With commit ce39ee28ec ("zynqmp: Do not place u-boot to reserved memory
location"), the function board_get_usable_ram_top() is allocating
MMU_SECTION_SIZE of about 2MB using lmb_alloc(). But we dont have this
much memory in case of mini U-Boot.
Keep these functions which use lmb under CONFIG_LMB so that they are
compiled and used only when LMB is enabled.
Signed-off-by: Ashok Reddy Soma <ashok.reddy.soma@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/75e52def75f573e554a6b177a78504c128cb0c4a.1657183534.git.michal.simek@amd.com
cpu:
- Add driver for microblaze cpu
net:
- Add support for DM_ETH_PHY to AXI emac and emaclite
xilinx:
- Switch platforms to DM_ETH_PHY
- DT chagnes in ZynqMP and Zynq
- Enable support for SquashFS
zynqmp:
- Add support for KR260 boards
- Move BSS from address 0
- Move platform identification from board code to soc driver
- Improve zynqmp_psu_init_minimize
versal:
- Enable loading app at EL1
serial:
- Setup default address and clock rates for DEBUG uarts
pinctrl:
- Add support for tri state and output enable properties
relocate-rela:
- Clean relocate-rela implementation for ARM64
- Add support for Microblaze
microblaze:
- Add support for runtime relocation
- Rework cache handling (wiring, Kconfig) based on cpuinfo
- Remove interrupt support
timer:
- Extract axi timer driver from Microblaze to generic location
-----BEGIN PGP SIGNATURE-----
iF0EABECAB0WIQQbPNTMvXmYlBPRwx7KSWXLKUoMIQUCYrlYngAKCRDKSWXLKUoM
ITgbAJ9S9xO2QqxtuodWAYMtJfvZ14c7mgCeKnyFTrrBnJkC0wPsGqE71oNJ49o=
=3gGm
-----END PGP SIGNATURE-----
Merge tag 'xilinx-for-v2022.10' of https://source.denx.de/u-boot/custodians/u-boot-microblaze into next
Xilinx changes for v2022.10
cpu:
- Add driver for microblaze cpu
net:
- Add support for DM_ETH_PHY to AXI emac and emaclite
xilinx:
- Switch platforms to DM_ETH_PHY
- DT chagnes in ZynqMP and Zynq
- Enable support for SquashFS
zynqmp:
- Add support for KR260 boards
- Move BSS from address 0
- Move platform identification from board code to soc driver
- Improve zynqmp_psu_init_minimize
versal:
- Enable loading app at EL1
serial:
- Setup default address and clock rates for DEBUG uarts
pinctrl:
- Add support for tri state and output enable properties
relocate-rela:
- Clean relocate-rela implementation for ARM64
- Add support for Microblaze
microblaze:
- Add support for runtime relocation
- Rework cache handling (wiring, Kconfig) based on cpuinfo
- Remove interrupt support
timer:
- Extract axi timer driver from Microblaze to generic location
Move axi timer driver from Microblaze to generic location.
Origin implementation was irq based with counting down timer.
CONFIG_TIMER drivers are designed differently that timer is free running up
timer with automatic reload without any interrupt.
Information about clock rates are find out in timer_pre_probe() that's why
there is no need to get any additional information from DT in the driver
itself (only register offset).
Signed-off-by: Michal Simek <michal.simek@amd.com>
Tested-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/6c12fc86bbc1f17d05c25018862e7b7b03346b36.1654684731.git.michal.simek@amd.com
Add a basic CPU driver that retrieves information about the microblaze CPU
core. cpu_ops handlers are implemented so that the "cpu" command can work
properly:
U-Boot-mONStR> cpu list
0: cpu@0 MicroBlaze @ 50MHz, Rev: 11.0, FPGA family: zynq7000
U-Boot-mONStR> cpu detail
0: cpu@0 MicroBlaze @ 50MHz, Rev: 11.0, FPGA family: zynq7000
ID = 0, freq = 50 MHz: L1 cache, MMU
Note: cpu_ver_lookup[] and family_string_lookup[] arrays were imported from
linux.
Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Link: https://lore.kernel.org/r/20220531181435.3473549-14-ovpanait@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Add helper code for PVR (Processor Version Register) data handling. It
will be used by the UCLASS_CPU driver to populate cpuinfo fields at
runtime.
Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Link: https://lore.kernel.org/r/20220531181435.3473549-13-ovpanait@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Align microblaze with the other architectures and provide an
implementation for flush_dcache_range(). Also, remove the microblaze
exception in drivers/core/device.c.
Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Link: https://lore.kernel.org/r/20220531181435.3473549-11-ovpanait@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Introduce a minimal cpuinfo structure to hold cache related info. The
instruction/data cache size and cache line size are initialized early in
the boot to default Kconfig values. They will be overwritten with data
from PVR/dtb if the microblaze UCLASS_CPU driver is enabled.
The cpuinfo struct was placed in global_data to allow the microblaze
UCLASS_CPU driver to also run before relocation (initialized global data
should be read-only before relocation).
gd_cpuinfo() helper macro was added to avoid volatile
"-Wdiscarded-qualifiers" warnings when using the pointer directly.
Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Link: https://lore.kernel.org/r/20220531181435.3473549-10-ovpanait@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com> (s/bralid/brlid/)
All flush_cache() calls in microblaze code are supposed to flush the
entire instruction and data caches, so introduce flush_cache_all()
helper to handle this.
Also, provide implementations for flush_dcache_all() and
invalidate_icache_all() so that icache and dcache u-boot commands can
work.
Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Link: https://lore.kernel.org/r/20220531181435.3473549-9-ovpanait@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Replace XILINX_DCACHE_BYTE_SIZE macro with two Kconfig symbols for
instruction and data caches sizes, respectively:
CONFIG_XILINX_MICROBLAZE0_ICACHE_SIZE
CONFIG_XILINX_MICROBLAZE0_DCACHE_SIZE
Also, get rid of the hardcoded value in icache_disable().
Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Link: https://lore.kernel.org/r/20220531181435.3473549-8-ovpanait@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com> (s/bralid/brlid/g)
Factor out icache/dcache components from flush_cache() function. Call the
newly added __flush_icache()/__flush_dcache() functions inside
icache_disable() and dcache_disable(), respectively. There is no need to
flush both caches when disabling a particular cache type.
Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Link: https://lore.kernel.org/r/20220531181435.3473549-7-ovpanait@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Replace CONFIG_ICACHE with a Kconfig option more limited in scope -
XILINX_MICROBLAZE0_USE_WIC. It should be enabled if the processor supports
the "wic" (Write to Instruction Cache) instruction. It will be used to
guard "wic" invocations in microblaze cache code.
Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Link: https://lore.kernel.org/r/20220531181435.3473549-6-ovpanait@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Replace CONFIG_DCACHE with a Kconfig option more limited in scope -
XILINX_MICROBLAZE0_USE_WDC. It should be enabled if the processor supports
the "wdc" (Write to Data Cache) instruction. It will be used to guard
"wdc" invocations in microblaze cache code.
Also, drop all ifdefs around flush_cache() calls and only keep one
CONFIG_IS_ENABLED() guard within flush_cache() itself.
Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Link: https://lore.kernel.org/r/20220531181435.3473549-5-ovpanait@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
XILINX_USE_DCACHE macro was removed in 7556fa09e0 ("microblaze: Simplify
cache handling"), but it was still used in a couple of places.
Replace those occurences with CONFIG_DCACHE.
Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Link: https://lore.kernel.org/r/20220531181435.3473549-4-ovpanait@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Microblaze is using NEEDS_MANUAL_RELOC from the beginnging. This is causing
issues with function pointer arrays which need to be updated manually after
relocation. Building code with -fPIC and linking with -pic will remove this
limitation and there is no longer need to run manual update.
By default still old option is enabled but by disabling NEEDS_MANUAL_RELOC
code will be compiled for full relocation.
The patch does couple of things which are connected to each other.
- Define STATIC_RELA dependency to call relocate-rela to fill sections.
