RE: [PATCH] mwifiex: add __GFP_REPEAT to skb allocation call

[David Laight]

From: Amitkumar Karwar
> Sent: 05 April 2016 06:48
...
> Our one time allocated 64k buffer read from firmware contains multiple data 
> chunks. We have a feature
> called single port aggregation in which firmware attaches an aggregated 
> buffer to single port. So
> sometimes a single data chunk can exceed 32k. dev_kfree_skb_any() is called 
> to free those data chunks.

Ah yes, which particular problem does aggregating data into a single buffer 
solve?

David

N�r��yb�X��ǧv�^�)޺{.n�+{��*ޕ�,�{ay�ʇڙ�,j��f���h���z��w���
���j:+v���w�j�mzZ+�ݢj"��!�i

RE: [PATCH] mwifiex: add __GFP_REPEAT to skb allocation call

[Amitkumar Karwar]

Hi Eric,

Thanks for the comments.

> From: Eric Dumazet [mailto:eric.duma...@gmail.com]
> Sent: Tuesday, March 29, 2016 6:29 PM
> To: Wei-Ning Huang
> Cc: Kalle Valo; Linux Wireless; LKML; Amitkumar Karwar; Nishant
> Sarmukadam; Sameer Nanda; net...@vger.kernel.org; Sonny Rao; Douglas
> Anderson
> Subject: Re: [PATCH] mwifiex: add __GFP_REPEAT to skb allocation call
> 
> On Tue, 2016-03-29 at 17:27 +0800, Wei-Ning Huang wrote:
> > Adding some chromium devs to the thread.
> >
> > In, http://lxr.free-electrons.com/source/mm/page_alloc.c#L3152
> >
> > The default mm retry allocation when 'order <=
> > PAGE_ALLOC_COSTLY_ORDER' of gfp_mask contains __GFP_REPEAT.
> > PAGE_ALLOC_COSTLY_ORDER is defined to be 3. On systems with page size
> > = 4K, this means memory compaction and retry is only done when the
> > size of allocation is <= 32K In mwifiex, the allocation size is 64K.
> 
> 
> 
> >  When we have system with
> > memory fragmentation and allocation failed, there will be no retry.
> > This is why we need to add __GFP_REPEAT here to allow the system to
> > perform memory compaction and retry allocation.
> >
> > Maybe Amit@marvell can comment on if this is a good fix on this issue.
> > I'm also aware that marvell is the progress of implementing
> > scatter/gatter for mwifiex, which can also fix the issue.
> 
> Before SG is implemented, you really need to copy incoming frames into
> smallest chunks (to get lowest skb->truesize) and leave the 64KB
> allocated stuff forever in the driver.

We do have a 64KB pre-allocated buffer for receiving Rx data in our driver.

> 
> __GFP_REPEAT wont really solve the issue.
> 
> It seems the problem comes from the fact that the drivers calls
> dev_kfree_skb_any() after calling mwifiex_deaggr_sdio_pkt(), instead of
> recycling this very precious 64KB skb once memory gets fragmented.

Our one time allocated 64k buffer read from firmware contains multiple data 
chunks. We have a feature called single port aggregation in which firmware 
attaches an aggregated buffer to single port. So sometimes a single data chunk 
can exceed 32k. dev_kfree_skb_any() is called to free those data chunks.

> 
> Another problem is that mwifiex_deaggr_sdio_pkt() uses
> mwifiex_alloc_dma_align_buf() with GFP_KERNEL | GFP_DMA
> 
> Really GFP_DMA makes no sense here, since the skb is going to be
> processed by the stack, which has no such requirement.
> 
> Please use normal skb allocations there.

Sure. I will submit a patch for this.

Regards,
Amitkumar

Re: [PATCH] mwifiex: add __GFP_REPEAT to skb allocation call

[Eric Dumazet]

On Tue, 2016-03-29 at 17:27 +0800, Wei-Ning Huang wrote:
> Adding some chromium devs to the thread.
> 
> In, http://lxr.free-electrons.com/source/mm/page_alloc.c#L3152
> 
> The default mm retry allocation when 'order <=
> PAGE_ALLOC_COSTLY_ORDER' of gfp_mask contains __GFP_REPEAT.
> PAGE_ALLOC_COSTLY_ORDER is defined to be 3. On systems with page size
> = 4K, this means memory compaction and retry is only done when the
> size of allocation is <= 32K
> In mwifiex, the allocation size is 64K.



>  When we have system with
> memory fragmentation and allocation failed, there will be no retry.
> This is why we need to add __GFP_REPEAT here to allow the system to
> perform memory compaction and retry allocation.
> 
> Maybe Amit@marvell can comment on if this is a good fix on this issue.
> I'm also aware that marvell is the progress of implementing
> scatter/gatter for mwifiex, which can also fix the issue.

Before SG is implemented, you really need to copy incoming frames into
smallest chunks (to get lowest skb->truesize) and leave the 64KB
allocated stuff forever in the driver.

__GFP_REPEAT wont really solve the issue.

It seems the problem comes from the fact that the drivers calls
dev_kfree_skb_any() after calling mwifiex_deaggr_sdio_pkt(), instead of
recycling this very precious 64KB skb once memory gets fragmented.

Another problem is that mwifiex_deaggr_sdio_pkt() uses
mwifiex_alloc_dma_align_buf() with GFP_KERNEL | GFP_DMA

Really GFP_DMA makes no sense here, since the skb is going to be
processed by the stack, which has no such requirement.

