Simon Glass [Wed, 25 Mar 2015 18:22:15 +0000 (12:22 -0600)]
dm: usb: Move storage device scanning into its own function
The usb_stor_scan() function is quite long, so split out the code that scans
each device into its own function. Also, rather than setting up the block
device list once at the start, set it up as each device is scanned. This
makes it possible to use this code from driver model.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Simon Glass [Wed, 25 Mar 2015 18:22:07 +0000 (12:22 -0600)]
dm: usb: Complete the splitting up of usb_new_device()
This function now calls usb_setup_device() to set up the device and
usb_hub_probe() to check if it is a hub. The XHCI special case is now a
parameter to usb_setup_device(). The latter will be used by the USB uclass
when it is added, since it does not rely on any CONFIGs or legacy data
structures.
Signed-off-by: Simon Glass <sjg@chromium.org>
Bug-fixes for descriptor reading and usb_new_device() return value Signed-off-by: Hans de Goede <hdegoede@redhat.com> Reviewed-by: Marek Vasut <marex@denx.de> Tested-by: Tom Rini <trini@konsulko.com>
Simon Glass [Wed, 25 Mar 2015 18:22:05 +0000 (12:22 -0600)]
dm: usb: Move descriptor setup code into its own function
usb_new_device() is far too long and does far too much. As a first step, move
the code that does initial setup and reads a descriptor into its own function
called usb_setup_descriptor().
For XHCI the init order is different - we set up the device but don't
actually read the descriptor until after we set an address. Support this
option as a parameter to usb_setup_descriptor().
Avoid changing this torturous code more than necessary to make it easy to
review.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Simon Glass [Wed, 25 Mar 2015 18:22:02 +0000 (12:22 -0600)]
dm: usb: Convert 'usb' command to support driver model
Adjust this command to work with the new driver model uclass. It needs to
iterate through multiple independent controllers to find hubs, and work
through their children recursively in a different way. Otherwise the
functionality is much the same.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Simon Glass [Wed, 25 Mar 2015 18:22:01 +0000 (12:22 -0600)]
dm: usb: Adjust usb_alloc_new_device() to return an error
This function returns NULL on error at present. Adjust it so that we can
return a real error, as is needed with driver model. Also improve the
error handling in its caller, usb_hub_port_connect_change(), and adjust
the code order to prepare for driver model.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Simon Glass [Wed, 25 Mar 2015 18:22:00 +0000 (12:22 -0600)]
dm: usb: Adjust usb command to prepare for driver model
Use 'udev' instead of 'dev' in a few places, reserving 'dev' for driver
model's struct udevice. Also adjust the code in a few minor ways to make
it easier to plumb in driver model.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Simon Glass [Wed, 25 Mar 2015 18:21:56 +0000 (12:21 -0600)]
dm: core: Mark device as active before calling uclass probe() methods
The uclass pre-probe functions may end up calling back into the device in
some circumstances. This can fail if recursion takes place. Adjust the
ordering so that we mark the device as active early, then retract this
later if needed.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Simon Glass [Wed, 25 Mar 2015 18:21:55 +0000 (12:21 -0600)]
dm: core: Rename driver data function to dev_get_driver_data()
The existing get_get_of_data() function provides access to both the driver's
compatible string and its driver data. However only the latter is actually
useful. Update the interface to reflect this and fix up existing users.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Simon Glass [Wed, 25 Mar 2015 18:21:53 +0000 (12:21 -0600)]
dm: core: Support allocating driver-private data for DMA
Some driver want to put DMA buffers in their private data. Add a flag
to tell driver model to align driver-private data to a cache boundary so
that DMA will work correctly in this case.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Simon Glass [Wed, 25 Mar 2015 18:21:49 +0000 (12:21 -0600)]
linker_lists: Add a function to access a linker list entry
Once declared, you cannot access a linker_list entry since you do not have
a symbol name for it. Add llsym() macro to provide this. This avoids
searching for the symbol at run-time based on name.
