The code assumes sorted mappings in descending address order. When
splitting a mapping, insert the new part next to the current mapping.
Signed-off-by: NStefan Brüns <stefan.bruens@rwth-aachen.de>
Reviewed-by: NAlexander Graf <agraf@suse.de>
Signed-off-by: NAlexander Graf <agraf@suse.de>

Currently each allocation creates a new mapping. Readding the mapping
as free memory (EFI_CONVENTIONAL_MEMORY) potentially allows to hand out
an existing mapping, thus limiting the number of mapping descriptors in
the memory map.

Mitigates a problem with current (4.8rc7) linux kernels when doing an
efi_get_memory map, resulting in an infinite loop. Space for the memory
map is reserved with allocate_pool (implicitly creating a new mapping) and
filled. If there is insufficient slack space (8 entries) in the map, the
space is freed and a new round is started, with space for one more entry.
As each round increases requirement and allocation by exactly one, there
is never enough slack space. (At least 32 entries are allocated, so as
long as there are less than 24 entries, there is enough slack).
Earlier kernels reserved no slack, and did less allocations, so this
problem was not visible.
Signed-off-by: NStefan Brüns <stefan.bruens@rwth-aachen.de>
Reviewed-by: NAlexander Graf <agraf@suse.de>
Signed-off-by: NAlexander Graf <agraf@suse.de>

We need a functional free_pool implementation, as otherwise each
allocate_pool causes growth of the memory descriptor table.

Different to free_pages, free_pool does not provide the size for the
to be freed allocation, thus we have to track the size ourselves.

As the only EFI requirement for pool allocation is an alignment of
8 bytes, we can keep allocating a range using the page allocator,
reserve the first 8 bytes for our bookkeeping and hand out the
remainder to the caller. This saves us from having to use any
independent data structures for tracking.

To simplify the conversion between pool allocations and the corresponding
page allocation, we create an auxiliary struct efi_pool_allocation.

Given the allocation size free_pool size can handoff freeing the page
range, which was indirectly allocated by a call to allocate_pool,
to free_pages.
Signed-off-by: NStefan Brüns <stefan.bruens@rwth-aachen.de>
Reviewed-by: NAlexander Graf <agraf@suse.de>
Signed-off-by: NAlexander Graf <agraf@suse.de>

We currently handle efi_allocate_pool() in our boot time service
file. In the following patch, pool allocation will receive additional
internal semantics that we should preserve inside efi_memory.c instead.

As foundation for those changes, split the function into an externally
facing efi_allocate_pool_ext() for use by payloads and an internal helper
efi_allocate_pool() in efi_memory.c that handles the actual allocation.

While at it, change the magic 0xfff / 12 constants to the more obvious
EFI_PAGE_MASK/SHIFT defines.
Signed-off-by: NStefan Brüns <stefan.bruens@rwth-aachen.de>
Reviewed-by: NAlexander Graf <agraf@suse.de>
Signed-off-by: NAlexander Graf <agraf@suse.de>

A type mismatch in the efi_allocate_pool boot service flow causes
hazardous memory scribbling on 32-bit systems.

This is efi_allocate_pool's prototype:

static efi_status_t EFIAPI efi_allocate_pool(int pool_type,
						    unsigned long size,
						    void **buffer);

Internally, it invokes efi_allocate_pages as follows:

efi_allocate_pages(0, pool_type, (size + 0xfff) >> 12,
					    (void*)buffer);

This is efi_allocate_pages' prototype:

efi_status_t efi_allocate_pages(int type, int memory_type,
					unsigned long pages,
					uint64_t *memory);

The problem: efi_allocate_pages does this internally:

    *memory = addr;

This fix in efi_allocate_pool uses a transitional uintptr_t cast to
ensure the correct outcome, irrespective of the system's native word
size.

