[stable] [PATCH] sfc: Fix efx_rx_buf_offset() for recycled pages

[stable] [PATCH] sfc: Fix efx_rx_buf_offset() for recycled pages

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 1323 bytes --]

[Re-sending with cc to the current stable address.  I still had the old
address in contacts, though I've definitely sent to the current address
previously...]

This bug fix is only for stable branches older than 3.10.  The bug was
fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
mapping/unmapping costs'), but that change is totally unsuitable for
stable.

Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
swiotlb') added an explicit page_offset member to struct
efx_rx_buffer, which must be set consistently with the u.page and
dma_addr fields.  However, it failed to add the necessary assignment
in efx_resurrect_rx_buffer().  It also did not correct the calculation
of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
that DMA-mapping a page will result in a page-aligned DMA address
(exactly what swiotlb violates).

Add the assignment of efx_rx_buffer::page_offset and change the
calculation of dma_addr to make use of it.

I'm attaching the patches for 3.0.y, 3.2.y and 3.4.y (the last of which
should also be applicable to the later unofficial stable branches).

Ben.

-- 
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


[-- Attachment #2: sfc-Fix-efx_rx_buf_offset-for-recycled-pages-3.0.patch --]
[-- Type: text/x-patch, Size: 1833 bytes --]

>From d4e70d42eff6938c12d7d0b12f657b10d06a156d Mon Sep 17 00:00:00 2001
From: Ben Hutchings <bhutchings@solarflare.com>
Date: Fri, 6 Sep 2013 22:39:20 +0100
Subject: sfc: Fix efx_rx_buf_offset() for recycled pages

This bug fix is only for stable branches older than 3.10.  The bug was
fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
mapping/unmapping costs'), but that change is totally unsuitable for
stable.

Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
swiotlb') added an explicit page_offset member to struct
efx_rx_buffer, which must be set consistently with the u.page and
dma_addr fields.  However, it failed to add the necessary assignment
in efx_resurrect_rx_buffer().  It also did not correct the calculation
of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
that DMA-mapping a page will result in a page-aligned DMA address
(exactly what swiotlb violates).

Add the assignment of efx_rx_buffer::page_offset and change the
calculation of dma_addr to make use of it.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/sfc/rx.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/sfc/rx.c b/drivers/net/sfc/rx.c
index d429f0a..26e4cca 100644
--- a/drivers/net/sfc/rx.c
+++ b/drivers/net/sfc/rx.c
@@ -311,8 +311,9 @@ static void efx_resurrect_rx_buffer(struct efx_rx_queue *rx_queue,
 
 	index = rx_queue->added_count & rx_queue->ptr_mask;
 	new_buf = efx_rx_buffer(rx_queue, index);
-	new_buf->dma_addr = rx_buf->dma_addr ^ (PAGE_SIZE >> 1);
 	new_buf->u.page = rx_buf->u.page;
+	new_buf->page_offset = rx_buf->page_offset ^ (PAGE_SIZE >> 1);
+	new_buf->dma_addr = state->dma_addr + new_buf->page_offset;
 	new_buf->len = rx_buf->len;
 	new_buf->is_page = true;
 	++rx_queue->added_count;
-- 
1.8.1.4


[-- Attachment #3: sfc-Fix-efx_rx_buf_offset-for-recycled-pages-3.2.patch --]
[-- Type: text/x-patch, Size: 1878 bytes --]

>From 43587737635de32882d9f69df36562e0d5b0e0bd Mon Sep 17 00:00:00 2001
From: Ben Hutchings <bhutchings@solarflare.com>
Date: Fri, 6 Sep 2013 22:39:20 +0100
Subject: sfc: Fix efx_rx_buf_offset() for recycled pages

This bug fix is only for stable branches older than 3.10.  The bug was
fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
mapping/unmapping costs'), but that change is totally unsuitable for
stable.

Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
swiotlb') added an explicit page_offset member to struct
efx_rx_buffer, which must be set consistently with the u.page and
dma_addr fields.  However, it failed to add the necessary assignment
in efx_resurrect_rx_buffer().  It also did not correct the calculation
of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
that DMA-mapping a page will result in a page-aligned DMA address
(exactly what swiotlb violates).

