RISC-V kernels support 3,4,5-level page tables at runtime by folding
upper levels.
In case of a 3-level page table, PGDIR is folded into P4D which in turn
is folded into PUD: PGDIR_SHIFT value is correctly set to the same value
as PUD_SHIFT, but P4D_SHIFT is not, then any use of P4D_SHIFT will access
invalid address bits (all set to 1).
Fix this by dynamically defining P4D_SHIFT value, like we already do for
PGDIR_SHIFT.
Fixes: d10efa21a9 ("riscv: mm: Control p4d's folding by pgtable_l5_enabled")
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Reviewed-by: Palmer Dabbelt <palmer@rivosinc.com>
Link: https://lore.kernel.org/r/20221201135128.1482189-2-alexghiti@rivosinc.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Andrew Jones <ajones@ventanamicro.com> says:
This series is a collection of cleanups for alternative-macros.h with
the main motivation being that adding new ALTERNATIVE_3, ALTERNATIVE_4,
... will be possible without lots of bloat.
* b4-shazam-merge:
riscv: Don't duplicate _ALTERNATIVE_CFG* macros
riscv: alternatives: Drop the underscores from the assembly macro names
riscv: alternatives: Don't name unused macro parameters
riscv: Don't duplicate __ALTERNATIVE_CFG in __ALTERNATIVE_CFG_2
Link: https://lore.kernel.org/r/20221129150053.50464-1-ajones@ventanamicro.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Andrew Jones <ajones@ventanamicro.com> says:
When a DT puts zicbom in the isa string, but does not provide a block
size, ALT_CMO_OP() will attempt to do cache operations on address
zero since the start address will be ANDed with zero. We can't simply
BUG() in riscv_init_cbom_blocksize() when we fail to find a block
size because the failure will happen before logging works, leaving
users to scratch their heads as to why the boot hung. Instead, ensure
Zicbom is disabled and output an error which will hopefully alert
people that the DT needs to be fixed. While at it, add a check that
the block size is a power-of-2 too.
* b4-shazam-merge:
RISC-V: Ensure Zicbom has a valid block size
RISC-V: Introduce riscv_isa_extension_check
RISC-V: Improve use of isa2hwcap[]
Link: https://lore.kernel.org/r/20221129143447.49714-1-ajones@ventanamicro.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
When a DT puts zicbom in the isa string, but does not provide a block
size, ALT_CMO_OP() will attempt to do cache operations on address
zero since the start address will be ANDed with zero. We can't simply
BUG() in riscv_init_cbom_blocksize() when we fail to find a block
size because the failure will happen before logging works, leaving
users to scratch their heads as to why the boot hung. Instead, ensure
Zicbom is disabled and output an error which will hopefully alert
people that the DT needs to be fixed. While at it, add a check that
the block size is a power-of-2 too.
Signed-off-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20221129143447.49714-4-ajones@ventanamicro.com
[Palmer: base on 5c20a3a9df ("RISC-V: Fix compilation without RISCV_ISA_ZICBOM"]
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Currently any isa extension found in the isa string is set in the
isa bitmap. An isa extension set in the bitmap indicates that the
extension is present and may be used (a.k.a is enabled). However,
when an extension cannot be used due to missing dependencies or
errata it should not be added to the bitmap. Introduce a function
where additional checks may be placed in order to determine if an
extension should be enabled or not.
Note, the checks may simply indicate an issue with the DT, but,
since extensions may be used in early boot, it's not always possible
to simply produce an error at the point the issue is determined.
It's best to keep the extension disabled and produce an error.
No functional change intended, as the function is only introduced
and always returns true. A later patch will provide checks for an
isa extension.
Signed-off-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Link: https://lore.kernel.org/r/20221129143447.49714-3-ajones@ventanamicro.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
These are fixes, but due to the possible early boot fallout they're
going in the merge window to get a bit more time to bake on linux-next.
* b4-shazam-merge
riscv: Move cast inside kernel_mapping_[pv]a_to_[vp]a
riscv: Fix crash during early errata patching
Link: https://lore.kernel.org/r/20221126060920.65009-1-samuel@sholland.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
The patch function for the T-Head PBMT errata calls __pa_symbol() before
relocation. This crashes when CONFIG_DEBUG_VIRTUAL is enabled, because
__pa_symbol() forwards to __phys_addr_symbol(), and __phys_addr_symbol()
checks against the absolute kernel start/end address.
