Roger Quadros [Fri, 12 Jan 2024 12:49:50 +0000 (14:49 +0200)]
arm: dts: k3-j721e-beagleboneai64: Fix USB operation
Without correct SERDES MUX and Lane control settings
USB0 will be broken. Set the MUX and Lane control devices
to be auto probed so they are configured correctly.
Roger Quadros [Fri, 12 Jan 2024 12:49:49 +0000 (14:49 +0200)]
arm: dts: k3-j721e: Fix USB0 operation
Without correct SERDES MUX and Lane control settings
USB0 will be broken. Set the MUX and Lane control devices
to be auto probed so they are configured correctly.
Fixes: 69b19ca67bcb ("arm: dts: k3-j721e: Sync with v6.6-rc1") Signed-off-by: Roger Quadros <rogerq@kernel.org>
Marek Vasut [Sat, 13 Jan 2024 17:57:27 +0000 (18:57 +0100)]
ARM: stm32: Set stdio to serial on DH STM32MP15xx DHSOM
In case CONSOLE_MUX and SYS_CONSOLE_IS_IN_ENV are enabled, the console
stdin, stdout, stderr must be defined in environment. Define the default
settings to fix the following warning on boot:
"
In: No input devices available!
Out: No output devices available!
Err: No error devices available!
"
Sort the default environment as well.
Signed-off-by: Marek Vasut <marex@denx.de> Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
Tom Rini [Fri, 19 Jan 2024 13:46:47 +0000 (08:46 -0500)]
Merge branch '2024-01-18-assorted-fixes'
- A number of OS boot related cleanups, a number of TI platform
fixes/cleanups, SMBIOS fixes, tweak get_maintainers.pl to report me
for more places, fix the "clean the build" pytest and add a bootstage
pytest, fix PKCS11 URI being omitted in some valid cases, make an iommu
problem easier to debug on new platforms, nvme and pci improvements,
refactor image-host code a bit, fix a typo in env setting, add a missing
dependency for CMD_LICENSE, and correct how we call getchar() in some
places.
Patrick Delaunay [Mon, 15 Jan 2024 14:05:56 +0000 (15:05 +0100)]
arm: Rename STM32MP13x
CONFIG options must not use lower-case letter. Convert this and related
ones to upper case.
Signed-off-by: Simon Glass <sjg@chromium.org> Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com> Reviewed-by: Igor Opaniuk <igor.opaniuk@foundries.io>
Patrick Delaunay [Mon, 15 Jan 2024 14:05:54 +0000 (15:05 +0100)]
board: st: stm32mp2: add checkboard
Implement the weak function checkboard to identify the used board with
compatible in device tree for the support of stm32mp2 STMicroelectronics
boards.
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com> Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Patrick Delaunay [Mon, 15 Jan 2024 14:05:53 +0000 (15:05 +0100)]
stm32mp: activate the command stboard for stm32mp25 boards
Activate the command stboard for stm32mp25 STMicroelectronics boards,
add the default used OTP identifier and the associated board identifier:
- stm32mp25xx-ev1 = MB1936
- stm32mp25xx-dk = MB1605
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com> Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
Patrick Delaunay [Mon, 15 Jan 2024 14:05:51 +0000 (15:05 +0100)]
smt32mp: add setup_mac_address for stm32mp25
Add a function setup_mac_address() to update the MAC address from the
default location in OTP for stm32mp2 platform.
The max number of OTP for MAC address is increased to 8 for STM32MP25,
defined with get_eth_nb() and checked in setup_mac_address.
The MAC address FF:FF:FF:FF:FF:FF, the broadcast ethaddr, is a invalid
value used for unused MAC address slot in OTP, for example for board
with STM32MP25x part number allows up to 5 ethernet ports but it is not
supported by the hardware, without switch; the associated variable
"enetaddr%d" is not created.
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com> Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Patrick Delaunay [Mon, 15 Jan 2024 14:05:46 +0000 (15:05 +0100)]
stm32mp: bsec: add driver data
Add driver data in BSEC driver to test presence of OP-TEE TA,
mandatory for STM32MP13 family and prepare the support of new device
with more OTP than 95.