An example usage is to declare a driver with U_BOOT_DRIVER(), then obtain
a pointer to that driver later.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Joe Hershberger [Sat, 4 Apr 2015 01:09:46 +0000 (20:09 -0500)]
dm: eth: Provide a way for drivers to manage packet buffers
Some drivers need a chance to manage their receive buffers after the
packet has been handled by the network stack. Add an operation that
will allow the driver to be called in that case.
Reported-by: Simon Glass <sjg@chromium.org> Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Acked-by: Simon Glass <sjg@chromium.org> Tested-by: Simon Glass <sjg@chromium.org>
Tested-on: pcduino3
dm: spi: Convert Freescale QSPI driver to driver model
Move the Freescale QSPI driver over to driver model.
Signed-off-by: Haikun Wang <Haikun.Wang@freescale.com> Signed-off-by: Peng Fan <Peng.Fan@freescale.com> Tested-by: Peng Fan <Peng.Fan@freescale.com> Acked-by: Simon Glass <sjg@chromium.org>
dm: ls1021a: dts: Update DSPI node to support DM SPI
Update DSPI controller node in ls1021a.dtsi.
Update flash device node in ls1021a-qds.dts.
Ls1021a-twr board doesn't support DSPI, so remove DSPI node
in ls1021a-twr.dts.
Signed-off-by: Haikun Wang <Haikun.Wang@freescale.com> Acked-by: Simon Glass <sjg@chromium.org>
haikun [Tue, 24 Mar 2015 13:16:31 +0000 (21:16 +0800)]
dm: ls1021a: dts: Change address_cells and size_cells from 2 to 1
Change address_cells and size_cells of root node and 'soc' node
from 2 to 1.
We backport ls1021a device tree source files from kernel to u-boot.
Kernel files set address_cells and size_cells to 2 in order to access
more than 4GB space.
But we don't have this requirement now and u-boot fdtdec_get_xxx interfaces
can't support property whose size is 'u64' completely.
So make this change.
Signed-off-by: Haikun Wang <Haikun.Wang@freescale.com> Acked-by: Simon Glass <sjg@chromium.org>
This commit adds driver model support to software emulated i2c bus driver.
This driver supports kernel-style device tree bindings. Fdt properties in use:
- compatible - "i2c-gpio"
- gpios - data and clock GPIO pin phandles
- delay-us - micro seconds delay between GPIOs toggle operations,
which is 1/4 of I2C speed clock period.
Driver base code is taken from: drivers/i2c/soft-i2c.c, changes:
- use "i2c-gpio" naming
- update comments style
- move preprocesor macros into functions
- add device tree support
- add driver model i2c support
- code cleanup,
- add Kconfig entry
Signed-off-by: Przemyslaw Marczak <p.marczak@samsung.com> Acked-by: Simon Glass <sjg@chromium.org>
Added braces in i2c_gpio_xfer() to fix style nit: Signed-off-by: Simon Glass <sjg@chromium.org>
dm: gpio: request list: return the count if requests max_count reached
The function gpio_request_list_by_name_nodev() returned -ENOSPC error,
when the loop count was greater than requested count. This was wrong,
because function should return the requested gpio count, when meets
the call request without errors. Now, the loop ends on requested
max_count.
Signed-off-by: Przemyslaw Marczak <p.marczak@samsung.com> Acked-by: Simon Glass <sjg@chromium.org>
Simon Glass [Thu, 26 Mar 2015 15:29:39 +0000 (09:29 -0600)]
cros_ec: exynos: Match up device tree with kernel version
The U-Boot device trees are slightly different in a few places. Adjust them
to remove most of the differences. Note that U-Boot does not support the
concept of interrupts as distinct from GPIOs, so this difference remains.
For sandbox, use the same keyboard file as for ARM boards and drop the
host emulation bus which seems redundant.
Simon Glass [Thu, 26 Mar 2015 15:29:29 +0000 (09:29 -0600)]
x86: chromebook_link: dts: Add PCH and LPC devices
The PCH (Platform Controller Hub) is on the PCI bus, so show it as such.
The LPC (Low Pin Count) and SPI bus are inside the PCH, so put these in the
right place also.
Rename the compatible strings to be more descriptive since this board is the
only user. Once we are using driver model fully on x86, these will be
dropped.
