In barebox network packets always go out at the current ethernet
device and are expected to be received from the current interface.
This has some side effects. When for example an NFS is mounted when
one interface is active and the interface is changed afterwards the
NFS packets leave the new interface, but the NFS server won't be
reachable there.
Instead of changing the whole network traffic to the current ethernet
interface we now initialize a network connection with the current
network interface, but then the connection will continue to use that
interface even when the current interface is changed.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
This adds a convenience function to register a MAC address device
parameter. The only current user is converted to use it.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Otherwise we loose memory on each device_unregister. The ethernet
code used to do this before calling unregister_device. This can
now be removed.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
free() already checks the pointer to be non
NULL. No need to do it again.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
A lot of files rely on include/driver.h including include/of.h (and
this including include/errno.h. include the files explicitly so we can
eventually get rid of including of.h from driver.h
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
If drivers want to fixup their specific instance they need some context
to know which instance they have to fixup.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
If a network device has not been registered from the devicetree, we may
still find it by its alias in the devicetree. This way also platform based
network devices can obtain a valid MAC address in the devicetree.
Signed-off-by: Renaud Barbier <renaud.barbier@ge.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
eth_of_fixup() is passed a devicetree, so use this one instead
of the internal devicetree. This makes sure it also works when
the tree to fixup is not the internal one.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Linux normally has no idea how to retrieve MAC Addresses, but instead
expects the MAC address in the devicetree. This patch adds the MAC
address to the devicetree for Linux if we find a valid one in barebox.
This mechanism is limited of course to devices barebox has a driver for
and which are probed themselves from the devicetree.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
We already have a possibility to register a MAC address provider
based on a ethernet device id. This adds a similar functionality
for devices probed from devicetree. Code can register itself to
be a MAC address provider for a certain devicetree node.
This helps on i.MX to let the IIM unit provide a MAC address for
the FEC.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Rather than storing the parameters globally and trying to keep them
in sync with the device parameters, store the parameters in the ethernet
device directly. Also, update to dev_add_param_ip().
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
When starting a network device wait until the link is up. Otherwise
autobooting does not work with little timeout and several attempts
have to be made until the network is finally up.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
dev_add_child is a very unsafe function. If called multiple times
it allows setting the same device to different parents thus corrupting
the siblings list. This happens regularly since:
| commit c2e568d19c
| Author: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
| Date: Sat Nov 3 16:11:05 2012 +0100
|
| bus: add bus device
|
| automatically add it as parent of any bus device if none already specified
|
| we have now a nice output per bus
If for example a FATfs is mounted this nice output per bus often ends with:
> `---- fat0
> `---- 0
> `---- 0x86f0000087020031-0x86f000410df27124: /dev/<NULL>
> `---- sram00
> `---- 0x00000000-0xffffffffffffffff: /dev/<NULL>
> `---- 0x00000000-0xffffffffffffffff: /dev/<NULL>
> unable to handle NULL pointer dereference at address 0x0000000c
> pc : [<87f08a20>] lr : [<87f08a04>]
> sp : 86eff8c0 ip : 87f3fbde fp : ffffffff
> r10: ffffffff r9 : 00000000 r8 : 00000003
> r7 : 86f075b8 r6 : 00000002 r5 : ffffffec r4 : 86f07544
> r3 : 00000000 r2 : 43f900b4 r1 : 00000020 r0 : 00000005
> Flags: Nzcv IRQs off FIQs off Mode SVC_32
> [<87f08a20>] (do_devinfo_subtree+0x90/0x130) from [<87f08a90>] (do_devinfo_subtree+0x100/0x130)
>
> [<87f3e070>] (unwind_backtrace+0x0/0x90) from [<87f28514>] (panic+0x28/0x3c)
> [<87f28514>] (panic+0x28/0x3c) from [<87f3e4b8>] (do_exception+0x10/0x14)
> [<87f3e4b8>] (do_exception+0x10/0x14) from [<87f3e544>] (do_data_abort+0x2c/0x38)
> [<87f3e544>] (do_data_abort+0x2c/0x38) from [<87f3e268>] (data_abort+0x48/0x60)
This patch fixes this by adding a device to its parents children list in
register_device so that dev_add_child is no longer needed. This function
is removed from the tree. Now callers of register_device have to clearly
set the parent *before* registering a device.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Reported-by: Jan Lübbe <jlu@pengutronix.de>
We call phy_update_status only once in 5 seconds. This makes
sure we do not have great overhead when using ethernet devices.
However, if phylib tells us the link is down anyway, there won't
be ethernet transfers, so it doesn't hurt to call phy_update_status
in this case. This makes sure we can use the ethernet device when
the link comes up and do not have an additional 5 second penalty
in this case.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
as we do not need to probe them and they have no driver or bus attached
Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
edev->init is called at eth_register time unconditionally and is supposed
to initialize the ethernet hardware. Since it's called unconditionally
this could be done by the driver without having an additional hook.
Some drivers need their initialization done earlier since they also register
a mdiobus which does hardware accesses on registration time.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Adapt phylib from linux
switch all the driver to it
reimplement mii bus
This will allow to have
- phy drivers
- to only connect the phy at then opening of the device
- if the phy is not ready or not up fail on open
Same behaviour as in linux and will allow to share code and simplify porting.
Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
The FSF address has changed in the past. Instead of updating it
each time the address changes, just drop it completely treewide.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Reduce confusion by making clear six bytes are passed,
not a string.
Signed-off-by: Johannes Stezenbach <js@sig21.net>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
We register a device in eth_register, thus we have to unregister it in
eth_unregister. Also, if the device we unregister is the current device,
we have to set the current eth_device to NULL so that it isn't used anymore.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
MAC addresses are sometimes stored at unusual places. This
patch makes it possible to give a MAC address to a ethernet
device id. This is independent of the device actually being
present.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
When retrieving the MAC address from EEPROM, we must notify the networking core
at eth registration time. Otherwise we get the 'No MAC address set' warning on
the first network access, and the MAC address is set to a random value instead
of the real one.
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
this rework is done in order to add a phylib and allow to have phy driver support
Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Change device parameters so that the memory management is in generic
code. This also removes the need of storing statically initialized
parameters as they are stored in a struct list_head for each device.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
We used to eth_open/eth_halt the network devices during NetLoopInit
which is called whenever the user enters a network command.
Change this behaviour so that the current network device gets opened
when making it the current one.
With this change it's always possible to send packages and we are able
to implement a new network stack in the next step.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Sascha Hauer
Lucas Stach
Renaud Barbier
Jean-Christophe PLAGNIOL-VILLARD
Johannes Stezenbach
Wolfram Sang
Baruch Siach