This was observed when bootefi'ing the EFI instance of FreeBSD's first
stage bootstrap (boot1.efi) on a 32-bit ARM platform (Qemu VExpress +
Cortex-a9).
Signed-off-by: NRobin Randhawa <robin.randhawa@arm.com>
Signed-off-by: NAlexander Graf <agraf@suse.de>

The current efi_get_memory_map() function overwrites the map_size
property before reading its value. That way the sanity check whether our
memory map fits into the given array always succeeds, potentially
overwriting arbitrary payload memory.

This patch moves the property update write after its sanity check, so
that the check actually verifies the correct value.

So far this has not triggered any known bugs, but we're better off safe
than sorry.

If the buffer is to small, the returned memory_map_size indicates the
required size to the caller.
Signed-off-by: NStefan Brüns <stefan.bruens@rwth-aachen.de>
Reviewed-by: NAlexander Graf <agraf@suse.de>
Signed-off-by: NAlexander Graf <agraf@suse.de>

In 74c16acc the return values where
changed, but the description was kept.
Signed-off-by: NStefan Brüns <stefan.bruens@rwth-aachen.de>
Reviewed-by: NAlexander Graf <agraf@suse.de>
Signed-off-by: NAlexander Graf <agraf@suse.de>

Signed-off-by: NTom Rini <trini@konsulko.com>

Move FDT_FIXUP_PARTITIONS to Kconfig and cleanup existing
uses.
Signed-off-by: NHeiko Schocher <hs@denx.de>

Move SYS_OS_BASE to Kconfig and cleanup existing
uses.
Signed-off-by: NHeiko Schocher <hs@denx.de>
[trini: Also migrate a4m2k]
Signed-off-by: NTom Rini <trini@konsulko.com>

Move SPL_OS_BOOT to Kconfig and cleanup existing
uses.
Signed-off-by: NHeiko Schocher <hs@denx.de>

We can only run this command if the sleep command is enabled and that
depends on CONFIG_CMD_MISC
Signed-off-by: NTom Rini <trini@konsulko.com>

Jon Master reports that QEMU refuses to load a U-Boot image built
with CONFIG_ARMV7_NONSEC, but without CONFIG_ARMV7_PSCI since
commit 5a3aae68 ("ARM: armv7: guard memory reserve for PSCI
with #ifdef CONFIG_ARMV7_PSCI").

It looks like only PSCI that needs the Secure stack, so move
the #ifdef to guard the whole of .secure_stack allocation in order
not to create the empty section.
Signed-off-by: NMasahiro Yamada <yamada.masahiro@socionext.com>
Reported-by: NJon Masters <jcm@redhat.com>
Link: http://patchwork.ozlabs.org/patch/664025/

lbc_clk is used to fixup dts as "bus frequency".
It is not being used by Linux IFC and eLBC driver.

So remove unused "bus frqeuency" fix-up of devicre tree.
Signed-off-by: NPrabhakar Kushwaha <prabhakar.kushwaha@nxp.com>
Reviewed-by: NYork Sun <york.sun@nxp.com>

PBL flush command is restricted to CCSR memory space. So use WAIT
PBI command to provide enough time for data to get flush in
target memory.
Signed-off-by: NZhao Qiang <qiang.zhao@nxp.com>
[York Sun: rewrap commit message]
Reviewed-by: NYork Sun <york.sun@nxp.com>

T series boards use unified RCW for sd, spi and nand boot.
Now split txxx_rcw.cfg to txxx_sd_rcw.cfg, txxx_spi_rcw.cfg
and txxx_nand_rcw.cfg for SPI/NAND/SD boot.
And modify RCW[PBI_SRC] for them:
	PBI_SRC=5            for SPI 24-bit addressing
	PBI_SRC=6            for SD boot
	PBI_SRC=14           for IFC NAND boot
Signed-off-by: NZhao Qiang <qiang.zhao@nxp.com>
Reviewed-by: NYork Sun <york.sun@nxp.com>

The fdt_path_offset() function is not inlined in upstream libfdt. Adjust
U-Boot's version to match.
Signed-off-by: NSimon Glass <sjg@chromium.org>

The signature for this macro has changed. Bring in the upstream version and
adjust U-Boot's usages to suit.
Signed-off-by: NSimon Glass <sjg@chromium.org>
Update to drivers/power/pmic/palmas.c:
Signed-off-by: NKeerthy <j-keerthy@ti.com>

Change-Id: I6cc9021339bfe686f9df21d61a1095ca2b3776e8

These have now landed upstream. The naming is different and in one case the
function signature has changed. Update the code to match.