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com> Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Patrick Delaunay [Mon, 15 Jan 2024 14:05:44 +0000 (15:05 +0100)]
arm64: dts: st: add bsec support to stm32mp25
Add BSEC support to STM32MP25 SoC family with SoC information:
- RPN = Device part number (BSEC_OTP_DATA9)
- PKG = package data register (Bits 2:0 of BSEC_OTP_DATA122)
Tom Rini [Fri, 19 Jan 2024 13:02:58 +0000 (08:02 -0500)]
Merge tag 'u-boot-rockchip-20240119' of https://source.denx.de/u-boot/custodians/u-boot-rockchip
- Add board: rk3328 FriendlyARM NanoPi R2C Plus, rk3588 Turing RK1 SoM;
- Enable SPI boot for rk3588 and rk3528;
- Set boot device in SPL as common code;
- other misc fixes;
Quentin Schulz [Thu, 18 Jan 2024 13:56:00 +0000 (14:56 +0100)]
rockchip: rk3128: remove noop file
arch_cpu_init is already returning 0 in its weak definition in
common/board_f.c so let's just remove the file entirely since nothing
else is done in it.
Quentin Schulz [Thu, 18 Jan 2024 13:55:58 +0000 (14:55 +0100)]
rockchip: find U-boot proper boot device by inverting the logic that sets it
BOOT_DEVICE_* is set by spl_node_to_boot_device() depending on the block
device number associated with the MMC device the SPL used to load U-Boot
proper from. It is NOT related to the mmc alias in the Device Tree.
For SPI flashes, all SPI flashes will return BOOT_DEVICE_SPI so there's
currently no way to know from which one the SPL loaded U-Boot proper
from. Therefore, let's just find the first valid candidate in
/chosen/u-boot,spl-boot-order that is a SPI flash and return that path.
This is a best effort.
While the original implementation may have worked, using the exact same
mechanism but in inverted fashion makes it less likely to have
surprising corner-cases or side-effects.
A nice side-effect is that all existing and future Rockchip SoCs now
automatically have their /chosen/u-boot,spl-boot-device set.
Quentin Schulz [Thu, 18 Jan 2024 13:55:57 +0000 (14:55 +0100)]
rockchip: factor out spl_perform_fixups into common spl-boot-order
All SoCs are susceptible to wanting to know which storage medium was
used to load U-Boot SPL. So instead of reimplementing the same functions
in SoCs over and over again (here just rk3399 and px30 but rk3588 is
coming), let's just put all this in common into spl-boot-order.c
allowing to support a new SoC just by defining the spl_boot_devices
array in the appropriate SoC file.
Note that spl_perform_fixups() now calls spl_image_fdt_addr() to get the
address of the fdt instead of directly reading the
spl_image_info->fdt_addr member, because that member is not guaranteed
to be present (guarded with compile flags). This is essential because we
move the logic away from px30 and rk3399 which had those compile flags
enabled to code run for all Rockchip SoCs.
Quentin Schulz [Thu, 18 Jan 2024 13:55:56 +0000 (14:55 +0100)]
rockchip: px30: simplify logic for getting SPL boot medium DT node
In preparation of moving spl_perform_fixups to spl-boot-order.c, let's
simplify the logic around mapping the BOOT_DEVICE_x enum index to a DT
node by using an instantiated array of chars instead of creating a new
data structure on the fly.
This will make it easier to factor out the code handling the SPL boot
medium detection by having spl_decode_boot_device common to all SoCs.
Quentin Schulz [Thu, 18 Jan 2024 13:55:55 +0000 (14:55 +0100)]
rockchip: rk3399: simplify logic for getting SPL boot medium DT node
In preparation of moving spl_perform_fixups to spl-boot-order.c, let's
simplify the logic around mapping the BOOT_DEVICE_x enum index to a DT
node by using an instantiated array of chars instead of creating a new
data structure on the fly.
