usb: gadget: allow to actually enable sparse fastboot support ...

When sparse support was (intended to be) made optional it was
effectively unconditionally disabled because

	IS_ENABLED(USB_GADGET_FASTBOOT_SPARSE)

always evaluates to 0. To actually make use of the introduced kconfig
symbol the CONFIG_ prefix must not be skipped.

Fixes: f4b5d3eeb607 ("usb: gadget: fastboot: Make sparse support optional")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Tested-by: Gavin Schenk <g.schenk@eckelmann.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: imx: implement support for limiting host to full speed ...

This is needed when host is known to not work properly in high speed
mode.

In linux, chipidea driver supports 'maximum-speed' device tree property.
When that is set to "full-speed", driver sets PFSC bit in PORTSC
register, which disallows use of high speed mode.

This patch implements same support for barebox.

Important technical detail is that PFSC bit is cleared by port reset,
thus setting it has to be done in ehci->init() callback which is called
after ehci_reset().

Signed-off-by: Nikita Yushchenko <nikita.yoush@cogentembedded.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: fix barebox update without using buffer ...

Commit e5098495 "usb: gadget: fastboot: Add option to download to a buffer"
breaks the use of temporary files for fastboot barebox_update.

In the patch the ".imagefile = FASTBOOT_TMPFILE" allocation gets
removed, but data.imagefile is still used. This leads to an NULL
pointer dereference. We fix it by using sourcefile instead.

Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: Add external command execution support ...

Custom projects may need vendor specific expansions to the fastboot
command execution. Allow these to be implemented without messing in
the fastboot code directly.
We have a hook for all commands and also one for the "flash" command.
Each hook can decide if the generic command parser is executed
afterwards (return value FASTBOOT_CMD_FALLTHROUGH) or if the generic
parser shall be skipped (return value 0 or negative error code).

This allows board code to implement vendor specific "oem" commands
or to handle the downloaded image in a special way (i.e. do signature
checks on them)

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: Add option to download to a buffer ...

This adds an option to download the image data to a temporary
buffer rather than to a file. While a file is generally the
better option, in some special cases a buffer is better for
memory usage.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: beautify fb_run_command() ...

Only some refactoring, no functional change intended.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: fix typo ...

s/correspoinding/corresponding/

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: Always remove temporary file ...

We should consistently remove the temporary image file, regardless
of the error code. To do so, always jump to the end of the function
where the temporary file is removed.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: use read_file_2 ...

read_file_2 gives us a proper error code and allows us
to specify a maximum size which in this case we happen to know.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: pass struct f_fastboot * around ...

The different fastboot commands mostly do not need the USB endpoint
or the USB request, they need the context data structure struct
f_fastboot * instead, so rather pass this one around between the
different functions.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: Make sparse support optional ...

Sparse support is not always desired, make it optional.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: allow data.image to be const ...

Do not directly store the dynamically allocated memory in data.image
as we want to make that pointer const and then no longer can call
free() on it.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: gadget: fastboot: Add sparse image support ...

Sparse images are needed for fastboot to flash images that do not
fit into memory.

Android fastboot is kind of a dumb protocol. It first sends you a
big image and afterwards it tells you where to put this image. This
of course limits fastboot to files smaller than the available memory.
To bypass this issue fastboot has so called Sparse image support. This
is a new image format that can contain holes in the data so a big image
can be written in multiple steps with holes in other areas each step.

Implementing this for UBI images is kind of tricky since ubiformat
normally expects to get a single image. Preparations to ubiformat
a device in multiple parts have been done in previous patches.
It is however, only for the first part possible to detect if it is
a UBI image. The other parts are just binary data and we cannot
know if we have to pass this data raw to the flash or use ubiformat.
For this reason this patch makes the 'u' flag in the partition
description mandatory for proper UBI image flashing on MTD devices.

fastboot passes a max_download_size variable to the host. This
variable contains the image size fastboot can take in one go.
We have to choose a good compromise here. On one hand we do not
want to risk that we actually do not have enough memory to hold the
transferred image, on the other hand we want images like bootloaders
not as sparse images, but in one go instead to make atomic exchange
of the bootloader possible. Right now we have chosen 8MiB which
should be big enough for all bootloaders. The value is configurable
via global.usbgadget.fastboot_max_download_size.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usbgadget: fastboot: select FILE_LIST ...

