Currently, if 64BIT and !XIP_KERNEL, the phys_ram_base is always 0,
no matter the real start of dram reported by memblock is.

Fixes: 6d7f91d9 ("riscv: Get rid of CONFIG_PHYS_RAM_BASE in kernel physical address conversion")
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Reviewed-by: Alexandre Ghiti <alex@ghiti.fr>
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

Use what is currently the SMP=y version of riscv_cpuid_to_hartid_mask()
for both SMP=y and SMP=n to fix a build failure with KVM=m and SMP=n due
to boot_cpu_hartid not being exported.  This also fixes a second bug
where the SMP=n version assumes the sole CPU in the system is in the
incoming mask, which may not hold true in kvm_riscv_vcpu_sbi_ecall() if
the KVM guest VM has multiple vCPUs (on a SMP=n system).

Fixes: 1ef46c23 ("RISC-V: Implement new SBI v0.2 extensions")
Reported-by: Adam Borowski <kilobyte@angband.pl>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

After commit 83e7b876 ("RISC-V: move empty_zero_page definition to
C and export it"), the empty_zero_page has been moved outside head.S,
the __PAGE_ALIGNED_BSS and .balign LoCs are useless, clean up them.
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

The function pointer vendor_patch_func is only used during init, so
mark it as __initdata.
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

There are no users outside head.S so make secondary_start_common()
static.
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

Except arch_cpu_idle_dead(), no users of this function. So remove
cpu_stop() and fold its code into arch_cpu_idle_dead().
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

When allocating crash kernel region without explicitly specifying its
base address/size, memblock_phys_alloc_range will attempt to allocate
memory top to bottom (memblock.bottom_up is false), so the crash
kernel region will end up in highmem on 64bit systems. This way
swiotlb can't work on the crash kernel, since there won't be any
32bit addressible memory available for the bounce buffers.

Try to allocate 32bit addressible memory if available, for the
crash kernel by restricting the top search address to be less
than SZ_4G. If that fails fallback to the previous behavior.

I tested this on HiFive Unmatched where the pci-e controller needs
swiotlb to work, with this patch it's possible to access the pci-e
controller on crash kernel and mount the rootfs from the nvme.
Signed-off-by: Nick Kossifidis <mick@ics.forth.gr>
Fixes: e53d2818 ("RISC-V: Add kdump support")
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

raw_smp_processor_id() doesn't return the hart id as stated in
arch/riscv/include/asm/smp.h, use smp_processor_id() instead
to get the cpu id, and cpuid_to_hartid_map() to pass the hart id
to the next kernel. This fixes kexec on HiFive Unleashed/Unmatched
where cpu ids and hart ids don't match (on qemu-virt they match).

Fixes: fba8a867 ("RISC-V: Add kexec support")
Signed-off-by: Nick Kossifidis <mick@ics.forth.gr>
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

On kdump instead of using an intermediate step to relocate the kernel,
that lives in a "control buffer" outside the current kernel's mapping,
we jump to the crash kernel directly by calling riscv_kexec_norelocate().
The current implementation uses va_pa_offset while switching to physical
addressing, however since we moved the kernel outside the linear mapping
this won't work anymore since riscv_kexec_norelocate() is part of the
kernel mapping and we should use kernel_map.va_kernel_pa_offset, and also
take XIP kernel into account.

We don't really need to use va_pa_offset on riscv_kexec_norelocate, we
can just set STVEC to the physical address of the new kernel instead and
let the hart jump to the new kernel on the next instruction after setting
SATP to zero. This fixes kdump and is also simpler/cleaner.

I tested this on the latest qemu and HiFive Unmatched and works as
expected.

Fixes: 2bfc6cd8 ("riscv: Move kernel mapping outside of linear mapping")
Signed-off-by: Nick Kossifidis <mick@ics.forth.gr>
Reviewed-by: Alexandre Ghiti <alex@ghiti.fr>
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

Fix the device node for the Platform-Level Interrupt Controller (PLIC):
  - Add missing "#address-cells" property,
  - Sort properties according to DT bindings.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

"make dtbs_check":

    arch/riscv/boot/dts/sifive/hifive-unleashed-a00.dt.yaml: soc: $nodename:0: '/' was expected
    	From schema: Documentation/devicetree/bindings/riscv/sifive.yaml
    arch/riscv/boot/dts/sifive/hifive-unleashed-a00.dt.yaml: soc: compatible: 'oneOf' conditional failed, one must be fixed:
    	'sifive,fu540-c000' is not one of ['sifive,hifive-unleashed-a00']
    	'sifive,fu540-c000' is not one of ['sifive,hifive-unmatched-a00']
    	'sifive,fu540-c000' was expected
    	'sifive,fu740-c000' was expected
    	'sifive,fu540' was expected
    	'sifive,fu740' was expected
    	From schema: Documentation/devicetree/bindings/riscv/sifive.yaml

This happens because the "soc" subnode declares compatibility with
"sifive,fu540-c000" and "sifive,fu540", while these are only intended
for the root node.