- REMAKE_ELF was already enabled but u-boot file can't be used because
sections are empty. relocate-rela will fill them and output file is
u-boot.elf which should be used.
- Add support for full relocation (u-boot.elf)
- Add support for early relocation when u-boot.bin is loaded to different
address then CONFIG_SYS_TEXT_BASE
- Add rela.dyn and dynsym sections
Disabling NEEDS_MANUAL_RELOC U-Boot size increased by 10% of it's original
size (550kB to 608kB).
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/a845670b34925859b2e321875f7588a29f6655f9.1655299267.git.michal.simek@amd.com
Symbol handling depends on compilation flags. Right now manual relocation
is used that's why symbols can be referenced just by name and there is no
need to find them out. But when position independent code (PIC) is used
symbols need to be described differently. That's why having one macro
change is easier than changing the whole code.
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/d704e9a267c8b536452fb999111dbfbc9d652be5.1655299267.git.michal.simek@amd.com
Right now U-Boot runs all the time from the same address where it is loaded
but going to full relocation code starting address doesn't need to be fixed
and can be simply discovered from reading PC register. That's why use r20
to get PC address and subtract offset from the beginning to get starting
address.
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/044b727c33dfbe662f68512d0da0775a4805f360.1655299267.git.michal.simek@amd.com
Enable u-boot.elf recreation from u-boot.bin to prepare for removing manul
relocation. Enable option for big endian configuration but it is not used
too much that's why it is completely untested.
By supporting this system there is a need to define LITTLE/BIG endian
Kconfig options to pass -EL/-EB flags.
Full command line for u-boot.elf recreation looks like this:
microblazeel-xilinx-linux-gnu-objcopy -I binary -B microblaze \
-O elf32-microblazeel u-boot.bin u-boot-elf.o
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/7e242a519fcd1c693b9103c5599b515af555ca43.1655299267.git.michal.simek@amd.com
CONFIG_SYS_INIT_SP_OFFSET macro place stack to TEXT_BASE - SYS_MALLOC_F_LEN
but there is no reason to do it now because board_init_f_alloc_reserve()
returns exact location where stack should be. That's why stack location is
calculated at run time and there is no need to hardcode it via macro. This
change will help with placing U-Boot to any address.
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/e9aee69646e022fd8a96cbee2d2a07ab81fb6e05.1655299267.git.michal.simek@amd.com
When U-Boot starts stack protection can be already enabled that's why setup
the lowest possible SLR value which is address 0. And the highest possible
stack in front of U-Boot. That's why you should never load U-Boot to the
beginning of DDR. There must be some space reserved. Code is using this
location for early malloc space, early global data and stack.
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/86b9748bad12142659804d6381bc6bbf20be44f1.1655299267.git.michal.simek@amd.com
Rename the sections used to implement linker lists so they begin with
'__u_boot_list' rather than '.u_boot_list'. The double underscore at the
start is still distinct from the single underscore used by the symbol
names.
Having a '.' in the section names conflicts with clang's ASAN
instrumentation which tries to add redzones between the linker list
elements, causing expected accesses to fail. However, clang doesn't try
to add redzones to user sections, which are names with all alphanumeric
and underscore characters.
Signed-off-by: Andrew Scull <ascull@google.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Move these two options to the arch Kconfig file.
Cc: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
Acked-by: Michal Simek <michal.simek@xilinx.com>
The correct mask for getting the source/destination register from ESR in
the case of an unaligned access exception is 0x3E0. With this change, a
dummy unaligned store produces the expected info:
"""
>> swi r5, r0, 0x111
...
Hardware exception at 0x111 address
Unaligned data access exception
Unaligned word access
Unaligned store access
Register R5
Return address from exception 0x7f99dfc
### ERROR ### Please RESET the board ###
"""
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/20220213080925.1548411-6-ovidiu.panait@windriver.com
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
The unaligned access messages are only valid in the case of an unaligned
data access exception. Do not print them for other types of hw exceptions.