The file list functions are used by the fastboot gadget since
a3976c3a84f2 (usbgadget: fastboot: Use function to find file_list
entry by name).

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: imx: implement support for "over-current-active-high" property ...

This property is already documented in the bindings and supported by
Linux. As on i.MX25 the HW default is active high and up to now barebox
didn't force this to low (which is the default for e.g. i.MX6), add the
property to barebox' imx25.dtsi to keep existing behaviour. If on a
board the OC pin is active low, you need to add

	/delete-property/ over-current-active-high;

in the board.dts.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Merge branch 'for-next/usbgadget'

usbgadget autostart: add usbgadget.fastboot_bbu to automatically export bbu handlers ...

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usbgadget: fastboot: Allow to automatically export the bbu handlers ...

We have a list of registered handlers which take a barebox update. Do
the next step and allow to automatically export them via fastboot so
that barebox can be updated via fastboot without manually exporting
the partition. Since this may not be desirable in all cases this
behaviour is configurable.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usbgadget: fastboot: Use function to find file_list entry by name ...

We have file_list_entry_by_name() now, so use it rather than open
coding it.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usbgadget: unregister when usb_composite_probe() fails ...

When usb_multi_register() returns an error it can't be called again
as we do not unregister the driver properly. Fix this.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usbgadget: only set to peripheral mode when error checking is done ...

Only begin to modify the hardware when the obious errors have been
checked.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usbgadget: do not register when no functions present ...

registering a multifunction device makes only sense when there's
at least one function configured. Just return otherwise.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usbgadget: autostart: Handle errors in file list gracefully ...

file_list_parse() can fail and returns an error pointer. Print
an error message when it fails and also set to NULL again so
that usb_multi_register() does not stumble upon the error pointer.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

remove checks for xzalloc() returning NULL ...

xzalloc() either returns memory or panics, so checking for NULL is useless.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Merge branch 'for-next/misc'

usb: gadget: autostart: properly release f_multi_opts ...

the same as for usb gadget autostarter.

|commit 2b9bcff
|Author: Sascha Hauer <s.hauer@pengutronix.de>
|Date:   Fri Jan 20 10:03:45 2017 +0100
|
|    usb: gadget: properly release f_multi_opts
|
|    The usbgadget commands uses statically allocated f_multi_opts and passes
|    this to usb_multi_register(). These f_multi_opts are of course no
|    longer valid when we leave the usbgadget command. Luckily we do not use
|    the data after we left the usbgadget command, so this never has been a
|    problem. However, f_multi_opts has some allocated members which we can
|    not free anymore on gadget unregistration because we no longer have the
|    pointer to them.
|
|    Fix this by adding a release function to struct f_multi_opts. This way
|    we can allocate all memory dynamically and properly free it when not
|    used anymore.
|
|    Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

at91_udc: add DT support ...

Based on the linux kernel version of the same driver.
Needs to adjust clock names as the clock names used in the device tree
does not match the clocknames used in platform_data.
The clocknames in the device tree are not unique, so it was
not an option to rename the clocks.

It boots and the driver is discovered - no further testing done.

$ devinfo fff78000.gadget
Resources:
  num: 0
  name: /ahb/apb/gadget@fff78000
  start: 0xfff78000
  size: 0x00004000
Driver: at91_udc
Bus: platform
Parameters:
  vbus: 1 (type: bool)
Device node: /ahb/apb/gadget@fff78000
gadget@fff78000 {
        compatible = "atmel,at91sam9263-udc";
        reg = <0xfff78000 0x4000>;
        interrupts = <0x18 0x4 0x2>;
        clocks = <0x26 0x27>;
        clock-names = "pclk", "hclk";
        status = "okay";
        atmel,vbus-gpio = <0x28 0x19 0x0>;
};

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

usb: imx: Only use clock when it exists ...

Fixes: bbe0df9

I assumed all chipidea devices actually have a clock. This is true
when looking at the device tree, but in barebox not all i.MX SoCs
use the clocks provided by the device tree. Most instead use
clkdev_add_physbase() to attach a clocks to devices. For these the
chipidea device does not have a clock.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Merge branch 'for-next/parameter-types'

usb: gadget: fastboot: delete temporary file ...

After we have copied the data to its final destination there is
no need to keep the temporary file around. Remove it when done
with it.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>