Simon Glass [Thu, 26 Mar 2015 15:29:26 +0000 (09:29 -0600)]
dm: x86: spi: Convert ICH SPI driver to driver model
Convert this driver over to use driver model. Since all x86 platforms use
it, move x86 to use driver model for SPI and SPI flash. Adjust all dependent
code and remove the old x86 spi_init() function.
Note that this does not make full use of the new PCI uclass as yet. We still
scan the bus looking for the device. It should move to finding its details
in the device tree.
Simon Glass [Thu, 26 Mar 2015 15:29:25 +0000 (09:29 -0600)]
dm: sf: Add driver model read/write/erase methods
Permit use of a udevice to talk to SPI flash. Ultimately we would like
to retire the use of 'struct spi_flash' for this purpose, so create the
new API for those who want to move to it.
Joe Hershberger [Sun, 22 Mar 2015 22:09:24 +0000 (17:09 -0500)]
net: Improve error handling
Take a pass at plumbing errors through to the users of the network stack
Currently only the start() function errors will be returned from
NetLoop(). recv() tends not to have errors, so that is likely not worth
adding. send() certainly can return errors, but this patch does not
attempt to plumb them yet. halt() is not expected to error.
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Joe Hershberger [Sun, 22 Mar 2015 22:09:23 +0000 (17:09 -0500)]
sandbox: eth: Add support for using the 'lo' interface
The 'lo' interface on Linux doesn't support thinks like ARP or
link-layer access like we use to talk to a normal network interface.
A higher-level network API must be used to access localhost.
As written, this interface is limited to not supporting ICMP since the
API doesn't allow the socket to be opened for all IP traffic and be able
to receive at the same time. UDP is far more useful to test with, so it
was selected over ICMP. Ping won't work, but things like TFTP should
work.
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Joe Hershberger [Sun, 22 Mar 2015 22:09:21 +0000 (17:09 -0500)]
sandbox: eth: Add a bridge to a real network for sandbox
Implement a bridge between U-Boot's network stack and Linux's raw packet
API allowing the sandbox to send and receive packets using the host
machine's network interface.
This raw Ethernet API requires elevated privileges. You can either run
as root, or you can add the capability needed like so:
Joe Hershberger [Sun, 22 Mar 2015 22:09:15 +0000 (17:09 -0500)]
test: dm: eth: Add tests for the eth dm implementation
Add a test for the eth uclass using the sandbox eth driver. Verify basic
functionality of the network stack / eth uclass by exercising the ping
function.
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Joe Hershberger [Sun, 22 Mar 2015 22:09:14 +0000 (17:09 -0500)]
sandbox: eth: Add ARP and PING response to sandbox driver
The sandbox driver will now generate response traffic to exercise the
ping command even when no network exists. This allows the basic data
pathways of the DM to be tested.
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Joe Hershberger [Sun, 22 Mar 2015 22:09:12 +0000 (17:09 -0500)]
dm: eth: Pass the packet pointer as a parameter to recv
Stop forcing drivers to call net_process_received_packet() - formerly
called NetReceive(). Now the uclass will handle calling the driver for
each packet until the driver errors or has nothing to return. The uclass
will then pass the good packets off to the network stack by calling
net_process_received_packet().
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com>
Joe Hershberger [Sun, 22 Mar 2015 22:09:07 +0000 (17:09 -0500)]
net: Make netretry actually do something
netretry previously would only retry in one specific case (your MAC
address is not set) and no other. This is basically useless. In the DM
implementation for eth it turns this into a completely useless case
since an un-configured MAC address results in not even entering the
NetLoop. The behavior is now changed to retry any failed command
(rotating through the eth adapters if ethrotate != no).
It also defaulted to retry forever. It is now changed to default to not
retry
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Joe Hershberger [Sun, 22 Mar 2015 22:09:06 +0000 (17:09 -0500)]
net: Remove the bd* parameter from net stack functions
This value is not used by the network stack and is available in the
global data, so stop passing it around. For the one legacy function
that still expects it (init op on old Ethernet drivers) pass in the
global pointer version directly to avoid changing that interface.
