* [PATCH v2] net: mv643xx_eth: Fix highmem support in non-TSO egress path
@ 2015-01-22 12:14 Ezequiel Garcia
2015-01-22 12:40 ` David Laight
2015-01-22 14:33 ` [PATCH v3] " Ezequiel Garcia
0 siblings, 2 replies; 7+ messages in thread
From: Ezequiel Garcia @ 2015-01-22 12:14 UTC (permalink / raw)
To: David Miller, Russell King; +Cc: netdev, B38611, fabio.estevam, Ezequiel Garcia
Commit 69ad0dd7af22b61d9e0e68e56b6290121618b0fb
Author: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Date: Mon May 19 13:59:59 2014 -0300
net: mv643xx_eth: Use dma_map_single() to map the skb fragments
caused a nasty regression by removing the support for highmem skb
fragments. By using page_address() to get the address of a fragment's
page, we are assuming a lowmem page. However, such assumption is incorrect,
as fragments can be in highmem pages, resulting in very nasty issues.
This commit fixes this by using the skb_frag_dma_map() helper,
which takes care of mapping the skb fragment properly. Additionally,
the type of mapping is now tracked, so it can be unmapped using
dma_unmap_page or dma_unmap_single when appropriate.
Fixes: 69ad0dd7af22 ("net: mv643xx_eth: Use dma_map_single() to map the skb fragments")
Reported-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
---
Changes from v1:
* Sending the mv643xx_eth fix for now.
* Fix DMA mapping type track, to use the correct unmap call.
drivers/net/ethernet/marvell/mv643xx_eth.c | 41 +++++++++++++++++++++++++-----
1 file changed, 34 insertions(+), 7 deletions(-)
diff --git a/drivers/net/ethernet/marvell/mv643xx_eth.c b/drivers/net/ethernet/marvell/mv643xx_eth.c
index a62fc38..e4c8461 100644
--- a/drivers/net/ethernet/marvell/mv643xx_eth.c
+++ b/drivers/net/ethernet/marvell/mv643xx_eth.c
@@ -192,6 +192,10 @@ static char mv643xx_eth_driver_version[] = "1.4";
#define IS_TSO_HEADER(txq, addr) \
((addr >= txq->tso_hdrs_dma) && \
(addr < txq->tso_hdrs_dma + txq->tx_ring_size * TSO_HEADER_SIZE))
+
+#define DESC_DMA_MAP_SINGLE 0
+#define DESC_DMA_MAP_PAGE 1
+
/*
* RX/TX descriptors.
*/
@@ -362,6 +366,7 @@ struct tx_queue {
dma_addr_t tso_hdrs_dma;
struct tx_desc *tx_desc_area;
+ char *tx_desc_mapping; /* array to track the type of the dma mapping */
dma_addr_t tx_desc_dma;
int tx_desc_area_size;
@@ -750,6 +755,7 @@ txq_put_data_tso(struct net_device *dev, struct tx_queue *txq,
if (txq->tx_curr_desc == txq->tx_ring_size)
txq->tx_curr_desc = 0;
desc = &txq->tx_desc_area[tx_index];
+ txq->tx_desc_mapping[tx_index] = DESC_DMA_MAP_SINGLE;
desc->l4i_chk = 0;
desc->byte_cnt = length;
@@ -879,14 +885,13 @@ static void txq_submit_frag_skb(struct tx_queue *txq, struct sk_buff *skb)
skb_frag_t *this_frag;
int tx_index;
struct tx_desc *desc;
- void *addr;
this_frag = &skb_shinfo(skb)->frags[frag];
- addr = page_address(this_frag->page.p) + this_frag->page_offset;
tx_index = txq->tx_curr_desc++;
if (txq->tx_curr_desc == txq->tx_ring_size)
txq->tx_curr_desc = 0;
desc = &txq->tx_desc_area[tx_index];
+ txq->tx_desc_mapping[tx_index] = DESC_DMA_MAP_PAGE;
/*
* The last fragment will generate an interrupt
@@ -902,8 +907,9 @@ static void txq_submit_frag_skb(struct tx_queue *txq, struct sk_buff *skb)
desc->l4i_chk = 0;
desc->byte_cnt = skb_frag_size(this_frag);
