diff mbox

[v4] net: add Faraday FTMAC100 10/100 Ethernet driver

Message ID 1295596533-1748-1-git-send-email-ratbert.chuang@gmail.com
State Changes Requested, archived
Delegated to: David Miller
Headers show

Commit Message

Po-Yu Chuang Jan. 21, 2011, 7:55 a.m. UTC
From: Po-Yu Chuang <ratbert@faraday-tech.com>

FTMAC100 Ethernet Media Access Controller supports 10/100 Mbps and
MII.  This driver has been working on some ARM/NDS32 SoC's including
Faraday A320 and Andes AG101.

Signed-off-by: Po-Yu Chuang <ratbert@faraday-tech.com>
---
v2:
always use NAPI
do not use our own net_device_stats structure
don't set trans_start and last_rx
stats.rx_packets and stats.rx_bytes include dropped packets
add missed netif_napi_del()
initialize spinlocks in probe function
remove rx_lock and hw_lock
use netdev_[err/info/dbg] instead of dev_* ones
use netdev_alloc_skb_ip_align()
remove ftmac100_get_stats()
use is_valid_ether_addr() instead of is_zero_ether_addr()
add const to ftmac100_ethtool_ops and ftmac100_netdev_ops
use net_ratelimit() instead of printk_ratelimit()
no explicit inline
use %pM to print MAC address
add comment before wmb
use napi poll() to handle all interrupts

v3:
undo "stats.rx_packets and stats.rx_bytes include dropped packets"
ftmac100_mdio_read() returns 0 if error
fix comment typos
use pr_fmt and pr_info
define INT_MASK_ALL_ENABLED
define MACCR_ENABLE_ALL
do not count length error many times
use bool/true/false
use cpu_to_le32/le32_to_cpu to access descriptors
indent style fix

v4:
should not access skb after netif_receive_skb()
use resource_size()
better way to use cpu_to_le32/le32_to_cpu
use spin_lock() for tx_lock
combine all netdev_info() together in ftmac100_poll()

 drivers/net/Kconfig    |    9 +
 drivers/net/Makefile   |    1 +
 drivers/net/ftmac100.c | 1207 ++++++++++++++++++++++++++++++++++++++++++++++++
 drivers/net/ftmac100.h |  180 +++++++
 4 files changed, 1397 insertions(+), 0 deletions(-)
 create mode 100644 drivers/net/ftmac100.c
 create mode 100644 drivers/net/ftmac100.h

Comments

Eric Dumazet Jan. 21, 2011, 9:08 a.m. UTC | #1
Le vendredi 21 janvier 2011 à 15:55 +0800, Po-Yu Chuang a écrit :
> From: Po-Yu Chuang <ratbert@faraday-tech.com>
> 
> FTMAC100 Ethernet Media Access Controller supports 10/100 Mbps and
> MII.  This driver has been working on some ARM/NDS32 SoC's including
> Faraday A320 and Andes AG101.
> 
> Signed-off-by: Po-Yu Chuang <ratbert@faraday-tech.com>


> +
> +static bool ftmac100_tx_complete_packet(struct ftmac100 *priv)
> +{
...
> +
> +	dma_unmap_single(priv->dev, map, skb_headlen(skb), DMA_TO_DEVICE);
> +
> +	dev_kfree_skb_irq(skb);
> +
> +	ftmac100_txdes_reset(txdes);
> +
> +	ftmac100_tx_clean_pointer_advance(priv);
> +
> +	priv->tx_pending--;
> +	netif_wake_queue(netdev);
> +
> +	return true;
> +}
> +

Thanks to NAPI, you can free skb directly, not queuing it via
NET_TX_SOFTIRQ softirq, using dev_kfree_skb() instead of
dev_kfree_skb_irq()



--
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
=?ISO-8859-2?Q?Micha=B3_Miros=B3aw?= Jan. 21, 2011, 12:26 p.m. UTC | #2
2011/1/21 Po-Yu Chuang <ratbert.chuang@gmail.com>:
> From: Po-Yu Chuang <ratbert@faraday-tech.com>
>
> FTMAC100 Ethernet Media Access Controller supports 10/100 Mbps and
> MII.  This driver has been working on some ARM/NDS32 SoC's including
> Faraday A320 and Andes AG101.
>
> Signed-off-by: Po-Yu Chuang <ratbert@faraday-tech.com>
[...]
> +static void ftmac100_txdes_reset(struct ftmac100_txdes *txdes)
> +{
> +       /* clear all except end of ring bit */
> +       txdes->txdes0 = 0;
> +       txdes->txdes1 &= FTMAC100_TXDES1_EDOTR;
> +       txdes->txdes2 = 0;
> +       txdes->txdes3 = 0;
> +}

This also probably needs cpu_to_le32().

[...]
> +static void ftmac100_free_buffers(struct ftmac100 *priv)
> +{
> +       int i;
> +
> +       for (i = 0; i < RX_QUEUE_ENTRIES; i += 2) {
> +               struct ftmac100_rxdes *rxdes = &priv->descs->rxdes[i];
> +               dma_addr_t d = ftmac100_rxdes_get_dma_addr(rxdes);
> +               void *page = ftmac100_rxdes_get_va(rxdes);
> +
> +               if (d)
> +                       dma_unmap_single(priv->dev, d, PAGE_SIZE,
> +                                        DMA_FROM_DEVICE);
> +
> +               if (page != NULL)
> +                       free_page((unsigned long)page);
> +       }
> +
[...]

> +static int ftmac100_alloc_buffers(struct ftmac100 *priv)
> +{
> +       int i;
> +
> +       priv->descs = dma_alloc_coherent(priv->dev,
> +                                        sizeof(struct ftmac100_descs),
> +                                        &priv->descs_dma_addr,
> +                                        GFP_KERNEL | GFP_DMA);
> +       if (priv->descs == NULL)
> +               return -ENOMEM;
> +
> +       memset(priv->descs, 0, sizeof(struct ftmac100_descs));
> +
> +       /* initialize RX ring */
> +
> +       ftmac100_rxdes_set_end_of_ring(&priv->descs->rxdes[RX_QUEUE_ENTRIES - 1]);
> +
> +       for (i = 0; i < RX_QUEUE_ENTRIES; i += 2) {
> +               struct ftmac100_rxdes *rxdes = &priv->descs->rxdes[i];
> +               void *page;
> +               dma_addr_t d;
> +
> +               page = (void *)__get_free_page(GFP_KERNEL | GFP_DMA);
> +               if (page == NULL)
> +                       goto err;
> +
> +               d = dma_map_single(priv->dev, page, PAGE_SIZE, DMA_FROM_DEVICE);
> +               if (unlikely(dma_mapping_error(priv->dev, d))) {
> +                       free_page((unsigned long)page);
> +                       goto err;
> +               }
> +
> +               /*
> +                * The hardware enforces a sub-2K maximum packet size, so we
> +                * put two buffers on every hardware page.
> +                */
> +               ftmac100_rxdes_set_va(rxdes, page);
> +               ftmac100_rxdes_set_va(rxdes + 1, page + PAGE_SIZE / 2);
> +
> +               ftmac100_rxdes_set_dma_addr(rxdes, d);
> +               ftmac100_rxdes_set_dma_addr(rxdes + 1, d + PAGE_SIZE / 2);
> +
> +               ftmac100_rxdes_set_buffer_size(rxdes, RX_BUF_SIZE);
> +               ftmac100_rxdes_set_buffer_size(rxdes + 1, RX_BUF_SIZE);
> +
> +               ftmac100_rxdes_set_dma_own(rxdes);
> +               ftmac100_rxdes_set_dma_own(rxdes + 1);
> +       }
[...]

Did you test this? This looks like it will result in double free after
packet RX, as you are giving the same page (referenced once) to two
distinct RX descriptors, that may be assigned different packets.

Since your not implementing any RX offloads, you might just allocate
fresh skb's with alloc_skb() and store skb pointer in rxdes3. Since
hardware doesn't touch it, you can skip cpu_to_le32()/le32_to_cpu()
there (leave a comment, though).

Unless this needs to work for ISA devices, you should drop GFP_DMA
allocation flag.

Best Regards,
Michał Mirosław
--
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
Po-Yu Chuang Jan. 24, 2011, 8:07 a.m. UTC | #3
Dear Eric,

On Fri, Jan 21, 2011 at 5:08 PM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
> Le vendredi 21 janvier 2011 à 15:55 +0800, Po-Yu Chuang a écrit :
> ...
>> +
>> +     dma_unmap_single(priv->dev, map, skb_headlen(skb), DMA_TO_DEVICE);
>> +
>> +     dev_kfree_skb_irq(skb);
>> +
>> +     ftmac100_txdes_reset(txdes);
>> +
>> +     ftmac100_tx_clean_pointer_advance(priv);
>> +
>> +     priv->tx_pending--;
>> +     netif_wake_queue(netdev);
>> +
>> +     return true;
>> +}
>> +
>
> Thanks to NAPI, you can free skb directly, not queuing it via
> NET_TX_SOFTIRQ softirq, using dev_kfree_skb() instead of
> dev_kfree_skb_irq()

Fixed. Thanks a lot.

best regards,
Po-Yu Chuang
--
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
Po-Yu Chuang Jan. 24, 2011, 8:26 a.m. UTC | #4
Dear Michał,

2011/1/21 Michał Mirosław <mirqus@gmail.com>:
> 2011/1/21 Po-Yu Chuang <ratbert.chuang@gmail.com>:
>> From: Po-Yu Chuang <ratbert@faraday-tech.com>
>>
>> FTMAC100 Ethernet Media Access Controller supports 10/100 Mbps and
>> MII.  This driver has been working on some ARM/NDS32 SoC's including
>> Faraday A320 and Andes AG101.
>>
>> Signed-off-by: Po-Yu Chuang <ratbert@faraday-tech.com>
> [...]
>> +static void ftmac100_txdes_reset(struct ftmac100_txdes *txdes)
>> +{
>> +       /* clear all except end of ring bit */
>> +       txdes->txdes0 = 0;
>> +       txdes->txdes1 &= FTMAC100_TXDES1_EDOTR;
>> +       txdes->txdes2 = 0;
>> +       txdes->txdes3 = 0;
>> +}
>
> This also probably needs cpu_to_le32().

Ah, I missed that. Fixed.