Fix this by checking against the kernel map instead of a symbol address.
Fixes: a35707c3d8 ("riscv: add memory-type errata for T-Head")
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Tested-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Link: https://lore.kernel.org/r/20221126060920.65009-1-samuel@sholland.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
The SiFive PLIC driver is used by all current implementations, including
those that do not have a SiFive PLIC. The current driver supports more
than just SiFive PLICs at present and, where possible, future PLIC
implementations will also use this driver. As every supported RISC-V SoC
selects the driver directly in Kconfig.socs there's no point in exposing
this kconfig option to users.
The Kconfig help text, in its current form, is misleading. There's no
point doing anything about that though, as it will no longer be user
selectable. Remove it.
Suggested-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Acked-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20221118104300.85016-2-conor@kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Anup Patel <apatel@ventanamicro.com> says:
The Linux NVDIMM PEM drivers require arch support to map and access the
persistent memory device. This series adds RISC-V PMEM support using
recently added Svpbmt and Zicbom support.
* b4-shazam-merge:
RISC-V: Enable PMEM drivers
RISC-V: Implement arch specific PMEM APIs
RISC-V: Fix MEMREMAP_WB for systems with Svpbmt
Link: https://lore.kernel.org/r/20221114090536.1662624-1-apatel@ventanamicro.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
The NVDIMM PMEM driver expects arch specific APIs for cache maintenance
and if arch does not provide these APIs then NVDIMM PMEM driver will
always use MEMREMAP_WT to map persistent memory which in-turn maps as
UC memory type defined by the RISC-V Svpbmt specification.
Now that the Svpbmt and Zicbom support is available in RISC-V kernel,
we implement PMEM APIs using ALT_CMO_OP() macros so that the NVDIMM
PMEM driver can use MEMREMAP_WB to map persistent memory.
Co-developed-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20221114090536.1662624-3-apatel@ventanamicro.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Currently, the memremap() called with MEMREMAP_WB maps memory using
the generic ioremap() function which breaks on system with Svpbmt
because memory mapped using _PAGE_IOREMAP page attributes is treated
as strongly-ordered non-cacheable IO memory.
To address this, we implement RISC-V specific arch_memremap_wb()
which maps memory using _PAGE_KERNEL page attributes resulting in
write-back cacheable mapping on systems with Svpbmt.
Fixes: ff689fd21c ("riscv: add RISC-V Svpbmt extension support")
Co-developed-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20221114090536.1662624-2-apatel@ventanamicro.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
The return to userspace path in entry.S may enable interrupts without the
corresponding lockdep annotation, producing a splat[0] when DEBUG_LOCKDEP
is enabled. Simply calling __trace_hardirqs_on() here gets a bit messy
due to the use of RA to point back to ret_from_exception, so just move
the whole slow-path loop into C. It's more readable and it lets us use
local_irq_{enable,disable}(), avoiding the need for manual annotations
altogether.
[0]:
------------[ cut here ]------------
DEBUG_LOCKS_WARN_ON(!lockdep_hardirqs_enabled())
WARNING: CPU: 2 PID: 1 at kernel/locking/lockdep.c:5512 check_flags+0x10a/0x1e0
Modules linked in:
CPU: 2 PID: 1 Comm: init Not tainted 6.1.0-rc4-00160-gb56b6e2b4f31 #53
Hardware name: riscv-virtio,qemu (DT)
epc : check_flags+0x10a/0x1e0
ra : check_flags+0x10a/0x1e0
<snip>
status: 0000000200000100 badaddr: 0000000000000000 cause: 0000000000000003
[<ffffffff808edb90>] lock_is_held_type+0x78/0x14e
[<ffffffff8003dae2>] __might_resched+0x26/0x22c
[<ffffffff8003dd24>] __might_sleep+0x3c/0x66
[<ffffffff80022c60>] get_signal+0x9e/0xa70
[<ffffffff800054a2>] do_notify_resume+0x6e/0x422
[<ffffffff80003c68>] ret_from_exception+0x0/0x10
irq event stamp: 44512
hardirqs last enabled at (44511): [<ffffffff808f901c>] _raw_spin_unlock_irqrestore+0x54/0x62
hardirqs last disabled at (44512): [<ffffffff80008200>] __trace_hardirqs_off+0xc/0x14
softirqs last enabled at (44472): [<ffffffff808f9fbe>] __do_softirq+0x3de/0x51e
softirqs last disabled at (44467): [<ffffffff80017760>] irq_exit+0xd6/0x104
---[ end trace 0000000000000000 ]---
possible reason: unannotated irqs-on.
