Add software TSO support for FEC.
This feature allows to improve outbound throughput performance.

Tested on imx6dl sabresd board, running iperf tcp tests shows:
- 16.2% improvement comparing with FEC SG patch
- 82% improvement comparing with NO SG & TSO patch

$ ethtool -K eth0 tso on
$ iperf -c 10.192.242.167 -t 3 &
[  3] local 10.192.242.108 port 35388 connected with 10.192.242.167 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0- 3.0 sec   181 MBytes   506 Mbits/sec

During the testing, CPU loading is 30%.
Since imx6dl FEC Bandwidth is limited to SOC system bus bandwidth, the
performance with SW TSO is a milestone.

CC: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: David Laight <David.Laight@ACULAB.COM>
CC: Li Frank <B20596@freescale.com>
Signed-off-by: NFugang Duan <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add Scatter/gather support for FEC.
This feature allows to improve outbound throughput performance.

Tested on imx6dl sabresd board:
Running iperf tests shows a 55.4% improvement.

$ ethtool -K eth0 sg off
$ iperf -c 10.192.242.167 -t 3 &
[  3] local 10.192.242.108 port 52618 connected with 10.192.242.167 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0- 3.0 sec  99.5 MBytes   278 Mbits/sec

$ ethtool -K eth0 sg on
$ iperf -c 10.192.242.167 -t 3 &
[  3] local 10.192.242.108 port 52617 connected with 10.192.242.167 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0- 3.0 sec   154 MBytes   432 Mbits/sec

CC: Li Frank <B20596@freescale.com>
Signed-off-by: NFugang Duan <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In order to support SG, software TSO, let's increase BD entry number.

CC: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: David Laight <David.Laight@ACULAB.COM>
Signed-off-by: NFugang Duan <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In order to enhance the code readable, let's factorize the
feature list.
Signed-off-by: NFugang Duan <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

IP header checksum is calcalated by network layer in default.
To support software TSO, it is better to use HW calculate the
IP header checksum.

FEC hw checksum feature request the checksum field in frame
is zero, otherwise the calculative CRC is not correct.

For segmentated TCP packet, HW calculate the IP header checksum again,
it doesn't bring any impact. For SW TSO, HW calculated checksum bring
better performance.
Signed-off-by: NFugang Duan <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Make the code more readable and easy to support other features like
SG, TSO, moving the common transmit function to one api.

And the patch also factorize the getting BD index to it own function.

CC: David Laight <David.Laight@ACULAB.COM>
Signed-off-by: NFugang Duan <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit 5bbde4d2 ("net: fec: use pinctrl PM helpers") caused the following
build error on m68k:

drivers/net/ethernet/freescale/fec_main.c: In function 'fec_enet_open':
drivers/net/ethernet/freescale/fec_main.c:1819:2: error: implicit declaration of function 'pinctrl_pm_select_default_state' [-Werror=implicit-function-declaration]
drivers/net/ethernet/freescale/fec_main.c: In function 'fec_enet_close':
drivers/net/ethernet/freescale/fec_main.c:1863:2: error: implicit declaration of function 'pinctrl_pm_select_sleep_state' [-Werror=implicit-function-declaration]

In order to fix the build error, include the linux/pinctrl/consumer.h header
file.
Reported-by: Nkbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

when system suspend, need to set pins to low power state to
save IO power consumption, there are three states of pinctrl:
"default", "idle" and "sleep". Currently enet supports default
and sleep state.
Signed-off-by: NFugang Duan <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since imx serials FEC/ENET MDIO clock source is internal ipg clock,
and "ahb" clock is defined as FEC/ENET bus clock, so the patch just
correct the fec driver MDIO clock source.
Signed-off-by: NFugang Duan <B38611@freescale.com>
Acked-by: NFrank Li <frank.li@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add below clock management to save fec power:
- After probe, disable all clocks incluing ipg, ahb, enet_out, ptp clock.
- Open ethx interface enable necessary clocks.
  Close ethx interface disable all clocks.

The patch also encapsulates the all enet clocks enable/disable to
.fec_enet_clk_enable(), which can reduce the repetitional code in
driver.
Signed-off-by: NFugang Duan <B38611@freescale.com>
Acked-by: NFrank Li <Frank.li@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Though we made sure to acquire a valid MAC for
the netdevice we never actually programmed it
into the hardware.
So if the bootloader did not set the MAC,
network operation would only work if userspace
explicitly asked to transfer the MAC to hardware.
Signed-off-by: NLucas Stach <l.stach@pengutronix.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

fec_enet_mdio_reset() does nothing useful and is optional for the MDIO
bus code, so let's just remove it.
Signed-off-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Replace kfree_skb with dev_kfree_skb_any in fec_enet_start_xmit that
can be called in hard irq and other contexts, when the packet is
dropped.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>

If the Freescale fec is in promiscuous mode and network cable is
reconnected then the promiscuous mode get lost. The problem is caused
by a too soon call of set_multicast_list to re-enable promisc mode.
The FEC_R_CNTRL register changes are overwritten by fec_restart.

