RISC-V: KVM: Fix compilation without RISCV_ISA_ZICBOM
Subsystems (2)
PATCH 1/1 RISC-V: KVM: Fix compilation without RISCV_ISA_ZICBOM
Fix undefined reference of riscv_cbom_block_size when compiling KVM without RISCV_ISA_ZICBOM. Note, RISCV_ISA_ZICBOM is a sufficient guard as it selects RISCV_DMA_NONCOHERENT, which is needed to compile dma-noncoherent.c (which is the file where riscv_cbom_block_size and its initializer live).
--- arch/riscv/kvm/vcpu.c | 2 ++ 1 file changed, 2 insertions(+)
arch/riscv/kvm/vcpu.c +2 -0
@@ -265,11 +265,13 @@ static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu,case KVM_REG_RISCV_CONFIG_REG (isa): reg_val = vcpu->arch.isa[0 ] &KVM_RISCV_BASE_ISA_MASK ;break ; +#ifdef CONFIG_RISCV_ISA_ZICBOM case KVM_REG_RISCV_CONFIG_REG (zicbom_block_size):if (!riscv_isa_extension_available (vcpu->arch.isa,ZICBOM ))return -EINVAL ; reg_val = riscv_cbom_block_size;break ; +#endif default :return -EINVAL ; }
Signature · 1 line
2.37.3
Palmer Dabbelt
maintainerPATCH 1/1 RISC-V: KVM: Fix compilation without RISCV_ISA_ZICBOM
Quoted 21 lines · ajones at ventanamicro.com
Fix undefined reference of
riscv_cbom_block_sizewhen compiling KVM withoutRISCV_ISA_ZICBOM. Note,RISCV_ISA_ZICBOMis a sufficient guard as it selectsRISCV_DMA_NONCOHERENT, which is needed to compiledma-noncoherent.c(which is the file whereriscv_cbom_block_sizeand its initializer live).Fixes afd5dde9a186 ("RISC-V: KVM: Provide UAPI for Zicbom block size")Reported-by kernel test robot <[email protected]>Signed-off-by Andrew Jones <[email protected]>--- arch/riscv/kvm/vcpu.c | 2 ++ 1 file changed, 2 insertions(+)arch/riscv/kvm/vcpu.c +0
-0@@ -265,11 +265,13 @@ static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu,case KVM_REG_RISCV_CONFIG_REG (isa): reg_val = vcpu->arch.isa[0 ] &KVM_RISCV_BASE_ISA_MASK ;break ;
arch/riscv/kvm/vcpu.c static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu, · line 277
Quoted 10 lines
arch/riscv/kvm/vcpu.c +2
-0+#ifdef CONFIG_RISCV_ISA_ZICBOM case KVM_REG_RISCV_CONFIG_REG (zicbom_block_size):if (!riscv_isa_extension_available (vcpu->arch.isa,ZICBOM ))return -EINVAL ; reg_val = riscv_cbom_block_size;break ; +#endif default :return -EINVAL ; }
Thanks, looks like this is tripping up the kernelci builds too:
https://linux.kernelci.org/build/id/6343b37c6fd2dcbb5dcab5f3/logs/ . This symbol is actually a bit odd, as we don't actually just use it under Zicbom because of the T-Head stuff. Looks like that's generically broken, though, so IMO we need something like this
arch/riscv/mm/cacheflush.c +3 -0
@@ -5,6 +5,9 @@#include <asm/cacheflush.h> +unsigned int riscv_cbom_block_size; +EXPORT_SYMBOL (riscv_cbom_block_size); +#ifdef CONFIG_SMP #include <asm/sbi.h>
arch/riscv/mm/dma-noncoherent.c +0 -1
@@ -12,7 +12,6 @@#include <linux/of_device.h> #include <asm/cacheflush.h> -unsigned int riscv_cbom_block_size;static bool noncoherent_supported; void arch_sync_dma_for_device(phys_addr_t paddr, size_t size,
which should be
We might also want the ifdef for KVM, as that would allow us to build KVM that doesn't support these VCPU configurations, but that's sort of a different question.
