Create a boot_from_devices() function to handle trying each device. This
helps to reduce the size of the already-large board_init_r() function.
Signed-off-by: NSimon Glass <sjg@chromium.org>

It is useful to name each method so that we can print out this name when
using the method. Currently this happens using a separate function. In
preparation for unifying this, add a name to each method.

The name is only available if we have libcommon support (i.e can use
printf()).
Signed-off-by: NSimon Glass <sjg@chromium.org>

A double underscore is normally reserved for compiler predefines. Use a
single underscore instead.
Signed-off-by: NSimon Glass <sjg@chromium.org>

The GPIO7 pad mux should be programmed to POWERHOLD value
as per board design. In cases where the PMIC is shut off the
mux is set to GPIO7 mode. So during initialization to be on the
safer side set the mode to POWERHOLD.
Signed-off-by: NKeerthy <j-keerthy@ti.com>
Reviewed-by: NTom Rini <trini@konsulko.com>

Enable Linear Physical Address Extension mode which is a
prerequisite for hypervisor mode.
Signed-off-by: NKeerthy <j-keerthy@ti.com>
Reviewed-by: NTom Rini <trini@konsulko.com>

solve issue when bootstage is used with armV7 generic timer
first call of timer_get_boot_us() use the function get_timer()
before timer initialization (arch.timer_rate_hz = 0)
=> div by 0

Commit-notes

When I activate bootstage on ARMV7 architecture with platform
using the generic armv7 timer defined in file
./arch/arm/cpu/armv7m/timer.c

I have a issue because gd->arch.timer_rate_hz = 0

For me the get_timer() function should not used before timer_init
(which initialize gd->arch.timer_rate_hz) at least for the ARMV7
timer.

But in the init sequence, the first bootstage fucntion is called
before timer_init and this function use the timer function.

For me it is a error in the generic init sequence :
mark_bootstage is called before timer_init.

static init_fnc_t init_sequence_f[] = {
....
    arch_cpu_init_dm,
    mark_bootstage,        /* need timer, go after init dm */
...
#if defined(CONFIG_ARM) || defined(CONFIG_MIPS) || \
        defined(CONFIG_BLACKFIN) || defined(CONFIG_NDS32) || \
        defined(CONFIG_SPARC)
    timer_init,        /* initialize timer */
#endif
.......

To solve the issue for all the paltform, we can move timer_init()
call just before mark_bootstage() in this array...

It should be ok for ARMV7 but I don't sure for other platform
impacted
- the other ARM platform or ARMV7 wich don't use generic timer
- MIPS BLACKFIN NDS32 or SPARC

and I don't sure of impact for other function called
(board_early_init_f for example....)

=> This patch solve issue only in timer armv7
   get_boot_us() can be called everytime without div by 0 issue
   (gd->arch.timer_rate_hz is not used)

END
Signed-off-by: NPatrick Delaunay <patrick.delaunay@st.com>
Signed-off-by: NPatrick Delaunay <patrick.delaunay73@gmail.com>

This reverts commit 3edc0c25, reversing
changes made to bb135a01.

The VSC8601 config tried to add an RGMII skew based on #defines that
no config defines. That's quite an ugly way to do it. Since the skew
is only needed on RGMII interfaces, check the interface mode at
runtime, and apply the settings accordingly.

Tested on custom board with AM3352 SOC and VSC801 PHY.
Signed-off-by: NAlexandru Gagniuc <alex.g@adaptrum.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

Testing on theadorable (Armada XP) has shown, that using this driver
results in many cache misaligned warning, such as:

CACHE: Misaligned operation at range [7fabd8fc, 7fabd900]

This patch now uses the ALLOC_CACHE_ALIGN_BUFFER() macro to allocate the
buffers on a cache aligned boundary. This fixes all warnings seen on the
Armada XP platform.
Signed-off-by: NStefan Roese <sr@denx.de>
Cc: Ted Chen <tedchen@realtek.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

The settings for 2.5G SGMII are wrong, which the 2.5G case is missed in
set_if_mode(), and the serdes PCS configuration are wrong, this patch uses
the correct settings took from Linux.
Signed-off-by: NShaohui Xie <Shaohui.Xie@nxp.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

Commit 525d187a ("net: phy: Optionally force master mode for RTL PHY")
added the define to force the PHY into master mode. Unfortunatly this is
an all or nothing switch. So it applies to either all PHY's or no PHY's.

The bug that define tried to solve was a buggy PLL in the RTL8211C only.