This applies the following upstream commits by
Thierry Reding <treding@nvidia.com> :

   604e61e fdt: Add functions to retrieve strings
   8702bd1 fdt: Add a function to get the index of a string
   2218387 fdt: Add a function to count strings
Signed-off-by: NSimon Glass <sjg@chromium.org>

This includes small changes to the following functions, from upstream
commit 6d1832c:

- fdt_get_max_phandle() (upstream commit 84e0e134)
- fdt_node_check_compatible (upstream commit 53bf130b)
- fdt_setprop_inplace_namelen_partial() to remove useless brackets and
     use idx instead of index
- _fdt_resize_property() to use idx instead of index
- _fdt_splice() (upstream commit d4c7c25c)

It also includes various typo fixes in libfdt.h
Signed-off-by: NSimon Glass <sjg@chromium.org>

Using pointer arithmetic to generate a pointer outside a known object is,
technically, undefined behaviour in C.  Unfortunately, we were using that
in fdt_offset_ptr() to detect overflows.

To fix this we need to do our bounds / overflow checking on the offsets
before constructing pointers from them.
Reported-by: NDavid Binderman <dcb314@hotmail.com>
Signed-off-by: NDavid Gibson <david@gibson.dropbear.id.au>
Signed-off-by: NSimon Glass <sjg@chromium.org>

Sometimes devicetree nodes and or properties are added out of the u-boot
console, maybe through some script or manual interaction.

The devicetree as loaded or embedded is quite small, so the devicetree
has to be resized to take up those new nodes/properties.

In original the devicetree was only extended by effective
4 * add_mem_rsv.

With this commit we can add an argument to the "fdt resize" command,
which takes the extrasize to be added.
Signed-off-by: NHannes Schmelzer <hannes.schmelzer@br-automation.com>
Signed-off-by: NHannes Schmelzer <oe5hpm@oevsv.at>
Acked-by: NSimon Glass <sjg@chromium.org>

eth-uclass.c expects DM-capable Ethernet adapters to implement ops->
read_rom_hwaddr(), or for some other mechanism to set pdata->enetaddr, or
for the user to set environment variable $usbethaddr. Without any of
these, it will refuse to initialize the device since no valid MAC address
is known. Implement this function for the smsc95xx driver.

With this feature implemented, there is no point smsc95xx_init_common()
re-reading the MAC address from ROM, so ifdef out this code when DM_ETH
is enabled.

This allows (at least) the built-in Ethernet on the NVIDIA Harmony board
to operate again.

Fixes: 0990fcb7 ("net: smsc95xx: Add driver-model support")
Signed-off-by: NStephen Warren <swarren@nvidia.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

Add a NFS download test, based on TFTP test.
Tested on i.MX6 SabreLite board.
Signed-off-by: NGuillaume GARDET <guillaume.gardet@free.fr>

Cc: Tom Rini <trini@konsulko.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Cc: Stephen Warren <swarren@wwwdotorg.org>
Cc: Simon Glass <sjg@chromium.org>
Reviewed-by: NStephen Warren <swarren@nvidia.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

After any operation that downloads a file (e.g., pxe get, or dhcp), the
buffer containing the downloaded data is flushed.  This is unnecessary
and annoying.  Unnecessary, because
the network driver should already have fliushed the cache for the DMAed area,
and annoying because it generates a cache misalignment message.
Signed-off-by: NPeter Chubb <peter.chubb@data61.csiro.au>
Acked-by: NHeiko Schocher <hs@denx.de>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>
Reviewed-by: NTom Rini <trini@konsulko.com>

The call to flush cache on the transmit buffer was misplaced (for very
short packets) and asked to flush less than a cacheline.