> [...]
>> +static void ftmac100_free_buffers(struct ftmac100 *priv)
>> +{
>> +       int i;
>> +
>> +       for (i = 0; i < RX_QUEUE_ENTRIES; i += 2) {
>> +               struct ftmac100_rxdes *rxdes = &priv->descs->rxdes[i];
>> +               dma_addr_t d = ftmac100_rxdes_get_dma_addr(rxdes);
>> +               void *page = ftmac100_rxdes_get_va(rxdes);
>> +
>> +               if (d)
>> +                       dma_unmap_single(priv->dev, d, PAGE_SIZE,
>> +                                        DMA_FROM_DEVICE);
>> +
>> +               if (page != NULL)
>> +                       free_page((unsigned long)page);
>> +       }
>> +
> [...]
>
>> +static int ftmac100_alloc_buffers(struct ftmac100 *priv)
>> +{
>> +       int i;
>> +
>> +       priv->descs = dma_alloc_coherent(priv->dev,
>> +                                        sizeof(struct ftmac100_descs),
>> +                                        &priv->descs_dma_addr,
>> +                                        GFP_KERNEL | GFP_DMA);
>> +       if (priv->descs == NULL)
>> +               return -ENOMEM;
>> +
>> +       memset(priv->descs, 0, sizeof(struct ftmac100_descs));
>> +
>> +       /* initialize RX ring */
>> +
>> +       ftmac100_rxdes_set_end_of_ring(&priv->descs->rxdes[RX_QUEUE_ENTRIES - 1]);
>> +
>> +       for (i = 0; i < RX_QUEUE_ENTRIES; i += 2) {
>> +               struct ftmac100_rxdes *rxdes = &priv->descs->rxdes[i];
>> +               void *page;
>> +               dma_addr_t d;
>> +
>> +               page = (void *)__get_free_page(GFP_KERNEL | GFP_DMA);
>> +               if (page == NULL)
>> +                       goto err;
>> +
>> +               d = dma_map_single(priv->dev, page, PAGE_SIZE, DMA_FROM_DEVICE);
>> +               if (unlikely(dma_mapping_error(priv->dev, d))) {
>> +                       free_page((unsigned long)page);
>> +                       goto err;
>> +               }
>> +
>> +               /*
>> +                * The hardware enforces a sub-2K maximum packet size, so we
>> +                * put two buffers on every hardware page.
>> +                */
>> +               ftmac100_rxdes_set_va(rxdes, page);
>> +               ftmac100_rxdes_set_va(rxdes + 1, page + PAGE_SIZE / 2);
>> +
>> +               ftmac100_rxdes_set_dma_addr(rxdes, d);
>> +               ftmac100_rxdes_set_dma_addr(rxdes + 1, d + PAGE_SIZE / 2);
>> +
>> +               ftmac100_rxdes_set_buffer_size(rxdes, RX_BUF_SIZE);
>> +               ftmac100_rxdes_set_buffer_size(rxdes + 1, RX_BUF_SIZE);
>> +
>> +               ftmac100_rxdes_set_dma_own(rxdes);
>> +               ftmac100_rxdes_set_dma_own(rxdes + 1);
>> +       }
> [...]
>
> Did you test this? This looks like it will result in double free after
> packet RX, as you are giving the same page (referenced once) to two
> distinct RX descriptors, that may be assigned different packets.

Yes, this is tested.

> Since your not implementing any RX offloads, you might just allocate
> fresh skb's with alloc_skb() and store skb pointer in rxdes3. Since

rxdes3 does not store virtual address of an skb.
It stores the address of the buffer allocated while open() and freed
only when stop().
The data in that buffer will be memcpy()ed to an skb allocated in
ftmac100_rx_packet().
No double free happens.

> hardware doesn't touch it, you can skip cpu_to_le32()/le32_to_cpu()
> there (leave a comment, though).

Agree. Thanks.

> Unless this needs to work for ISA devices, you should drop GFP_DMA
> allocation flag.

Ben mentioned about this in the previous mail. I thought that it is OK
to keep GFP_DMA
on ARM platform, but since you point out this flag is for ISA, I will drop it.

best regards,
Po-Yu Chuang
--
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
=?ISO-8859-2?Q?Micha=B3_Miros=B3aw?= Jan. 24, 2011, 8:22 p.m. UTC | #5
W dniu 24 stycznia 2011 09:26 użytkownik Po-Yu Chuang
<ratbert.chuang@gmail.com> napisał:
> 2011/1/21 Michał Mirosław <mirqus@gmail.com>:
>> 2011/1/21 Po-Yu Chuang <ratbert.chuang@gmail.com>:
>>> +static void ftmac100_free_buffers(struct ftmac100 *priv)
>>> +{
>>> +       int i;
>>> +
>>> +       for (i = 0; i < RX_QUEUE_ENTRIES; i += 2) {
>>> +               struct ftmac100_rxdes *rxdes = &priv->descs->rxdes[i];
>>> +               dma_addr_t d = ftmac100_rxdes_get_dma_addr(rxdes);
>>> +               void *page = ftmac100_rxdes_get_va(rxdes);
>>> +
>>> +               if (d)
>>> +                       dma_unmap_single(priv->dev, d, PAGE_SIZE,
>>> +                                        DMA_FROM_DEVICE);
>>> +
>>> +               if (page != NULL)
>>> +                       free_page((unsigned long)page);
>>> +       }
>>> +
>> [...]
>>
>>> +static int ftmac100_alloc_buffers(struct ftmac100 *priv)
>>> +{
>>> +       int i;
>>> +
>>> +       priv->descs = dma_alloc_coherent(priv->dev,
>>> +                                        sizeof(struct ftmac100_descs),
>>> +                                        &priv->descs_dma_addr,
>>> +                                        GFP_KERNEL | GFP_DMA);
>>> +       if (priv->descs == NULL)
>>> +               return -ENOMEM;
>>> +
>>> +       memset(priv->descs, 0, sizeof(struct ftmac100_descs));
>>> +
>>> +       /* initialize RX ring */
>>> +
>>> +       ftmac100_rxdes_set_end_of_ring(&priv->descs->rxdes[RX_QUEUE_ENTRIES - 1]);
>>> +
>>> +       for (i = 0; i < RX_QUEUE_ENTRIES; i += 2) {
>>> +               struct ftmac100_rxdes *rxdes = &priv->descs->rxdes[i];
>>> +               void *page;
>>> +               dma_addr_t d;
>>> +
>>> +               page = (void *)__get_free_page(GFP_KERNEL | GFP_DMA);
>>> +               if (page == NULL)
>>> +                       goto err;
>>> +
>>> +               d = dma_map_single(priv->dev, page, PAGE_SIZE, DMA_FROM_DEVICE);
>>> +               if (unlikely(dma_mapping_error(priv->dev, d))) {
>>> +                       free_page((unsigned long)page);
>>> +                       goto err;
>>> +               }
>>> +
>>> +               /*
>>> +                * The hardware enforces a sub-2K maximum packet size, so we
>>> +                * put two buffers on every hardware page.
>>> +                */
>>> +               ftmac100_rxdes_set_va(rxdes, page);
>>> +               ftmac100_rxdes_set_va(rxdes + 1, page + PAGE_SIZE / 2);
>>> +
>>> +               ftmac100_rxdes_set_dma_addr(rxdes, d);
>>> +               ftmac100_rxdes_set_dma_addr(rxdes + 1, d + PAGE_SIZE / 2);
>>> +
>>> +               ftmac100_rxdes_set_buffer_size(rxdes, RX_BUF_SIZE);
>>> +               ftmac100_rxdes_set_buffer_size(rxdes + 1, RX_BUF_SIZE);
>>> +
>>> +               ftmac100_rxdes_set_dma_own(rxdes);
>>> +               ftmac100_rxdes_set_dma_own(rxdes + 1);
>>> +       }
>> [...]
>>
>> Did you test this? This looks like it will result in double free after
>> packet RX, as you are giving the same page (referenced once) to two
>> distinct RX descriptors, that may be assigned different packets.
>
> Yes, this is tested.
>
>> Since your not implementing any RX offloads, you might just allocate
>> fresh skb's with alloc_skb() and store skb pointer in rxdes3. Since
>
> rxdes3 does not store virtual address of an skb.
> It stores the address of the buffer allocated while open() and freed
> only when stop().
> The data in that buffer will be memcpy()ed to an skb allocated in
> ftmac100_rx_packet().
> No double free happens.

Ah, I blindly assumed that you're just appending the buffers to the
skb (using skb_fill_page_desc() and friends). Since you have to mark
descriptors for the device anyway, it might be faster to allocate new
skbs and map those as rx buffers (changing the descriptor's buffer
address after every RX) instead of keeping static buffer and copying
every time. (For small packets it wastes lot of memory, though - so
the right choice depends on the expected workload.)

Best Regards,
Michał Mirosław
--
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
Po-Yu Chuang Jan. 25, 2011, 2:46 a.m. UTC | #6
Dear Michał,

2011/1/25 Michał Mirosław <mirqus@gmail.com>:
> W dniu 24 stycznia 2011 09:26 użytkownik Po-Yu Chuang
> <ratbert.chuang@gmail.com> napisał:
>> 2011/1/21 Michał Mirosław <mirqus@gmail.com>:
>>> Did you test this? This looks like it will result in double free after
>>> packet RX, as you are giving the same page (referenced once) to two
>>> distinct RX descriptors, that may be assigned different packets.
>>
>> Yes, this is tested.
>>
>>> Since your not implementing any RX offloads, you might just allocate
>>> fresh skb's with alloc_skb() and store skb pointer in rxdes3. Since
>>
>> rxdes3 does not store virtual address of an skb.
>> It stores the address of the buffer allocated while open() and freed
>> only when stop().
>> The data in that buffer will be memcpy()ed to an skb allocated in
>> ftmac100_rx_packet().
>> No double free happens.
>
> Ah, I blindly assumed that you're just appending the buffers to the
> skb (using skb_fill_page_desc() and friends). Since you have to mark
> descriptors for the device anyway, it might be faster to allocate new
> skbs and map those as rx buffers (changing the descriptor's buffer
> address after every RX) instead of keeping static buffer and copying
> every time. (For small packets it wastes lot of memory, though - so
> the right choice depends on the expected workload.)