The Olimex OLinuXino Lime2 has gotten an upgrade where the PHY was
replaced with an RTL8211E. With this define however, both lime2 boards
are either forced to master mode or not. We could of course have a
binary for each board, but the following patch fixes this by adding a
'quirk' to the flags to the rtl8211b and rtl8211c only. It is now
possible to force master mode, but only have it apply to the rtl8211b
and rtl8211c.
Signed-off-by: NOlliver Schinagl <oliver@schinagl.nl>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

All internal defines in the realtek phy are with a small X,
except MIIM_RTL8211X_CTRL1000T_MASTER. Make this more consistent
Signed-off-by: NOlliver Schinagl <oliver@schinagl.nl>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

The BIT macro is the preferred method to set bits.
This patch adds the bit macro and converts bit invocations.
Signed-off-by: NOlliver Schinagl <oliver@schinagl.nl>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

Fix the following error, the $ret variable handling must
be part of the loop, while due to the missing parenthesis
it was not.

drivers/net/phy/micrel.c: In function ‘ksz9021_of_config’:
drivers/net/phy/micrel.c:303:2: warning: this ‘for’ clause does not guard... [-Wmisleading-indentation]
  for (i = 0; i < ARRAY_SIZE(ofcfg); i++)
  ^~~
drivers/net/phy/micrel.c:305:3: note: ...this statement, but the latter is misleadingly indented as if it is guarded by the ‘for’
   if (ret)
   ^~
drivers/net/phy/micrel.c: In function ‘ksz9031_of_config’:
drivers/net/phy/micrel.c:411:2: warning: this ‘for’ clause does not guard... [-Wmisleading-indentation]
  for (i = 0; i < ARRAY_SIZE(ofcfg); i++)
  ^~~
drivers/net/phy/micrel.c:413:3: note: ...this statement, but the latter is misleadingly indented as if it is guarded by the ‘for’
   if (ret)
   ^~
Signed-off-by: NMarek Vasut <marex@denx.de>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Acked-by: NJoe Hershberger <joe.hershberger@ni.com>

axi_emac, emaclite and gem have the same issue with registering
multiple instances with mdio busses. mdio bus name has to be uniq but
drivers are setting up only one name for all.
Use mdio_register_seq() and pass dev->seq number to allow multiple
mdio instances registration.
Reported-by: NPhani Kiran Kara <phanikiran.kara@gmail.com>
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
Series-to: u-boot
Series-cc: Phani Kiran Kara <phanikiran.kara@gmail.com>

The most of ethernet drivers are using this mdio registration sequence.
strcpy(priv->bus->name, "emac");
mdio_register(priv->bus);
Where driver can be used only with one MDIO bus because only unique
name should be used.

Other drivers are using unique device name for MDIO registration to
support multiple instances.
snprintf(priv->bus->name, sizeof(bus->name), "%s", name);

With DM dev->seq is used more even in logs
(like random MAC address generation:
printf("\nWarning: %s (eth%d) using random MAC address - %pM\n",
       dev->name, dev->seq, pdata->enetaddr);
)
where eth%d prefix is used.

Simplify driver code to register mdio device with dev->seq number
to simplify mdio registration and reduce code duplication across
all drivers. With DM_SEQ_ALIAS enabled dev->seq reflects alias setting.
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
---
For example:

Board: Xilinx Zynq
Net:   ZYNQ GEM: e000b000, phyaddr 7, interface rgmii-id

Warning: ethernet@e000b000 (eth0) using random MAC address -
7a:fc:90:53:6a:41
eth0: ethernet@e000b000ZYNQ GEM: e000c000, phyaddr ffffffff, interface
rgmii-id

Warning: ethernet@e000c000 (eth3) using random MAC address -
1a:ff:d7:1a:a1:b2
, eth3: ethernet@e000c000
** Bad device size - mmc 0 **
Checking if uenvcmd is set ...
Hit any key to stop autoboot:  0
Zynq> mdio list
eth0:
17 - Marvell 88E1111S <--> ethernet@e000b000
eth3:
17 - Marvell 88E1111S <--> ethernet@e000c000
Zynq>

With origin files there was an issue with serdes setting for SCSI.
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>

The dram_init* functions for the zynq board are not safe for use from
the board_init_f stage due to its use of the 'tmp' static variable.

This incorrect use of a static variable was causing rare issues where
the dram_init function would overwrite some parts the __rel_dyn section
which caused obscure failures.

Using the zynq_zybo configuration, U-Boot would generate the following
error during image load. This was caused due to dram_init overwriting
the relocations for the "image" variable within the do_bootm function.
Out of coincidence the un-initialized memory has a compression type
which is the same as the value for the relocation type R_ARM_RELATIVE.

   Uncompressing Invalid Image ... Unimplemented compression type 23

It should be noted that this is just one way the issue could surface,
other cases my not be observed in normal boot flow.