PATCH 1/1 RISC-V: KVM: Fix compilation without RISCV_ISA_ZICBOM
Quoted 66 lines · Palmer Dabbelt
Quoted 21 lines · ajones at ventanamicro.com
Fix undefined reference of
riscv_cbom_block_sizewhen compiling KVM withoutRISCV_ISA_ZICBOM. Note,RISCV_ISA_ZICBOMis a sufficient guard as it selectsRISCV_DMA_NONCOHERENT, which is needed to compiledma-noncoherent.c(which is the file whereriscv_cbom_block_sizeand its initializer live).Fixes afd5dde9a186 ("RISC-V: KVM: Provide UAPI for Zicbom block size")Reported-by kernel test robot <[email protected]>Signed-off-by Andrew Jones <[email protected]>--- arch/riscv/kvm/vcpu.c | 2 ++ 1 file changed, 2 insertions(+)arch/riscv/kvm/vcpu.c +0
-0@@ -265,11 +265,13 @@ static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu,case KVM_REG_RISCV_CONFIG_REG (isa): reg_val = vcpu->arch.isa[0 ] &KVM_RISCV_BASE_ISA_MASK ;break ;arch/riscv/kvm/vcpu.c static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu, · line 277
Quoted 10 lines
arch/riscv/kvm/vcpu.c +2
-0+#ifdef CONFIG_RISCV_ISA_ZICBOM case KVM_REG_RISCV_CONFIG_REG (zicbom_block_size):if (!riscv_isa_extension_available (vcpu->arch.isa,ZICBOM ))return -EINVAL ; reg_val = riscv_cbom_block_size;break ; +#endif default :return -EINVAL ; }Thanks, looks like this is tripping up the kernelci builds too:
https://linux.kernelci.org/build/id/6343b37c6fd2dcbb5dcab5f3/logs/ . This symbol is actually a bit odd, as we don't actually just use it under Zicbom because of the T-Head stuff. Looks like that's generically broken, though, so IMO we need something like thisarch/riscv/mm/cacheflush.c +3
-0@@ -5,6 +5,9 @@#include <asm/cacheflush.h> +unsigned int riscv_cbom_block_size; +EXPORT_SYMBOL (riscv_cbom_block_size); +#ifdef CONFIG_SMP #include <asm/sbi.h> arch/riscv/mm/dma-noncoherent.c +0
-1@@ -12,7 +12,6 @@#include <linux/of_device.h> #include <asm/cacheflush.h> -unsigned int riscv_cbom_block_size;static bool noncoherent_supported; void arch_sync_dma_for_device(phys_addr_t paddr, size_t size,
Quoted 7 lines
which should be
Fixes 8f7e001e0325 ("RISC-V: Clean up the Zicbom block size probing").We might also want the ifdef for KVM, as that would allow us to build KVM that doesn't support these VCPU configurations, but that's sort of a different question.
I assume (or hope?) this approach has the advantage of also fixing:
https://lore.kernel.org/all/202210102012.iShn6U6Q-lkp at intel.com/
PATCH 1/1 RISC-V: KVM: Fix compilation without RISCV_ISA_ZICBOM
In reply to Palmer DabbeltQuoted 67 lines · Palmer Dabbelt
Quoted 21 lines · ajones at ventanamicro.com
Fix undefined reference of
riscv_cbom_block_sizewhen compiling KVM withoutRISCV_ISA_ZICBOM. Note,RISCV_ISA_ZICBOMis a sufficient guard as it selectsRISCV_DMA_NONCOHERENT, which is needed to compiledma-noncoherent.c(which is the file whereriscv_cbom_block_sizeand its initializer live).Fixes afd5dde9a186 ("RISC-V: KVM: Provide UAPI for Zicbom block size")Reported-by kernel test robot <[email protected]>Signed-off-by Andrew Jones <[email protected]>--- arch/riscv/kvm/vcpu.c | 2 ++ 1 file changed, 2 insertions(+)arch/riscv/kvm/vcpu.c +0
-0@@ -265,11 +265,13 @@ static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu,case KVM_REG_RISCV_CONFIG_REG (isa): reg_val = vcpu->arch.isa[0 ] &KVM_RISCV_BASE_ISA_MASK ;break ;arch/riscv/kvm/vcpu.c static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu, · line 277
Quoted 10 lines
arch/riscv/kvm/vcpu.c +2
-0+#ifdef CONFIG_RISCV_ISA_ZICBOM case KVM_REG_RISCV_CONFIG_REG (zicbom_block_size):if (!riscv_isa_extension_available (vcpu->arch.isa,ZICBOM ))return -EINVAL ; reg_val = riscv_cbom_block_size;break ; +#endif default :return -EINVAL ; }Thanks, looks like this is tripping up the kernelci builds too:
https://linux.kernelci.org/build/id/6343b37c6fd2dcbb5dcab5f3/logs/ . This symbol is actually a bit odd, as we don't actually just use it under Zicbom because of the T-Head stuff. Looks like that's generically broken, though, so IMO we need something like thisarch/riscv/mm/cacheflush.c +3
-0@@ -5,6 +5,9 @@#include <asm/cacheflush.h> +unsigned int riscv_cbom_block_size; +EXPORT_SYMBOL (riscv_cbom_block_size); +#ifdef CONFIG_SMP #include <asm/sbi.h> arch/riscv/mm/dma-noncoherent.c +0
-1@@ -12,7 +12,6 @@#include <linux/of_device.h> #include <asm/cacheflush.h> -unsigned int riscv_cbom_block_size;static bool noncoherent_supported; void arch_sync_dma_for_device(phys_addr_t paddr, size_t size,
Quoted 3 lines
which should be
Fixes 8f7e001e0325 ("RISC-V: Clean up the Zicbom block size probing").