The reason I chose to use memcpy rx buffer to skb is that I didn't know
how to deal with multi-segment packets if I preallocated skb for each
rx descriptor. This function seems to be what I need.
Let me rework this. Thanks.

best regards,
Po-Yu Chuang
--
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
Po-Yu Chuang Feb. 1, 2011, 3:56 a.m. UTC | #7
Dear Michał

2011/1/25 Po-Yu Chuang <ratbert.chuang@gmail.com>:
> 2011/1/25 Michał Mirosław <mirqus@gmail.com>:
>>
>> Ah, I blindly assumed that you're just appending the buffers to the
>> skb (using skb_fill_page_desc() and friends). Since you have to mark
>> descriptors for the device anyway, it might be faster to allocate new
>> skbs and map those as rx buffers (changing the descriptor's buffer
>> address after every RX) instead of keeping static buffer and copying
>> every time. (For small packets it wastes lot of memory, though - so
>> the right choice depends on the expected workload.)
>
> The reason I chose to use memcpy rx buffer to skb is that I didn't know
> how to deal with multi-segment packets if I preallocated skb for each
> rx descriptor. This function seems to be what I need.
> Let me rework this. Thanks.

After looking at some drivers using skb_fill_page_desc(), I am still
confused.

It seems that this function is mainly for jumbo packet and the
beginning of data of received packet should still be in skb->data.

Scheme 1)
If I simply allocate a page for each rx ring entry, I still need to allocate
an skb and copy at least packet header in first page to skb->data. Then
add the page of rest of payload to skb by skb_fill_page_desc().

Scheme 2)
If I simply allocate an skb for each rx ring entry, zero copy can be easily
achieved if the packet contains only one segment (one rx ring entry).
However, if the received packet has more than one segment, I guess I
might need to allocate a bigger skb and copy data of all segments
(skb->data) to the new skb?


The problem is, in most cases, the received packets are one-segment only.
(TBH, I don't know when will multi-segment packets appear.)

If all packets are one-segment, I can just allocate an skb for each rx ring
enry and achieve zero-copy. However, since there might be multi-segment
packets, I need to deal with them.

How can I do?

best regards,
Po-Yu Chuang
--
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 Feb. 1, 2011, 4:35 a.m. UTC | #8
From: Po-Yu Chuang <ratbert.chuang@gmail.com>
Date: Tue, 1 Feb 2011 11:56:16 +0800

> If I simply allocate a page for each rx ring entry, I still need to allocate
> an skb and copy at least packet header in first page to skb->data. Then
> add the page of rest of payload to skb by skb_fill_page_desc().

You should attach the pages, the use __pskb_pull_tail() to bring in the
headers to the linear skb->data area.

See drivers/net/niu.c:niu_process_rx_pkt().
--
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
Po-Yu Chuang Feb. 24, 2011, 7:27 a.m. UTC | #9
Hi David,

On Tue, Feb 1, 2011 at 12:35 PM, David Miller <davem@davemloft.net> wrote:
> From: Po-Yu Chuang <ratbert.chuang@gmail.com>
> Date: Tue, 1 Feb 2011 11:56:16 +0800
>
>> If I simply allocate a page for each rx ring entry, I still need to allocate
>> an skb and copy at least packet header in first page to skb->data. Then
>> add the page of rest of payload to skb by skb_fill_page_desc().
>
> You should attach the pages, the use __pskb_pull_tail() to bring in the
> headers to the linear skb->data area.
>
> See drivers/net/niu.c:niu_process_rx_pkt().

I tried two ways to implement zero-copy.
One is to preallocate skb big enough for any rx packet and use the skb
as rx buffer.
The other is use page as rx buffer, use skb_fill_page_desc() to add a
data page to
skb and then pull only header to skb by __pskb_pull_tail() as you suggested.

Two implementations are slower than the original memcpy version.
(benchmarked with iperf)

I guess the problem is because a HW restriction that the rx buffer must be
64 bits aligned. Since I cannot make rx buffer starts at offset 2 bytes, the
IP header, TCP header and data are not 4 bytes aligned. The performance
drops drastically.

Therefore, I will submit later a v6 which is still using memcpy().

best regards,
Po-Yu Chuang
--
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 Feb. 24, 2011, 7:51 a.m. UTC | #10
From: Po-Yu Chuang <ratbert.chuang@gmail.com>
Date: Thu, 24 Feb 2011 15:27:55 +0800

> I guess the problem is because a HW restriction that the rx buffer must be
> 64 bits aligned. Since I cannot make rx buffer starts at offset 2 bytes, the
> IP header, TCP header and data are not 4 bytes aligned. The performance
> drops drastically.

I cannot believe that after 20 years of commodity ethernet networking
chips were first designed, people are still designing hardware that
doesn't do this right.

Just emit garbage bytes into the sub-word alignment padding if the chip
wants to word align it's DMA writes.

Even the 15 year old Dec Tulip chips do this properly.

--
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
Po-Yu Chuang Feb. 24, 2011, 8:07 a.m. UTC | #11
Hi David,

On Thu, Feb 24, 2011 at 3:51 PM, David Miller <davem@davemloft.net> wrote:
> From: Po-Yu Chuang <ratbert.chuang@gmail.com>
> Date: Thu, 24 Feb 2011 15:27:55 +0800
>
>> I guess the problem is because a HW restriction that the rx buffer must be
>> 64 bits aligned. Since I cannot make rx buffer starts at offset 2 bytes, the
>> IP header, TCP header and data are not 4 bytes aligned. The performance
>> drops drastically.
>
> I cannot believe that after 20 years of commodity ethernet networking
> chips were first designed, people are still designing hardware that
> doesn't do this right.

Ha ha...
Well, this restriction was removed in the later IPs of our company. :-p

>
> Just emit garbage bytes into the sub-word alignment padding if the chip
> wants to word align it's DMA writes.

Not sure what do you mean. The problem is that HW does not accept a
base address of RX buffer which is not 8 bytes aligned.

> Even the 15 year old Dec Tulip chips do this properly.

best regards,
Po-Yu Chuang
--
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
Eric Dumazet Feb. 24, 2011, 8:22 a.m. UTC | #12
Le jeudi 24 février 2011 à 16:07 +0800, Po-Yu Chuang a écrit :
> Hi David,
> 
> On Thu, Feb 24, 2011 at 3:51 PM, David Miller <davem@davemloft.net> wrote:
> > From: Po-Yu Chuang <ratbert.chuang@gmail.com>
> > Date: Thu, 24 Feb 2011 15:27:55 +0800
> >
> >> I guess the problem is because a HW restriction that the rx buffer must be
> >> 64 bits aligned. Since I cannot make rx buffer starts at offset 2 bytes, the
> >> IP header, TCP header and data are not 4 bytes aligned. The performance
> >> drops drastically.
> >
> > I cannot believe that after 20 years of commodity ethernet networking
> > chips were first designed, people are still designing hardware that
> > doesn't do this right.
> 
> Ha ha...
> Well, this restriction was removed in the later IPs of our company. :-p
> 
> >
> > Just emit garbage bytes into the sub-word alignment padding if the chip
> > wants to word align it's DMA writes.
> 
> Not sure what do you mean. The problem is that HW does not accept a
> base address of RX buffer which is not 8 bytes aligned.

I still dont understand the problem, maybe you should post your work as
RFC

Why not using for the first part the skb buffer (eventually copy the
first 128 bytes to get aligned IP/TCP header), and add frags for
following parts ?



--
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 Feb. 24, 2011, 6:43 p.m. UTC | #13
From: Po-Yu Chuang <ratbert.chuang@gmail.com>
Date: Thu, 24 Feb 2011 16:07:48 +0800

> On Thu, Feb 24, 2011 at 3:51 PM, David Miller <davem@davemloft.net> wrote:
>> Just emit garbage bytes into the sub-word alignment padding if the chip
>> wants to word align it's DMA writes.
> 
> Not sure what do you mean. The problem is that HW does not accept a
> base address of RX buffer which is not 8 bytes aligned.

I am saying this is what hardware should do if it has such a
restriction.
--
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

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 4f1755b..26da0ee 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -2014,6 +2014,15 @@  config BCM63XX_ENET
 	  This driver supports the ethernet MACs in the Broadcom 63xx
 	  MIPS chipset family (BCM63XX).
 
+config FTMAC100
+	tristate "Faraday FTMAC100 10/100 Ethernet support"
+	depends on ARM
+	select MII
+	help
+	  This driver supports the FTMAC100 Ethernet controller from
+	  Faraday. It is used on Faraday A320, Andes AG101, AG101P
+	  and some other ARM/NDS32 SoC's.
+
 source "drivers/net/fs_enet/Kconfig"
 
 source "drivers/net/octeon/Kconfig"
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index b90738d..7c21711 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -147,6 +147,7 @@  obj-$(CONFIG_FORCEDETH) += forcedeth.o
 obj-$(CONFIG_NE_H8300) += ne-h8300.o 8390.o
 obj-$(CONFIG_AX88796) += ax88796.o
 obj-$(CONFIG_BCM63XX_ENET) += bcm63xx_enet.o
+obj-$(CONFIG_FTMAC100) += ftmac100.o
 