Add the assignment of efx_rx_buffer::page_offset and change the
calculation of dma_addr to make use of it.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/ethernet/sfc/rx.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/sfc/rx.c b/drivers/net/ethernet/sfc/rx.c
index 9ce8665..c231b3f 100644
--- a/drivers/net/ethernet/sfc/rx.c
+++ b/drivers/net/ethernet/sfc/rx.c
@@ -312,8 +312,9 @@ static void efx_resurrect_rx_buffer(struct efx_rx_queue *rx_queue,
 
 	index = rx_queue->added_count & rx_queue->ptr_mask;
 	new_buf = efx_rx_buffer(rx_queue, index);
-	new_buf->dma_addr = rx_buf->dma_addr ^ (PAGE_SIZE >> 1);
 	new_buf->u.page = rx_buf->u.page;
+	new_buf->page_offset = rx_buf->page_offset ^ (PAGE_SIZE >> 1);
+	new_buf->dma_addr = state->dma_addr + new_buf->page_offset;
 	new_buf->len = rx_buf->len;
 	new_buf->is_page = true;
 	++rx_queue->added_count;
-- 
1.8.1.4


[-- Attachment #4: sfc-Fix-efx_rx_buf_offset-for-recycled-pages-3.4.patch --]
[-- Type: text/x-patch, Size: 1888 bytes --]

>From 55961f9a914e637f143da275da42154a49c9743a Mon Sep 17 00:00:00 2001
From: Ben Hutchings <bhutchings@solarflare.com>
Date: Fri, 6 Sep 2013 22:39:20 +0100
Subject: sfc: Fix efx_rx_buf_offset() for recycled pages

This bug fix is only for stable branches older than 3.10.  The bug was
fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
mapping/unmapping costs'), but that change is totally unsuitable for
stable.

Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
swiotlb') added an explicit page_offset member to struct
efx_rx_buffer, which must be set consistently with the u.page and
dma_addr fields.  However, it failed to add the necessary assignment
in efx_resurrect_rx_buffer().  It also did not correct the calculation
of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
that DMA-mapping a page will result in a page-aligned DMA address
(exactly what swiotlb violates).

Add the assignment of efx_rx_buffer::page_offset and change the
calculation of dma_addr to make use of it.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/ethernet/sfc/rx.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/sfc/rx.c b/drivers/net/ethernet/sfc/rx.c
index 7b3c068..62311ad3 100644
--- a/drivers/net/ethernet/sfc/rx.c
+++ b/drivers/net/ethernet/sfc/rx.c
@@ -311,8 +311,9 @@ static void efx_resurrect_rx_buffer(struct efx_rx_queue *rx_queue,
 
 	index = rx_queue->added_count & rx_queue->ptr_mask;
 	new_buf = efx_rx_buffer(rx_queue, index);
-	new_buf->dma_addr = rx_buf->dma_addr ^ (PAGE_SIZE >> 1);
 	new_buf->u.page = rx_buf->u.page;
+	new_buf->page_offset = rx_buf->page_offset ^ (PAGE_SIZE >> 1);
+	new_buf->dma_addr = state->dma_addr + new_buf->page_offset;
 	new_buf->len = rx_buf->len;
 	new_buf->flags = EFX_RX_BUF_PAGE;
 	++rx_queue->added_count;
-- 
1.8.1.4

Re: [PATCH] sfc: Fix efx_rx_buf_offset() for recycled pages

[David Miller]

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Thu, 19 Sep 2013 17:13:51 +0100

> This bug fix is only for stable branches older than 3.10.  The bug was
> fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
> mapping/unmapping costs'), but that change is totally unsuitable for
> stable.
> 
> Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
> swiotlb') added an explicit page_offset member to struct
> efx_rx_buffer, which must be set consistently with the u.page and
> dma_addr fields.  However, it failed to add the necessary assignment
> in efx_resurrect_rx_buffer().  It also did not correct the calculation
> of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
> that DMA-mapping a page will result in a page-aligned DMA address
> (exactly what swiotlb violates).
> 
> Add the assignment of efx_rx_buffer::page_offset and change the
> calculation of dma_addr to make use of it.

No objections.

Re: [PATCH] sfc: Fix efx_rx_buf_offset() for recycled pages

[Luis Henriques]

Ben Hutchings <bhutchings@solarflare.com> writes:

> [Re-sending with cc to the current stable address.  I still had the old
> address in contacts, though I've definitely sent to the current address
> previously...]
>
> This bug fix is only for stable branches older than 3.10.  The bug was
> fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
> mapping/unmapping costs'), but that change is totally unsuitable for
> stable.
>
> Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
> swiotlb') added an explicit page_offset member to struct
> efx_rx_buffer, which must be set consistently with the u.page and
> dma_addr fields.  However, it failed to add the necessary assignment
> in efx_resurrect_rx_buffer().  It also did not correct the calculation
> of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
> that DMA-mapping a page will result in a page-aligned DMA address
> (exactly what swiotlb violates).
>
> Add the assignment of efx_rx_buffer::page_offset and change the
> calculation of dma_addr to make use of it.
>
> I'm attaching the patches for 3.0.y, 3.2.y and 3.4.y (the last of which
> should also be applicable to the later unofficial stable branches).
>
> Ben.