I agree with your change.
May be also move riscv_init_cbom_blocksize() toarch/riscv/mm/cacheflush.c ?
Quoted 4 lines
We might also want the ifdef for KVM, as that would allow us to build KVM that doesn't support these VCPU configurations, but that's sort of a different question.
Yes, I will be sending a PR including the KVM fix this week itself.
Quoted 5 lines
--
kvm-riscv mailing list
kvm-riscv at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
Regards,
Anup
PATCH 1/1 RISC-V: KVM: Fix compilation without RISCV_ISA_ZICBOM
In reply to Palmer DabbeltQuoted 66 lines · Palmer Dabbelt
Quoted 21 lines · ajones at ventanamicro.com
Fix undefined reference of
riscv_cbom_block_sizewhen compiling KVM withoutRISCV_ISA_ZICBOM. Note,RISCV_ISA_ZICBOMis a sufficient guard as it selectsRISCV_DMA_NONCOHERENT, which is needed to compiledma-noncoherent.c(which is the file whereriscv_cbom_block_sizeand its initializer live).Fixes afd5dde9a186 ("RISC-V: KVM: Provide UAPI for Zicbom block size")Reported-by kernel test robot <[email protected]>Signed-off-by Andrew Jones <[email protected]>--- arch/riscv/kvm/vcpu.c | 2 ++ 1 file changed, 2 insertions(+)arch/riscv/kvm/vcpu.c +0
-0@@ -265,11 +265,13 @@ static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu,case KVM_REG_RISCV_CONFIG_REG (isa): reg_val = vcpu->arch.isa[0 ] &KVM_RISCV_BASE_ISA_MASK ;break ;arch/riscv/kvm/vcpu.c static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu, · line 277
Quoted 10 lines
arch/riscv/kvm/vcpu.c +2
-0+#ifdef CONFIG_RISCV_ISA_ZICBOM case KVM_REG_RISCV_CONFIG_REG (zicbom_block_size):if (!riscv_isa_extension_available (vcpu->arch.isa,ZICBOM ))return -EINVAL ; reg_val = riscv_cbom_block_size;break ; +#endif default :return -EINVAL ; }Thanks, looks like this is tripping up the kernelci builds too:
https://linux.kernelci.org/build/id/6343b37c6fd2dcbb5dcab5f3/logs/ . This symbol is actually a bit odd, as we don't actually just use it under Zicbom because of the T-Head stuff. Looks like that's generically broken, though, so IMO we need something like thisarch/riscv/mm/cacheflush.c +3
-0@@ -5,6 +5,9 @@#include <asm/cacheflush.h> +unsigned int riscv_cbom_block_size; +EXPORT_SYMBOL (riscv_cbom_block_size); +#ifdef CONFIG_SMP #include <asm/sbi.h> arch/riscv/mm/dma-noncoherent.c +0
-1@@ -12,7 +12,6 @@#include <linux/of_device.h> #include <asm/cacheflush.h> -unsigned int riscv_cbom_block_size;static bool noncoherent_supported; void arch_sync_dma_for_device(phys_addr_t paddr, size_t size,
Quoted 3 lines
which should be
Fixes 8f7e001e0325 ("RISC-V: Clean up the Zicbom block size probing").