 obj-$(CONFIG_TSI108_ETH) += tsi108_eth.o
 obj-$(CONFIG_MV643XX_ETH) += mv643xx_eth.o
diff --git a/drivers/net/ftmac100.c b/drivers/net/ftmac100.c
new file mode 100644
index 0000000..58b2d5f
--- /dev/null
+++ b/drivers/net/ftmac100.c
@@ -0,0 +1,1207 @@ 
+/*
+ * Faraday FTMAC100 10/100 Ethernet
+ *
+ * (C) Copyright 2009-2011 Faraday Technology
+ * Po-Yu Chuang <ratbert@faraday-tech.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#define pr_fmt(fmt)	KBUILD_MODNAME ": " fmt
+
+#include <linux/dma-mapping.h>
+#include <linux/etherdevice.h>
+#include <linux/ethtool.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/kernel.h>
+#include <linux/mii.h>
+#include <linux/module.h>
+#include <linux/moduleparam.h>
+#include <linux/netdevice.h>
+#include <linux/platform_device.h>
+
+#include "ftmac100.h"
+
+#define DRV_NAME	"ftmac100"
+#define DRV_VERSION	"0.2"
+
+#define RX_QUEUE_ENTRIES	128	/* must be power of 2 */
+#define TX_QUEUE_ENTRIES	16	/* must be power of 2 */
+
+#define MAX_PKT_SIZE		1518
+#define RX_BUF_SIZE		2044	/* must be smaller than 0x7ff */
+
+/******************************************************************************
+ * private data
+ *****************************************************************************/
+struct ftmac100_descs {
+	struct ftmac100_rxdes	rxdes[RX_QUEUE_ENTRIES];
+	struct ftmac100_txdes	txdes[TX_QUEUE_ENTRIES];
+};
+
+struct ftmac100 {
+	struct resource		*res;
+	void			*base;
+	int			irq;
+
+	struct ftmac100_descs	*descs;
+	dma_addr_t		descs_dma_addr;
+
+	unsigned int		rx_pointer;
+	unsigned int		tx_clean_pointer;
+	unsigned int		tx_pointer;
+	unsigned int		tx_pending;
+
+	spinlock_t		tx_lock;
+
+	struct net_device	*netdev;
+	struct device		*dev;
+	struct napi_struct	napi;
+
+	struct mii_if_info	mii;
+};
+
+/******************************************************************************
+ * internal functions (hardware register access)
+ *****************************************************************************/
+#define INT_MASK_ALL_ENABLED	(FTMAC100_INT_RPKT_FINISH	| \
+				 FTMAC100_INT_NORXBUF		| \
+				 FTMAC100_INT_XPKT_OK		| \
+				 FTMAC100_INT_XPKT_LOST		| \
+				 FTMAC100_INT_RPKT_LOST		| \
+				 FTMAC100_INT_AHB_ERR		| \
+				 FTMAC100_INT_PHYSTS_CHG)
+
+static void ftmac100_enable_all_int(struct ftmac100 *priv)
+{
+	iowrite32(INT_MASK_ALL_ENABLED, priv->base + FTMAC100_OFFSET_IMR);
+}
+
+static void ftmac100_disable_all_int(struct ftmac100 *priv)
+{
+	iowrite32(0, priv->base + FTMAC100_OFFSET_IMR);
+}
+
+static void ftmac100_set_rx_ring_base(struct ftmac100 *priv, dma_addr_t addr)
+{
+	iowrite32(addr, priv->base + FTMAC100_OFFSET_RXR_BADR);
+}
+
+static void ftmac100_set_tx_ring_base(struct ftmac100 *priv, dma_addr_t addr)
+{
+	iowrite32(addr, priv->base + FTMAC100_OFFSET_TXR_BADR);
+}
+
+static void ftmac100_txdma_start_polling(struct ftmac100 *priv)
+{
+	iowrite32(1, priv->base + FTMAC100_OFFSET_TXPD);
+}
+
+static int ftmac100_reset(struct ftmac100 *priv)
+{
+	struct net_device *netdev = priv->netdev;
+	int i;
+
+	/* NOTE: reset clears all registers */
+	iowrite32(FTMAC100_MACCR_SW_RST, priv->base + FTMAC100_OFFSET_MACCR);
+
+	for (i = 0; i < 5; i++) {
+		unsigned int maccr;
+
+		maccr = ioread32(priv->base + FTMAC100_OFFSET_MACCR);
+		if (!(maccr & FTMAC100_MACCR_SW_RST)) {
+			/*
+			 * FTMAC100_MACCR_SW_RST cleared does not indicate
+			 * that hardware reset completed (what the f*ck).
+			 * We still need to wait for a while.
+			 */
+			usleep_range(500, 1000);
+			return 0;
+		}
+
+		usleep_range(1000, 10000);
+	}
+
+	netdev_err(netdev, "software reset failed\n");
+	return -EIO;
+}
+
+static void ftmac100_set_mac(struct ftmac100 *priv, const unsigned char *mac)
+{
+	unsigned int maddr = mac[0] << 8 | mac[1];
+	unsigned int laddr = mac[2] << 24 | mac[3] << 16 | mac[4] << 8 | mac[5];
+
+	iowrite32(maddr, priv->base + FTMAC100_OFFSET_MAC_MADR);
+	iowrite32(laddr, priv->base + FTMAC100_OFFSET_MAC_LADR);
+}
+
+#define MACCR_ENABLE_ALL	(FTMAC100_MACCR_XMT_EN	| \
+				 FTMAC100_MACCR_RCV_EN	| \
+				 FTMAC100_MACCR_XDMA_EN	| \
+				 FTMAC100_MACCR_RDMA_EN	| \
+				 FTMAC100_MACCR_CRC_APD	| \
+				 FTMAC100_MACCR_FULLDUP	| \
+				 FTMAC100_MACCR_RX_RUNT	| \
+				 FTMAC100_MACCR_RX_BROADPKT)
+
+static int ftmac100_start_hw(struct ftmac100 *priv)
+{
+	struct net_device *netdev = priv->netdev;
+
+	if (ftmac100_reset(priv))
+		return -EIO;
+
+	/* setup ring buffer base registers */
+
+	ftmac100_set_rx_ring_base(priv,
+				  priv->descs_dma_addr +
+				  offsetof(struct ftmac100_descs, rxdes));
+	ftmac100_set_tx_ring_base(priv,
+				  priv->descs_dma_addr +
+				  offsetof(struct ftmac100_descs, txdes));
+
+	iowrite32(FTMAC100_APTC_RXPOLL_CNT(1), priv->base + FTMAC100_OFFSET_APTC);
+
+	ftmac100_set_mac(priv, netdev->dev_addr);
+
+	iowrite32(MACCR_ENABLE_ALL, priv->base + FTMAC100_OFFSET_MACCR);
+	return 0;
+}
+
+static void ftmac100_stop_hw(struct ftmac100 *priv)
+{
+	iowrite32(0, priv->base + FTMAC100_OFFSET_MACCR);
+}
+
+/******************************************************************************
+ * internal functions (receive descriptor)
+ *****************************************************************************/
+static bool ftmac100_rxdes_first_segment(struct ftmac100_rxdes *rxdes)
+{
+	return rxdes->rxdes0 & cpu_to_le32(FTMAC100_RXDES0_FRS);
+}
+
+static bool ftmac100_rxdes_last_segment(struct ftmac100_rxdes *rxdes)
+{
+	return rxdes->rxdes0 & cpu_to_le32(FTMAC100_RXDES0_LRS);
+}
+
+static bool ftmac100_rxdes_owned_by_dma(struct ftmac100_rxdes *rxdes)
+{
+	return rxdes->rxdes0 & cpu_to_le32(FTMAC100_RXDES0_RXDMA_OWN);
+}
+
+static void ftmac100_rxdes_set_dma_own(struct ftmac100_rxdes *rxdes)
+{
+	/* clear status bits */
+	rxdes->rxdes0 = cpu_to_le32(FTMAC100_RXDES0_RXDMA_OWN);
+}
+
+static bool ftmac100_rxdes_rx_error(struct ftmac100_rxdes *rxdes)
+{
+	return rxdes->rxdes0 & cpu_to_le32(FTMAC100_RXDES0_RX_ERR);
+}
+
+static bool ftmac100_rxdes_crc_error(struct ftmac100_rxdes *rxdes)
+{
+	return rxdes->rxdes0 & cpu_to_le32(FTMAC100_RXDES0_CRC_ERR);
+}
+
+static bool ftmac100_rxdes_frame_too_long(struct ftmac100_rxdes *rxdes)
+{
+	return rxdes->rxdes0 & cpu_to_le32(FTMAC100_RXDES0_FTL);
+}
+
+static bool ftmac100_rxdes_runt(struct ftmac100_rxdes *rxdes)
+{
+	return rxdes->rxdes0 & cpu_to_le32(FTMAC100_RXDES0_RUNT);
+}
+
+static bool ftmac100_rxdes_odd_nibble(struct ftmac100_rxdes *rxdes)
+{
+	return rxdes->rxdes0 & cpu_to_le32(FTMAC100_RXDES0_RX_ODD_NB);
+}
+
+static unsigned int ftmac100_rxdes_frame_length(struct ftmac100_rxdes *rxdes)
+{
+	return rxdes->rxdes0 & cpu_to_le32(FTMAC100_RXDES0_RFL);
+}
+
+static bool ftmac100_rxdes_multicast(struct ftmac100_rxdes *rxdes)
+{
+	return rxdes->rxdes0 & cpu_to_le32(FTMAC100_RXDES0_MULTICAST);
+}
+
+static void ftmac100_rxdes_set_buffer_size(struct ftmac100_rxdes *rxdes,
+					   unsigned int size)
+{
+	rxdes->rxdes1 &= cpu_to_le32(FTMAC100_RXDES1_EDORR);
+	rxdes->rxdes1 |= cpu_to_le32(FTMAC100_RXDES1_RXBUF_SIZE(size));
+}
+
+static void ftmac100_rxdes_set_end_of_ring(struct ftmac100_rxdes *rxdes)
+{
+	rxdes->rxdes1 |= cpu_to_le32(FTMAC100_RXDES1_EDORR);
+}
+
+static void ftmac100_rxdes_set_dma_addr(struct ftmac100_rxdes *rxdes,
+					dma_addr_t addr)
+{
+	rxdes->rxdes2 = cpu_to_le32(addr);
+}
+
+static dma_addr_t ftmac100_rxdes_get_dma_addr(struct ftmac100_rxdes *rxdes)
+{
+	return le32_to_cpu(rxdes->rxdes2);
+}
+
+/* rxdes3 is not used by hardware, we use it to keep track of buffer */
