>  
>  	spin_unlock(&resource_alignment_lock);

Why not move the allocation outside the lock? Something like this
seems much more sensible:


diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 484e35349565..fe205829f676 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -6145,14 +6145,16 @@ static ssize_t resource_alignment_show(struct bus_type *bus, char *buf)
 static ssize_t resource_alignment_store(struct bus_type *bus,
 					const char *buf, size_t count)
 {
-	spin_lock(&resource_alignment_lock);
+	char *param = kstrndup(buf, count, GFP_KERNEL);
 
-	kfree(resource_alignment_param);
-	resource_alignment_param = kstrndup(buf, count, GFP_KERNEL);
+	if (!param)
+		return -ENOMEM;
 
+	spin_lock(&resource_alignment_lock);
+	kfree(resource_alignment_param);
+	resource_alignment_param = param;
 	spin_unlock(&resource_alignment_lock);
-
-	return resource_alignment_param ? count : -ENOMEM;
+	return count;
 }
 
 static BUS_ATTR_RW(resource_alignment);

^ permalink raw reply related	[flat|nested] 5+ messages in thread
* [PATCH v2 -next] PCI: Don't use GFP_KERNEL for kstrbdup in resource_alignment_store
  2019-08-31 12:49 [PATCH -next] PCI: Use GFP_ATOMIC in resource_alignment_store() YueHaibing
  2019-09-02  7:50 ` Christoph Hellwig
@ 2019-09-03  8:22 ` YueHaibing
  1 sibling, 0 replies; 5+ messages in thread
From: YueHaibing @ 2019-09-03  8:22 UTC (permalink / raw)
  To: bhelgaas, hch; +Cc: linux-pci, linux-kernel, YueHaibing

When allocating memory, the GFP_KERNEL cannot be used during the
spin_lock period. It may cause scheduling when holding spin_lock.

Suggested-by: Christoph Hellwig <hch@infradead.org>
Fixes: f13755318675 ("PCI: Move pci_[get|set]_resource_alignment_param() into their callers")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
---
v2: move alloc out of spinlock
---
 drivers/pci/pci.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 484e353..a3d5920 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -6145,14 +6145,17 @@ static ssize_t resource_alignment_show(struct bus_type *bus, char *buf)
 static ssize_t resource_alignment_store(struct bus_type *bus,
 					const char *buf, size_t count)
 {
-	spin_lock(&resource_alignment_lock);
+	char *param = kstrndup(buf, count, GFP_KERNEL);
 
-	kfree(resource_alignment_param);
-	resource_alignment_param = kstrndup(buf, count, GFP_KERNEL);
+	if (!param)
+		return -ENOMEM;
 
+	spin_lock(&resource_alignment_lock);
+	kfree(resource_alignment_param);
+	resource_alignment_param = param;
 	spin_unlock(&resource_alignment_lock);
 
-	return resource_alignment_param ? count : -ENOMEM;
+	return count;
 }
 
 static BUS_ATTR_RW(resource_alignment);
-- 
2.7.4



^ permalink raw reply related	[flat|nested] 5+ messages in thread
* Re: [PATCH -next] PCI: Use GFP_ATOMIC in resource_alignment_store()
  2019-09-02  7:50 ` Christoph Hellwig
@ 2019-09-03 15:51   ` Logan Gunthorpe
  2019-09-05 21:47     ` Bjorn Helgaas
  0 siblings, 1 reply; 5+ messages in thread
From: Logan Gunthorpe @ 2019-09-03 15:51 UTC (permalink / raw)
  To: Christoph Hellwig, YueHaibing
  Cc: Bjorn Helgaas, linux-pci, linux-kernel, kernel-janitors



On 2019-09-02 1:50 a.m., Christoph Hellwig wrote:
> On Sat, Aug 31, 2019 at 12:49:32PM +0000, YueHaibing wrote:
>> When allocating memory, the GFP_KERNEL cannot be used during the
>> spin_lock period. It may cause scheduling when holding spin_lock.
>>
>> Fixes: f13755318675 ("PCI: Move pci_[get|set]_resource_alignment_param() into their callers")
>> Signed-off-by: YueHaibing <yuehaibing@huawei.com>
>> ---
>>  drivers/pci/pci.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
>> index 484e35349565..0b5fc6736f3f 100644
>> --- a/drivers/pci/pci.c
>> +++ b/drivers/pci/pci.c
>> @@ -6148,7 +6148,7 @@ static ssize_t resource_alignment_store(struct bus_type *bus,
>>  	spin_lock(&resource_alignment_lock);
>>  
>>  	kfree(resource_alignment_param);
>> -	resource_alignment_param = kstrndup(buf, count, GFP_KERNEL);
>> +	resource_alignment_param = kstrndup(buf, count, GFP_ATOMIC);
>>  
>>  	spin_unlock(&resource_alignment_lock);
> 
> Why not move the allocation outside the lock? Something like this
> seems much more sensible:

Yes, that seems like a good way to do it. Bjorn, can you squash
Christoph's patch or do you want me to resend a new one?