Signed-off-by: Andrew Bresticker <abrestic@rivosinc.com>
Fixes: 3c46979829 ("riscv: Enable LOCKDEP_SUPPORT & fixup TRACE_IRQFLAGS_SUPPORT")
Link: https://lore.kernel.org/r/20221111223108.1976562-1-abrestic@rivosinc.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Current implementation of update_mmu_cache function performs local TLB
flush. It does not take into account ASID information. Besides, it does
not take into account other harts currently running the same mm context
or possible migration of the running context to other harts. Meanwhile
TLB flush is not performed for every context switch if ASID support
is enabled.
Patch [1] proposed to add ASID support to update_mmu_cache to avoid
flushing local TLB entirely. This patch takes into account other
harts currently running the same mm context as well as possible
migration of this context to other harts.
For this purpose the approach from flush_icache_mm is reused. Remote
harts currently running the same mm context are informed via SBI calls
that they need to flush their local TLBs. All the other harts are marked
as needing a deferred TLB flush when this mm context runs on them.
[1] https://lore.kernel.org/linux-riscv/20220821013926.8968-1-tjytimi@163.com/
Signed-off-by: Sergey Matyukevich <sergey.matyukevich@syntacore.com>
Fixes: 65d4b9c530 ("RISC-V: Implement ASID allocator")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/linux-riscv/20220829205219.283543-1-geomatsi@gmail.com/#t
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
This is reported by kmemleak detector:
unreferenced object 0xff2000000403d000 (size 4096):
comm "kexec", pid 146, jiffies 4294900633 (age 64.792s)
hex dump (first 32 bytes):
7f 45 4c 46 02 01 01 00 00 00 00 00 00 00 00 00 .ELF............
04 00 f3 00 01 00 00 00 00 00 00 00 00 00 00 00 ................
backtrace:
[<00000000566ca97c>] kmemleak_vmalloc+0x3c/0xbe
[<00000000979283d8>] __vmalloc_node_range+0x3ac/0x560
[<00000000b4b3712a>] __vmalloc_node+0x56/0x62
[<00000000854f75e2>] vzalloc+0x2c/0x34
[<00000000e9a00db9>] crash_prepare_elf64_headers+0x80/0x30c
[<0000000067e8bf48>] elf_kexec_load+0x3e8/0x4ec
[<0000000036548e09>] kexec_image_load_default+0x40/0x4c
[<0000000079fbe1b4>] sys_kexec_file_load+0x1c4/0x322
[<0000000040c62c03>] ret_from_syscall+0x0/0x2
In elf_kexec_load(), a buffer is allocated via vzalloc() to store elf
headers. While it's not freed back to system when kdump kernel is
reloaded or unloaded, or when image->elf_header is successfully set and
then fails to load kdump kernel for some reason. Fix it by freeing the
buffer in arch_kimage_file_post_load_cleanup().
Fixes: 8acea455fa ("RISC-V: Support for kexec_file on panic")
Signed-off-by: Li Huafei <lihuafei1@huawei.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20221104095658.141222-2-lihuafei1@huawei.com
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
This is reported by kmemleak detector:
unreferenced object 0xff60000082864000 (size 9588):
comm "kexec", pid 146, jiffies 4294900634 (age 64.788s)
hex dump (first 32 bytes):
d0 0d fe ed 00 00 12 ed 00 00 00 48 00 00 11 40 ...........H...@
00 00 00 28 00 00 00 11 00 00 00 02 00 00 00 00 ...(............