This patch fixes this by moving the call behind the init of FEC_R_CNTRL
register in fec_restart.

Successful tested on a i.MX28 board.
Signed-off-by: NStefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The current flow: Set TX BD ready, and then set "INT" and "PINS" bit to
enable tx interrupt generation and crc checksum.

There has potential issue like as:
CPU			fec uDMA
Set tx ready bit
			uDMA start the BD transmission
Set "INT" bit
Set "PINS" bit
...

Above situation cause fec tx interrupt lost and fec MAC don't do
CRC checksum. The patch fix the potential issue.
Signed-off-by: NFugang Duan <B38611@freescale.com>
Acked-by: NFrank Li <Frank.li@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If napi is left enabled after a failed attempt to bring the interface
up, we BUG:

fec 2188000.ethernet eth0: no PHY, assuming direct connection to switch
libphy: PHY fixed-0:00 not found
fec 2188000.ethernet eth0: could not attach to PHY
------------[ cut here ]------------
kernel BUG at include/linux/netdevice.h:502!
Internal error: Oops - BUG: 0 [#1] SMP ARM
...
PC is at fec_enet_open+0x4d0/0x500
LR is at __dev_open+0xa4/0xfc

Only enable napi after we are past all the failure paths.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

None of these files are actually using any __init type directives
and hence don't need to include <linux/init.h>.   Most are just a
left over from __devinit and __cpuinit removal, or simply due to
code getting copied from one driver to the next.

This covers everything under drivers/net except for wireless, which
has been submitted separately.
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In order to keep DT compatibility we need to revert this, otherwise the original
dts files will no longer work with this driver change.

This reverts commit 7a399e3a.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We should not assume that the PHY reset is always active low.

Retrieve this information from the device tree instead, so that the PHY reset
can work on both cases.
Reported-by: NPhilipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

skb_tx_timestamp(skb) should be called _before_ TX completion
has a chance to trigger, otherwise it is too late and we access
freed memory.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Fixes: de5fb0a0 ("net: fec: put tx to napi poll function to fix dead lock")
Cc: Frank Li <Frank.Li@freescale.com>
Cc: Richard Cochran <richardcochran@gmail.com>
Acked-by: NRichard Cochran <richardcochran@gmail.com>
Acked-by: NFrank Li <Frank.Li@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit 03191656 worked around
errata ERR006358, but comment contains duplicated lines, impairing
the readability.  Remove them.
Signed-off-by: NPhilippe De Muyter <phdm@macqel.be>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

On tx submit the driver always dma_map_single() FEC_ENET_TX_FRSIZE (=2048)
bytes. This works because we don't overwrite any memory after the data buffer,
we remove it from cache if it was there. So we hurt performace in case the
mapping of a smaller area makes a difference.
There is also a bug: If the data area starts shortly before the end of
RAM say 0xc7fffa10 and the RAM ends at 0xc8000000 then we have enough
space to fit the data area (according to skb->len) but we would map beyond
end of ram if we are using 2048. In v2.6.31 (against which kernel this patch
made) there is the following check in dma_cache_maint():

|BUG_ON(!virt_addr_valid(start) || !virt_addr_valid(start + size - 1));

Since the area starting at 0xc8000000 is no longer virt_addr_valid() we
BUG() during dma_map_single(). The BUG() statement was removed in v3.5-rc1 as
per 2dc6a016 ("ARM: dma-mapping: use asm-generic/dma-mapping-common.h").

This patch was tested on v2.6.31 and then forward-ported and compile
tested only against the net tree. I think it is still worth fixing
mainline even after the BUG() statement is gone.
Tested-by: NFugang Duan <B38611@freescale.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: NSebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: NFugang Duan <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This is untested.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

The driver fails to check the results of DMA mapping and results in
the following warning: (with kernel config "CONFIG_DMA_API_DEBUG" enable)

------------[ cut here ]------------
WARNING: at lib/dma-debug.c:937 check_unmap+0x43c/0x7d8()
fec 2188000.ethernet: DMA-API: device driver failed to check map
error[device address=0x00000000383a8040] [size=2048 bytes] [mapped as single]