This will make it easier to factor out the code handling the SPL boot
medium detection by having spl_decode_boot_device common to all SoCs.
Chris Morgan [Tue, 2 Jan 2024 15:46:54 +0000 (09:46 -0600)]
doc: board: anbernic: Update rgxx3 to add new boards
Update the RGxx3 documentation to note that it now supports the
RG-ARC-D, RG-ARC-S, Powkiddy RK2023, and Powkiddy RGB30. Also update
verbiage around panel detection to note that it is no longer hard coded
to the RG503.
Signed-off-by: Chris Morgan <macromorgan@hotmail.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Chris Morgan [Tue, 2 Jan 2024 15:46:53 +0000 (09:46 -0600)]
board: rockchip: Add support for new boards to RGxx3
Add support for the Anbernic RG-ARC-D, Anbernic RG-ARC-S, Powkiddy
RK2023, and Powkiddy RGB30 to the Anbernic RGxx3. While the Powkiddy
devices are manufactured by Powkiddy instead of Anbernic,
the hardware is so similar they can all use the same bootloader.
Signed-off-by: Chris Morgan <macromorgan@hotmail.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Chris Morgan [Tue, 2 Jan 2024 15:46:48 +0000 (09:46 -0600)]
board: rockchip: Refactor panel auto-detect code
Make the inability to detect a panel using the auto detection code not
fail the entire boot process. This means that if the panel ID cannot
be read we don't set an environment variable for the panel, and if an
environment variable for the panel is not set we don't attempt to
update the compatible string. Changes to the code also ensure that
when there are multiple compatible strings required for the panel
we use them both, which solves some issues that will pop up soon
for the Linux driver.
Signed-off-by: Chris Morgan <macromorgan@hotmail.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
U-Boot proper automatically modifies boot_targets to swap the order in
which MMC storage media are used for standard boot based on which MMC
storage medium was used to load U-Boot proper. This is however only done
if the user has not changed it manually, therefore a check between the
default and current value is done.
This used to work fine until the migration to standard boot where
boot_targets value size in the default environment went above the 32
characters that env_get_default function can return, thus resulting in a
truncated variable.
Therefore the check between default and current value would always fail.
By using the newly added env_get_default_into function, a buffer of
the appropriate size can be allocated on the stack to get the whole
value of boot_targets in the default environment and thus fixing the
check.
Quentin Schulz [Wed, 17 Jan 2024 17:59:12 +0000 (18:59 +0100)]
env: migrate env_get_default to call env_get_default_into
Since both functions share a similar goal and env_get_default_into can
do what env_get_default wants to do with specific arguments, let's make
env_get_default call env_get_default_into so as to avoid code
duplication.
Cc: Quentin Schulz <foss+uboot@0leil.net> Reviewed-by: Tom Rini <trini@konsulko.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com> Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Quentin Schulz [Wed, 17 Jan 2024 17:59:11 +0000 (18:59 +0100)]
env: allow to copy value from default environment into a buffer
env_get_default suffers from a particular issue int that it can only
return a value truncated to gd->env_buf (32) characters. This may be
enough for most variables but it isn't for others, so let's allow users
to provide a preallocated buffer to copy the value into instead,
allowing for more control, though it'll still be truncated if the value
size is bigger than the preallocated buffer.
Cc: Quentin Schulz <foss+uboot@0leil.net> Reviewed-by: Tom Rini <trini@konsulko.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com> Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Quentin Schulz [Wed, 17 Jan 2024 17:59:10 +0000 (18:59 +0100)]
rockchip: ringneck-px30/puma-rk3399: factor out storage medium selection
Ringneck PX30 and Puma RK3399 both have the same expectation with regard
to bootstd device order and U-Boot environment storage device, except
that Puma RK3399 also supports SPI Flash.
Let's move all of this into a common file where common logic can be put.
Quentin Schulz [Wed, 17 Jan 2024 17:59:09 +0000 (18:59 +0100)]
rockchip: puma-rk3399: do not hardcode MMC controller paths
To prepare to put the similar logic around storage medium selection for
Ringneck PX30 and Puma RK3399 in common, let's not use hardcoded paths
but use uclass functions instead to find udevice based on their DT node
full path.
