Add PMEM support for RISC-V
v4 cover Add PMEM support for RISC-V
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.
First two patches are fixes and remaining two patches add the required PMEM support for Linux RISC-V.
These patches can also be found in riscv_pmem_v4 branch at:
https://github.com/avpatel/linux.git
Changes since v3:
- Pickup correct version of Drew's patch as PATCH1
Changes since v2:
- Rebased on Linux-6.1-rc1
- Replaced PATCH1 with the patch proposed by Drew
Changes since v1:
- Fix error reported by test bot
https://lore.kernel.org/all/[email protected]/
Andrew Jones (1):
RISC-V: Fix compilation without RISCV_ISA_ZICBOM
Anup Patel (3):
RISC-V: Fix ioremap_cache() and ioremap_wc() for systems with Svpbmt
RISC-V: Implement arch specific PMEM APIs
RISC-V: Enable PMEM drivers
arch/riscv/Kconfig | 1 + arch/riscv/configs/defconfig | 1 + arch/riscv/include/asm/cacheflush.h | 8 ------ arch/riscv/include/asm/io.h | 10 +++++++ arch/riscv/include/asm/pgtable.h | 2 ++ arch/riscv/mm/Makefile | 1 + arch/riscv/mm/cacheflush.c | 38 ++++++++++++++++++++++++++ arch/riscv/mm/dma-noncoherent.c | 41 ----------------------------- arch/riscv/mm/pmem.c | 21 +++++++++++++++ 9 files changed, 74 insertions(+), 49 deletions(-) create mode 100644 arch/riscv/mm/pmem.c
Signature · 7 lines
2.34.1 _______________________________________________ linux-riscv mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-riscv
PATCH v4 1/4 RISC-V: Fix compilation without RISCV_ISA_ZICBOM
From: Andrew Jones <[email protected]>
riscv_cbom_block_size and riscv_init_cbom_blocksize() should always be available and riscv_init_cbom_blocksize() should always be invoked, even when compiling without RISCV_ISA_ZICBOM enabled. This is because disabling RISCV_ISA_ZICBOM means "don't use zicbom instructions in the kernel" not "pretend there isn't zicbom, even when there is". When zicbom is available, whether the kernel enables its use with RISCV_ISA_ZICBOM or not, KVM will offer it to guests. Ensure we can build KVM and that the block size is initialized even when compiling without RISCV_ISA_ZICBOM.
--- arch/riscv/include/asm/cacheflush.h | 8 ------ arch/riscv/mm/cacheflush.c | 38 ++++++++++++++++++++++++++ arch/riscv/mm/dma-noncoherent.c | 41 ----------------------------- 3 files changed, 38 insertions(+), 49 deletions(-)
arch/riscv/include/asm/cacheflush.h +0 -8
@@ -42,16 +42,8 @@ void flush_icache_mm(struct mm_struct *mm, bool local);#endif /* CONFIG_SMP */ -/* -* The T-Head CMO errata internally probe the CBOM block size, but otherwise -* don't depend on Zicbom. -*/ extern unsigned int riscv_cbom_block_size; -#ifdef CONFIG_RISCV_ISA_ZICBOM void riscv_init_cbom_blocksize (void ); -#else -static inline void riscv_init_cbom_blocksize (void ) { } -#endif #ifdef CONFIG_RISCV_DMA_NONCOHERENT void riscv_noncoherent_supported (void );
arch/riscv/mm/cacheflush.c +38 -0
@@ -3,6 +3,7 @@* Copyright (C) 2017 SiFive */ +#include <linux/of.h> #include <asm/cacheflush.h> #ifdef CONFIG_SMP
@@ -86,3 +87,40 @@ void flush_icache_pte(pte_t pte) flush_icache_all(); }#endif /* CONFIG_MMU */ + +unsigned int riscv_cbom_block_size; +EXPORT_SYMBOL_GPL (riscv_cbom_block_size); + +void riscv_init_cbom_blocksize (void ) +{ +struct device_node *node; +unsigned long cbom_hartid; +u32 val, probed_block_size; +int ret; + + probed_block_size =0 ; +for_each_of_cpu_node (node) { +unsigned long hartid; + + ret =riscv_of_processor_hartid (node, &hartid); +if (ret) +continue ; + +/* set block-size for cbom extension if available */ + ret =of_property_read_u32 (node,"riscv,cbom-block-size" , &val); +if (ret) +continue ; + +if (!probed_block_size) { + probed_block_size = val; + cbom_hartid = hartid; + }else { +if (probed_block_size != val) +pr_warn ("cbom-block-size mismatched between harts %lu and %lu\n" , + cbom_hartid, hartid); + } + } + +if (probed_block_size) + riscv_cbom_block_size = probed_block_size; +}