Fix this by removing the bogus compatible values from the "soc" node.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

To improve human readability and enable automatic validation, the tuples
in "reg" properties containing register blocks should be grouped using
angle brackets.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

To improve human readability and enable automatic validation, the tuples
in the various properties containing interrupt specifiers should be
grouped.

Fix this by grouping the tuples of "interrupts" and
"interrupts-extended" properties using angle brackets.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

To improve human readability and enable automatic validation, the tuples
in the various properties containing interrupt specifiers should be
grouped.

Fix this by grouping the tuples of "interrupts" and
"interrupts-extended" properties using angle brackets.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

Fix the device node for the clock controller:
  - Remove bogus "reg-names" property,
  - Remove unneeded "clock-output-names" property.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

"make dtbs_check" reports:

    arch/riscv/boot/dts/microchip/microchip-mpfs-icicle-kit.dt.yaml: soc: refclk: {'compatible': ['fixed-clock'], '#clock-cells': [[0]], 'clock-frequency': [[600000000]], 'clock-output-names': ['msspllclk'], 'phandle': [[7]]} should not be valid under {'type': 'object'}
	From schema: dtschema/schemas/simple-bus.yaml

Fix this by moving the node out of the "soc" subnode.
While at it, rename it to "msspllclk", and drop the now superfluous
"clock-output-names" property.
Move the actual clock-frequency value to the board DTS, since it is not
set until bitstream programming time.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

Fix the device node for the Platform-Level Interrupt Controller (PLIC):
  - Add missing "#address-cells" property,
  - Sort properties according to DT bindings.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

It does not make sense to have an (empty) chosen node in an SoC-specific
.dtsi, as chosen is meant for system-specific configuration.
It is already provided in microchip-mpfs-icicle-kit.dts anyway.

Fixes: 0fa6107e ("RISC-V: Initial DTS for Microchip ICICLE board")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

To improve human readability and enable automatic validation, the tuples
in the various properties containing interrupt specifiers should be
grouped.

Fix this by grouping the tuples of "interrupts" and
"interrupts-extended" properties using angle brackets.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Tested-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

"make dtbs_check":

    arch/riscv/boot/dts/canaan/sipeed_maix_bit.dt.yaml: spi-flash@0: $nodename:0: 'spi-flash@0' does not match '^flash(@.*)?$'
	    From schema: Documentation/devicetree/bindings/mtd/jedec,spi-nor.yaml

Fix this by renaming all SPI FLASH nodes to "flash".
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Tested-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

Add two THP helpers required to create PMD migration swap entries,
and enable THP migration via ARCH_ENABLE_THP_MIGRATION. This can
reduce time of THP migration without splitting and guarantee the
migrated pages are still contiguous.
Signed-off-by: Nanyong Sun <sunnanyong@huawei.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

This is a preparation for enabling THP migration.
As the commit b65399f6("arm64/mm: Change THP helpers
to comply with generic MM semantics") mentioned, pmd_present()
and pmd_trans_huge() are expected to behave in the following
manner:
-------------------------------------------------------------------------
|	PMD states	|	pmd_present	|	pmd_trans_huge	|
-------------------------------------------------------------------------
|	Mapped		|	Yes		|	Yes		|
-------------------------------------------------------------------------
|	Splitting	|	Yes		|	Yes		|
-------------------------------------------------------------------------
|	Migration/Swap	|	No		|	No		|
-------------------------------------------------------------------------

At present the PROT_NONE bit reuses the READ bit could not comply with
above semantics with two problems:
1. When splitting a PMD THP, PMD is first invalidated with
pmdp_invalidate()->pmd_mkinvalid(), which clears the PRESENT bit
and PROT_NONE bit/READ bit, if the PMD is read-only, then the PAGE_LEAF
property is also cleared, which results in pmd_present() return false.
2. When migrating, the swap entry only clear the PRESENT bit
and PROT_NONE bit/READ bit, the W/X bit may be set, so _PAGE_LEAF may be
true which results in pmd_present() return true.

