In parallel mode, the current implementation assumes that a maximum of
two flashes are connected. The QSPI controller splits the data evenly
between both the flashes so, both the flashes that are connected in
parallel mode should be identical.
During each operation SPI-NOR sets 0th bit for CS0 & 1st bit for CS1 in
nor->flags.

In stacked mode the current implementation assumes that a maximum of two
flashes are connected and both the flashes are of same make but can
differ in sizes. So, except the sizes all other flash parameters of both
the flashes are identical

Spi-nor will pass on the appropriate flash select flag to low level
driver, and it will select pass all the data to that particular flash.

Write operation in parallel mode are performed in page size * 2 chunks as
each write operation results in writing both the flashes. For doubling
the address space each operation is performed at addr/2 flash offset,
where addr is the address specified by the user.

Similarly for read and erase operations it will read from both flashes,
so size and offset are divided by 2 and send to flash.

Adding the config option SPI_ADVANCE for non SPL code.

Signed-off-by: Ashok Reddy Soma <ashok.reddy.soma@amd.com>
Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>

By default flash lock option is enabled, enable this option only
when it is required. By disabling the lock config will save some
amount of memory.

Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>

Enable simple-pm-bus driver to handle case where axi bus coming between PS
(fixed) part to PL (programmable) part has own clock or power domain.
That's why enable driver to be ready for this configuration.

Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/b9f4bb85be502616edf3be2b79e52a0e2c03e821.1725349691.git.michal.simek@amd.com

To boot into the firmware recovery tool, the user currently
needs to press a button on the board while powering the
system up. To simplify this process, a U-Boot command
was added to allow booting directly into the recovery tool.

For example:
ZynqMP> zynqmp reboot <multiboot offset in hex>

Co-develop-by: Prasad Kummari <prasad.kummari@amd.com>
Signed-off-by: Prasad Kummari <prasad.kummari@amd.com>
Co-develop-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20240827115529.2931334-1-prasad.kummari@amd.com

Included the SPI header to resolve the no previous prototypes
for function. Removed unused mode variable.
sparse warnings
warning: no previous prototype for 'spi_get_env_dev'
[-Wmissing-prototypes]
warning: variable 'mode' set but not used
[-Wunused-but-set-variable]

Signed-off-by: Prasad Kummari <prasad.kummari@amd.com>
Link: https://lore.kernel.org/r/20240905115758.999936-1-prasad.kummari@amd.com


Signed-off-by: Michal Simek <michal.simek@amd.com>

This patch fixes timeout issues seen on large NOR flash.
For full-chip erase, where we use the SPINOR_OP_CHIP_ERASE (0xc7)
opcode. Use a different timeout for full-chip erase than for other
commands.

 [Ported from Linux kernel commit
                09b6a377687b ("mtd: spi-nor: scale up timeout for
                               full-chip erase") ]

Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>

Merge tag 'u-boot-imx-next-20240919' of https://gitlab.denx.de/u-boot/custodians/u-boot-imx into next

CI: https://source.denx.de/u-boot/custodians/u-boot-imx/-/pipelines/22363

- Several updates to i.MX9 SOC and i.MX93 EVK.
- Power domain fixes.
- TRDC cleanup and update.
- MAC address layout update.
- Add support for the i.MX9301/9302 variants.
- Add runtime detection of voltage mode.
- Generalize some code for i.MX8M and i.MX9.
- Add support for Comvetia imx6q-lxr board.

This PR contains various improvements in the A/B update logic for EFI

- Read both copies of metadata, in case one of the is corrupted
- Check the metadata version against the running firmware to make sure it's
  allowed
- Limit the use of a revert capsule if the board is on a trial state and
  make sure it's not applied if the max counter has expired

Sughosh Ganu authored 6 months ago and Ilias Apalodimas committed 6 months ago

When dealing with processing of the empty capsule, the capsule gets
applied only when the checks for the empty capsule pass. Print a
message to highlight if empty capsule checks fail, and return an error
value, similar to the normal capsules.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Tested-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

Sughosh Ganu authored 6 months ago and Ilias Apalodimas committed 6 months ago

When in Trial State, the platform keeps a count of the number of times
it has booted in the Trial State. Once the threshold of the maximum
allowed count exceeds, the platform reverts to boot from a different
bank on subsequent boot, thus coming out of the Trial State. It is
expected that all the updated images would be accepted or rejected
while the platform is in Trial State. Put in checks so that it is not
possible to apply an empty capsule once the max Trial Count exceeds.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Tested-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