arch/riscv/mm/dma-noncoherent.c +0 -41
55 lines not on this page · raw message
Signature · 7 lines
2.34.1 _______________________________________________ linux-riscv mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-riscv
Re: [PATCH v4 1/4] RISC-V: Fix compilation without RISCV_ISA_ZICBOM
Am Mittwoch, 19. Oktober 2022, 15:11:25 CEST schrieb Anup Patel:
Quoted 12 lines
From: Andrew Jones <[email protected]>
riscv_cbom_block_sizeandriscv_init_cbom_blocksize()should always be available andriscv_init_cbom_blocksize()should always be invoked, even when compiling withoutRISCV_ISA_ZICBOMenabled. This is because disablingRISCV_ISA_ZICBOMmeans "don't use zicbom instructions in the kernel" not "pretend there isn't zicbom, even when there is". When zicbom is available, whether the kernel enables its use withRISCV_ISA_ZICBOMor not, KVM will offer it to guests. Ensure we can build KVM and that the block size is initialized even when compiling withoutRISCV_ISA_ZICBOM.
Quoted 5 lines
Fixes 8f7e001e0325 ("RISC-V: Clean up the Zicbom block size probing")Reported-by kernel test robot <[email protected]>Signed-off-by Andrew Jones <[email protected]>Signed-off-by Anup Patel <[email protected]>Reviewed-by Conor Dooley <[email protected]>
_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv
PATCH v4 2/4 RISC-V: Fix ioremap_cache() and ioremap_wc() for systems with Svpbmt
In reply to Anup PatelCurrently, all flavors of ioremap_xyz() function maps to the generic ioremap() which means any ioremap_xyz() call will always map the target memory as IO using _PAGE_IOREMAP page attributes. This breaks ioremap_cache() and ioremap_wc() on systems with Svpbmt because memory remapped using ioremap_cache() and ioremap_wc() will use _PAGE_IOREMAP page attributes.
To address above (just like other architectures), we implement RISC-V specific ioremap_cache() and ioremap_wc() which maps memory using page attributes as defined by the Svpbmt specification.
--- arch/riscv/include/asm/io.h | 10 ++++++++++ arch/riscv/include/asm/pgtable.h | 2 ++ 2 files changed, 12 insertions(+)
arch/riscv/include/asm/io.h +10 -0
@@ -133,6 +133,16 @@ __io_writes_outs(outs, u64, q, __io_pbr(), __io_paw())#define outsq (addr, buffer, count) __outsq(PCI_IOBASE + (addr), buffer, count)#endif +#ifdef CONFIG_MMU +#define ioremap_wc (addr, size) \ + ioremap_prot((addr), (size), _PAGE_IOREMAP_WC) +#endif +#include <asm-generic/io.h> +#ifdef CONFIG_MMU +#define ioremap_cache (addr, size) \ + ioremap_prot((addr), (size), _PAGE_KERNEL) +#endif +#endif /* _ASM_RISCV_IO_H */
arch/riscv/include/asm/pgtable.h +2 -0
@@ -182,6 +182,8 @@ extern struct pt_alloc_ops pt_ops __initdata;#define PAGE_TABLE __pgprot(_PAGE_TABLE)#define _PAGE_IOREMAP ((_PAGE_KERNEL & ~_PAGE_MTMASK) | _PAGE_IO) +#define _PAGE_IOREMAP_WC ((_PAGE_KERNEL & ~_PAGE_MTMASK) | \ + _PAGE_NOCACHE)#define PAGE_KERNEL_IO __pgprot(_PAGE_IOREMAP)extern pgd_t swapper_pg_dir[];
Signature · 7 lines
2.34.1 _______________________________________________ linux-riscv mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-riscv
Re: [PATCH v4 2/4] RISC-V: Fix ioremap_cache() and ioremap_wc() for systems with Svpbmt
Am Mittwoch, 19. Oktober 2022, 15:11:26 CEST schrieb Anup Patel:
Quoted 7 lines
Currently, all flavors of
ioremap_xyz()function maps to the genericioremap()which means anyioremap_xyz()call will always map the target memory as IO using_PAGE_IOREMAPpage attributes. This breaksioremap_cache()andioremap_wc()on systems with Svpbmt because memory remapped usingioremap_cache()andioremap_wc()will use_PAGE_IOREMAPpage attributes.