+static void ftmac100_rxdes_set_va(struct ftmac100_rxdes *rxdes, void *addr)
+{
+	rxdes->rxdes3 = cpu_to_le32(addr);
+}
+
+static void *ftmac100_rxdes_get_va(struct ftmac100_rxdes *rxdes)
+{
+	return (void *)le32_to_cpu(rxdes->rxdes3);
+}
+
+/******************************************************************************
+ * internal functions (receive)
+ *****************************************************************************/
+static int ftmac100_next_rx_pointer(int pointer)
+{
+	return (pointer + 1) & (RX_QUEUE_ENTRIES - 1);
+}
+
+static void ftmac100_rx_pointer_advance(struct ftmac100 *priv)
+{
+	priv->rx_pointer = ftmac100_next_rx_pointer(priv->rx_pointer);
+}
+
+static struct ftmac100_rxdes *ftmac100_current_rxdes(struct ftmac100 *priv)
+{
+	return &priv->descs->rxdes[priv->rx_pointer];
+}
+
+static struct ftmac100_rxdes *
+ftmac100_rx_locate_first_segment(struct ftmac100 *priv)
+{
+	struct ftmac100_rxdes *rxdes = ftmac100_current_rxdes(priv);
+
+	while (!ftmac100_rxdes_owned_by_dma(rxdes)) {
+		if (ftmac100_rxdes_first_segment(rxdes))
+			return rxdes;
+
+		ftmac100_rxdes_set_dma_own(rxdes);
+		ftmac100_rx_pointer_advance(priv);
+		rxdes = ftmac100_current_rxdes(priv);
+	}
+
+	return NULL;
+}
+
+static bool ftmac100_rx_packet_error(struct ftmac100 *priv,
+				     struct ftmac100_rxdes *rxdes)
+{
+	struct net_device *netdev = priv->netdev;
+	bool error = false;
+
+	if (unlikely(ftmac100_rxdes_rx_error(rxdes))) {
+		if (net_ratelimit())
+			netdev_info(netdev, "rx err\n");
+
+		netdev->stats.rx_errors++;
+		error = true;
+	}
+
+	if (unlikely(ftmac100_rxdes_crc_error(rxdes))) {
+		if (net_ratelimit())
+			netdev_info(netdev, "rx crc err\n");
+
+		netdev->stats.rx_crc_errors++;
+		error = true;
+	}
+
+	if (unlikely(ftmac100_rxdes_frame_too_long(rxdes))) {
+		if (net_ratelimit())
+			netdev_info(netdev, "rx frame too long\n");
+
+		netdev->stats.rx_length_errors++;
+		error = true;
+	} else if (unlikely(ftmac100_rxdes_runt(rxdes))) {
+		if (net_ratelimit())
+			netdev_info(netdev, "rx runt\n");
+
+		netdev->stats.rx_length_errors++;
+		error = true;
+	} else if (unlikely(ftmac100_rxdes_odd_nibble(rxdes))) {
+		if (net_ratelimit())
+			netdev_info(netdev, "rx odd nibble\n");
+
+		netdev->stats.rx_length_errors++;
+		error = true;
+	}
+
+	return error;
+}
+
+static void ftmac100_rx_drop_packet(struct ftmac100 *priv)
+{
+	struct net_device *netdev = priv->netdev;
+	struct ftmac100_rxdes *rxdes = ftmac100_current_rxdes(priv);
+	bool done = false;
+
+	if (net_ratelimit())
+		netdev_dbg(netdev, "drop packet %p\n", rxdes);
+
+	do {
+		if (ftmac100_rxdes_last_segment(rxdes))
+			done = true;
+
+		ftmac100_rxdes_set_dma_own(rxdes);
+		ftmac100_rx_pointer_advance(priv);
+		rxdes = ftmac100_current_rxdes(priv);
+	} while (!done && !ftmac100_rxdes_owned_by_dma(rxdes));
+
+	netdev->stats.rx_dropped++;
+}
+
+static bool ftmac100_rx_packet(struct ftmac100 *priv, int *processed)
+{
+	struct net_device *netdev = priv->netdev;
+	struct ftmac100_rxdes *rxdes;
+	struct sk_buff *skb;
+	int length;
+	bool copied = false;
+	bool done = false;
+
+	rxdes = ftmac100_rx_locate_first_segment(priv);
+	if (!rxdes)
+		return false;
+
+	if (unlikely(ftmac100_rx_packet_error(priv, rxdes))) {
+		ftmac100_rx_drop_packet(priv);
+		return true;
+	}
+
+	/* start processing */
+
+	length = ftmac100_rxdes_frame_length(rxdes);
+
+	skb = netdev_alloc_skb_ip_align(netdev, length);
+	if (unlikely(!skb)) {
+		if (net_ratelimit())
+			netdev_err(netdev, "rx skb alloc failed\n");
+
+		ftmac100_rx_drop_packet(priv);
+		return true;
+	}
+
+	if (unlikely(ftmac100_rxdes_multicast(rxdes)))
+		netdev->stats.multicast++;
+
+	do {
+		dma_addr_t d = ftmac100_rxdes_get_dma_addr(rxdes);
+		void *buf = ftmac100_rxdes_get_va(rxdes);
+		int size;
+
+		size = min(length - copied, RX_BUF_SIZE);
+
+		dma_sync_single_for_cpu(priv->dev, d, RX_BUF_SIZE,
+					DMA_FROM_DEVICE);
+		memcpy(skb_put(skb, size), buf, size);
+
+		copied += size;
+
+		if (ftmac100_rxdes_last_segment(rxdes))
+			done = true;
+
+		dma_sync_single_for_device(priv->dev, d, RX_BUF_SIZE,
+					   DMA_FROM_DEVICE);
+
+		ftmac100_rxdes_set_dma_own(rxdes);
+
+		ftmac100_rx_pointer_advance(priv);
+		rxdes = ftmac100_current_rxdes(priv);
+	} while (!done && copied < length);
+
+	skb->protocol = eth_type_trans(skb, netdev);
+
+	netdev->stats.rx_packets++;
+	netdev->stats.rx_bytes += skb->len;
+
+	/* push packet to protocol stack */
+	netif_receive_skb(skb);
+
+	(*processed)++;
+	return true;
+}
+
+/******************************************************************************
+ * internal functions (transmit descriptor)
+ *****************************************************************************/
+static void ftmac100_txdes_reset(struct ftmac100_txdes *txdes)
+{
+	/* clear all except end of ring bit */
+	txdes->txdes0 = 0;
+	txdes->txdes1 &= FTMAC100_TXDES1_EDOTR;
+	txdes->txdes2 = 0;
+	txdes->txdes3 = 0;
+}
+
+static bool ftmac100_txdes_owned_by_dma(struct ftmac100_txdes *txdes)
+{
+	return txdes->txdes0 & cpu_to_le32(FTMAC100_TXDES0_TXDMA_OWN);
+}
+
+static void ftmac100_txdes_set_dma_own(struct ftmac100_txdes *txdes)
+{
+	/*
+	 * Make sure dma own bit will not be set before any other
+	 * descriptor fields.
+	 */
+	wmb();
+	txdes->txdes0 |= cpu_to_le32(FTMAC100_TXDES0_TXDMA_OWN);
+}
+
+static bool ftmac100_txdes_excessive_collision(struct ftmac100_txdes *txdes)
+{
+	return txdes->txdes0 & cpu_to_le32(FTMAC100_TXDES0_TXPKT_EXSCOL);
+}
+
+static bool ftmac100_txdes_late_collision(struct ftmac100_txdes *txdes)
+{
+	return txdes->txdes0 & cpu_to_le32(FTMAC100_TXDES0_TXPKT_LATECOL);
+}
+
+static void ftmac100_txdes_set_end_of_ring(struct ftmac100_txdes *txdes)
+{
+	txdes->txdes1 |= cpu_to_le32(FTMAC100_TXDES1_EDOTR);
+}
+
+static void ftmac100_txdes_set_first_segment(struct ftmac100_txdes *txdes)
+{
+	txdes->txdes1 |= cpu_to_le32(FTMAC100_TXDES1_FTS);
+}
+
+static void ftmac100_txdes_set_last_segment(struct ftmac100_txdes *txdes)
+{
+	txdes->txdes1 |= cpu_to_le32(FTMAC100_TXDES1_LTS);
+}
+
+static void ftmac100_txdes_set_txint(struct ftmac100_txdes *txdes)
+{
+	txdes->txdes1 |= cpu_to_le32(FTMAC100_TXDES1_TXIC);
+}
+
+static void ftmac100_txdes_set_buffer_size(struct ftmac100_txdes *txdes,
+					   unsigned int len)
+{
+	txdes->txdes1 |= cpu_to_le32(FTMAC100_TXDES1_TXBUF_SIZE(len));
+}
+
+static void ftmac100_txdes_set_dma_addr(struct ftmac100_txdes *txdes,
+					dma_addr_t addr)
+{
+	txdes->txdes2 = cpu_to_le32(addr);
+}
+
+static dma_addr_t ftmac100_txdes_get_dma_addr(struct ftmac100_txdes *txdes)
+{
+	return le32_to_cpu(txdes->txdes2);
+}
+
+/* txdes3 is not used by hardware, we use it to keep track of socket buffer */
+static void ftmac100_txdes_set_skb(struct ftmac100_txdes *txdes,
+				   struct sk_buff *skb)
+{
+	txdes->txdes3 = cpu_to_le32(skb);
+}
+
+static struct sk_buff *ftmac100_txdes_get_skb(struct ftmac100_txdes *txdes)
+{
+	return (struct sk_buff *)le32_to_cpu(txdes->txdes3);
+}
+
+/******************************************************************************
+ * internal functions (transmit)
+ *****************************************************************************/
+static int ftmac100_next_tx_pointer(int pointer)
+{
+	return (pointer + 1) & (TX_QUEUE_ENTRIES - 1);
+}
+
+static void ftmac100_tx_pointer_advance(struct ftmac100 *priv)
+{
+	priv->tx_pointer = ftmac100_next_tx_pointer(priv->tx_pointer);
+}
+
+static void ftmac100_tx_clean_pointer_advance(struct ftmac100 *priv)
+{
+	priv->tx_clean_pointer = ftmac100_next_tx_pointer(priv->tx_clean_pointer);
