When testing the FEC driver on a mx6solox we noticed that the TDAR bit gets
always cleared prior then the READY bit is cleared in the last BD, which causes
FEC packets reception to always fail.
As explained by Ye Li:
"The TDAR bit is cleared when the descriptors are all out from TX ring, but on
mx6solox we noticed that the READY bit is still not cleared right after TDAR.
These are two distinct signals, and in IC simulation, we found that TDAR always
gets cleared prior than the READY bit of last BD becomes cleared.
In mx6solox, we use a later version of FEC IP. It looks like that this
intrinsic behaviour of TDAR bit has changed in this newer FEC version."
Fix this by polling the READY bit of BD after the TDAR polling, which covers the
mx6solox case and does not harm the other SoCs.
No performance drop has been noticed with this patch applied when testing TFTP
transfers on several boards of different i.mx SoCs.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Marek Vasut <marex@denx.de>
mx6solox has a requirement for 64 bytes alignment for RX DMA transfer.
Other SoCs work with the standard 32 bytes alignment.
Adjust it accordingly by using 64 bytes aligment in the FEC RX DMA buffers,
which addresses the needs from mx6solox and also works for the other SoCs.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Marek Vasut <marex@denx.de>
The prescaler value for MDC is calculated wrong for
MX6S. Implement the same logic as linux here which handles
ENET_MAC different then original FEC.
Tested on a custom board with i.MX6S and 100MBit PHY
Signed-off-by: Markus Niebel <Markus.Niebel@tqs.de>
The fec_halt() never free'd both RX and TX DMA descriptors that
were allocated in fec_init(), nor did it free the RX buffers.
Rework the FEC driver so that these descriptors and buffers are
allocated only once in fec_probe().
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Fabio Estevam <fabio.estevam@freescale.com>
Cc: Stefano Babic <sbabic@denx.de>
Performing tftp transfers on mx28 results in random timeouts.
Hector Palacios and Robert Hodaszi analyzed the root cause being related to the
wrong alignment of the 'buff' buffer inside fec_recv().
Benoît Thébaudeau provided an excellent analysis of the alignment bug that is
present on older versions, such as GCC 4.5.4:
http://marc.info/?l=u-boot&m=137942904906131&w=2
Use ALLOC_CACHE_ALIGN_BUFFER() to avoid alignment issues from older GCC
versions.
Reported-by: Hector Palacios <hector.palacios@digi.com>
Tested-by: Oliver Metz <oliver@freetz.org>
Tested-by: Hector Palacios <hector.palacios@digi.com>
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Marek Vasut <marex@denx.de>
Tested-by: Marek Vasut <marex@denx.de>
Similarly as mx25 and mx53, mx6solo-lite needs to setup the MII gasket for RMII
mode.
Add support for mx6solo-lite.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Without this change, the following message is generated:
Warning: FEC using MAC address from net device
See doc/README.enetaddr for details.
Signed-off-by: Eric Nelson <eric.nelson@boundarydevices.com>
The MX28 multi-layer AHB bus can be too slow and trigger the
FEC DMA too early, before all the data hit the DRAM. This patch
ensures the data are written in the RAM before the DMA starts.
Please see the comment in the patch for full details.
This patch was produced with an amazing help from Albert Aribaud,
who pointed out it can possibly be such a bus synchronisation
issue.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Albert ARIBAUD <albert.u.boot@aribaud.net>
Cc: Fabio Estevam <fabio.estevam@freescale.com>
Cc: Stefano Babic <sbabic@denx.de>
Tested-by: Fabio Estevam <fabio.estevam@freescale.com>
Tested-by: Alexandre Pereira da Silva <aletes.xgr@gmail.com>
Remove incorrectly called and duplicate flush_dcache_range() call
from fec_mxc driver. The call is not needed, since the caches are
already flushed in fec_tbd_init(), moreover the second argument should
be the ending address, not size.
Signed-off-by: Marek Vasut <marex@denx.de>
Reported-by: Albert Aribaud <albert.u.boot@aribaud.net>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Tom Rini <trini@ti.com>
This patch adds FEC support for Vybrid VF610 platform.
In function fec_open(), RCR register is only set as RGMII mode. But RCR
register should be set as RMII mode for VF610 platform.
This configuration is already done in fec_reg_setup(), so this piece of
code could just leave untouched the FEC_RCNTRL_RGMII / FEC_RCNTRL_RMII /
FEC_RCNTRL_MII_MODE bits.
Signed-off-by: Alison Wang <b18965@freescale.com>
Reviewed-by: Benoit Thebaudeau <benoit.thebaudeau@advansee.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Only the hardware ethernet registers are needed
for this function, so don't pass the more general
structure. I'm trying to separate MII and fec.
This also fixes MX28 fec_mii_setspeed use on secondary ethernet port
This was found by inspection of the code and should be
checked on real hardware.