Move the flush cache call to after a short packet has been padded
to minimum length (so the padding is flushed too), and round the size
up to a cacheline.
Signed-off-by: NPeter Chubb <peter.chubb@data61.csiro.au>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

If mac-address is changed using "setenv ethaddr ...." command the new
mac-adress also must be written into the responsible ethernet driver.
Signed-off-by: NHannes Schmelzer <oe5hpm@oevsv.at>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

Signed-off-by: NChris Packham <judge.packham@gmail.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

On some boards these switches are wired directly into a SERDES
interface on another Ethernet MAC. Add the ability to specify
these kinds of boards using CONFIG_MV88E61XX_FIXED_PORTS which defines
a bit mask of these fixed ports.
Signed-off-by: NChris Packham <judge.packham@gmail.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

The Marvell Link Street mv88e60xx is a series of FastEthernet switch
chips, some of which also support Gigabit ports. It is similar to the
mv88e61xx series which support Gigabit on all ports.

The main difference is the number of ports. Which affects the
PORT_COUNT define and the size of the mask passed to
mv88e61xx_port_set_vlan().

Other than that it's just a matter of adding the appropriate chip
IDs.
Signed-off-by: NChris Packham <judge.packham@gmail.com>
Cc: Joshua Scott <joshua.scott@alliedtelesis.co.nz>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

Ethernet ports might be used in the kernel even if CPSW driver
is disabled at u-boot. So always set ethaddr and eth1addr
environment variable from efuse.

Retain usbnet_devaddr as it is required for SPL USB eth boot.
Signed-off-by: NRoger Quadros <rogerq@ti.com>
Reviewed-by: NTom Rini <trini@konsulko.com>
Reviewed-by: NTom Rini <trini@konsulko.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

Both ethernet ports can be used as CPSW ethernet (RMII mode)
or PRU ethernet (MII mode) by setting the jumper near the port.
Read the jumper value and set the pinmux, external mux and
PHY clock accordingly.

As jumper line is overridden by PHY RX_DV pin immediately
after bootstrap (power-up/reset), we have to use GPIO edge
detection to capture the jumper line status.

As u-boot doesn't provide any infrastructure for GPIO edge
detection, we directly access the GPIO registers.
Signed-off-by: NRoger Quadros <rogerq@ti.com>
Reviewed-by: NTom Rini <trini@konsulko.com>
Reviewed-by: NTom Rini <trini@konsulko.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

cpsw driver supports only selection of phy mode in control module
but control module has more setting like RGMII ID mode selection,
RMII clock source selection. So ported to cpsw-phy-sel driver
from kernel to u-boot.
Signed-off-by: NMugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: NLokesh Vutla <lokeshvutla@ti.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

In AM335x GP EVM, Atheros 8031 phy is used, enable the driver as
AM335x SoC RGMII delay mode has to be enabled in phy as mentioned
in the silicon errata Advisory 1.0.10
Signed-off-by: NMugunthan V N <mugunthanvnm@ti.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

In the current driver implementation, config() callback is common
for AR8035 and AR8031 phy. In config() callback, driver tries to
configure MMD Access Control Register and MMD Access Address Data
Register unconditionally for both phy versions which leads to
auto negotiation failure in AM335x EVMsk second port which uses
AR8031 Giga bit RGMII phy. Fixing this by adding separate config
for AR8031 phy.
Reviewed-by: NSekhar Nori <nsekhar@ti.com>
Signed-off-by: NMugunthan V N <mugunthanvnm@ti.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

This adds support for internal delay on RX and TX on RGMII interface for the
AR8035 phy.

This is basically the same Linux driver do. Tested on a Zynq Zturn board (for
which u-boot support in is my tree; first patch waiting ML approval)
Signed-off-by: NAndrea Merello <andrea.merello@gmail.com>
Reviewed-by: NMarek Vasut <marex@denx.de>
Acked-by: NMarek Vasut <marex@denx.de>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>