Modules linked in:
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.10.17-16827-g9cdb0ba-dirty #188
[<80013c4c>] (unwind_backtrace+0x0/0xf8) from [<80011704>] (show_stack+0x10/0x14)
[<80011704>] (show_stack+0x10/0x14) from [<80025614>] (warn_slowpath_common+0x4c/0x6c)
[<80025614>] (warn_slowpath_common+0x4c/0x6c) from [<800256c8>] (warn_slowpath_fmt+0x30/0x40)
[<800256c8>] (warn_slowpath_fmt+0x30/0x40) from [<8026bfdc>] (check_unmap+0x43c/0x7d8)
[<8026bfdc>] (check_unmap+0x43c/0x7d8) from [<8026c584>] (debug_dma_unmap_page+0x6c/0x78)
[<8026c584>] (debug_dma_unmap_page+0x6c/0x78) from [<8038049c>] (fec_enet_rx_napi+0x254/0x8a8)
[<8038049c>] (fec_enet_rx_napi+0x254/0x8a8) from [<804dc8c0>] (net_rx_action+0x94/0x160)
[<804dc8c0>] (net_rx_action+0x94/0x160) from [<8002c758>] (__do_softirq+0xe8/0x1d0)
[<8002c758>] (__do_softirq+0xe8/0x1d0) from [<8002c8e8>] (do_softirq+0x4c/0x58)
[<8002c8e8>] (do_softirq+0x4c/0x58) from [<8002cb50>] (irq_exit+0x90/0xc8)
[<8002cb50>] (irq_exit+0x90/0xc8) from [<8000ea88>] (handle_IRQ+0x3c/0x94)
[<8000ea88>] (handle_IRQ+0x3c/0x94) from [<8000855c>] (gic_handle_irq+0x28/0x5c)
[<8000855c>] (gic_handle_irq+0x28/0x5c) from [<8000de00>] (__irq_svc+0x40/0x50)
Exception stack(0x815a5f38 to 0x815a5f80)
5f20:                                                       815a5f80 3b9aca00
5f40: 0fe52383 00000002 0dd8950e 00000002 81e7b080 00000000 00000000 815ac4d8
5f60: 806032ec 00000000 00000017 815a5f80 80059028 8041fc4c 60000013 ffffffff
[<8000de00>] (__irq_svc+0x40/0x50) from [<8041fc4c>] (cpuidle_enter_state+0x50/0xf0)
[<8041fc4c>] (cpuidle_enter_state+0x50/0xf0) from [<8041fd94>] (cpuidle_idle_call+0xa8/0x14c)
[<8041fd94>] (cpuidle_idle_call+0xa8/0x14c) from [<8000edac>] (arch_cpu_idle+0x10/0x4c)
[<8000edac>] (arch_cpu_idle+0x10/0x4c) from [<800582f8>] (cpu_startup_entry+0x60/0x130)
[<800582f8>] (cpu_startup_entry+0x60/0x130) from [<80bc7a48>] (start_kernel+0x2d0/0x328)
[<80bc7a48>] (start_kernel+0x2d0/0x328) from [<10008074>] (0x10008074)
---[ end trace c6edec32436e0042 ]---

Because dma-debug add new interfaces to debug dma mapping errors, pls refer
to: http://lwn.net/Articles/516640/

After dma mapping, it must call dma_mapping_error() to check mapping error,
otherwise the map_err_type alway is MAP_ERR_NOT_CHECKED, check_unmap() define
the mapping is not checked and dump the error msg. So,add dma_mapping_error()
checking to fix the WARNING

And RX DMA buffers are used repeatedly and the driver copies it into an skb,
fec_enet_rx() should not map or unmap, use dma_sync_single_for_cpu()/dma_sync_single_for_device()
instead of dma_map_single()/dma_unmap_single().

There have another potential issue:  fec_enet_rx() passes the DMA address to __va().
Physical and DMA addresses are *not* the same thing. They may differ if the device
is behind an IOMMU or bounce buffering was required, or just because there is a fixed
offset between the device and host physical addresses. Also fix it in this patch.

=============================================
V2: add net_ratelimit() to limit map err message.
    use dma_sync_single_for_cpu() instead of dma_map_single().
    fix the issue that pass DMA addresses to __va() to get virture address.
V1: initial send
=============================================
Signed-off-by: NFugang Duan <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch proposes to remove the IRQF_DISABLED flag from
drivers/net/ethernet/freescale/fec_main.c

It's a NOOP since 2.6.35 and it will be removed one day.
Signed-off-by: NMichael Opdenacker <michael.opdenacker@free-electrons.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Bug: error to get the previous BD entry. When the current BD
is the first BD, the previous BD entry must be the last BD,
not "bdp - 1" in current logic.

V4:
  * Optimize fec_enet_get_nextdesc() for code clean.
    Replace "ex_new_bd - ring_size" with "ex_base".
    Replace "new_bd - ring_size" with "base".

V3:
  * Restore the API name because David suggest to use fec_enet_
    prefix for all function in fec driver.
    So, change next_bd() -> fec_enet_get_nextdesc()
        change pre_bd()  -> fec_enet_get_prevdesc()
  * Reduce the two APIs parameters for easy to call.