Quoted 8 lines
To address above (just like other architectures), we implement RISC-V specific
ioremap_cache()andioremap_wc()which maps memory using page attributes as defined by the Svpbmt specification.Fixes ff689fd21cb1 ("riscv: add RISC-V Svpbmt extension support")Co-developed-by Mayuresh Chitale <[email protected]>Signed-off-by Mayuresh Chitale <[email protected]>Signed-off-by Anup Patel <[email protected]>
Wasn't there discussion around those functions in general in v2?
In any case, the patch doesn't break anything on qemu and d1, so
Quoted 40 lines
--- arch/riscv/include/asm/io.h | 10 ++++++++++ arch/riscv/include/asm/pgtable.h | 2 ++ 2 files changed, 12 insertions(+)arch/riscv/include/asm/io.h +10
-0@@ -133,6 +133,16 @@ __io_writes_outs(outs, u64, q, __io_pbr(), __io_paw())#define outsq (addr, buffer, count) __outsq(PCI_IOBASE + (addr), buffer, count)#endif +#ifdef CONFIG_MMU +#define ioremap_wc (addr, size) \ + ioremap_prot((addr), (size), _PAGE_IOREMAP_WC) +#endif +#include <asm-generic/io.h> +#ifdef CONFIG_MMU +#define ioremap_cache (addr, size) \ + ioremap_prot((addr), (size), _PAGE_KERNEL) +#endif +#endif /* _ASM_RISCV_IO_H */ arch/riscv/include/asm/pgtable.h +2
-0@@ -182,6 +182,8 @@ extern struct pt_alloc_ops pt_ops __initdata;#define PAGE_TABLE __pgprot(_PAGE_TABLE)#define _PAGE_IOREMAP ((_PAGE_KERNEL & ~_PAGE_MTMASK) | _PAGE_IO) +#define _PAGE_IOREMAP_WC ((_PAGE_KERNEL & ~_PAGE_MTMASK) | \ + _PAGE_NOCACHE)#define PAGE_KERNEL_IO __pgprot(_PAGE_IOREMAP)extern pgd_t swapper_pg_dir[];
Re: [PATCH v4 2/4] RISC-V: Fix ioremap_cache() and ioremap_wc() for systems with Svpbmt
Quoted 13 lines · Heiko Stuebner
Am Mittwoch, 19. Oktober 2022, 15:11:26 CEST schrieb Anup Patel:
Quoted 11 lines
Currently, all flavors of
ioremap_xyz()function maps to the genericioremap()which means anyioremap_xyz()call will always map the target memory as IO using_PAGE_IOREMAPpage attributes. This breaksioremap_cache()andioremap_wc()on systems with Svpbmt because memory remapped usingioremap_cache()andioremap_wc()will use_PAGE_IOREMAPpage attributes.To address above (just like other architectures), we implement RISC-V specific
ioremap_cache()andioremap_wc()which maps memory using page attributes as defined by the Svpbmt specification.
Quoted 6 lines
Quoted 4 lines
Fixes ff689fd21cb1 ("riscv: add RISC-V Svpbmt extension support")Co-developed-by Mayuresh Chitale <[email protected]>Signed-off-by Mayuresh Chitale <[email protected]>Signed-off-by Anup Patel <[email protected]>Wasn't there discussion around those functions in general in v2?
Yes, there was discussion about a few drivers using ioremap_xyz() which is discouraged and drivers should use memremap().
We still need the arch specific ioremap_xyz() functions/macros added by this patch because these are required by the generic kernel memremap() implementation (refer, kernel/iomem.c).