+}
+
+static struct ftmac100_txdes *ftmac100_current_txdes(struct ftmac100 *priv)
+{
+	return &priv->descs->txdes[priv->tx_pointer];
+}
+
+static struct ftmac100_txdes *
+ftmac100_current_clean_txdes(struct ftmac100 *priv)
+{
+	return &priv->descs->txdes[priv->tx_clean_pointer];
+}
+
+static bool ftmac100_tx_complete_packet(struct ftmac100 *priv)
+{
+	struct net_device *netdev = priv->netdev;
+	struct ftmac100_txdes *txdes;
+	struct sk_buff *skb;
+	dma_addr_t map;
+
+	if (priv->tx_pending == 0)
+		return false;
+
+	txdes = ftmac100_current_clean_txdes(priv);
+
+	if (ftmac100_txdes_owned_by_dma(txdes))
+		return false;
+
+	skb = ftmac100_txdes_get_skb(txdes);
+	map = ftmac100_txdes_get_dma_addr(txdes);
+
+	if (unlikely(ftmac100_txdes_excessive_collision(txdes) ||
+		     ftmac100_txdes_late_collision(txdes))) {
+		/*
+		 * packet transmitted to ethernet lost due to late collision
+		 * or excessive collision
+		 */
+		netdev->stats.tx_aborted_errors++;
+	} else {
+		netdev->stats.tx_packets++;
+		netdev->stats.tx_bytes += skb->len;
+	}
+
+	dma_unmap_single(priv->dev, map, skb_headlen(skb), DMA_TO_DEVICE);
+
+	dev_kfree_skb_irq(skb);
+
+	ftmac100_txdes_reset(txdes);
+
+	ftmac100_tx_clean_pointer_advance(priv);
+
+	priv->tx_pending--;
+	netif_wake_queue(netdev);
+
+	return true;
+}
+
+static void ftmac100_tx_complete(struct ftmac100 *priv)
+{
+	spin_lock(&priv->tx_lock);
+	while (ftmac100_tx_complete_packet(priv))
+		;
+	spin_unlock(&priv->tx_lock);
+}
+
+static int ftmac100_xmit(struct ftmac100 *priv, struct sk_buff *skb,
+			 dma_addr_t map)
+{
+	struct net_device *netdev = priv->netdev;
+	struct ftmac100_txdes *txdes;
+	unsigned int len = (skb->len < ETH_ZLEN) ? ETH_ZLEN : skb->len;
+
+	txdes = ftmac100_current_txdes(priv);
+	ftmac100_tx_pointer_advance(priv);
+
+	/* setup TX descriptor */
+
+	spin_lock(&priv->tx_lock);
+	ftmac100_txdes_set_skb(txdes, skb);
+	ftmac100_txdes_set_dma_addr(txdes, map);
+
+	ftmac100_txdes_set_first_segment(txdes);
+	ftmac100_txdes_set_last_segment(txdes);
+	ftmac100_txdes_set_txint(txdes);
+	ftmac100_txdes_set_buffer_size(txdes, len);
+
+	priv->tx_pending++;
+	if (priv->tx_pending == TX_QUEUE_ENTRIES) {
+		if (net_ratelimit())
+			netdev_info(netdev, "tx queue full\n");
+
+		netif_stop_queue(netdev);
+	}
+
+	/* start transmit */
+	ftmac100_txdes_set_dma_own(txdes);
+	spin_unlock(&priv->tx_lock);
+
+	ftmac100_txdma_start_polling(priv);
+
+	return NETDEV_TX_OK;
+}
+
+/******************************************************************************
+ * internal functions (buffer)
+ *****************************************************************************/
+static void ftmac100_free_buffers(struct ftmac100 *priv)
+{
+	int i;
+
+	for (i = 0; i < RX_QUEUE_ENTRIES; i += 2) {
+		struct ftmac100_rxdes *rxdes = &priv->descs->rxdes[i];
+		dma_addr_t d = ftmac100_rxdes_get_dma_addr(rxdes);
+		void *page = ftmac100_rxdes_get_va(rxdes);
+
+		if (d)
+			dma_unmap_single(priv->dev, d, PAGE_SIZE,
+					 DMA_FROM_DEVICE);
+
+		if (page != NULL)
+			free_page((unsigned long)page);
+	}
+
+	for (i = 0; i < TX_QUEUE_ENTRIES; i++) {
+		struct ftmac100_txdes *txdes = &priv->descs->txdes[i];
+		struct sk_buff *skb = ftmac100_txdes_get_skb(txdes);
+
+		if (skb) {
+			dma_addr_t map;
+
+			map = ftmac100_txdes_get_dma_addr(txdes);
+			dma_unmap_single(priv->dev, map, skb_headlen(skb),
+					 DMA_TO_DEVICE);
+			dev_kfree_skb(skb);
+		}
+	}
+
+	dma_free_coherent(priv->dev, sizeof(struct ftmac100_descs),
+			  priv->descs, priv->descs_dma_addr);
+}
+
+static int ftmac100_alloc_buffers(struct ftmac100 *priv)
+{
+	int i;
+
+	priv->descs = dma_alloc_coherent(priv->dev,
+					 sizeof(struct ftmac100_descs),
+					 &priv->descs_dma_addr,
+					 GFP_KERNEL | GFP_DMA);
+	if (priv->descs == NULL)
+		return -ENOMEM;
+
+	memset(priv->descs, 0, sizeof(struct ftmac100_descs));
+
+	/* initialize RX ring */
+
+	ftmac100_rxdes_set_end_of_ring(&priv->descs->rxdes[RX_QUEUE_ENTRIES - 1]);
+
+	for (i = 0; i < RX_QUEUE_ENTRIES; i += 2) {
+		struct ftmac100_rxdes *rxdes = &priv->descs->rxdes[i];
+		void *page;
+		dma_addr_t d;
+
+		page = (void *)__get_free_page(GFP_KERNEL | GFP_DMA);
+		if (page == NULL)
+			goto err;
+
+		d = dma_map_single(priv->dev, page, PAGE_SIZE, DMA_FROM_DEVICE);
+		if (unlikely(dma_mapping_error(priv->dev, d))) {
+			free_page((unsigned long)page);
+			goto err;
+		}
+
+		/*
+		 * The hardware enforces a sub-2K maximum packet size, so we
+		 * put two buffers on every hardware page.
+		 */
+		ftmac100_rxdes_set_va(rxdes, page);
+		ftmac100_rxdes_set_va(rxdes + 1, page + PAGE_SIZE / 2);
+
+		ftmac100_rxdes_set_dma_addr(rxdes, d);
+		ftmac100_rxdes_set_dma_addr(rxdes + 1, d + PAGE_SIZE / 2);
+
+		ftmac100_rxdes_set_buffer_size(rxdes, RX_BUF_SIZE);
+		ftmac100_rxdes_set_buffer_size(rxdes + 1, RX_BUF_SIZE);
+
+		ftmac100_rxdes_set_dma_own(rxdes);
+		ftmac100_rxdes_set_dma_own(rxdes + 1);
+	}
+
+	/* initialize TX ring */
+
+	ftmac100_txdes_set_end_of_ring(&priv->descs->txdes[TX_QUEUE_ENTRIES - 1]);
+	return 0;
+
+err:
+	ftmac100_free_buffers(priv);
+	return -ENOMEM;
+}
+
+/******************************************************************************
+ * struct mii_if_info functions
+ *****************************************************************************/
+static int ftmac100_mdio_read(struct net_device *netdev, int phy_id, int reg)
+{
+	struct ftmac100 *priv = netdev_priv(netdev);
+	unsigned int phycr;
+	int i;
+
+	phycr = FTMAC100_PHYCR_PHYAD(phy_id) |
+		FTMAC100_PHYCR_REGAD(reg) |
+		FTMAC100_PHYCR_MIIRD;
+
+	iowrite32(phycr, priv->base + FTMAC100_OFFSET_PHYCR);
+	for (i = 0; i < 10; i++) {
+		phycr = ioread32(priv->base + FTMAC100_OFFSET_PHYCR);
+
+		if ((phycr & FTMAC100_PHYCR_MIIRD) == 0)
+			return phycr & FTMAC100_PHYCR_MIIRDATA;
+
+		usleep_range(100, 1000);
+	}
+
+	netdev_err(netdev, "mdio read timed out\n");
+	return 0;
+}
+
+static void ftmac100_mdio_write(struct net_device *netdev, int phy_id, int reg,
+				int data)
+{
+	struct ftmac100 *priv = netdev_priv(netdev);
+	unsigned int phycr;
+	int i;
+
+	phycr = FTMAC100_PHYCR_PHYAD(phy_id) |
+		FTMAC100_PHYCR_REGAD(reg) |
+		FTMAC100_PHYCR_MIIWR;
+
+	data = FTMAC100_PHYWDATA_MIIWDATA(data);
+
+	iowrite32(data, priv->base + FTMAC100_OFFSET_PHYWDATA);
+	iowrite32(phycr, priv->base + FTMAC100_OFFSET_PHYCR);
+
+	for (i = 0; i < 10; i++) {
+		phycr = ioread32(priv->base + FTMAC100_OFFSET_PHYCR);
+
+		if ((phycr & FTMAC100_PHYCR_MIIWR) == 0)
+			return;
+
+		usleep_range(100, 1000);
+	}
+
+	netdev_err(netdev, "mdio write timed out\n");
+}
+
+/******************************************************************************
+ * struct ethtool_ops functions
+ *****************************************************************************/
+static void ftmac100_get_drvinfo(struct net_device *netdev,
+				 struct ethtool_drvinfo *info)
+{
+	strcpy(info->driver, DRV_NAME);
+	strcpy(info->version, DRV_VERSION);
+	strcpy(info->bus_info, dev_name(&netdev->dev));
+}
+
+static int ftmac100_get_settings(struct net_device *netdev,
+				 struct ethtool_cmd *cmd)
+{
+	struct ftmac100 *priv = netdev_priv(netdev);
+	return mii_ethtool_gset(&priv->mii, cmd);
+}
+
+static int ftmac100_set_settings(struct net_device *netdev,
+				 struct ethtool_cmd *cmd)
+{
+	struct ftmac100 *priv = netdev_priv(netdev);
+	return mii_ethtool_sset(&priv->mii, cmd);
+}
+
+static int ftmac100_nway_reset(struct net_device *netdev)
+{
+	struct ftmac100 *priv = netdev_priv(netdev);