V2:
  * Add tx_ring_size and rx_ring_size to struct fec_enet_private.
  * Replace api fec_enet_get_nextdesc() with next_bd().
    Replace api fec_enet_get_prevdesc() with pre_bd().

  * Move all ring size check logic to next_bd() and pre_bd(), which
    simplifies the code redundancy.

V1:
  * Add BD ring size check to get the previous BD entry in correctly.
Reviewed-by: NLi Frank <B20596@freescale.com>
Signed-off-by: NFugang Duan  <B38611@freescale.com>
Acked-by: NFrank Li <frank.li@freescale.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit dc975382 "net: fec: add napi support to improve proformance"
converted the fec driver to the napi model. However, that commit
forgot to remove the call to skb_defer_rx_timestamp which is only
needed in non-napi drivers.

(The function napi_gro_receive eventually calls netif_receive_skb,
which in turn calls skb_defer_rx_timestamp.)

This patch should also be applied to the 3.9 and 3.10 kernels.
Signed-off-by: NRichard Cochran <richardcochran@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

My static checker complains that on some arches unsigned longs can be 8
characters which is larger than the buffer is only 6 chars.
Additionally, Ben Hutchings points out that the buffer actually holds
big endian data and the buffer we are reading from is CPU endian.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: NBen Hutchings <bhutchings@solarflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use the wrapper function for retrieving the platform data instead of
accessing dev->platform_data directly. This is a cosmetic change
to make the code simpler and enhance the readability.
Signed-off-by: NJingoo Han <jg1.han@samsung.com>
Acked-by: NFugang Duan  <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Instead of using a custom 'FEC_NAPI_WEIGHT', just use the generic
'NAPI_POLL_WEIGHT' definition instead.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Don't test for having link and let hardware deal with this situation.

Without this patch I see a machine running an -rt patched Linux being
stuck in sch_direct_xmit when it looses link while there is still a
packet to be sent. In this case the fec_enet_start_xmit routine returned
NETDEV_TX_BUSY which makes the network stack reschedule the packet and
so sch_direct_xmit calls fec_enet_start_xmit again.
I failed to reproduce a complete hang without -rt, but I think the
problem exists there, too.
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If the ready bit in the transmit buffer descriptor (TxBD[R])
is previously detected as not set during a prior frame transmission,
then the ENET_TDAR[TDAR] bit is cleared at a later time, even if
additional TxBDs were added to the ring and the ENET_TDAR[TDAR]
bit is set. This results in frames not being transmitted until
there is a 0-to-1 transition on ENET_TDAR[TDAR].

Workarounds:
code can use the transmit frame interrupt flag (ENET_EIR[TXF])
as a method to detect whether the ENET has completed transmission
and the ENET_TDAR[TDAR] has been cleared. If ENET_TDAR[TDAR] is
detected as cleared when packets are queued and waiting for transmit,
then a write to the TDAR bit will restart TxBD processing.

This case main happen when loading is light. A ethernet package may
not send out utile next package put into tx queue.

How to test:
while [ true ]
do
	ping <IP> -s 10000 -w 4
	ping <IP> -s 6000 -w 2
	ping <IP> -s 4000 -w 2
	ping <IP> -s 10000 -w 2
done

You will see below result in overnight test.

6008 bytes from 10.192.242.116: seq=1 ttl=128 time=0.722 ms
4008 bytes from 10.192.242.116: seq=0 ttl=128 time=1001.008 ms
4008 bytes from 10.192.242.116: seq=1 ttl=128 time=1.010 ms
10008 bytes from 10.192.242.116: seq=0 ttl=128 time=0.896 ms

After apply this patch, >1000ms delay disappear.
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Acked-by: NFugang Duan  <B38611@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add MODULE_ALIAS, so that auto module loading can work.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Using devm_request_irq() can make the code smaller and cleaner.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As devm_ioremap_resource() is used, there is no need to explicitely check the
return value from platform_get_resource(), as this is something that
devm_ioremap_resource() takes care by itself.

Also, place platform_get_resource() prior to devm_ioremap_resource() for
better code readability.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

clk_prepare_enable() may fail, so let's check its return value and propagate it
in the case of error.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

clk_ptp should also be enabled in fec_resume() and disabled in fec_suspend().
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

On fec_probe the clocks are enabled in the following order:

clk_ahb -> clk_ipg -> clk_enet_out -> clk_ptp

, so in the error and remove paths we should disabled them in the opposite
order.

Also fix the order in the suspend/resume functions.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

clk_enet_out and clk_ptp are optional clocks, so we should not enable/disable
them unconditionally.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This enables the driver to take advantage of the FEC VLAN
indicator to improve performance.
Signed-off-by: NJim Baxter <jim_baxter@mentor.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>