Signed-off-by: Troy Kisky <troy.kisky@boundarydevices.com>
It is more logical to test for CONFIG_FEC_MXC_PHYADDR
to determine whether to define the function fecmxc_initialize.
Signed-off-by: Troy Kisky <troy.kisky@boundarydevices.com>
The mechanism waiting for transmission to finish in fec_send() now
relies on the E-bit being cleared in the TX buffer descriptor. In
case of data cache being on, this means invalidation of data cache
above this TX buffer descriptor on each test for the E-bit being
cleared.
Apparently, there is another way to check if the transmission did
complete. This is by checking the TDAR bit in the X_DES_ACTIVE
register. Reading a register does not need any data cache invalidation,
which is beneficial.
Rework the sequence that wait for completion of the transmission so that
the TDAR bit is tested first and afterwards check the E-bit being clear.
This cuts down the number of cache invalidation calls to one.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Otavio Salvador <otavio@ossystems.com.br>
Cc: Stefano Babic <sbabic@denx.de>
The FEC hardware sometimes errors out on data transfer and hangs in
the tightloop adjusted by this patch. So add timeout into the tightloop
to make such a hang recoverable.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Otavio Salvador <otavio@ossystems.com.br>
Cc: Stefano Babic <sbabic@denx.de>
Align the address that's to be invalidated/flushed properly.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Benoit Thebaudeau <benoit.thebaudeau@advans>
Cc: Eric Nelson <eric.nelson@boundarydevices.com>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Do not pass unaligned RX buffer to the upper layers. The upper layer,
especially in the ARP case, recycles the buffer and passes it back into
the FEC, into it's TX path. With caches enabled, the FEC hangs on this
from time to time.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Benoit Thebaudeau <benoit.thebaudeau@advans>
Cc: Eric Nelson <eric.nelson@boundarydevices.com>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Tested-by: Fabio Estevam <fabio.estevam@freescale.com>
The MDC signal is available on all xMII (i.e. 'not 7-wire') interfaces, so
mii_speed has to be set for all these interfaces, and not only for MII.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Cc: Joe Hershberger <joe.hershberger@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
At least on i.MX25, the RMII mode did not work, which is fixed by this patch.
The MII_MODE bit of the FEC RCR register means xMII, i.e. 'not 7-wire', so set
it accordingly.
According to the xMII and 7-wire (aka GPSI) standards, full duplex should be
available on xMII, but not on 7-wire, so set FCE accordingly. The FEC may
support full duplex for 7-wire too, but the reference manual does not say that,
so avoid an invalid assumption. Actually, the choice between half and full
duplex also depends on the endpoint/switch/repeater configuration, so a config
option could be added for that, but there has been no need for it so far.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Cc: Joe Hershberger <joe.hershberger@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Now that phy_startup() can return an actual error code, check for that error
code and abort network initialization if the PHY fails.
Signed-off-by: Timur Tabi <timur@freescale.com>
Acked-by: Nobuhiro Iwamamatsu <nobuhiro.iwamatsu.yj@renesas.com> (sh_eth part)
Acked-by: Stephan Linz <linz@li-pro.net> (Xilinx part, xilinx_axi_emac and xilinx_ll_temac)
Reviewed-by: Marek Vasut <marex@denx.de> (FEC part)
Fix this:
fec_mxc.c: In function 'fec_probe':
fec_mxc.c:926:13: warning: assignment from incompatible pointer type
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com>
Cc: Joe Hershberger <joe.hershberger@gmail.com>
Abstract out common register setup. This also configured r_cntrl
to correct value at registration time.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
Cc: Fabio Estevam <fabio.estevam@freescale.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Ensure that transmit and receive buffers are cache-line aligned.
Invalidate cache for each packet as received, update receive buffer
descriptors one cache line at a time, flush cache before transmitting.
Original patch by Marek:
http://lists.denx.de/pipermail/u-boot/2012-February/117695.html
Signed-off-by: Eric Nelson <eric.nelson at boundarydevices.com>
Acked-by: Marek Vasut <marex@denx.de>
Tested-by: Marek Vasut <marex@denx.de>
Surround non PHYLIB routines miiphy_restart_aneg
and miiphy_wait_aneg with ifndef CONFIG_PHYLIB.
When later PHYLIB is required it is easy to delete
the non-PHYLIB code. This separation idea
came from Andy Fleming.
fec_miiphy_read, and fec_miiphy_write changed to
fec_phy_read, and fec_phy_write with argument changes.
Deprecated miiphy_register is no longer used. Both
old and new PHYLIB code now use mdio_register.
Cleanup some debug prints.
Inline function fec_miiphy_fec_to_eth is no longer necessary.
Moved to the single call location.
Signed-off-by: Troy Kisky <troy.kisky@boundarydevices.com>
Add multi-FEC support for imx_get_mac_from_fuse by passing dev_id as a parameter.