Sughosh Ganu authored 6 months ago and Ilias Apalodimas committed 6 months ago

The FWU code supports both versions of the FWU metadata, i.e. v1 and
v2. A platform can then select one of the two versions through a
config symbol. Put a dependency in the FWU metadata version selection
config symbol to ensure that both versions of the metadata cannot be
enabled.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Reviewed-by: Michal Simek <michal.simek@amd.com>
Tested-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

Sughosh Ganu authored 6 months ago and Ilias Apalodimas committed 6 months ago

The platform transitions out of Trial State into the Regular State
only when all the images in the update bank have been accepted. Check
for this condition before transitioning out of Trial State.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Tested-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

Sughosh Ganu authored 6 months ago and Ilias Apalodimas committed 6 months ago

Do a sanity check that the version of the FWU metadata that has been
read aligns with the version enabled in the image. This allows to
indicate an early failure as part of the FWU module initialisation.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Tested-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

Sughosh Ganu authored 6 months ago and Ilias Apalodimas committed 6 months ago

In the version 2 of the FWU metadata, the metadata is broken into two
parts, a top-level structure, which provides information on the total
size of the structure among other things. Try reading the primary
partition first, and if that fails, try reading the secondary
partition. This will help in the scenario where the primary metadata
partition has been corrupted, but the secondary partition is intact.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Tested-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

Add support for the Comvetia i.MX6Q LXR2 board, which is
uses the Phytec PFLA02 SoM.

Based on the original work from Stefano Babic <sbabic@denx.de>.

The Phytec PFLA02 devicetrees are taken from kernel 6.11-rc7.

The imx6q-lxr.dts has been submitted upstream:

https://lore.kernel.org/linux-devicetree/20240913200906.1753458-3-festevam@gmail.com/



After it gets accepted in mainline (most likely in kernel 6.13),
the lxr2 board can then be switched to OF_UPSTREAM and these device trees
can be removed from U-Boot.

Signed-off-by: Fabio Estevam <festevam@denx.de>

Add back low drive mode 1866mts ddr timing file, no need
CONFIG_IMX9_LOW_DRIVE_MODE anymore, using runtime selection.

Signed-off-by: Peng Fan <peng.fan@nxp.com>

Remove unused CONFIG_IMX9_LOW_DRIVE_MODE kconfig and
imx93_11x11_evk_ld_defconfig.
Remove the ld timing file.
The LD mode support will be added back with runtime detection later.

Signed-off-by: Peng Fan <peng.fan@nxp.com>

1. Use runtime voltage selection for LD/OD/ND mode.
2. According to latest PE/TE report, the voltages of VDD_SOC for
   LD and ND mode need add 50mv margin, so LD voltage is 0.75v->0.8v,
   ND voltage is 0.8v->0.85v.
3. Use TOFF_DEB to differentiate new trimmed pmic and old pmic

Signed-off-by: Peng Fan <peng.fan@nxp.com>

Add trdc_mbc_blk_num to get num blks in a MBC mem slot, then drop
the hardcoded value '40' for NIC OCRAM configuration.

Signed-off-by: Peng Fan <peng.fan@nxp.com>

Replace magic number with meaningful macros.

Signed-off-by: Peng Fan <peng.fan@nxp.com>

i.MX8M and i.MX9 have duplicated fixup_thermal_trips, so move it
to arch/arm/mach-imx/fdt.c to avoid duplicated code.

The critial temperature point for i.MX9 set to "maxc - 5" back to give
some margin.

Signed-off-by: Peng Fan <peng.fan@nxp.com>

Add support for iMX93 low performance parts 9302 and 9301 which
restrict to low drive voltage only.
The parts run A55 max speed at 900Mhz and M33 at 133Mhz, have NPU
and A55 core1 (9301) disabled.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>

Disale CPU1 for i.MX93 variants that only has one A55 core and update
cooling maps.

Signed-off-by: Peng Fan <peng.fan@nxp.com>

disable_cpu_nodes could be reused by i.MX9, so move disable_cpu_nodes
out from mach-imx/imx8m/soc.c to mach-imx/fdt.c and update
disable_cpu_nodes to make it easy to support different socs.

Signed-off-by: Peng Fan <peng.fan@nxp.com>

i.MX8M use PPI for PMU interrupts, there is no reason to update
interrupt-affinity for PMU even interrupt-affinity was wrongly added
to device tree before.