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Reported-by: Simon Glass <sjg@chromium.org> Reviewed-by: Simon Glass <sjg@chromium.org> Signed-off-by: Simon Glass <sjg@chromium.org>
(Trival fix to remove an unneeded variable declaration in 4xx_enet.c)
Joe Hershberger [Sun, 22 Mar 2015 22:09:00 +0000 (17:09 -0500)]
net: Provide a function to get the current MAC address
The current implementation exposes the eth_device struct to code that
needs to access the MAC address. Add a wrapper function for this to
abstract away the pointer for this operation.
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Joe Hershberger [Sun, 22 Mar 2015 22:08:59 +0000 (17:08 -0500)]
common: Make sure arch-specific map_sysmem() is defined
In the case where the arch defines a custom map_sysmem(), make sure that
including just mapmem.h is sufficient to have these functions as they
are when the arch does not override it.
Also split the non-arch specific functions out of common.h
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Simon Glass [Thu, 5 Mar 2015 19:25:33 +0000 (12:25 -0700)]
dm: x86: pci: Convert chromebook_link to use driver model for pci
Move chromebook_link over to driver model for PCI.
This involves:
- adding a uclass for platform controller hub
- removing most of the existing PCI driver
- adjusting how CPU init works to use driver model instead
- rename the lpc compatible string (it will be removed later)
This does not really take advantage of driver model fully, but it does work.
Furture work will improve the code structure to remove many of the explicit
calls to init the board.
Simon Glass [Thu, 5 Mar 2015 19:25:29 +0000 (12:25 -0700)]
dm: sandbox: Add a emulated PCI device as an example
This device sits on the sandbox PCI bus and provides a case-swapping
service for sandbox. It illustrates the use of both PCI I/O and PCI
memory accesses.
Simon Glass [Thu, 5 Mar 2015 19:25:28 +0000 (12:25 -0700)]
dm: sandbox: pci: Add a PCI emulation uclass
Since sandbox does not have real devices (unless it borrows those from the
host) it must use emulations. Provide a uclass which permits PCI operations
to be passed through to an emulation device.
Simon Glass [Thu, 5 Mar 2015 19:25:26 +0000 (12:25 -0700)]
dm: sandbox: pci: Add PCI support for sandbox
Add the required header information, device tree nodes and I/O accessor
functions to support PCI on sandbox. All devices are emulated by drivers
which can be added as required for testing or development.
Simon Glass [Thu, 5 Mar 2015 19:25:25 +0000 (12:25 -0700)]
dm: pci: Add a uclass for PCI
Add a uclass for PCI controllers and a generic one for PCI devices. Adjust
the 'pci' command and the existing PCI support to work with this new uclass.
Keep most of the compatibility code in a separate file so that it can be
removed one day.
TODO: Add more header file comments to the new parts of pci.h
Simon Glass [Thu, 5 Mar 2015 19:25:22 +0000 (12:25 -0700)]
dm: core: Add a uclass pre_probe() method for devices
Some uclasses want to set up a device before it is probed. Add a method
for this.
An example is with PCI, where a PCI uclass wants to set up its private
data for later use. This allows the device's uclass() method to make calls
whcih use that data (for example, read PCI memory regions from device
tree, set up bus numbers).
Simon Glass [Thu, 5 Mar 2015 19:25:21 +0000 (12:25 -0700)]
dm: core: Mark device as active before calling its probe() method
At present the device is not active when the probe() method is called. But
some probe() methods want to set up the device and this can involve
accessing it through normal methods. For example a PCI bus may wish to
set up its PCI parameters using calls to pci_hose_write_config_dword() and
similar.
At present this does not work because every such call within the probe()
method sees that the device is not active and attempts to probe it.
Already we mark the device as probed before calling the uclass post_probe()
method. This is a subtle change but I believe the new approach is better.
Since the scope of the change is only the probe() method and all its callees
it should still be within the control of the board author.
Simon Glass [Thu, 5 Mar 2015 19:25:20 +0000 (12:25 -0700)]
dm: core: Add dev_get_uclass_priv() to access uclass private data
Add a convenience function to access the private data that a uclass stores
for each of its devices. Convert over most existing uses for consistency
and to provide an example for others.
projects / u-boot.git / log