+	return mii_nway_restart(&priv->mii);
+}
+
+static u32 ftmac100_get_link(struct net_device *netdev)
+{
+	struct ftmac100 *priv = netdev_priv(netdev);
+	return mii_link_ok(&priv->mii);
+}
+
+static const struct ethtool_ops ftmac100_ethtool_ops = {
+	.set_settings		= ftmac100_set_settings,
+	.get_settings		= ftmac100_get_settings,
+	.get_drvinfo		= ftmac100_get_drvinfo,
+	.nway_reset		= ftmac100_nway_reset,
+	.get_link		= ftmac100_get_link,
+};
+
+/******************************************************************************
+ * interrupt handler
+ *****************************************************************************/
+static irqreturn_t ftmac100_interrupt(int irq, void *dev_id)
+{
+	struct net_device *netdev = dev_id;
+	struct ftmac100 *priv = netdev_priv(netdev);
+
+	if (likely(netif_running(netdev))) {
+		/* Disable interrupts for polling */
+		ftmac100_disable_all_int(priv);
+		napi_schedule(&priv->napi);
+	}
+
+	return IRQ_HANDLED;
+}
+
+/******************************************************************************
+ * struct napi_struct functions
+ *****************************************************************************/
+static int ftmac100_poll(struct napi_struct *napi, int budget)
+{
+	struct ftmac100 *priv = container_of(napi, struct ftmac100, napi);
+	struct net_device *netdev = priv->netdev;
+	unsigned int status;
+	bool completed = true;
+	int rx = 0;
+
+	status = ioread32(priv->base + FTMAC100_OFFSET_ISR);
+
+	if (status & (FTMAC100_INT_RPKT_FINISH | FTMAC100_INT_NORXBUF)) {
+		/*
+		 * FTMAC100_INT_RPKT_FINISH:
+		 *	RX DMA has received packets into RX buffer successfully
+		 *
+		 * FTMAC100_INT_NORXBUF:
+		 *	RX buffer unavailable
+		 */
+		bool retry;
+
+		do {
+			retry = ftmac100_rx_packet(priv, &rx);
+		} while (retry && rx < budget);
+
+		if (retry && rx == budget)
+			completed = false;
+	}
+
+	if (status & (FTMAC100_INT_XPKT_OK | FTMAC100_INT_XPKT_LOST)) {
+		/*
+		 * FTMAC100_INT_XPKT_OK:
+		 *	packet transmitted to ethernet successfully
+		 *
+		 * FTMAC100_INT_XPKT_LOST:
+		 *	packet transmitted to ethernet lost due to late
+		 *	collision or excessive collision
+		 */
+		ftmac100_tx_complete(priv);
+	}
+
+	if (status & (FTMAC100_INT_NORXBUF | FTMAC100_INT_RPKT_LOST |
+		      FTMAC100_INT_AHB_ERR | FTMAC100_INT_PHYSTS_CHG)) {
+		if (net_ratelimit())
+			netdev_info(netdev, "[ISR] = 0x%x: %s%s%s%s\n", status,
+				    status & FTMAC100_INT_NORXBUF ? "NORXBUF " : "",
+				    status & FTMAC100_INT_RPKT_LOST ? "RPKT_LOST " : "",
+				    status & FTMAC100_INT_AHB_ERR ? "AHB_ERR " : "",
+				    status & FTMAC100_INT_PHYSTS_CHG ? "PHYSTS_CHG" : "");
+
+		if (status & FTMAC100_INT_NORXBUF) {
+			/* RX buffer unavailable */
+			netdev->stats.rx_over_errors++;
+		}
+
+		if (status & FTMAC100_INT_RPKT_LOST) {
+			/* received packet lost due to RX FIFO full */
+			netdev->stats.rx_fifo_errors++;
+		}
+
+		if (status & FTMAC100_INT_PHYSTS_CHG) {
+			/* PHY link status change */
+			mii_check_link(&priv->mii);
+		}
+	}
+
+	if (completed) {
+		/* stop polling */
+		napi_complete(napi);
+		ftmac100_enable_all_int(priv);
+	}
+
+	return rx;
+}
+
+/******************************************************************************
+ * struct net_device_ops functions
+ *****************************************************************************/
+static int ftmac100_open(struct net_device *netdev)
+{
+	struct ftmac100 *priv = netdev_priv(netdev);
+	int err;
+
+	err = ftmac100_alloc_buffers(priv);
+	if (err) {
+		netdev_err(netdev, "failed to allocate buffers\n");
+		goto err_alloc;
+	}
+
+	err = request_irq(priv->irq, ftmac100_interrupt, 0, netdev->name,
+		netdev);
+	if (err) {
+		netdev_err(netdev, "failed to request irq %d\n", priv->irq);
+		goto err_irq;
+	}
+
+	priv->rx_pointer = 0;
+	priv->tx_clean_pointer = 0;
+	priv->tx_pointer = 0;
+	priv->tx_pending = 0;
+
+	err = ftmac100_start_hw(priv);
+	if (err)
+		goto err_hw;
+
+	napi_enable(&priv->napi);
+	netif_start_queue(netdev);
+
+	ftmac100_enable_all_int(priv);
+	return 0;
+
+err_hw:
+	free_irq(priv->irq, netdev);
+err_irq:
+	ftmac100_free_buffers(priv);
+err_alloc:
+	return err;
+}
+
+static int ftmac100_stop(struct net_device *netdev)
+{
+	struct ftmac100 *priv = netdev_priv(netdev);
+
+	ftmac100_disable_all_int(priv);
+	netif_stop_queue(netdev);
+	napi_disable(&priv->napi);
+	ftmac100_stop_hw(priv);
+	free_irq(priv->irq, netdev);
+	ftmac100_free_buffers(priv);
+
+	return 0;
+}
+
+static int ftmac100_hard_start_xmit(struct sk_buff *skb,
+				    struct net_device *netdev)
+{
+	struct ftmac100 *priv = netdev_priv(netdev);
+	dma_addr_t map;
+
+	if (unlikely(skb->len > MAX_PKT_SIZE)) {
+		if (net_ratelimit())
+			netdev_dbg(netdev, "tx packet too big\n");
+
+		netdev->stats.tx_dropped++;
+		dev_kfree_skb(skb);
+		return NETDEV_TX_OK;
+	}
+
+	map = dma_map_single(priv->dev, skb->data, skb_headlen(skb), DMA_TO_DEVICE);
+	if (unlikely(dma_mapping_error(priv->dev, map))) {
+		/* drop packet */
+		if (net_ratelimit())
+			netdev_err(netdev, "map socket buffer failed\n");
+
+		netdev->stats.tx_dropped++;
+		dev_kfree_skb(skb);
+		return NETDEV_TX_OK;
+	}
+
+	return ftmac100_xmit(priv, skb, map);
+}
+
+/* optional */
+static int ftmac100_do_ioctl(struct net_device *netdev, struct ifreq *ifr,
+			     int cmd)
+{
+	struct ftmac100 *priv = netdev_priv(netdev);
+	struct mii_ioctl_data *data = if_mii(ifr);
+
+	return generic_mii_ioctl(&priv->mii, data, cmd, NULL);
+}
+
+static const struct net_device_ops ftmac100_netdev_ops = {
+	.ndo_open		= ftmac100_open,
+	.ndo_stop		= ftmac100_stop,
+	.ndo_start_xmit		= ftmac100_hard_start_xmit,
+	.ndo_set_mac_address	= eth_mac_addr,
+	.ndo_validate_addr	= eth_validate_addr,
+	.ndo_do_ioctl		= ftmac100_do_ioctl,
+};
+
+/******************************************************************************
+ * struct platform_driver functions
+ *****************************************************************************/
+static int ftmac100_remove(struct platform_device *pdev)
+{
+	struct net_device *netdev;
+	struct ftmac100 *priv;
+
+	netdev = platform_get_drvdata(pdev);
+	if (netdev == NULL)
+		return 0;
+
+	platform_set_drvdata(pdev, NULL);
+
+	priv = netdev_priv(netdev);
+
+	netif_napi_del(&priv->napi);
+	unregister_netdev(netdev);
+
+	if (priv->base != NULL)
+		iounmap(priv->base);
+
+	if (priv->res != NULL)
+		release_resource(priv->res);
+
+	free_netdev(netdev);
+	return 0;
+}
+
+static int ftmac100_probe(struct platform_device *pdev)
+{
+	struct resource *res;
+	int irq;
+	struct net_device *netdev;
+	struct ftmac100 *priv;
+	int err;
+
+	if (!pdev)
+		return -ENODEV;
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (!res)
+		return -ENXIO;
+
+	irq = platform_get_irq(pdev, 0);
+	if (irq < 0)
+		return irq;
+
+	/* setup net_device */
+
+	netdev = alloc_etherdev(sizeof(struct ftmac100));
+	if (netdev == NULL) {
+		err = -ENOMEM;
+		goto err_out;
+	}
+
+	SET_NETDEV_DEV(netdev, &pdev->dev);
+	SET_ETHTOOL_OPS(netdev, &ftmac100_ethtool_ops);
+	netdev->netdev_ops = &ftmac100_netdev_ops;
+
+	platform_set_drvdata(pdev, netdev);
+
+	/* setup private data */
+
+	priv = netdev_priv(netdev);
+	priv->netdev = netdev;
+	priv->dev = &pdev->dev;
+
+	spin_lock_init(&priv->tx_lock);
+
+	/* initialize NAPI */
+	netif_napi_add(netdev, &priv->napi, ftmac100_poll, 64);
+
+	/* map io memory */
+	priv->res = request_mem_region(res->start, resource_size(res),
+				       dev_name(&pdev->dev));
+	if (priv->res == NULL) {