- desc->buf_ptr = dma_map_single(mp->dev->dev.parent, addr,
- desc->byte_cnt, DMA_TO_DEVICE);
+ desc->buf_ptr = skb_frag_dma_map(mp->dev->dev.parent,
+ this_frag, 0, desc->byte_cnt,
+ DMA_TO_DEVICE);
}
}
@@ -936,6 +942,7 @@ static int txq_submit_skb(struct tx_queue *txq, struct sk_buff *skb,
if (txq->tx_curr_desc == txq->tx_ring_size)
txq->tx_curr_desc = 0;
desc = &txq->tx_desc_area[tx_index];
+ txq->tx_desc_mapping[tx_index] = DESC_DMA_MAP_SINGLE;
if (nr_frags) {
txq_submit_frag_skb(txq, skb);
@@ -1047,9 +1054,12 @@ static int txq_reclaim(struct tx_queue *txq, int budget, int force)
int tx_index;
struct tx_desc *desc;
u32 cmd_sts;
+ char desc_dma_map;
tx_index = txq->tx_used_desc;
desc = &txq->tx_desc_area[tx_index];
+ desc_dma_map = txq->tx_desc_mapping[tx_index];
+
cmd_sts = desc->cmd_sts;
if (cmd_sts & BUFFER_OWNED_BY_DMA) {
@@ -1065,9 +1075,19 @@ static int txq_reclaim(struct tx_queue *txq, int budget, int force)
reclaimed++;
txq->tx_desc_count--;
- if (!IS_TSO_HEADER(txq, desc->buf_ptr))
- dma_unmap_single(mp->dev->dev.parent, desc->buf_ptr,
- desc->byte_cnt, DMA_TO_DEVICE);
+ if (!IS_TSO_HEADER(txq, desc->buf_ptr)) {
+
+ if (desc_dma_map == DESC_DMA_MAP_PAGE)
+ dma_unmap_page(mp->dev->dev.parent,
+ desc->buf_ptr,
+ desc->byte_cnt,
+ DMA_TO_DEVICE);
+ else
+ dma_unmap_single(mp->dev->dev.parent,
+ desc->buf_ptr,
+ desc->byte_cnt,
+ DMA_TO_DEVICE);
+ }
if (cmd_sts & TX_ENABLE_INTERRUPT) {
struct sk_buff *skb = __skb_dequeue(&txq->tx_skb);
@@ -2048,6 +2068,11 @@ static int txq_init(struct mv643xx_eth_private *mp, int index)
nexti * sizeof(struct tx_desc);
}
+ txq->tx_desc_mapping = kcalloc(txq->tx_ring_size, sizeof(char),
+ GFP_KERNEL);
+ if (!txq->tx_desc_mapping)
+ return -ENOMEM;
+
/* Allocate DMA buffers for TSO MAC/IP/TCP headers */
txq->tso_hdrs = dma_alloc_coherent(mp->dev->dev.parent,
txq->tx_ring_size * TSO_HEADER_SIZE,
@@ -2077,6 +2102,8 @@ static void txq_deinit(struct tx_queue *txq)
else
dma_free_coherent(mp->dev->dev.parent, txq->tx_desc_area_size,
txq->tx_desc_area, txq->tx_desc_dma);
+ kfree(txq->tx_desc_mapping);
+
if (txq->tso_hdrs)
dma_free_coherent(mp->dev->dev.parent,
txq->tx_ring_size * TSO_HEADER_SIZE,
--
2.2.1
^ permalink raw reply related [flat|nested] 7+ messages in thread
* RE: [PATCH v2] net: mv643xx_eth: Fix highmem support in non-TSO egress path
2015-01-22 12:14 [PATCH v2] net: mv643xx_eth: Fix highmem support in non-TSO egress path Ezequiel Garcia
@ 2015-01-22 12:40 ` David Laight
2015-01-22 13:26 ` Ezequiel Garcia
2015-01-22 14:33 ` [PATCH v3] " Ezequiel Garcia
1 sibling, 1 reply; 7+ messages in thread
From: David Laight @ 2015-01-22 12:40 UTC (permalink / raw)
To: 'Ezequiel Garcia', David Miller, Russell King
Cc: netdev, B38611, fabio.estevam
From: Ezequiel Garcia
> Commit 69ad0dd7af22b61d9e0e68e56b6290121618b0fb
> Author: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> Date: Mon May 19 13:59:59 2014 -0300
>
> net: mv643xx_eth: Use dma_map_single() to map the skb fragments
>
> caused a nasty regression by removing the support for highmem skb
> fragments. By using page_address() to get the address of a fragment's
> page, we are assuming a lowmem page. However, such assumption is incorrect,
> as fragments can be in highmem pages, resulting in very nasty issues.