Quoted 4 lines
In any case, the patch doesn't break anything on qemu and d1, so
Tested-by Heiko Stuebner <[email protected]>
Regards,
Anup
Quoted 46 lines
Quoted 42 lines
--- arch/riscv/include/asm/io.h | 10 ++++++++++ arch/riscv/include/asm/pgtable.h | 2 ++ 2 files changed, 12 insertions(+)arch/riscv/include/asm/io.h +10
-0@@ -133,6 +133,16 @@ __io_writes_outs(outs, u64, q, __io_pbr(), __io_paw())#define outsq (addr, buffer, count) __outsq(PCI_IOBASE + (addr), buffer, count)#endif +#ifdef CONFIG_MMU +#define ioremap_wc (addr, size) \ + ioremap_prot((addr), (size), _PAGE_IOREMAP_WC) +#endif +#include <asm-generic/io.h> +#ifdef CONFIG_MMU +#define ioremap_cache (addr, size) \ + ioremap_prot((addr), (size), _PAGE_KERNEL) +#endif +#endif /* _ASM_RISCV_IO_H */ arch/riscv/include/asm/pgtable.h +2
-0@@ -182,6 +182,8 @@ extern struct pt_alloc_ops pt_ops __initdata;#define PAGE_TABLE __pgprot(_PAGE_TABLE)#define _PAGE_IOREMAP ((_PAGE_KERNEL & ~_PAGE_MTMASK) | _PAGE_IO) +#define _PAGE_IOREMAP_WC ((_PAGE_KERNEL & ~_PAGE_MTMASK) | \ + _PAGE_NOCACHE)#define PAGE_KERNEL_IO __pgprot(_PAGE_IOREMAP)extern pgd_t swapper_pg_dir[];
Re: [PATCH v4 2/4] RISC-V: Fix ioremap_cache() and ioremap_wc() for systems with Svpbmt
Quoted 19 lines · Anup Patel
Quoted 18 lines · Heiko Stuebner
Am Mittwoch, 19. Oktober 2022, 15:11:26 CEST schrieb Anup Patel:
Quoted 15 lines
Currently, all flavors of
ioremap_xyz()function maps to the genericioremap()which means anyioremap_xyz()call will always map the target memory as IO using_PAGE_IOREMAPpage attributes. This breaksioremap_cache()andioremap_wc()on systems with Svpbmt because memory remapped usingioremap_cache()andioremap_wc()will use_PAGE_IOREMAPpage attributes.To address above (just like other architectures), we implement RISC-V specific
ioremap_cache()andioremap_wc()which maps memory using page attributes as defined by the Svpbmt specification.Fixes ff689fd21cb1 ("riscv: add RISC-V Svpbmt extension support")Co-developed-by Mayuresh Chitale <[email protected]>Signed-off-by Mayuresh Chitale <[email protected]>Signed-off-by Anup Patel <[email protected]>
Quoted 8 lines
Quoted 1 line
Wasn't there discussion around those functions in general in v2?
Yes, there was discussion about a few drivers using
ioremap_xyz()which is discouraged and drivers should usememremap().We still need the arch specific
ioremap_xyz()functions/macros added by this patch because these are required by the generic kernelmemremap()implementation (refer,kernel/iomem.c).
There is a difference between the strongly discouraged ioremap_cache() that pretty much has no valid users, and the ioremap_wt/ioremap_wc functions that are sometimes used for mapping video framebuffer or similar.
It should be sufficient to provide a arch_memremap_wb() and no ioremap_cache() to make memremap() work correctly.
Arnd
Re: [PATCH v4 2/4] RISC-V: Fix ioremap_cache() and ioremap_wc() for systems with Svpbmt
Quoted 30 lines · Arnd Bergmann
Quoted 19 lines · Anup Patel
Quoted 18 lines · Heiko Stuebner
Am Mittwoch, 19. Oktober 2022, 15:11:26 CEST schrieb Anup Patel:
Quoted 15 lines
Currently, all flavors of
ioremap_xyz()function maps to the genericioremap()which means anyioremap_xyz()call will always map the target memory as IO using_PAGE_IOREMAPpage attributes. This breaksioremap_cache()andioremap_wc()on systems with Svpbmt because memory remapped usingioremap_cache()andioremap_wc()will use_PAGE_IOREMAPpage attributes.To address above (just like other architectures), we implement RISC-V specific
ioremap_cache()andioremap_wc()which maps memory using page attributes as defined by the Svpbmt specification.Fixes ff689fd21cb1 ("riscv: add RISC-V Svpbmt extension support")Co-developed-by Mayuresh Chitale <[email protected]>Signed-off-by Mayuresh Chitale <[email protected]>Signed-off-by Anup Patel <[email protected]>Quoted 8 lines
Quoted 1 line
Wasn't there discussion around those functions in general in v2?