Quentin Schulz [Wed, 17 Jan 2024 17:59:08 +0000 (18:59 +0100)]
rockchip: ringneck-px30: do not hardcode MMC controller paths
To prepare to put the similar logic around storage medium selection for
Ringneck PX30 and Puma RK3399 in common, let's not use hardcoded paths
but use uclass functions instead to find udevice based on their DT node
full path.
Jonas Karlman [Fri, 17 Nov 2023 23:24:34 +0000 (23:24 +0000)]
rockchip: rk3588: Fix boot from SPI flash
The commit fd6e425be243 ("rockchip: rk3588-rock-5b: Enable boot from SPI
NOR flash") added a new BROM_BOOTSOURCE_SPINOR_RK3588 with value 6.
At the time the reason for this new bootsource id value 6 was unknown.
We now know that the BootRom on RK3588 use different bootsource id
values depending on the iomux used by the flash spi controller, and not
by the type of spi nor or spi nand flash used.
Add the following enum values and use them for RK3588 boot_devices.
Hugh Cole-Baker [Tue, 26 Dec 2023 16:43:30 +0000 (16:43 +0000)]
rockchip: rk35xx: expand space for decompressed kernel
An uncompressed 6.7.0-rc1 Linux kernel Image built with the arm64
defconfig is about 40MB. This does not fit in to the space between
kernel_comp_addr_r and fdt_addr_r, so when uncompressing an Image.gz
to this size, the FDT will be overwritten. Rearrange addresses to have
128MiB for the kernel and its decompression buffer, then devicetree,
overlay and ramdisk at the end.
Signed-off-by: Hugh Cole-Baker <sigmaris@gmail.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Jonas Karlman [Sun, 12 Nov 2023 17:48:58 +0000 (17:48 +0000)]
rockchip: board: Remove dwc3 usb init and gadget handler functions
Remove board_usb_init() and dm_usb_gadget_handle_interrupts() functions
related to dwc3, they use e.g. a hard-coded reg address for RK3399 and
are obsolete with use of DM_USB_GADGET.
Use of DM_USB_GADGET, USB_DWC3_GENERIC and USB_DWC3_GADGET have replaced
same feature provided by the removed functions on RK3399 boards.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Jonas Karlman [Sun, 12 Nov 2023 17:48:57 +0000 (17:48 +0000)]
configs: rockchip: Use dwc3-generic driver on RK3328 and RK3399
Complete the transition away from xhci-dwc3 and dwc3-of-simple drivers
and change to use the dwc3-generic driver on remaining RK3328 and RK3399
boards.
MISC, USB_DWC3 and USB_DWC3_GENERIC is enabled on boards that used to
enable USB_XHCI_DWC3. USB_XHCI_DWC3 is dropped from updated boards along
with the default y of USB_XHCI_DWC3_OF_SIMPLE.
There is no intended change in functionality with this changes, USB 3.0
is expected to continue same as before this change.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se> Reviewed-by: Marek Vasut <marex@denx.de> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Anthony Loiseau [Thu, 11 Jan 2024 16:51:27 +0000 (17:51 +0100)]
get_maintainer.pl: slightly modify penguin_chief
Penguin chiefs are actually stripped out from get_maintainer results unless
--git-chief-penguins is provided, which appends them to all results instead.
This is a issue for U-Boot since Tom Rini (penguin chief) is also maintainer
of some sub-trees ("ARM", "ARM TI" and "THE REST").
Hopefully, this match is performed case-senditive over name and email,
therefore upper-casing Tom Rini last name workarounds the issue.
Important note: This also fixes "THE REST" catch all section resulting
in Tom Rini listed as (maintainer:THE REST) for all files. Hope there is not
too much developers scripting their patch email header from get_maintainer.pl
output, otherwise Tom Rini may be flooded.