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/20220213080925.1548411-5-ovidiu.panait@windriver.com
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
According to the MicroBlaze reference manual (xilinx2021.2/ug984/page-37):
"""
If an exception is caused by an instruction in a delay slot (that is,
ESR[DS]=1), the exception handler should return execution to
the address stored in BTR instead of the normal exception return
address stored in R17.
"""
Adjust the code to print the proper return address for delay slot
exceptions.
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/20220213080925.1548411-4-ovidiu.panait@windriver.com
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
The switch statement in _hw_exception_handler() only covers the
rightmost 5 bits that encode the exception cause:
switch (state & 0x1f)
{
...
}
For this reason, the "0x1000" case will never be reached, because the 13th
bit was zeroed out. To fix this, move delay slot exception handling before
the switch statement (delay slot (DS) bit in Exception Status Register is
independent of the exception cause (EC)).
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/20220213080925.1548411-3-ovidiu.panait@windriver.com
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
The privileged instruction exception seems to have been introduced in
microblaze v7.00 along with MMU support, so having it wrapped in
MICROBLAZE_v5 ifdefs seems incorrect. Move it out of the ifdef, since all
recent microblaze versions support it.
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/20220213080925.1548411-1-ovidiu.panait@windriver.com
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
When building for a custom board, it is quite common to maintain a
private branch which include some defconfig and .dts files. But to
hook up those .dts files requires modifying a file "belonging" to
upstream U-Boot, the arch/*/dts/Makefile. Forward-porting that branch
to a newer upstream then often results in a conflict which, while it
is trivial to resolve by hand, makes it harder to have a CI do "try to
build our board against latest upstream".
The .config usually includes information on precisely what .dtb(s) are
needed, so to avoid having to modify the Makefile, simply add the
files in (SPL_)OF_LIST to dtb-y.
A technicality is that (SPL_)OF_LIST is not always defined, so rework
the Kconfig symbols so that (SPL_)OF_LIST is always defined (when
(SPL_)OF_CONTROL), but only prompted for in the cases which used to be
their "depends on".
nios2 and microblaze already have something like this in their
dts/Makefile, and the rationale in commit 41f59f6853 is similar to
the above. So this simply generalizes existing practice. Followup
patches could remove the logic in those two makefiles, just as there's
potential for moving some common boilerplate from all the
arch/*/dts/Makefile files to the new scripts/Makefile.dts.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Reviewed-by: Simon Glass <sjg@chromium.org>
Current code assumes that the vector base address is always at 0x0.
However, this value is configurable for MicroBlaze using the
CONFIG_XILINX_MICROBLAZE0_VECTOR_BASE_ADDR Kconfig option. Update the
reset routines to branch to this location instead.
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/20211130163358.2531677-10-ovidiu.panait@windriver.com
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Current code assumes that the vector base address is always at 0x0.
However, this value is configurable for MicroBlaze, so update the
__setup_exceptions routine to work with any vector base address.
The r4 register is reserved for the vector base address inside
__setup_exceptions and the function prologe/epilogue are also updated to
save and restore r4.
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/20211130163358.2531677-9-ovidiu.panait@windriver.com
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Migrate CONFIG_SYS_USR_EXCEP to Kconfig. Also, rename it to
XILINX_MICROBLAZE0_USR_EXCEP in order to match the naming convention of
microblaze-generic Kconfig options.
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/20211130163358.2531677-7-ovidiu.panait@windriver.com
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Microblaze is one the last two users of the CONFIG_SYS_RESET_ADDRESS
macro (the other is arch/powerpc/cpu/mpc8xx/cpu.c, but the macro is not
defined anywhere in powerpc code, so it should be removed there too).
Replace CONFIG_SYS_RESET_ADDRESS usage in start.S with
CONFIG_SYS_TEXT_BASE. If the reset address should really be
user-configurable, a new Kconfig option could be added.
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/20211130163358.2531677-6-ovidiu.panait@windriver.com
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
To simpify the code, use stack space as scratch memory for endian offset
calculation, rather than saving/restoring the first unused MB vector.
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Link: https://lore.kernel.org/r/20211130163358.2531677-5-ovidiu.panait@windriver.com
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