Yes, there was discussion about a few drivers using
ioremap_xyz()which is discouraged and drivers should usememremap().We still need the arch specific
ioremap_xyz()functions/macros added by this patch because these are required by the generic kernelmemremap()implementation (refer,kernel/iomem.c).
Quoted 8 lines
There is a difference between the strongly discouraged
ioremap_cache()that pretty much has no valid users, and the ioremap_wt/ioremap_wc functions that are sometimes used for mapping video framebuffer or similar.It should be sufficient to provide a
arch_memremap_wb()and noioremap_cache()to makememremap()work correctly.
Okay, I will simplify this patch to only implement arch_memremap_wb().
This will make the MEMREMAP_WB flag to work correctly but other MEMREMAP_xyz flags will always use non-cacheable IO mappings.
Regards,
Anup
_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv
PATCH v4 3/4 RISC-V: Implement arch specific PMEM APIs
In reply to Anup PatelThe 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.
--- arch/riscv/Kconfig | 1 + arch/riscv/mm/Makefile | 1 + arch/riscv/mm/pmem.c | 21 +++++++++++++++++++++ 3 files changed, 23 insertions(+) create mode 100644 arch/riscv/mm/pmem.c
arch/riscv/Kconfig +1 -0
@@ -25,6 +25,7 @@ config RISCV select ARCH_HAS_GIGANTIC_PAGE select ARCH_HAS_KCOV select ARCH_HAS_MMIOWB + select ARCH_HAS_PMEM_API select ARCH_HAS_PTE_SPECIAL select ARCH_HAS_SET_DIRECT_MAP if MMU select ARCH_HAS_SET_MEMORY if MMU
arch/riscv/mm/Makefile +1 -0
@@ -31,3 +31,4 @@ endif obj-$(CONFIG_DEBUG_VIRTUAL) += physaddr.o obj-$(CONFIG_RISCV_DMA_NONCOHERENT) += dma-noncoherent.o +obj-$(CONFIG_ARCH_HAS_PMEM_API) += pmem.o
arch/riscv/mm/pmem.c new file +21 -0
@@ -0,0 +1,21 @@ +// SPDX-License-Identifier: GPL-2.0 +/* +* Copyright (C) 2022 Ventana Micro Systems Inc. +*/ + +#include <linux/export.h> +#include <linux/libnvdimm.h> + +#include <asm/cacheflush.h> + +void arch_wb_cache_pmem (void *addr,size_t size) +{ +ALT_CMO_OP (clean, addr, size, riscv_cbom_block_size); +} +EXPORT_SYMBOL_GPL (arch_wb_cache_pmem); + +void arch_invalidate_pmem (void *addr,size_t size) +{ +ALT_CMO_OP (inval, addr, size, riscv_cbom_block_size); +} +EXPORT_SYMBOL_GPL (arch_invalidate_pmem);
Signature · 7 lines
2.34.1 _______________________________________________ linux-riscv mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-riscv
PATCH v4 4/4 RISC-V: Enable PMEM drivers
In reply to Anup PatelWe now have PMEM arch support available in RISC-V kernel so let us enable relevant drivers in defconfig.
--- arch/riscv/configs/defconfig | 1 + 1 file changed, 1 insertion(+)
arch/riscv/configs/defconfig +1 -0
@@ -159,6 +159,7 @@ CONFIG_VIRTIO_MMIO=y CONFIG_RPMSG_CHAR=y CONFIG_RPMSG_CTRL=y CONFIG_RPMSG_VIRTIO=y +CONFIG_LIBNVDIMM=y CONFIG_EXT4_FS=y CONFIG_EXT4_FS_POSIX_ACL=y CONFIG_EXT4_FS_SECURITY=y
Signature · 7 lines
2.34.1 _______________________________________________ linux-riscv mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-riscv