Cc: Tom Rini <trini@konsulko.com> Signed-off-by: Anthony Loiseau <anthony.loiseau@allcircuits.com>
Ilias Apalodimas [Thu, 11 Jan 2024 14:39:36 +0000 (16:39 +0200)]
smbios: fix matching issues for table types
commit 738b34668f28 ("smbios: Fallback to the default DT if sysinfo nodes are missing")
allowed the code to fallback and fill in SMBIOS tables with properties
from the compatible and product nodes of the DT, in case the
'smbios,sysinfo' node is missing.
That works fine for Type1/2 tables, but for other types we need to
match the smbios,sysinfo subnode name as well. So add it to the
smbios_ctx and check it during the sysinfo <-> DT mathcing
Ivan Orlov [Thu, 11 Jan 2024 12:58:19 +0000 (12:58 +0000)]
cmd: license: Add CONFIG_GZIP dependency
'License' command processing code could be successfully compiled only
when CONFIG_GZIP option is enabled, otherwise it can't find the 'gunzip'
function definition (it is defined in lib/gunzip.c).
Add CONFIG_GZIP dependency to 'license' command config option in the
corresponding Kconfig.
Signed-off-by: Ivan Orlov <ivan.orlov@codethink.co.uk>
cmd: bootflow: remove dead code in do_bootflow_cmdline()
Checking argc < 3 twice is redundant.
Addresses-Coverity-ID: 477198 Logically dead code Fixes: 921f63e57238 ("bootflow: Allow setting a cmdline arg with no value") Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
A SMBIOS 3 entry point has a different length than an SMBIOS 2.1 entry
point.
Fixes: 70924294f375 ("smbios: Use SMBIOS 3.0 to support an address above 4GB") Fixes: 1c5f6fa3883d ("smbios: Drop support for SMBIOS2 tables")
Addresses-Coverity-ID: 477212 ("Wrong sizeof argument") Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Tom Rini [Tue, 9 Jan 2024 22:57:16 +0000 (17:57 -0500)]
getchar(): Correct usage
The function getchar() returns an 'int' and not a 'char'. Coverity notes
that "Assigning the return value of getchar to char ... truncates its value."
and so for the most part we can resolve this easily by using 'int' as
intended, and often used throughout the codebase. A few places are not
so simple and would require further re-architecting of the code in order
to change this, so we leave them be.
Sekhar Nori [Tue, 9 Jan 2024 08:45:51 +0000 (14:15 +0530)]
arch: mach-k3: Fix incorrect mapping of higher DDR addresses as device memory
Entry for physical address 0x500000000 in memory map table for MMU
configuration is spilling over and inadvertently making DDR available at
higher address (above 4GB address space) get mapped as device memory
(nGnRnE).
Fix this by adjusting entry size. Tested on AM62A SK. Before this patch:
With commit 42c0e5bb054d ("test: Find leftovers after clean/mrproper")
we assume that we are performing out of tree builds. test_clean and
test_mrproper fail if the assumption isn't met.
However there are valid scenarios where tests run with in source builds
(i.e., test.py --build-dir=.).
Automatically skip cleanup tests in such scenarios to prevent false
positives.
Signed-off-by: Tobias Deiminger <tdmg@linutronix.de> Reviewed-by: Harald Seiler <u-boot@rahix.de>
Hugo Cornelis [Mon, 8 Jan 2024 14:24:30 +0000 (15:24 +0100)]
image-host: refactor and protect for very long filenames
This patch adds a function fit_image_read_key_iv_data that checks the
return value of snprintf and allows to generate a sensible error
message when generating binary images using filenames that are too
long for the OS to handle.
This is especially relevant for automated builds such as Buildroot and
Yocto builds.
Signed-off-by: Hugo Cornelis <hugo.cornelis@essensium.com>
Clean up all configuration files to adhere to yamllint rules.