>
> This commit fixes this by using the skb_frag_dma_map() helper,
> which takes care of mapping the skb fragment properly. Additionally,
> the type of mapping is now tracked, so it can be unmapped using
> dma_unmap_page or dma_unmap_single when appropriate.
>
> Fixes: 69ad0dd7af22 ("net: mv643xx_eth: Use dma_map_single() to map the skb fragments")
> Reported-by: Russell King <rmk+kernel@arm.linux.org.uk>
> Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
...
> @@ -2048,6 +2068,11 @@ static int txq_init(struct mv643xx_eth_private *mp, int index)
> nexti * sizeof(struct tx_desc);
> }
>
> + txq->tx_desc_mapping = kcalloc(txq->tx_ring_size, sizeof(char),
> + GFP_KERNEL);
> + if (!txq->tx_desc_mapping)
> + return -ENOMEM;
> +
> /* Allocate DMA buffers for TSO MAC/IP/TCP headers */
> txq->tso_hdrs = dma_alloc_coherent(mp->dev->dev.parent,
> txq->tx_ring_size * TSO_HEADER_SIZE,
I'm guessing there is an error path for dma_alloc_coherent() failing.
You've not modified it to free tx_desc_mapping.
David
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v2] net: mv643xx_eth: Fix highmem support in non-TSO egress path
2015-01-22 12:40 ` David Laight
@ 2015-01-22 13:26 ` Ezequiel Garcia
0 siblings, 0 replies; 7+ messages in thread
From: Ezequiel Garcia @ 2015-01-22 13:26 UTC (permalink / raw)
To: David Laight, David Miller, Russell King; +Cc: netdev, B38611, fabio.estevam
On 01/22/2015 09:40 AM, David Laight wrote:
> From: Ezequiel Garcia
>> Commit 69ad0dd7af22b61d9e0e68e56b6290121618b0fb
>> Author: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
>> Date: Mon May 19 13:59:59 2014 -0300
>>
>> net: mv643xx_eth: Use dma_map_single() to map the skb fragments
>>
>> caused a nasty regression by removing the support for highmem skb
>> fragments. By using page_address() to get the address of a fragment's
>> page, we are assuming a lowmem page. However, such assumption is incorrect,
>> as fragments can be in highmem pages, resulting in very nasty issues.
>>
>> This commit fixes this by using the skb_frag_dma_map() helper,
>> which takes care of mapping the skb fragment properly. Additionally,
>> the type of mapping is now tracked, so it can be unmapped using
>> dma_unmap_page or dma_unmap_single when appropriate.
>>
>> Fixes: 69ad0dd7af22 ("net: mv643xx_eth: Use dma_map_single() to map the skb fragments")
>> Reported-by: Russell King <rmk+kernel@arm.linux.org.uk>
>> Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> ...