Hi Palmer,
Should I post your suggestion as a v2 or were you planning to post it? I'm happy to do so, but I'd prefer to also move riscv_init_cbom_blocksize() and drop its #ifdef CONFIG_RISCV_ISA_ZICBOM guard. The reason I'd prefer we always try to parse the block size is that we always parse the ISA string, and, when zicbom is present, we set it in the ISA extension bitmap, even when CONFIG_RISCV_ISA_ZICBOM isn't enabled. This is a reasonable behavior if CONFIG_RISCV_ISA_ZICBOM is meant to decide whether or not the kernel uses the feature for itself, as its Kconfig text implies, rather than to mean we should pretend the feature isn't present, whether it is or not. So, if we're not trying to completely avoid the feature by disabling CONFIG_RISCV_ISA_ZICBOM, then KVM can still inform the guest about it. When KVM does that it also needs the block size.
Thanks,
drew
PATCH 1/1 RISC-V: KVM: Fix compilation without RISCV_ISA_ZICBOM
Quoted 73 lines · Andrew Jones
Quoted 66 lines · Palmer Dabbelt
Quoted 21 lines · ajones at ventanamicro.com
Fix undefined reference of
riscv_cbom_block_sizewhen compiling KVM withoutRISCV_ISA_ZICBOM. Note,RISCV_ISA_ZICBOMis a sufficient guard as it selectsRISCV_DMA_NONCOHERENT, which is needed to compiledma-noncoherent.c(which is the file whereriscv_cbom_block_sizeand its initializer live).Fixes afd5dde9a186 ("RISC-V: KVM: Provide UAPI for Zicbom block size")Reported-by kernel test robot <[email protected]>Signed-off-by Andrew Jones <[email protected]>--- arch/riscv/kvm/vcpu.c | 2 ++ 1 file changed, 2 insertions(+)arch/riscv/kvm/vcpu.c +0
-0@@ -265,11 +265,13 @@ static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu,case KVM_REG_RISCV_CONFIG_REG (isa): reg_val = vcpu->arch.isa[0 ] &KVM_RISCV_BASE_ISA_MASK ;break ;arch/riscv/kvm/vcpu.c static int kvm_riscv_vcpu_get_reg_config(struct kvm_vcpu *vcpu, · line 277
Quoted 10 lines
arch/riscv/kvm/vcpu.c +2
-0+#ifdef CONFIG_RISCV_ISA_ZICBOM case KVM_REG_RISCV_CONFIG_REG (zicbom_block_size):if (!riscv_isa_extension_available (vcpu->arch.isa,ZICBOM ))return -EINVAL ; reg_val = riscv_cbom_block_size;break ; +#endif default :return -EINVAL ; }Thanks, looks like this is tripping up the kernelci builds too:
https://linux.kernelci.org/build/id/6343b37c6fd2dcbb5dcab5f3/logs/ . This symbol is actually a bit odd, as we don't actually just use it under Zicbom because of the T-Head stuff. Looks like that's generically broken, though, so IMO we need something like thisarch/riscv/mm/cacheflush.c +3
-0@@ -5,6 +5,9 @@#include <asm/cacheflush.h> +unsigned int riscv_cbom_block_size; +EXPORT_SYMBOL (riscv_cbom_block_size); +#ifdef CONFIG_SMP #include <asm/sbi.h> arch/riscv/mm/dma-noncoherent.c +0
-1@@ -12,7 +12,6 @@#include <linux/of_device.h> #include <asm/cacheflush.h> -unsigned int riscv_cbom_block_size;static bool noncoherent_supported; void arch_sync_dma_for_device(phys_addr_t paddr, size_t size,Quoted 3 lines
which should be
Fixes 8f7e001e0325 ("RISC-V: Clean up the Zicbom block size probing").Hi Palmer,
Quoted 13 lines
Should I post your suggestion as a v2 or were you planning to post it? I'm happy to do so, but I'd prefer to also move
riscv_init_cbom_blocksize()and drop its #ifdefCONFIG_RISCV_ISA_ZICBOMguard. The reason I'd prefer we always try to parse the block size is that we always parse the ISA string, and, when zicbom is present, we set it in the ISA extension bitmap, even whenCONFIG_RISCV_ISA_ZICBOMisn't enabled. This is a reasonable behavior ifCONFIG_RISCV_ISA_ZICBOMis meant to decide whether or not the kernel uses the feature for itself, as itsKconfigtext implies, rather than to mean we should pretend the feature isn't present, whether it is or not. So, if we're not trying to completely avoid the feature by disablingCONFIG_RISCV_ISA_ZICBOM, then KVM can still inform the guest about it. When KVM does that it also needs the block size.
I went ahead and posted a v2 since it may be easier to communicate the idea by patch.
Thanks,
drew