Please use normal skb allocations there.

diff --git a/drivers/net/wireless/marvell/mwifiex/sdio.c 
b/drivers/net/wireless/marvell/mwifiex/sdio.c
index b2c839a..8404db5 100644
--- a/drivers/net/wireless/marvell/mwifiex/sdio.c
+++ b/drivers/net/wireless/marvell/mwifiex/sdio.c
@@ -1123,8 +1123,8 @@ static void mwifiex_deaggr_sdio_pkt(struct 
mwifiex_adapter *adapter,
__func__, pkt_len, blk_size);
break;
}
-   skb_deaggr = mwifiex_alloc_dma_align_buf(pkt_len,
-GFP_KERNEL | GFP_DMA);
+   skb_deaggr = __netdev_alloc_skb_ip_align(NULL, pkt_len,
+GFP_KERNEL);
if (!skb_deaggr)
break;
skb_put(skb_deaggr, pkt_len);




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

RE: [PATCH] mwifiex: add __GFP_REPEAT to skb allocation call

[Amitkumar Karwar]

> From: Wei-Ning Huang [mailto:wnhu...@google.com]
> Sent: Tuesday, March 29, 2016 2:57 PM
> To: Kalle Valo
> Cc: Linux Wireless; LKML; Amitkumar Karwar; Nishant Sarmukadam; Sameer
> Nanda; net...@vger.kernel.org; Sonny Rao; Douglas Anderson
> Subject: Re: [PATCH] mwifiex: add __GFP_REPEAT to skb allocation call
> 
> Adding some chromium devs to the thread.
> 
> In, http://lxr.free-electrons.com/source/mm/page_alloc.c#L3152
> 
> The default mm retry allocation when 'order <= PAGE_ALLOC_COSTLY_ORDER'
> of gfp_mask contains __GFP_REPEAT.
> PAGE_ALLOC_COSTLY_ORDER is defined to be 3. On systems with page size =
> 4K, this means memory compaction and retry is only done when the size of
> allocation is <= 32K In mwifiex, the allocation size is 64K. When we
> have system with memory fragmentation and allocation failed, there will
> be no retry.
> This is why we need to add __GFP_REPEAT here to allow the system to
> perform memory compaction and retry allocation.
> 
> Maybe Amit@marvell can comment on if this is a good fix on this issue.
> I'm also aware that marvell is the progress of implementing
> scatter/gatter for mwifiex, which can also fix the issue.
> 
> Wei-Ning
> 

This fix would be useful. We have a feature called single port aggregation in 
which sometimes data received from SDIO interface can be >32k (but less than 
64k). This feature improves throughput performance. We are preparing patches 
for scatter/gather feature. but scatter/gather won't be supported by some 
platforms. Hence this fix would still be needed.

Regards,
Amitkumar

Re: [PATCH] mwifiex: add __GFP_REPEAT to skb allocation call

[Wei-Ning Huang]

Adding some chromium devs to the thread.

In, http://lxr.free-electrons.com/source/mm/page_alloc.c#L3152

The default mm retry allocation when 'order <=
PAGE_ALLOC_COSTLY_ORDER' of gfp_mask contains __GFP_REPEAT.
PAGE_ALLOC_COSTLY_ORDER is defined to be 3. On systems with page size
= 4K, this means memory compaction and retry is only done when the
size of allocation is <= 32K
In mwifiex, the allocation size is 64K. When we have system with
memory fragmentation and allocation failed, there will be no retry.
This is why we need to add __GFP_REPEAT here to allow the system to
perform memory compaction and retry allocation.

Maybe Amit@marvell can comment on if this is a good fix on this issue.
I'm also aware that marvell is the progress of implementing
scatter/gatter for mwifiex, which can also fix the issue.

Wei-Ning

On Tue, Mar 29, 2016 at 4:37 PM, Kalle Valo  wrote:
> Wei-Ning Huang  writes:
>
>> "single skb allocation failure" happens when system is under heavy
>> memory pressure.  Add __GFP_REPEAT to skb allocation call so kernel
>> attempts to reclaim pages and retry the allocation.
>>
>> Signed-off-by: Wei-Ning Huang 
>
> Is this really a proper way to fix the issue? This is the first time I'm
> hearing about the flag and there isn't even a single user in
> drivers/net. I would like to get confirmation from others that
> __GFP_REPEAT is really ok to use in a wireless driver before I can take
> this.
>
> --
> Kalle Valo



-- 
Wei-Ning Huang, 黃偉寧 | Software Engineer, Google Inc., Taiwan |
wnhu...@google.com | Cell: +886 910-380678
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] mwifiex: add __GFP_REPEAT to skb allocation call

[Kalle Valo]

Wei-Ning Huang  writes:

> "single skb allocation failure" happens when system is under heavy
> memory pressure.  Add __GFP_REPEAT to skb allocation call so kernel
> attempts to reclaim pages and retry the allocation.
>
> Signed-off-by: Wei-Ning Huang 

Is this really a proper way to fix the issue? This is the first time I'm
hearing about the flag and there isn't even a single user in
drivers/net. I would like to get confirmation from others that
__GFP_REPEAT is really ok to use in a wireless driver before I can take
this.

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

Re: [PATCH] mwifiex: add __GFP_REPEAT to skb allocation call

[James Cameron]

On Tue, Mar 29, 2016 at 12:47:20PM +0800, Wei-Ning Huang wrote:
> "single skb allocation failure" happens when system is under heavy
> memory pressure.  Add __GFP_REPEAT to skb allocation call so kernel
> attempts to reclaim pages and retry the allocation.

Oh, that's interesting, we're back to this symptom again.

Nice to see this fix.

Heavy memory pressure on 3.5 caused dev_alloc_skb failure in this
driver.  Tracked at OLPC as #12694.

-- 
James Cameron
http://quozl.netrek.org/
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html