Signed-off-by: Neha Malcom Francis <n-francis@ti.com>
[trini: Update more yaml files added since this was posted] Signed-off-by: Tom Rini <trini@konsulko.com> Suggested-by: Nishanth Menon <nm@ti.com>
Caleb Connolly [Thu, 4 Jan 2024 17:12:22 +0000 (17:12 +0000)]
iommu: dont fail silently
When attempting to probe a device which has an associated IOMMU, if the
IOMMU device can't be found (no driver, disabled driver, driver failed
to probe, etc) then we currently fail to probe the device with no
discernable error.
If we fail to hook the device up to its IOMMU, we should make sure that
the user knows about it. Write some better error messages for
dev_iommu_enable() to facilitate this.
Caleb Connolly [Thu, 4 Jan 2024 16:03:35 +0000 (16:03 +0000)]
bootdev: avoid infinite probe loop
Sometimes, when only one bootdev is available, and it fails to probe, we
end up in an infinite loop calling probe() on the same device over and
over. With only debug level log output.
Break the loop if we fail to probe the same device twice in a row, and
promote the probe failure message to log_warning().
Tom Rini [Thu, 18 Jan 2024 16:38:25 +0000 (11:38 -0500)]
Revert "test: hush: dollar: fix bugous behavior"
What we were doing here amounts to simplifying the code for the new hush
parser, and the workarounds are required when using the old one. As we
have returned to having the old parser be default for now, we must undo
this so that the test passes again.
Tom Rini [Thu, 18 Jan 2024 14:07:25 +0000 (09:07 -0500)]
Revert "cmd: Set modern hush as default shell"
Currently at least two SoC families are able to cause a crash using
their regular boot scripts, with the new parser. For now, revert to the
old parser as default.
Tom Rini [Wed, 17 Jan 2024 22:12:36 +0000 (17:12 -0500)]
test: test_trace.py: Have trace-cmd report always use -l
With newer versions of trace-cmd the report subcommand will have
different output from before if we do or do not pass -l. However, with
the -l flag passed our put is consistent here. This updates our regular
expressions and comments for this as well now.
QEMU commit 0c7ffc977195 ("hw/net: cadence_gem: Fix MDIO_OP_xxx values")
is needed to fix the Ethernet PHY driver used by the emulated SiFive
Unleashed Board emulation.
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Tom Rini [Wed, 17 Jan 2024 14:27:43 +0000 (09:27 -0500)]
Merge tag 'xilinx-for-v2024.04-rc1-v2' of https://source.denx.de/u-boot/custodians/u-boot-microblaze
Xilinx changes for v2024.04-rc1-v2
xilinx:
- Enable NFS, WGET, DNS and BLKMAP by default
zynqmp:
- Support new power-management node
- Remove multiple blank lines from DTSes
- Wire multiboot with DFU infrastructure
- Fix i2c-gpio pinctrl group name
- SOM DT changes (phy on kd240, kv260 cleanups
- Cleanup i2c bus on zcu1285
- DT cleanup (fix node names not to use _)
- Fix USB interrupts
- Cleanup zcu100 DT
- Add support for kaslr-seed
Tom Rini [Wed, 17 Jan 2024 14:27:23 +0000 (09:27 -0500)]
Merge tag 'efi-2024-04-rc1-2' of https://source.denx.de/u-boot/custodians/u-boot-efi
Documentation:
* add generated index to table of contents
* create index entries for commands
* update Python packages used to build the documentation
* fix typos in dfu documentation
- Add more pytests to exercise functionality on real hardware, cleanup
and add tests around "cp"/memmove, add phyCORE-AM62x and Phytium
Pe2201 platforms, Nuvoton NPCM BMC reset driver and improve QEMU
SMBIOS support
The first sentence is half-way true; the macro is always defined, but
has the value 0 or 1.
The second is outright false. A lot of code guarded by
MEM_SUPPORT_64BIT_DATA uses a "ulong" to store values, so if
sizeof(long) is not 8, that code would probably compile, but not work
at all as expected.
It would probably be possible to make all such code explicitly use u64
and thus make it work on 32 bit targets, but until that is done, do
not pretend that it's ok to override the automatic value of
MEM_SUPPORT_64BIT_DATA.