Thanks,

Logan

> 
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 484e35349565..fe205829f676 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -6145,14 +6145,16 @@ static ssize_t resource_alignment_show(struct bus_type *bus, char *buf)
>  static ssize_t resource_alignment_store(struct bus_type *bus,
>  					const char *buf, size_t count)
>  {
> -	spin_lock(&resource_alignment_lock);
> +	char *param = kstrndup(buf, count, GFP_KERNEL);
>  
> -	kfree(resource_alignment_param);
> -	resource_alignment_param = kstrndup(buf, count, GFP_KERNEL);
> +	if (!param)
> +		return -ENOMEM;
>  
> +	spin_lock(&resource_alignment_lock);
> +	kfree(resource_alignment_param);
> +	resource_alignment_param = param;
>  	spin_unlock(&resource_alignment_lock);
> -
> -	return resource_alignment_param ? count : -ENOMEM;
> +	return count;
>  }
>  
>  static BUS_ATTR_RW(resource_alignment);
> 

^ permalink raw reply	[flat|nested] 5+ messages in thread
* Re: [PATCH -next] PCI: Use GFP_ATOMIC in resource_alignment_store()
  2019-09-03 15:51   ` Logan Gunthorpe
@ 2019-09-05 21:47     ` Bjorn Helgaas
  0 siblings, 0 replies; 5+ messages in thread
From: Bjorn Helgaas @ 2019-09-05 21:47 UTC (permalink / raw)
  To: Logan Gunthorpe
  Cc: Christoph Hellwig, YueHaibing, linux-pci, linux-kernel, kernel-janitors

On Tue, Sep 03, 2019 at 09:51:05AM -0600, Logan Gunthorpe wrote:
> 
> 
> On 2019-09-02 1:50 a.m., Christoph Hellwig wrote:
> > On Sat, Aug 31, 2019 at 12:49:32PM +0000, YueHaibing wrote:
> >> When allocating memory, the GFP_KERNEL cannot be used during the
> >> spin_lock period. It may cause scheduling when holding spin_lock.
> >>
> >> Fixes: f13755318675 ("PCI: Move pci_[get|set]_resource_alignment_param() into their callers")
> >> Signed-off-by: YueHaibing <yuehaibing@huawei.com>
> >> ---
> >>  drivers/pci/pci.c | 2 +-
> >>  1 file changed, 1 insertion(+), 1 deletion(-)
> >>
> >> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> >> index 484e35349565..0b5fc6736f3f 100644
> >> --- a/drivers/pci/pci.c
> >> +++ b/drivers/pci/pci.c
> >> @@ -6148,7 +6148,7 @@ static ssize_t resource_alignment_store(struct bus_type *bus,
> >>  	spin_lock(&resource_alignment_lock);
> >>  
> >>  	kfree(resource_alignment_param);
> >> -	resource_alignment_param = kstrndup(buf, count, GFP_KERNEL);
> >> +	resource_alignment_param = kstrndup(buf, count, GFP_ATOMIC);
> >>  
> >>  	spin_unlock(&resource_alignment_lock);
> > 
> > Why not move the allocation outside the lock? Something like this
> > seems much more sensible:
> 
> Yes, that seems like a good way to do it. Bjorn, can you squash
> Christoph's patch or do you want me to resend a new one?

I folded Christoph's fix into it, thanks!

> > diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> > index 484e35349565..fe205829f676 100644
> > --- a/drivers/pci/pci.c
> > +++ b/drivers/pci/pci.c
> > @@ -6145,14 +6145,16 @@ static ssize_t resource_alignment_show(struct bus_type *bus, char *buf)
> >  static ssize_t resource_alignment_store(struct bus_type *bus,
> >  					const char *buf, size_t count)
> >  {
> > -	spin_lock(&resource_alignment_lock);
> > +	char *param = kstrndup(buf, count, GFP_KERNEL);
> >  
> > -	kfree(resource_alignment_param);
> > -	resource_alignment_param = kstrndup(buf, count, GFP_KERNEL);
> > +	if (!param)
> > +		return -ENOMEM;
> >  
> > +	spin_lock(&resource_alignment_lock);
> > +	kfree(resource_alignment_param);
> > +	resource_alignment_param = param;
> >  	spin_unlock(&resource_alignment_lock);
> > -
> > -	return resource_alignment_param ? count : -ENOMEM;
> > +	return count;
> >  }
> >  
> >  static BUS_ATTR_RW(resource_alignment);
> > 

^ permalink raw reply	[flat|nested] 5+ messages in thread
end of thread, other threads:[~2019-09-05 21:47 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-08-31 12:49 [PATCH -next] PCI: Use GFP_ATOMIC in resource_alignment_store() YueHaibing
2019-09-02  7:50 ` Christoph Hellwig
2019-09-03 15:51   ` Logan Gunthorpe
2019-09-05 21:47     ` Bjorn Helgaas
2019-09-03  8:22 ` [PATCH v2 -next] PCI: Don't use GFP_KERNEL for kstrbdup in resource_alignment_store YueHaibing

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).