+		dev_err(&pdev->dev, "Could not reserve memory region\n");
+		err = -ENOMEM;
+		goto err_out;
+	}
+
+	priv->base = ioremap(res->start, res->end - res->start);
+	if (priv->base == NULL) {
+		dev_err(&pdev->dev, "Failed to ioremap ethernet registers\n");
+		err = -EIO;
+		goto err_out;
+	}
+
+	priv->irq = irq;
+
+	/* initialize struct mii_if_info */
+
+	priv->mii.phy_id	= 0;
+	priv->mii.phy_id_mask	= 0x1f;
+	priv->mii.reg_num_mask	= 0x1f;
+	priv->mii.dev		= netdev;
+	priv->mii.mdio_read	= ftmac100_mdio_read;
+	priv->mii.mdio_write	= ftmac100_mdio_write;
+
+	/* register network device */
+
+	err = register_netdev(netdev);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to register netdev\n");
+		goto err_out;
+	}
+
+	netdev_info(netdev, "irq %d, mapped at %p\n", priv->irq, priv->base);
+
+	if (!is_valid_ether_addr(netdev->dev_addr)) {
+		random_ether_addr(netdev->dev_addr);
+		netdev_info(netdev, "generated random MAC address %pM\n",
+			    netdev->dev_addr);
+	}
+
+	return 0;
+
+err_out:
+	ftmac100_remove(pdev);
+	return err;
+}
+
+static struct platform_driver ftmac100_driver = {
+	.probe		= ftmac100_probe,
+	.remove		= ftmac100_remove,
+	.driver		= {
+		.name	= DRV_NAME,
+		.owner	= THIS_MODULE,
+	},
+};
+
+/******************************************************************************
+ * initialization / finalization
+ *****************************************************************************/
+static int __init ftmac100_init(void)
+{
+	pr_info("Loading version " DRV_VERSION " ...\n");
+	return platform_driver_register(&ftmac100_driver);
+}
+
+static void __exit ftmac100_exit(void)
+{
+	platform_driver_unregister(&ftmac100_driver);
+}
+
+module_init(ftmac100_init);
+module_exit(ftmac100_exit);
+
+MODULE_AUTHOR("Po-Yu Chuang <ratbert@faraday-tech.com>");
+MODULE_DESCRIPTION("FTMAC100 driver");
+MODULE_LICENSE("GPL");
diff --git a/drivers/net/ftmac100.h b/drivers/net/ftmac100.h
new file mode 100644
index 0000000..46a0c47
--- /dev/null
+++ b/drivers/net/ftmac100.h
@@ -0,0 +1,180 @@ 
+/*
+ * Faraday FTMAC100 10/100 Ethernet
+ *
+ * (C) Copyright 2009-2011 Faraday Technology
+ * Po-Yu Chuang <ratbert@faraday-tech.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#ifndef __FTMAC100_H
+#define __FTMAC100_H
+
+#define	FTMAC100_OFFSET_ISR		0x00
+#define	FTMAC100_OFFSET_IMR		0x04
+#define	FTMAC100_OFFSET_MAC_MADR	0x08
+#define	FTMAC100_OFFSET_MAC_LADR	0x0c
+#define	FTMAC100_OFFSET_MAHT0		0x10
+#define	FTMAC100_OFFSET_MAHT1		0x14
+#define	FTMAC100_OFFSET_TXPD		0x18
+#define	FTMAC100_OFFSET_RXPD		0x1c
+#define	FTMAC100_OFFSET_TXR_BADR	0x20
+#define	FTMAC100_OFFSET_RXR_BADR	0x24
+#define	FTMAC100_OFFSET_ITC		0x28
+#define	FTMAC100_OFFSET_APTC		0x2c
+#define	FTMAC100_OFFSET_DBLAC		0x30
+#define	FTMAC100_OFFSET_MACCR		0x88
+#define	FTMAC100_OFFSET_MACSR		0x8c
+#define	FTMAC100_OFFSET_PHYCR		0x90
+#define	FTMAC100_OFFSET_PHYWDATA	0x94
+#define	FTMAC100_OFFSET_FCR		0x98
+#define	FTMAC100_OFFSET_BPR		0x9c
+#define	FTMAC100_OFFSET_TS		0xc4
+#define	FTMAC100_OFFSET_DMAFIFOS	0xc8
+#define	FTMAC100_OFFSET_TM		0xcc
+#define	FTMAC100_OFFSET_TX_MCOL_SCOL	0xd4
+#define	FTMAC100_OFFSET_RPF_AEP		0xd8
+#define	FTMAC100_OFFSET_XM_PG		0xdc
+#define	FTMAC100_OFFSET_RUNT_TLCC	0xe0
+#define	FTMAC100_OFFSET_CRCER_FTL	0xe4
+#define	FTMAC100_OFFSET_RLC_RCC		0xe8
+#define	FTMAC100_OFFSET_BROC		0xec
+#define	FTMAC100_OFFSET_MULCA		0xf0
+#define	FTMAC100_OFFSET_RP		0xf4
+#define	FTMAC100_OFFSET_XP		0xf8
+
+/*
+ * Interrupt status register & interrupt mask register
+ */
+#define	FTMAC100_INT_RPKT_FINISH	(1 << 0)
+#define	FTMAC100_INT_NORXBUF		(1 << 1)
+#define	FTMAC100_INT_XPKT_FINISH	(1 << 2)
+#define	FTMAC100_INT_NOTXBUF		(1 << 3)
+#define	FTMAC100_INT_XPKT_OK		(1 << 4)
+#define	FTMAC100_INT_XPKT_LOST		(1 << 5)
+#define	FTMAC100_INT_RPKT_SAV		(1 << 6)
+#define	FTMAC100_INT_RPKT_LOST		(1 << 7)
+#define	FTMAC100_INT_AHB_ERR		(1 << 8)
+#define	FTMAC100_INT_PHYSTS_CHG		(1 << 9)
+
+/*
+ * Interrupt timer control register
+ */
+#define FTMAC100_ITC_RXINT_CNT(x)	(((x) & 0xf) << 0)
+#define FTMAC100_ITC_RXINT_THR(x)	(((x) & 0x7) << 4)
+#define FTMAC100_ITC_RXINT_TIME_SEL	(1 << 7)
+#define FTMAC100_ITC_TXINT_CNT(x)	(((x) & 0xf) << 8)
+#define FTMAC100_ITC_TXINT_THR(x)	(((x) & 0x7) << 12)
+#define FTMAC100_ITC_TXINT_TIME_SEL	(1 << 15)
+
+/*
+ * Automatic polling timer control register
+ */
+#define	FTMAC100_APTC_RXPOLL_CNT(x)	(((x) & 0xf) << 0)
+#define	FTMAC100_APTC_RXPOLL_TIME_SEL	(1 << 4)
+#define	FTMAC100_APTC_TXPOLL_CNT(x)	(((x) & 0xf) << 8)
+#define	FTMAC100_APTC_TXPOLL_TIME_SEL	(1 << 12)
+
+/*
+ * DMA burst length and arbitration control register
+ */
+#define FTMAC100_DBLAC_INCR4_EN		(1 << 0)
+#define FTMAC100_DBLAC_INCR8_EN		(1 << 1)
+#define FTMAC100_DBLAC_INCR16_EN	(1 << 2)
+#define FTMAC100_DBLAC_RXFIFO_LTHR(x)	(((x) & 0x7) << 3)
+#define FTMAC100_DBLAC_RXFIFO_HTHR(x)	(((x) & 0x7) << 6)
+#define FTMAC100_DBLAC_RX_THR_EN	(1 << 9)
+
+/*
+ * MAC control register
+ */
+#define	FTMAC100_MACCR_XDMA_EN		(1 << 0)
+#define	FTMAC100_MACCR_RDMA_EN		(1 << 1)
+#define	FTMAC100_MACCR_SW_RST		(1 << 2)
+#define	FTMAC100_MACCR_LOOP_EN		(1 << 3)
+#define	FTMAC100_MACCR_CRC_DIS		(1 << 4)
+#define	FTMAC100_MACCR_XMT_EN		(1 << 5)
+#define	FTMAC100_MACCR_ENRX_IN_HALFTX	(1 << 6)
+#define	FTMAC100_MACCR_RCV_EN		(1 << 8)
+#define	FTMAC100_MACCR_HT_MULTI_EN	(1 << 9)
+#define	FTMAC100_MACCR_RX_RUNT		(1 << 10)
+#define	FTMAC100_MACCR_RX_FTL		(1 << 11)
+#define	FTMAC100_MACCR_RCV_ALL		(1 << 12)
+#define	FTMAC100_MACCR_CRC_APD		(1 << 14)
+#define	FTMAC100_MACCR_FULLDUP		(1 << 15)
+#define	FTMAC100_MACCR_RX_MULTIPKT	(1 << 16)
+#define	FTMAC100_MACCR_RX_BROADPKT	(1 << 17)
+
+/*
+ * PHY control register
+ */
+#define FTMAC100_PHYCR_MIIRDATA		0xffff
+#define FTMAC100_PHYCR_PHYAD(x)		(((x) & 0x1f) << 16)
+#define FTMAC100_PHYCR_REGAD(x)		(((x) & 0x1f) << 21)
+#define FTMAC100_PHYCR_MIIRD		(1 << 26)
+#define FTMAC100_PHYCR_MIIWR		(1 << 27)
+
+/*
+ * PHY write data register
+ */
+#define FTMAC100_PHYWDATA_MIIWDATA(x)	((x) & 0xffff)
+
+/*
+ * Transmit descriptor, aligned to 16 bytes
+ */
+struct ftmac100_txdes {
+	unsigned int	txdes0;
+	unsigned int	txdes1;
+	unsigned int	txdes2;	/* TXBUF_BADR */
+	unsigned int	txdes3;	/* not used by HW */
+} __attribute__ ((aligned(16)));
+
+#define	FTMAC100_TXDES0_TXPKT_LATECOL	(1 << 0)
+#define	FTMAC100_TXDES0_TXPKT_EXSCOL	(1 << 1)
+#define	FTMAC100_TXDES0_TXDMA_OWN	(1 << 31)
+
+#define	FTMAC100_TXDES1_TXBUF_SIZE(x)	((x) & 0x7ff)
+#define	FTMAC100_TXDES1_LTS		(1 << 27)
+#define	FTMAC100_TXDES1_FTS		(1 << 28)
+#define	FTMAC100_TXDES1_TX2FIC		(1 << 29)
+#define	FTMAC100_TXDES1_TXIC		(1 << 30)
+#define	FTMAC100_TXDES1_EDOTR		(1 << 31)
+
+/*
+ * Receive descriptor, aligned to 16 bytes
+ */
+struct ftmac100_rxdes {
+	unsigned int	rxdes0;
+	unsigned int	rxdes1;
+	unsigned int	rxdes2;	/* RXBUF_BADR */
+	unsigned int	rxdes3;	/* not used by HW */
+} __attribute__ ((aligned(16)));
+
+#define	FTMAC100_RXDES0_RFL		0x7ff
+#define	FTMAC100_RXDES0_MULTICAST	(1 << 16)
+#define	FTMAC100_RXDES0_BROADCAST	(1 << 17)
+#define	FTMAC100_RXDES0_RX_ERR		(1 << 18)
+#define	FTMAC100_RXDES0_CRC_ERR		(1 << 19)
+#define	FTMAC100_RXDES0_FTL		(1 << 20)
+#define	FTMAC100_RXDES0_RUNT		(1 << 21)
+#define	FTMAC100_RXDES0_RX_ODD_NB	(1 << 22)
+#define	FTMAC100_RXDES0_LRS		(1 << 28)
+#define	FTMAC100_RXDES0_FRS		(1 << 29)
+#define	FTMAC100_RXDES0_RXDMA_OWN	(1 << 31)
+
+#define	FTMAC100_RXDES1_RXBUF_SIZE(x)	((x) & 0x7ff)
+#define	FTMAC100_RXDES1_EDORR		(1 << 31)
+
+#endif /* __FTMAC100_H */