This change removes the existing code and copies the implementation of
the dram_init and dram_init_banksize from the
arch/arm/mach-uniphier/dram_init.c source. This version of these
functions does not use static variables and behaves the same (reading
banks from fdt, and using the first bank as the ram_size).
Signed-off-by: NNathan Rossi <nathan@nathanrossi.com>
Fixes: 758f29d0 ("ARM: zynq: Support systems with more memory banks")
Cc: Michal Simek <monstr@monstr.eu>
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>

Signed-off-by: NMichal Simek <michal.simek@xilinx.com>

Use embded option because of qemu

Use my repo till Stephen merge it.
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>

The patch is fixing:
"tools: mkimage: Check if file is regular file"
(sha1: 56c7e801)
which contains two issues reported by Coverity
Unchecked return value from stat and incorrect calling sequence where
attack can happen between calling stat and fopen.
Using pair in opposite order (fopen and fstat) is fixing this issue
because fstat is using the same file descriptor (FILE *).

Also fixing issue with:
"tools: mkimage: Add support for initialization table for Zynq and
ZynqMP" (sha1: 3b646080)
where file wasn't checked that it is regular file.

Reported-by: Coverity (CID: 154711, 154712)
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
Reviewed-by: NTom Rini <trini@konsulko.com>
Series-to: trini
Series-cc: u-boot

This patch also includes ARM64 zynqmp changes:
- Remove platform non DM initialization
- Remove hardcoded sata base address
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
Reviewed-by: NSimon Glass <sjg@chromium.org>
Series-to: sjg, agraf@suse.de
Series-cc: uboot
Series-version: 4
Series-changes: 2
- make ceva_init_sata static
- Move SATA_CEVA to defconfig
- Initalized max_lun and max_id platdata

Series-changes: 3
- Extend Kconfig help description
- sort dm.h
- Remove SPL undefinition from board file
- Fix Kconfig dependecies

All sata based drivers are bind and corresponding block
device is created. Based on this find_scsi_device() is able
to get back block device based on scsi_curr_dev pointer.

intr_scsi() is commented now but it can be replaced by calling
find_scsi_device() and scsi_scan().

scsi_dev_desc[] is commented out but common/scsi.c heavily depends on
it. That's why CONFIG_SYS_SCSI_MAX_DEVICE is hardcoded to 1 and symbol
is reassigned to a block description allocated by uclass.
There is only one block description by device now but it doesn't need to
be correct when more devices are present.

scsi_bind() ensures corresponding block device creation.
uclass post_probe (scsi_post_probe()) is doing low level init.

SCSI/SATA DM based drivers requires to have 64bit base address as
the first entry in platform data structure to setup mmio_base.
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
Reviewed-by: NSimon Glass <sjg@chromium.org>
Series-changes: 2
- Use CONFIG_DM_SCSI instead of mix of DM_SCSI and DM_SATA
  Ceva sata has never used sata commands that's why keep it in
  SCSI part only.
- Separate scsi_scan() for DM_SCSI and do not change cmd/scsi.c
- Extend platdata

Series-changes: 3
- Fix scsi_scan return path
- Fix header location uclass-internal.h
- Add scsi_max_devs under !DM_SCSI
- Add new header device-internal because of device_probe()
- Redesign block device creation algorithm
- Use device_unbind in error path
- Create block device with id and lun numbers (lun was there in v2)
- Cleanup dev_num initialization in block device description
  with fixing parameters in blk_create_devicef
- Create new Kconfig menu for SATA/SCSI drivers
- Extend description for DM_SCSI
- Fix Kconfig dependencies
- Fix kernel doc format in scsi_platdata
- Fix ahci_init_one - vendor variable

Series-changes: 4
- Fix Kconfig entry
- Remove SPL ifdef around SCSI uclass
- Clean ahci_print_info() ifdef logic

This patch adds DM support to the xHCI PCI driver. Enabling its use
e.g. in x86 platforms.

Status: On the congatec BayTrail SoM, xHCI still does not work
correctly with this patch. Some internal timeouts lead to resets (BUG).
Additional work is needed here. I'm posting this version as WIP so that
other developers interested in this support might use it as a start.
I might get back to it in a few weeks as well.
Signed-off-by: NStefan Roese <sr@denx.de>
Cc: George McCollister <george.mccollister@gmail.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Marek Vasut <marex@denx.de>
Reviewed-by: NSimon Glass <sjg@chromium.org>

Replace arch/arm/cpu/armv7/socfpga/ path with
arch/arm/mach-socfpga/ and removed board file path
since board/altera has different boards with relevant
board maintainers.