>> @@ -2048,6 +2068,11 @@ static int txq_init(struct mv643xx_eth_private *mp, int index)
>> nexti * sizeof(struct tx_desc);
>> }
>>
>> + txq->tx_desc_mapping = kcalloc(txq->tx_ring_size, sizeof(char),
>> + GFP_KERNEL);
>> + if (!txq->tx_desc_mapping)
>> + return -ENOMEM;
>> +
>> /* Allocate DMA buffers for TSO MAC/IP/TCP headers */
>> txq->tso_hdrs = dma_alloc_coherent(mp->dev->dev.parent,
>> txq->tx_ring_size * TSO_HEADER_SIZE,
>
> I'm guessing there is an error path for dma_alloc_coherent() failing.
> You've not modified it to free tx_desc_mapping.
>
Yeah, you guess right.
--
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH v3] net: mv643xx_eth: Fix highmem support in non-TSO egress path
2015-01-22 12:14 [PATCH v2] net: mv643xx_eth: Fix highmem support in non-TSO egress path Ezequiel Garcia
2015-01-22 12:40 ` David Laight
@ 2015-01-22 14:33 ` Ezequiel Garcia
2015-01-24 13:45 ` Ezequiel Garcia
2015-01-27 0:15 ` David Miller
1 sibling, 2 replies; 7+ messages in thread
From: Ezequiel Garcia @ 2015-01-22 14:33 UTC (permalink / raw)
To: David Miller, Russell King
Cc: netdev, B38611, fabio.estevam, David.Laight, Ezequiel Garcia
Commit 69ad0dd7af22b61d9e0e68e56b6290121618b0fb
Author: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Date: Mon May 19 13:59:59 2014 -0300
net: mv643xx_eth: Use dma_map_single() to map the skb fragments
caused a nasty regression by removing the support for highmem skb
fragments. By using page_address() to get the address of a fragment's
page, we are assuming a lowmem page. However, such assumption is incorrect,
as fragments can be in highmem pages, resulting in very nasty issues.
This commit fixes this by using the skb_frag_dma_map() helper,
which takes care of mapping the skb fragment properly. Additionally,
the type of mapping is now tracked, so it can be unmapped using
dma_unmap_page or dma_unmap_single when appropriate.
This commit also fixes the error path in txq_init() to release the
resources properly.
Fixes: 69ad0dd7af22 ("net: mv643xx_eth: Use dma_map_single() to map the skb fragments")
Reported-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
---
Changes from v2:
* Add proper resource release in txq_init() error path
Changes from v1:
* Sending the mv643xx_eth fix for now.
* Fix DMA mapping type track, to use the correct unmap call.
---
drivers/net/ethernet/marvell/mv643xx_eth.c | 59 +++++++++++++++++++++++++-----
1 file changed, 49 insertions(+), 10 deletions(-)
diff --git a/drivers/net/ethernet/marvell/mv643xx_eth.c b/drivers/net/ethernet/marvell/mv643xx_eth.c
index a62fc38..1c75829 100644
--- a/drivers/net/ethernet/marvell/mv643xx_eth.c
+++ b/drivers/net/ethernet/marvell/mv643xx_eth.c
@@ -192,6 +192,10 @@ static char mv643xx_eth_driver_version[] = "1.4";
#define IS_TSO_HEADER(txq, addr) \
((addr >= txq->tso_hdrs_dma) && \
(addr < txq->tso_hdrs_dma + txq->tx_ring_size * TSO_HEADER_SIZE))
+
+#define DESC_DMA_MAP_SINGLE 0
+#define DESC_DMA_MAP_PAGE 1
+
/*
* RX/TX descriptors.