backtrace:
[<00000000f95b17c4>] kmemleak_alloc+0x34/0x3e
[<00000000b9ec8e3e>] kmalloc_order+0x9c/0xc4
[<00000000a95cf02e>] kmalloc_order_trace+0x34/0xb6
[<00000000f01e68b4>] __kmalloc+0x5c2/0x62a
[<000000002bd497b2>] kvmalloc_node+0x66/0xd6
[<00000000906542fa>] of_kexec_alloc_and_setup_fdt+0xa6/0x6ea
[<00000000e1166bde>] elf_kexec_load+0x206/0x4ec
[<0000000036548e09>] kexec_image_load_default+0x40/0x4c
[<0000000079fbe1b4>] sys_kexec_file_load+0x1c4/0x322
[<0000000040c62c03>] ret_from_syscall+0x0/0x2
In elf_kexec_load(), a buffer is allocated via kvmalloc() to store fdt.
While it's not freed back to system when kexec kernel is reloaded or
unloaded. Then memory leak is caused. Fix it by introducing riscv
specific function arch_kimage_file_post_load_cleanup(), and freeing the
buffer there.
Fixes: 6261586e0c ("RISC-V: Add kexec_file support")
Signed-off-by: Li Huafei <lihuafei1@huawei.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Liao Chang <liaochang1@huawei.com>
Link: https://lore.kernel.org/r/20221104095658.141222-1-lihuafei1@huawei.com
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Add arch_crash_save_vmcoreinfo(), which exports VM layout(MODULES,
VMALLOC, VMEMMAP ranges and KERNEL_LINK_ADDR), va bits and ram base for
vmcore.
* b4-shazam-merge:
Documentation: kdump: describe VMCOREINFO export for RISCV64
RISC-V: Add arch_crash_save_vmcoreinfo support
Link: https://lore.kernel.org/r/20221026144208.373504-1-xianting.tian@linux.alibaba.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Add arch_crash_save_vmcoreinfo(), which exports VM layout(MODULES,
VMALLOC, VMEMMAP ranges and KERNEL_LINK_ADDR), va bits and ram base for
vmcore.
Default pagetable levels and PAGE_OFFSET aren't same for different
kernel version as below. For pagetable levels, it sets sv57 by default
and falls back to setting sv48 at boot time if sv57 is not supported by
the hardware.
For ram base, the default value is 0x80200000 for qemu riscv64 env and,
for example, is 0x200000 on the XuanTie 910 CPU.
* Linux Kernel 5.18 ~
* PGTABLE_LEVELS = 5
* PAGE_OFFSET = 0xff60000000000000
* Linux Kernel 5.17 ~
* PGTABLE_LEVELS = 4
* PAGE_OFFSET = 0xffffaf8000000000
* Linux Kernel 4.19 ~
* PGTABLE_LEVELS = 3
* PAGE_OFFSET = 0xffffffe000000000
Since these configurations change from time to time and version to
version, it is preferable to export them via vmcoreinfo than to change
the crash's code frequently, it can simplify the development of crash
tool.
Signed-off-by: Xianting Tian <xianting.tian@linux.alibaba.com>
Tested-by: Deepak Gupta <debug@rivosinc.com>
Tested-by: Guo Ren <guoren@kernel.org>
Acked-by: Baoquan He <bhe@redhat.com>
Link: https://lore.kernel.org/r/20221026144208.373504-2-xianting.tian@linux.alibaba.com
[Palmer: wrap commit text]
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
With the PG_arch_1 we keep track if the page's data cache is clean,
architecture rely on this property to treat new pages as dirty with
respect to the data cache and perform the flushing before mapping the pages
into userspace.
This patch adds a new architecture hook, arch_clear_hugepage_flags,so that
architectures which rely on the page flags being in a particular state for
fresh allocations can adjust the flags accordingly when a page is freed
into the pool.
Fixes: 9e953cda5c ("riscv: Introduce huge page support for 32/64bit kernel")
Signed-off-by: Tong Tiangen <tongtiangen@huawei.com>
Link: https://lore.kernel.org/r/20221024094725.3054311-3-tongtiangen@huawei.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Jamie Iles <jamie@jamieiles.com> says:
This series enables dynamic ftrace support for RV32I bringing it to
parity with RV64I. Most of the work is already there, this is largely
just assembly fixes to handle register sizes, correct handling of the
psABI calling convention and Kconfig change.