Cc: Marek Vasut <marex@denx.de>
Signed-off-by: NJagan Teki <jagan@openedev.com>

With the acquisition of Altera by Intel, my Altera email may be going
away soon. Update the contact to a more reliable address.
Signed-off-by: NDinh Nguyen <dinguyen@kernel.org>

Some Altera Quartus generated files have long lines that are split with a '\' at
the end of the line. It also wOn Windows, rites files in DOS format, which can
confuse some of the processing scripts in this file. This patch solves both issues.
Signed-off-by: NBill Randle <bill.randle@gmail.com>
Cc: Marek Vasut <marex@denx.de>

Valid Altera SoCFPGA preloader image must contain special data at
offsets 0x40, 0x44, 0x48 and valid instructions at address 0x4c or
0x50. These addresses are by default used by U-Boot's vector table
and a piece of reset handler, thus a valid preloader corrupts those
addresses slightly. While this works most of the time, this can and
does prevent the board from rebooting sometimes and triggering this
issue may even depend on compiler.

The problem is that when SoCFPGA performs warm reset, it checks the
addresses 0x40..0x4b in SRAM for a valid preloader signature and
header checksum. If those are found, it jumps to address 0x4c or
0x50 (this is unclear). These addresses are populated by the first
few instructions of arch/arm/cpu/armv7/start.S:

ffff0040 <data_abort>:
ffff0040:       ebfffffe        bl      ffff0040 <data_abort>

ffff0044 <reset>:
ffff0044:       ea000012        b       ffff0094 <save_boot_params>

ffff0048 <save_boot_params_ret>:
ffff0048:       e10f0000        mrs     r0, CPSR
ffff004c:       e200101f        and     r1, r0, #31
ffff0050:       e331001a        teq     r1, #26

Without this patch, the CPU will enter the code at 0xffff004c or
0xffff0050 , at which point the value of r0 and r1 registers is
undefined. Moreover, jumping directly to the preloader entry point
at address 0xffff0000 will also fail, because address 0xffff004.
is invalid and contains the preloader magic.

Add BOOT0 hook which reserves the area at offset 0x40..0x5f and
populates offset 0x50 with jump to the entry point. This way, the
preloader signature is stored in reserved space and can not corrupt
the SPL code.
Signed-off-by: NMarek Vasut <marex@denx.de>
Cc: Chin Liang See <clsee@altera.com>
Cc: Dinh Nguyen <dinguyen@opensource.altera.com>
Cc: Stefan Roese <sr@denx.de>
Tested-by: NDinh Nguyen <dinguyen@opensource.altera.com>

Add CycloneV based Terasic DE1-SoC board. The board boots
from SD/MMC. Ethernet and USB host is supported.
Signed-off-by: NAnatolij Gustschin <agust@denx.de>
Cc: Marek Vasut <marex@denx.de>

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

The default configuration for QSPI AHB bus can't support 16MB+.
But some flash on NXP layerscape board are more than 16MB.
Signed-off-by: NYuan Yao <yao.yuan@nxp.com>
Reviewed-by: NYork Sun <york.sun@nxp.com>

Enable the errata-a005697 for ls1012a
Signed-off-by: NChangming Huang <jerry.huang@nxp.com>
Reviewed-by: NYork Sun <york.sun@nxp.com>

Add the name for register space and memory space.
<0x1550000 0x10000 > is the QSPI register space.
<0x40000000 0x4000000> is the QSPI memory space.
Signed-off-by: NYuan Yao <yao.yuan@nxp.com>
Reviewed-by: NYork Sun <york.sun@nxp.com>

Signed-off-by: NPriyanka Jain <priyanka.jain@nxp.com>
Reviewed-by: NYork Sun <york.sun@nxp.com>

- add additional function erratum_a009942_check_cpo to check if the
  board needs tuning CPO calibration for optimal setting.
- move ERRATUM_A009942(with revision to check cpo_sample option) from
  fsl_ddr_gen4.c to ctrl_regs.c for reuse on all DDR4/DDR3 parts.
- move ERRATUM_A008378 from fsl_ddr_gen4.c to ctrl_regs.c
- remove obsolete ERRATUM_A004934 which is replaced with ERRATUM_A009942.
Signed-off-by: NShengzhou Liu <Shengzhou.Liu@nxp.com>
[YS: Replaced CONFIG_QEMU_E500 with CONFIG_ARCH_QEMU_E500]
Reviewed-by: NYork Sun <york.sun@nxp.com>

Fix following warning in case multiple erratum macro was not defined.
warning: unused variable 'tmp'
warning: unused variable 'ddr_freq'
Signed-off-by: NShengzhou Liu <Shengzhou.Liu@nxp.com>
Reviewed-by: NYork Sun <york.sun@nxp.com>

Its easier to watch the output of the build process when the platforms
specific boards are grouped in a separate job. This patch adds a job
for all mvebu boards (arm and aarch64).
Signed-off-by: NStefan Roese <sr@denx.de>
Cc: Tom Rini <trini@konsulko.com>
Reviewed-by: NTom Rini <trini@konsulko.com>