diff mbox

[00/11] DMA sync errors

Message ID 20100120152832.06a905a5@nehalam
State Rejected, archived
Headers show

Commit Message

stephen hemminger Jan. 20, 2010, 11:28 p.m. UTC 
This should fix the dma-debug API code (and documentation), to
avoid false positives when sync is done on a partial map.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

David Miller Jan. 21, 2010, 12:18 a.m. UTC | #1 
From: Stephen Hemminger <shemminger@vyatta.com>
Date: Wed, 20 Jan 2010 15:28:32 -0800

> This should fix the dma-debug API code (and documentation), to
> avoid false positives when sync is done on a partial map.
> 
> Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
 ...
> --- a/lib/dma-debug.c	2010-01-20 15:22:55.919519883 -0800
> +++ b/lib/dma-debug.c	2010-01-20 15:26:31.648895638 -0800
> @@ -285,11 +285,9 @@ static struct dma_debug_entry *hash_buck
>  	}
>  
>  	/*
> -	 * If we have multiple matches but no perfect-fit, just return
> -	 * NULL.
> +	 * If we have multiple matches but no perfect-fit
> +	 * return best value and let caller deal with it.
>  	 */
> -	ret = (matches == 1) ? ret : NULL;
> -
>  	return ret;
>  }
>  

I think you have to enforce a perfect-fit match for the
non-SYNC cases.

check_sync() can pass in 'true' for a new bool argument "partial_ok"
added to hash_bucket_find() whereas check_unmap() can pass 'false'.

get_nr_mapped_entries() should pass 'false' as well.

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
stephen hemminger Jan. 21, 2010, 12:34 a.m. UTC | #2 
On Wed, 20 Jan 2010 16:18:05 -0800 (PST)
David Miller <davem@davemloft.net> wrote:

> From: Stephen Hemminger <shemminger@vyatta.com>
> Date: Wed, 20 Jan 2010 15:28:32 -0800
> 
> > This should fix the dma-debug API code (and documentation), to
> > avoid false positives when sync is done on a partial map.
> > 
> > Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
>  ...
> > --- a/lib/dma-debug.c	2010-01-20 15:22:55.919519883 -0800
> > +++ b/lib/dma-debug.c	2010-01-20 15:26:31.648895638 -0800
> > @@ -285,11 +285,9 @@ static struct dma_debug_entry *hash_buck
> >  	}
> >  
> >  	/*
> > -	 * If we have multiple matches but no perfect-fit, just return
> > -	 * NULL.
> > +	 * If we have multiple matches but no perfect-fit
> > +	 * return best value and let caller deal with it.
> >  	 */
> > -	ret = (matches == 1) ? ret : NULL;
> > -
> >  	return ret;
> >  }
> >  
> 
> I think you have to enforce a perfect-fit match for the
> non-SYNC cases.
> 
> check_sync() can pass in 'true' for a new bool argument "partial_ok"
> added to hash_bucket_find() whereas check_unmap() can pass 'false'.
> 
> get_nr_mapped_entries() should pass 'false' as well.

check_unmap has checks for different size, different type,
call_ents, direction etc.

The code for get_nr_mapped_entries() is the only question;
it is used by unmap_sg() and probably needs perfect match only.



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller Jan. 21, 2010, 12:36 a.m. UTC | #3 
From: Stephen Hemminger <shemminger@vyatta.com>
Date: Wed, 20 Jan 2010 16:34:10 -0800

> check_unmap has checks for different size, different type,
> call_ents, direction etc.
> 
> The code for get_nr_mapped_entries() is the only question;
> it is used by unmap_sg() and probably needs perfect match only.

Ok.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Jarek Poplawski Jan. 21, 2010, 12:44 p.m. UTC | #4 
On Wed, Jan 20, 2010 at 04:34:10PM -0800, Stephen Hemminger wrote:
> On Wed, 20 Jan 2010 16:18:05 -0800 (PST)
> David Miller <davem@davemloft.net> wrote:
> 
> > From: Stephen Hemminger <shemminger@vyatta.com>
> > Date: Wed, 20 Jan 2010 15:28:32 -0800
> > 
> > > This should fix the dma-debug API code (and documentation), to
> > > avoid false positives when sync is done on a partial map.
> > > 
> > > Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
> >  ...
> > > --- a/lib/dma-debug.c	2010-01-20 15:22:55.919519883 -0800
> > > +++ b/lib/dma-debug.c	2010-01-20 15:26:31.648895638 -0800
> > > @@ -285,11 +285,9 @@ static struct dma_debug_entry *hash_buck
> > >  	}
> > >  
> > >  	/*
> > > -	 * If we have multiple matches but no perfect-fit, just return
> > > -	 * NULL.
> > > +	 * If we have multiple matches but no perfect-fit
> > > +	 * return best value and let caller deal with it.
> > >  	 */
> > > -	ret = (matches == 1) ? ret : NULL;
> > > -
> > >  	return ret;
> > >  }
> > >  
> > 
> > I think you have to enforce a perfect-fit match for the
> > non-SYNC cases.
> > 
> > check_sync() can pass in 'true' for a new bool argument "partial_ok"
> > added to hash_bucket_find() whereas check_unmap() can pass 'false'.
> > 
> > get_nr_mapped_entries() should pass 'false' as well.
> 
> check_unmap has checks for different size, different type,
> call_ents, direction etc.
> 
> The code for get_nr_mapped_entries() is the only question;
> it is used by unmap_sg() and probably needs perfect match only.
> 

Hmm... After looking at it more, mainly dma_sync_single_range()
implementations, I didn't find any reasons for these comments in
DMA-API.txt, but even if I missed something, this patch definitely
shows these lib/dma-debug warnings are buggy if they depend on the
number of mapped entries (while otherwise ref->size < entry->size
isn't reported). Btw, it seems code should try to return minimum
entry->size for such a check.

Thanks Stephen, now I feel 99.9% safe ;-)

Sorry for mistrust, David!
Jarek P.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller Jan. 21, 2010, 1:12 p.m. UTC | #5 
From: Jarek Poplawski <jarkao2@gmail.com>
Date: Thu, 21 Jan 2010 12:44:23 +0000

> Sorry for mistrust, David!

That's twice in the past week Jarek.

One more and you will be in real troubles :-)
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

--- a/Documentation/DMA-API.txt	2010-01-20 15:17:01.390143729 -0800
+++ b/Documentation/DMA-API.txt	2010-01-20 15:18:48.967875255 -0800
@@ -377,9 +377,10 @@  void
 pci_dma_sync_sg(struct pci_dev *hwdev, struct scatterlist *sg,
 		       int nelems, int direction)
 
-Synchronise a single contiguous or scatter/gather mapping.  All the
-parameters must be the same as those passed into the single mapping
-API.
+Synchronise a single contiguous or scatter/gather mapping.  The
+device and handle must be the same as those passed into the single mapping
+API. The size can be less than the original mapping if only part
+of the mapping needs to be accessed.
 
 Notes:  You must do this:
 
--- a/lib/dma-debug.c	2010-01-20 15:22:55.919519883 -0800
+++ b/lib/dma-debug.c	2010-01-20 15:26:31.648895638 -0800
@@ -285,11 +285,9 @@  static struct dma_debug_entry *hash_buck
 	}
 
 	/*
-	 * If we have multiple matches but no perfect-fit, just return
-	 * NULL.
+	 * If we have multiple matches but no perfect-fit
+	 * return best value and let caller deal with it.
 	 */
-	ret = (matches == 1) ? ret : NULL;
-
 	return ret;
 }