*/
@@ -362,6 +366,7 @@ struct tx_queue {
dma_addr_t tso_hdrs_dma;
struct tx_desc *tx_desc_area;
+ char *tx_desc_mapping; /* array to track the type of the dma mapping */
dma_addr_t tx_desc_dma;
int tx_desc_area_size;
@@ -750,6 +755,7 @@ txq_put_data_tso(struct net_device *dev, struct tx_queue *txq,
if (txq->tx_curr_desc == txq->tx_ring_size)
txq->tx_curr_desc = 0;
desc = &txq->tx_desc_area[tx_index];
+ txq->tx_desc_mapping[tx_index] = DESC_DMA_MAP_SINGLE;
desc->l4i_chk = 0;
desc->byte_cnt = length;
@@ -879,14 +885,13 @@ static void txq_submit_frag_skb(struct tx_queue *txq, struct sk_buff *skb)
skb_frag_t *this_frag;
int tx_index;
struct tx_desc *desc;
- void *addr;
this_frag = &skb_shinfo(skb)->frags[frag];
- addr = page_address(this_frag->page.p) + this_frag->page_offset;
tx_index = txq->tx_curr_desc++;
if (txq->tx_curr_desc == txq->tx_ring_size)
txq->tx_curr_desc = 0;
desc = &txq->tx_desc_area[tx_index];
+ txq->tx_desc_mapping[tx_index] = DESC_DMA_MAP_PAGE;
/*
* The last fragment will generate an interrupt
@@ -902,8 +907,9 @@ static void txq_submit_frag_skb(struct tx_queue *txq, struct sk_buff *skb)
desc->l4i_chk = 0;
desc->byte_cnt = skb_frag_size(this_frag);
- desc->buf_ptr = dma_map_single(mp->dev->dev.parent, addr,
- desc->byte_cnt, DMA_TO_DEVICE);
+ desc->buf_ptr = skb_frag_dma_map(mp->dev->dev.parent,
+ this_frag, 0, desc->byte_cnt,
+ DMA_TO_DEVICE);
}
}
@@ -936,6 +942,7 @@ static int txq_submit_skb(struct tx_queue *txq, struct sk_buff *skb,
if (txq->tx_curr_desc == txq->tx_ring_size)
txq->tx_curr_desc = 0;
desc = &txq->tx_desc_area[tx_index];
+ txq->tx_desc_mapping[tx_index] = DESC_DMA_MAP_SINGLE;
if (nr_frags) {
txq_submit_frag_skb(txq, skb);
@@ -1047,9 +1054,12 @@ static int txq_reclaim(struct tx_queue *txq, int budget, int force)
int tx_index;
struct tx_desc *desc;
u32 cmd_sts;
+ char desc_dma_map;
tx_index = txq->tx_used_desc;
desc = &txq->tx_desc_area[tx_index];
+ desc_dma_map = txq->tx_desc_mapping[tx_index];
+
cmd_sts = desc->cmd_sts;
if (cmd_sts & BUFFER_OWNED_BY_DMA) {
@@ -1065,9 +1075,19 @@ static int txq_reclaim(struct tx_queue *txq, int budget, int force)
reclaimed++;
txq->tx_desc_count--;
- if (!IS_TSO_HEADER(txq, desc->buf_ptr))
- dma_unmap_single(mp->dev->dev.parent, desc->buf_ptr,
- desc->byte_cnt, DMA_TO_DEVICE);
+ if (!IS_TSO_HEADER(txq, desc->buf_ptr)) {
+
+ if (desc_dma_map == DESC_DMA_MAP_PAGE)
+ dma_unmap_page(mp->dev->dev.parent,
+ desc->buf_ptr,
+ desc->byte_cnt,
+ DMA_TO_DEVICE);
+ else
+ dma_unmap_single(mp->dev->dev.parent,
+ desc->buf_ptr,
+ desc->byte_cnt,
+ DMA_TO_DEVICE);
+ }
if (cmd_sts & TX_ENABLE_INTERRUPT) {
struct sk_buff *skb = __skb_dequeue(&txq->tx_skb);
@@ -1996,6 +2016,7 @@ static int txq_init(struct mv643xx_eth_private *mp, int index)
struct tx_queue *txq = mp->txq + index;
struct tx_desc *tx_desc;
int size;
+ int ret;
int i;
txq->index = index;
@@ -2048,18 +2069,34 @@ static int txq_init(struct mv643xx_eth_private *mp, int index)