Thanks Ben.  I'm queuing it for the 3.5 kernel.

Cheers,
-- 
Luis

Re: [PATCH] sfc: Fix efx_rx_buf_offset() for recycled pages

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 1362 bytes --]

On Thu, 2013-09-19 at 17:13 +0100, Ben Hutchings wrote:
> [Re-sending with cc to the current stable address.  I still had the old
> address in contacts, though I've definitely sent to the current address
> previously...]
> 
> This bug fix is only for stable branches older than 3.10.  The bug was
> fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
> mapping/unmapping costs'), but that change is totally unsuitable for
> stable.
> 
> Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
> swiotlb') added an explicit page_offset member to struct
> efx_rx_buffer, which must be set consistently with the u.page and
> dma_addr fields.  However, it failed to add the necessary assignment
> in efx_resurrect_rx_buffer().  It also did not correct the calculation
> of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
> that DMA-mapping a page will result in a page-aligned DMA address
> (exactly what swiotlb violates).
> 
> Add the assignment of efx_rx_buffer::page_offset and change the
> calculation of dma_addr to make use of it.
> 
> I'm attaching the patches for 3.0.y, 3.2.y and 3.4.y (the last of which
> should also be applicable to the later unofficial stable branches).

Queued up for 3.2.

Ben.

-- 
Ben Hutchings
Tomorrow will be cancelled due to lack of interest.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

[stable] [PATCH] sfc: Fix efx_rx_buf_offset() for recycled pages

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 1162 bytes --]

This bug fix is only for stable branches older than 3.10.  The bug was
fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
mapping/unmapping costs'), but that change is totally unsuitable for
stable.

Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
swiotlb') added an explicit page_offset member to struct
efx_rx_buffer, which must be set consistently with the u.page and
dma_addr fields.  However, it failed to add the necessary assignment
in efx_resurrect_rx_buffer().  It also did not correct the calculation
of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
that DMA-mapping a page will result in a page-aligned DMA address
(exactly what swiotlb violates).

Add the assignment of efx_rx_buffer::page_offset and change the
calculation of dma_addr to make use of it.

I'm attaching the patches for 3.0.y, 3.2.y and 3.4.y (the last of which
should also be applicable to the later unofficial stable branches).

Ben.

-- 
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.

[-- Attachment #2: sfc-Fix-efx_rx_buf_offset-for-recycled-pages-3.0.patch --]
[-- Type: text/x-patch, Size: 1833 bytes --]

>From d4e70d42eff6938c12d7d0b12f657b10d06a156d Mon Sep 17 00:00:00 2001
From: Ben Hutchings <bhutchings@solarflare.com>
Date: Fri, 6 Sep 2013 22:39:20 +0100
Subject: sfc: Fix efx_rx_buf_offset() for recycled pages

This bug fix is only for stable branches older than 3.10.  The bug was
fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
mapping/unmapping costs'), but that change is totally unsuitable for
stable.

Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
swiotlb') added an explicit page_offset member to struct
efx_rx_buffer, which must be set consistently with the u.page and
dma_addr fields.  However, it failed to add the necessary assignment
in efx_resurrect_rx_buffer().  It also did not correct the calculation
of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
that DMA-mapping a page will result in a page-aligned DMA address
(exactly what swiotlb violates).

Add the assignment of efx_rx_buffer::page_offset and change the
calculation of dma_addr to make use of it.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/sfc/rx.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/sfc/rx.c b/drivers/net/sfc/rx.c
index d429f0a..26e4cca 100644
--- a/drivers/net/sfc/rx.c
+++ b/drivers/net/sfc/rx.c
@@ -311,8 +311,9 @@ static void efx_resurrect_rx_buffer(struct efx_rx_queue *rx_queue,
 
 	index = rx_queue->added_count & rx_queue->ptr_mask;
 	new_buf = efx_rx_buffer(rx_queue, index);
-	new_buf->dma_addr = rx_buf->dma_addr ^ (PAGE_SIZE >> 1);
 	new_buf->u.page = rx_buf->u.page;
+	new_buf->page_offset = rx_buf->page_offset ^ (PAGE_SIZE >> 1);
+	new_buf->dma_addr = state->dma_addr + new_buf->page_offset;
 	new_buf->len = rx_buf->len;
 	new_buf->is_page = true;
 	++rx_queue->added_count;
-- 
1.8.1.4


[-- Attachment #3: sfc-Fix-efx_rx_buf_offset-for-recycled-pages-3.2.patch --]
[-- Type: text/x-patch, Size: 1878 bytes --]

>From 43587737635de32882d9f69df36562e0d5b0e0bd Mon Sep 17 00:00:00 2001
From: Ben Hutchings <bhutchings@solarflare.com>
Date: Fri, 6 Sep 2013 22:39:20 +0100
Subject: sfc: Fix efx_rx_buf_offset() for recycled pages

This bug fix is only for stable branches older than 3.10.  The bug was
fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
mapping/unmapping costs'), but that change is totally unsuitable for
stable.

Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
swiotlb') added an explicit page_offset member to struct
efx_rx_buffer, which must be set consistently with the u.page and
dma_addr fields.  However, it failed to add the necessary assignment
in efx_resurrect_rx_buffer().  It also did not correct the calculation
of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
that DMA-mapping a page will result in a page-aligned DMA address
(exactly what swiotlb violates).

Add the assignment of efx_rx_buffer::page_offset and change the
calculation of dma_addr to make use of it.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/ethernet/sfc/rx.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/sfc/rx.c b/drivers/net/ethernet/sfc/rx.c
index 9ce8665..c231b3f 100644
--- a/drivers/net/ethernet/sfc/rx.c
+++ b/drivers/net/ethernet/sfc/rx.c
@@ -312,8 +312,9 @@ static void efx_resurrect_rx_buffer(struct efx_rx_queue *rx_queue,
 
 	index = rx_queue->added_count & rx_queue->ptr_mask;
 	new_buf = efx_rx_buffer(rx_queue, index);
-	new_buf->dma_addr = rx_buf->dma_addr ^ (PAGE_SIZE >> 1);
 	new_buf->u.page = rx_buf->u.page;
+	new_buf->page_offset = rx_buf->page_offset ^ (PAGE_SIZE >> 1);
+	new_buf->dma_addr = state->dma_addr + new_buf->page_offset;
 	new_buf->len = rx_buf->len;
 	new_buf->is_page = true;
 	++rx_queue->added_count;
-- 
1.8.1.4


[-- Attachment #4: sfc-Fix-efx_rx_buf_offset-for-recycled-pages-3.4.patch --]
[-- Type: text/x-patch, Size: 1888 bytes --]

>From 55961f9a914e637f143da275da42154a49c9743a Mon Sep 17 00:00:00 2001
From: Ben Hutchings <bhutchings@solarflare.com>
Date: Fri, 6 Sep 2013 22:39:20 +0100
Subject: sfc: Fix efx_rx_buf_offset() for recycled pages

This bug fix is only for stable branches older than 3.10.  The bug was
fixed upstream by commit 2768935a4660 ('sfc: reuse pages to avoid DMA
mapping/unmapping costs'), but that change is totally unsuitable for
stable.

Commit b590ace09d51 ('sfc: Fix efx_rx_buf_offset() in the presence of
swiotlb') added an explicit page_offset member to struct
efx_rx_buffer, which must be set consistently with the u.page and
dma_addr fields.  However, it failed to add the necessary assignment
in efx_resurrect_rx_buffer().  It also did not correct the calculation
of efx_rx_buffer::dma_addr in efx_resurrect_rx_buffer(), which assumes
that DMA-mapping a page will result in a page-aligned DMA address
(exactly what swiotlb violates).

Add the assignment of efx_rx_buffer::page_offset and change the
calculation of dma_addr to make use of it.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/ethernet/sfc/rx.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/sfc/rx.c b/drivers/net/ethernet/sfc/rx.c
index 7b3c068..62311ad3 100644
--- a/drivers/net/ethernet/sfc/rx.c
+++ b/drivers/net/ethernet/sfc/rx.c
@@ -311,8 +311,9 @@ static void efx_resurrect_rx_buffer(struct efx_rx_queue *rx_queue,
 
 	index = rx_queue->added_count & rx_queue->ptr_mask;
 	new_buf = efx_rx_buffer(rx_queue, index);
-	new_buf->dma_addr = rx_buf->dma_addr ^ (PAGE_SIZE >> 1);
 	new_buf->u.page = rx_buf->u.page;
+	new_buf->page_offset = rx_buf->page_offset ^ (PAGE_SIZE >> 1);
+	new_buf->dma_addr = state->dma_addr + new_buf->page_offset;
 	new_buf->len = rx_buf->len;
 	new_buf->flags = EFX_RX_BUF_PAGE;
 	++rx_queue->added_count;
-- 
1.8.1.4