Validated with all ftrace boot time self test with qemu for RV32I and
RV64I in addition to real tracing on an RV32I FPGA design.
* b4-shazam-merge:
RISC-V: enable dynamic ftrace for RV32I
RISC-V: preserve a1 in mcount
RISC-V: reduce mcount save space on RV32
RISC-V: use REG_S/REG_L for mcount
Link: https://lore.kernel.org/r/20221115200832.706370-1-jamie@jamieiles.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Currently, when detecting vmap stack overflow, riscv firstly switches
to the so called shadow stack, then use this shadow stack to call the
get_overflow_stack() to get the overflow stack. However, there's
a race here if two or more harts use the same shadow stack at the same
time.
To solve this race, we introduce spin_shadow_stack atomic var, which
will be swap between its own address and 0 in atomic way, when the
var is set, it means the shadow_stack is being used; when the var
is cleared, it means the shadow_stack isn't being used.
Fixes: 31da94c25a ("riscv: add VMAP_STACK overflow detection")
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Suggested-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Guo Ren <guoren@kernel.org>
Link: https://lore.kernel.org/r/20221030124517.2370-1-jszhang@kernel.org
[Palmer: Add AQ to the swap, and also some comments.]
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
on an arch level, RISC-V defaults to FLATMEM. On PolarFire SoC, the
memory layout is almost always sparse, with a maximum of 1 GiB at
0x8000_0000 & a possible 16 GiB range at 0x10_0000_0000. The Icicle kit,
for example, has 2 GiB of DDR - so there's a big hole in the memory map
between the two gigs. Prior to v6.1-rc1, boot times from defconfig
builds were pretty bad on Icicle but enabling sparsemem would fix those
issues. As of v6.1-rc1, the Icicle kit no longer boots from defconfig
builds with the in-kernel devicetree. A change to the memory map
resulted in a futher "sparse-ification", producing a splat on boot:
OF: fdt: Ignoring memory range 0x80000000 - 0x80200000
Machine model: Microchip PolarFire-SoC Icicle Kit
earlycon: ns16550a0 at MMIO32 0x0000000020100000 (options '115200n8')
printk: bootconsole [ns16550a0] enabled
printk: debug: skip boot console de-registration.
efi: UEFI not found.
Zone ranges:
DMA32 [mem 0x0000000080200000-0x00000000ffffffff]
Normal [mem 0x0000000100000000-0x000000107fffffff]
Movable zone start for each node
Early memory node ranges
node 0: [mem 0x0000000080200000-0x00000000bfbfffff]
node 0: [mem 0x00000000bfc00000-0x00000000bfffffff]
node 0: [mem 0x0000001040000000-0x000000107fffffff]
Initmem setup node 0 [mem 0x0000000080200000-0x000000107fffffff]
Kernel panic - not syncing: Failed to allocate 1073741824 bytes for node 0 memory map
CPU: 0 PID: 0 Comm: swapper Not tainted 5.19.0-dirty #1
Hardware name: Microchip PolarFire-SoC Icicle Kit (DT)
Call Trace:
[<ffffffff800057f0>] show_stack+0x30/0x3c
[<ffffffff807d5802>] dump_stack_lvl+0x4a/0x66
[<ffffffff807d5836>] dump_stack+0x18/0x20
[<ffffffff807d1ae8>] panic+0x124/0x2c6
[<ffffffff80814064>] free_area_init_core+0x0/0x11e
[<ffffffff80813720>] free_area_init_node+0xc2/0xf6
[<ffffffff8081331e>] free_area_init+0x222/0x260
[<ffffffff808064d6>] misc_mem_init+0x62/0x9a
[<ffffffff80803cb2>] setup_arch+0xb0/0xea
[<ffffffff8080039a>] start_kernel+0x88/0x4ee
---[ end Kernel panic - not syncing: Failed to allocate 1073741824 bytes for node 0 memory map ]---
With the aim of keeping defconfig builds booting on icicle, enable
SPARSEMEM_MANUAL.
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20221021160028.4042304-1-conor@kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