Solution:
Adjust PROT_NONE bit from READ to GLOBAL bit can satisfy the above rules:
1. GLOBAL bit has no other meanings, not like the R/W/X bit, which is
also relative with _PAGE_LEAF property.
2. GLOBAL bit is at bit 5, making swap entry start from bit 6, bit 0-5
are zero, which means the PRESENT, PROT_NONE, and PAGE_LEAF are
all false, then the pmd_present() and pmd_trans_huge() return false when
in migration/swap.
Signed-off-by: Nanyong Sun <sunnanyong@huawei.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

Fixes misspelling of guaranteed in comment describing why fetching fence
is guaranteed to work when switching to kernel page tables.
Signed-off-by: hasheddan <georgedanielmangum@gmail.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

It's been a while since cleaning up the defconfigs, so I manually
checked up on each change.  This found a handful of minor issues, which
have been fixed in-line.

The "k210_generic" DT has been the default in Kconfig since 67d96729
("riscv: Update Canaan Kendryte K210 device tree"), so drop it from the
defconfigs to avoid diff with savedefconfig.
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

We've always had CONFIG_EFI as "def_bool y" so this has always been
redundant.  It's removed by savedefconfig, so drop it to keep things
clean.
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

As of ab7fbad0 ("riscv: Fix unmet direct dependencies built based on
SOC_VIRT") we select CONFIG_POWER_RESET=y along with CONFIG_SOC_VIRT,
which is already in defconfig.  This make setting CONFIG_POWER_RESET in
the defconfigs redundant, so remove it to remain consistent with
savedefconfig.
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

This should have no functional change, it just sorts CONFIG_BLK_DEV_BSG
the same way savedefconfig does.
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

This should have no functional change, it just sorts
CONFIG_SURFACE_PLATFORMS the same way savedefconfig does.
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

This should have no functional change, it just sorts CONFIG_MMC the same
way savedefconfig does.  This only touches the rv64 defconfig because
rv32_defconfig was already sorted correctly.
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

This should have no functional change, it just sorts
CONFIG_PTP_1588_CLOCK the same way savedefconfig does.  This only
touches the rv64 defconfig because rv32_defconfig was already sorted
correctly.
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

This should have no functional change, it just sorts
CONFIG_SOC_POLARFIRE the same way savedefconfig does.
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

This should have no functional change, it just sorts
CONFIG_SYSFS_SYSCALL the same way savedefconfig does.
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

This should have no functional change, it just sorts CONFIG_BPF_SYSCALL
the same way savedefconfig does.
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

For non-relocatable kernels we need to be able to link the kernel at
approximately PAGE_OFFSET, thus requiring medany (as medlow requires the
code to be linked within 2GiB of 0).  The inverse doesn't apply, though:
since medany code can be linked anywhere it's fine to link it close to
0, so we can support the smaller memory config.

Fixes: de5f4b8f ("RISC-V: Define MAXPHYSMEM_1GB only for RV32")
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

We have CONFIG_FRAMEBUFFER_CONSOLE=y in the defconfigs, but that depends
on CONFIG_FB so it's not actually getting set.  I'm assuming most users
on real systems want a framebuffer console, so this enables CONFIG_FB to
allow that to take effect.

Fixes: 33c57c0d ("RISC-V: Add a basic defconfig")
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

After commit 1355c31e ("asm-generic: pgalloc: provide generic
pmd_alloc_one() and pmd_free_one()"), the main part to support
PMD split page table lock is in asm-generic/pgalloc.h.

The only change is add pgtable_pmd_page_ctor() into alloc_pmd_late(),
then we could enable ARCH_ENABLE_SPLIT_PMD_PTLOCK for RV64.
Reviewed-by: Alexandre Ghiti <alex@ghiti.fr>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

We used to define VMALLOC_END equal to the start of the next region
*minus one* which is inconsistent with the use of this define in the
core code (for example, see the definitions of VMALLOC_TOTAL and
is_vmalloc_addr).

And then make the definition of VMEMMAP_END consistent with VMALLOC_END
and all other regions actually.
Signed-off-by: Alexandre Ghiti <alexandre.ghiti@canonical.com>
Reviewed-by: Jisheng Zhang <jszhang@kernel.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

Similar as other architectures such as arm64, x86 and so on, use
offsets relative to the exception table entry values rather than
absolute addresses for both the exception locationand the fixup.
And recently, arm64 and x86 remove anonymous out-of-line fixups, we
want to acchieve the same result.

These are no longer necessary now that we have a more standard extable
mechanism.
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>