nexti * sizeof(struct tx_desc);
}
+ txq->tx_desc_mapping = kcalloc(txq->tx_ring_size, sizeof(char),
+ GFP_KERNEL);
+ if (!txq->tx_desc_mapping) {
+ ret = -ENOMEM;
+ goto err_free_desc_area;
+ }
+
/* Allocate DMA buffers for TSO MAC/IP/TCP headers */
txq->tso_hdrs = dma_alloc_coherent(mp->dev->dev.parent,
txq->tx_ring_size * TSO_HEADER_SIZE,
&txq->tso_hdrs_dma, GFP_KERNEL);
if (txq->tso_hdrs == NULL) {
- dma_free_coherent(mp->dev->dev.parent, txq->tx_desc_area_size,
- txq->tx_desc_area, txq->tx_desc_dma);
- return -ENOMEM;
+ ret = -ENOMEM;
+ goto err_free_desc_mapping;
}
skb_queue_head_init(&txq->tx_skb);
return 0;
+
+err_free_desc_mapping:
+ kfree(txq->tx_desc_mapping);
+err_free_desc_area:
+ if (index == 0 && size <= mp->tx_desc_sram_size)
+ iounmap(txq->tx_desc_area);
+ else
+ dma_free_coherent(mp->dev->dev.parent, txq->tx_desc_area_size,
+ txq->tx_desc_area, txq->tx_desc_dma);
+ return ret;
}
static void txq_deinit(struct tx_queue *txq)
@@ -2077,6 +2114,8 @@ static void txq_deinit(struct tx_queue *txq)
else
dma_free_coherent(mp->dev->dev.parent, txq->tx_desc_area_size,
txq->tx_desc_area, txq->tx_desc_dma);
+ kfree(txq->tx_desc_mapping);
+
if (txq->tso_hdrs)
dma_free_coherent(mp->dev->dev.parent,
txq->tx_ring_size * TSO_HEADER_SIZE,
--
2.2.1
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH v3] net: mv643xx_eth: Fix highmem support in non-TSO egress path
2015-01-22 14:33 ` [PATCH v3] " Ezequiel Garcia
@ 2015-01-24 13:45 ` Ezequiel Garcia
2015-01-26 23:11 ` Russell King - ARM Linux
2015-01-27 0:15 ` David Miller
1 sibling, 1 reply; 7+ messages in thread
From: Ezequiel Garcia @ 2015-01-24 13:45 UTC (permalink / raw)
To: David Miller, Russell King; +Cc: netdev, B38611, fabio.estevam, David.Laight
Hi all,
On 01/22/2015 11:33 AM, Ezequiel Garcia wrote:
> Commit 69ad0dd7af22b61d9e0e68e56b6290121618b0fb
> Author: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> Date: Mon May 19 13:59:59 2014 -0300
>
> net: mv643xx_eth: Use dma_map_single() to map the skb fragments
>
> caused a nasty regression by removing the support for highmem skb
> fragments. By using page_address() to get the address of a fragment's
> page, we are assuming a lowmem page. However, such assumption is incorrect,
> as fragments can be in highmem pages, resulting in very nasty issues.
>
> This commit fixes this by using the skb_frag_dma_map() helper,
> which takes care of mapping the skb fragment properly. Additionally,
> the type of mapping is now tracked, so it can be unmapped using
> dma_unmap_page or dma_unmap_single when appropriate.
>
> This commit also fixes the error path in txq_init() to release the
> resources properly.
>
> Fixes: 69ad0dd7af22 ("net: mv643xx_eth: Use dma_map_single() to map the skb fragments")
> Reported-by: Russell King <rmk+kernel@arm.linux.org.uk>
Russell, when you have some time, please test this patch. I think it
solves the regression and it manages to unmap the descriptors properly.
Thanks!