This feature is important on mx28 SoC for example that has two FEC ports.
Cc: Ben Warren <biggerbadderben@gmail.com>
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Stefano Babic <sbabic@denx.de>
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
fec_mxc.c: In function 'fec_mii_setspeed':
fec_mxc.c:112:2: warning: format '%#lx' expects type 'long unsigned int', but
argument 2 has type 'u32'
fec_mxc.c: In function 'fec_recv':
fec_mxc.c:632:2: warning: format '%x' expects type 'unsigned int', but argument
2 has type 'long unsigned int'
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Simon Glass <sjg@chromium.org>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
This patch allows user to register multiple FEC controllers. To preserve
compatibility with older boards, the mxcfec_register() call is still in place.
To use multiple controllers, new macro is in place, the mxcfec_register_multi(),
which takes more arguments. The syntax is:
mxcfec_register_multi(bd, FEC ID, FEC PHY ID on the MII bus, base address);
To disable the fecmxc_register() compatibility stuff, define the macro
CONFIG_FEC_MXC_MULTI. This will remove the requirement for defining IMX_FEC_BASE
and CONFIG_FEC_MXC_PHYADDR.
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
The default is MII100, which was hardcoded previously in the driver.
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Ben Warren <biggerbadderben@gmail.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Detlev Zundel <dzu@denx.de>
The include/miiphy.h header duplicates a lot of things from linux/mii.h.
So punt all the things that overlap to keep the API simple and to make
merging between U-Boot and Linux simpler.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
The patch is to support getting FEC MAC address from fuse bank.
Signed-off-by: Jason Liu <r64343@freescale.com>
Tested-by: Stefano Babic <sbabic@denx.de>
This prevents access to the member of eth_device which is not initialized.
Signed-off-by: Nobuhiro Iwamatsu <iwamatsu@nigauri.org>
CC: Ben Warren <biggerbadderben@gmail.com>
Call fec_set_hwaddr in init routine to setup MAC address so when ethaddr is set
late via setenv the change will propagate to the hw.
Signed-off-by: John Rigby <john.rigby@linaro.org>
CC: Ben Warren <biggerbadderben@gmail.com>
The driver name does not need to be writable, so constify it.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
After discussion on the ML it is suggested to drop unrequired
and not useful characters from the device name.
This patch changes the name for the fec_mxc driver from
"FEC_MXC" to "FEC".
Signed-off-by: Stefano Babic <sbabic@denx.de>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Fix MX27 FEC logic to check validity of the MAC address in fuse.
Only null (empty fuse) or invalid MAC address was retrieved from mx27 fuses before this change.
Signed-off-by: Eric Jarrige <jorasse@armadeus.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Avoid using the internal eeprom on MX25 like MX51 already does.
Signed-off-by: John Rigby <jcrigby@gmail.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Only fill the device enetaddr with the contents of the eeprom,
do not program it in MAC address registers
Signed-off-by: Heiko Schocher <hs@denx.de>
Acked-by: Ben Warren <biggerbadderben@gmail.com>
general cleanup
move clock init to cpu_eth_init in cpu/arm926ejs/mx27/generic.c
make MX27 specific phy init conditional on CONFIG_MX27
replace call to imx_get_ahbclk with one to imx_get_fecclk
and define imx_get_fecclk in include/asm-arm/arch-mx27/clock.h
Signed-off-by: John Rigby <jcrigby@gmail.com>
CC: Ben Warren <biggerbadderben@gmail.com>
CC: Fred Fan <fanyefeng@gmail.com>
CC: Tom <Tom.Rix@windriver.com>
The patch add support for the Freescale mx51 processor
to the FEC ethernet driver.
Signed-off-by: Stefano Babic <sbabic@denx.de>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
fec_init() will only allocate fec->base_ptr if it is non-NULL. But
the cleanup routine on error will free the pointer without setting
it to NULL. This means that a later call to fec_init() would result
in using an invalid pointer.
Signed-off-by: John Ogness <john.ogness@linutronix.de>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Sometimes, inside NetLoop, eth_halt() is called before eth_init() has
been called. This is harmless except for free() calls to pointers
which have not been allocated yet.
This patch initializes those pointers to NULL and allocates them only
the first time. This way we can get rid of free calls in halt callback.
This has been tested in i.MX27 Litekit board and eldk-4.2 toolchains.
Signed-off-by: Javier Martin <javier.martin@vista-silicon.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
This patch fixes erroneous access to the ethernet PHY which broke the driver.
1. Selector field in the auto-negotiation register must be 0x00001 for
using 802.3, not 0x00000 which is reseved.
2. Access to the PHY address specified by CONFIG_FEC_MXC_PHYADDR, not
0x0 fixed address.
This has been tested in i.MX27 Litekit board and eldk-4.2 toolchains.
Now using proper defines for auto-negotiation register.
Signed-off-by: Javier Martin <javier.martin@vista-silicon.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