Signed-off-by: Peng Fan <peng.fan@nxp.com>

To support 1.866GTS LPDDR4x timing script, need to add 233Mhz freq
to DDR PLL for second mission point at 933MTS. Otherwise DDR training
will fail.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>

Normally, the wdog will be used for trigger external PMIC reset
through the WDOG_ANY pin. If the PMIC chip has debounce logic for
the reset signal, in some corner case the wdog can NOT trigger
external PMIC reset if the SoC has been reset internal before the
PMIC captures the WDOG_ANY pin reset, so need to keep the WDOG3-5
reset masked in the SRC to let the PMIC to do the reset safely.

Reviewed-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Jacky Bai <ping.bai@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>

Since we use SPEED GRADE fuse to set A55 frequency, remove the
set_arm_core_low_drive_clk function which has hard coded frequency.
And adjust clock_init called sequence and split it to early and late
functions.
Set the authen register in early function, because CCF driver checks
NS bit.
Set bus and core clock in late function, because the fuse read and
SoC type/rev depend on ELE.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>

Replace the static CONFIG_IMX9_LOW_DRIVE_MODE with runtime target
voltage mode by checking the part's SPEED GRADE fuse.
SPL will configure to highest A55 speed which is indicated by the SPEED
fuse and select corresponding voltage mode.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>

The boot image includes Edgelock Enclave(ELE) Firmware. Print the
information out to let user know which version firmware is being used.

Signed-off-by: Peng Fan <peng.fan@nxp.com>

The second Ethernet MAC (eQOS) fuse layout is changed since i.MX93 A1
following other i.MX platforms, for example i.MX8MP.

Order for A0:
MAC1_ADDR[15:0]
MAC1_ADDR[31:16]
MAC1_ADDR[47:32]
MAC2_ADDR[47:32]
MAC2_ADDR[15:0]
MAC2_ADDR[31:16]

Order since A1:
MAC1_ADDR[15:0]
MAC1_ADDR[31:16]
MAC1_ADDR[47:32]
MAC2_ADDR[15:0]
MAC2_ADDR[31:16]
MAC2_ADDR[47:32]

Signed-off-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>

To support OSCCA enabled part which has disabled FSB access from SOC,
change directly read from FSB to use fuse_read API.

Signed-off-by: Peng Fan <peng.fan@nxp.com>

Print UID in big endian format and as one buffer of bytes, so customer
can directly use it for EdgeLock 2GO.

Before:
UID: 0xf6c8ae93 0x0f46b326 0x10d61eb3 0x0583c2d2

Become:
UID: 93aec8f626b3460fb31ed610d2c28305

Signed-off-by: Ye Li <ye.li@nxp.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>

ROM use UID[0] and UID[1] as serial number with big endian when usb serial
download.

After update this, uuu(>1.6) can use below command to filter out devices
when multi boards connected.

uuu -ms <serial#> ...

[sudo] uuu -lsusb can list known devices with serial# informaiton.

Signed-off-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>

On OSCCA part, M33 TCM is used for ROM PATCH and protected by ELE ROM.
So after release TRDC, we need to configure TRDC for M33 TCM,
otherwise A55 can't access the TCM.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Ye Li <ye.li@nxp.com>

SSAR handshake done means power on finished, not ISO done. so correct
the waiting mask.

Fixes: 0256577a ("imx: imx9: Add MIX power init")
Signed-off-by: Peng Fan <peng.fan@nxp.com>

Simon Glass <sjg@chromium.org> says:

This series includes the patches needed to make make the EFI 'boot' test
work. That test has now been split off into a separate series along with
the EFI patches.

This series fixes these problems:
- sandbox memory-mapping conflict with PCI
- the fix for that causes the mbr test to crash as it sets up pointers
  instead of addresses for its 'mmc' commands
- the mmc and read commands which cast addresses to pointers
- a tricky bug to do with USB keyboard and stdio
- a few other minor things

The test currently runs twice as it is declared twice. Unwind this.

Signed-off-by: Simon Glass <sjg@chromium.org>

The normal approach with sandbox is to use a fixed memory address in the
RAM, to avoid needing to create a map for transient local variables.

Update this test to use this approach.

Signed-off-by: Simon Glass <sjg@chromium.org>

It isn't that important to factor out constants in tests, but in this
case we have 0x200 and 512 used. The commands don't use the constant
as they use a block count ('1'). It doesn't create more code to use a
constant, so create one.

Signed-off-by: Simon Glass <sjg@chromium.org>