--
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v3] net: mv643xx_eth: Fix highmem support in non-TSO egress path
2015-01-24 13:45 ` Ezequiel Garcia
@ 2015-01-26 23:11 ` Russell King - ARM Linux
0 siblings, 0 replies; 7+ messages in thread
From: Russell King - ARM Linux @ 2015-01-26 23:11 UTC (permalink / raw)
To: Ezequiel Garcia; +Cc: David Miller, netdev, B38611, fabio.estevam, David.Laight
On Sat, Jan 24, 2015 at 10:45:29AM -0300, Ezequiel Garcia wrote:
> Hi all,
>
> On 01/22/2015 11:33 AM, Ezequiel Garcia wrote:
> > Commit 69ad0dd7af22b61d9e0e68e56b6290121618b0fb
> > Author: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> > Date: Mon May 19 13:59:59 2014 -0300
> >
> > net: mv643xx_eth: Use dma_map_single() to map the skb fragments
> >
> > caused a nasty regression by removing the support for highmem skb
> > fragments. By using page_address() to get the address of a fragment's
> > page, we are assuming a lowmem page. However, such assumption is incorrect,
> > as fragments can be in highmem pages, resulting in very nasty issues.
> >
> > This commit fixes this by using the skb_frag_dma_map() helper,
> > which takes care of mapping the skb fragment properly. Additionally,
> > the type of mapping is now tracked, so it can be unmapped using
> > dma_unmap_page or dma_unmap_single when appropriate.
> >
> > This commit also fixes the error path in txq_init() to release the
> > resources properly.
> >
> > Fixes: 69ad0dd7af22 ("net: mv643xx_eth: Use dma_map_single() to map the skb fragments")
> > Reported-by: Russell King <rmk+kernel@arm.linux.org.uk>
>
> Russell, when you have some time, please test this patch. I think it
> solves the regression and it manages to unmap the descriptors properly.
Yes, this seems to work as well, thanks.
--
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v3] net: mv643xx_eth: Fix highmem support in non-TSO egress path
2015-01-22 14:33 ` [PATCH v3] " Ezequiel Garcia
2015-01-24 13:45 ` Ezequiel Garcia
@ 2015-01-27 0:15 ` David Miller
1 sibling, 0 replies; 7+ messages in thread
From: David Miller @ 2015-01-27 0:15 UTC (permalink / raw)
To: ezequiel.garcia; +Cc: linux, netdev, B38611, fabio.estevam, David.Laight
From: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Date: Thu, 22 Jan 2015 11:33:02 -0300
> Commit 69ad0dd7af22b61d9e0e68e56b6290121618b0fb
> Author: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> Date: Mon May 19 13:59:59 2014 -0300
>
> net: mv643xx_eth: Use dma_map_single() to map the skb fragments
>
> caused a nasty regression by removing the support for highmem skb
> fragments. By using page_address() to get the address of a fragment's
> page, we are assuming a lowmem page. However, such assumption is incorrect,
> as fragments can be in highmem pages, resulting in very nasty issues.
>
> This commit fixes this by using the skb_frag_dma_map() helper,
> which takes care of mapping the skb fragment properly. Additionally,
> the type of mapping is now tracked, so it can be unmapped using
> dma_unmap_page or dma_unmap_single when appropriate.
>
> This commit also fixes the error path in txq_init() to release the
> resources properly.
>
> Fixes: 69ad0dd7af22 ("net: mv643xx_eth: Use dma_map_single() to map the skb fragments")
> Reported-by: Russell King <rmk+kernel@arm.linux.org.uk>
> Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Applied, thanks.
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2015-01-27 0:15 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-01-22 12:14 [PATCH v2] net: mv643xx_eth: Fix highmem support in non-TSO egress path Ezequiel Garcia
2015-01-22 12:40 ` David Laight
2015-01-22 13:26 ` Ezequiel Garcia
2015-01-22 14:33 ` [PATCH v3] " Ezequiel Garcia
2015-01-24 13:45 ` Ezequiel Garcia
2015-01-26 23:11 ` Russell King - ARM Linux
2015-01-27 0:15 ` David Miller
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.