RISC-V: KVM: AIA: Convert HGEI management to fully per-HART
Subsystems (5)
v2 cover RISC-V: KVM: AIA: Convert HGEI management to fully per-HART
From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
This short series converts RISC-V KVM AIA's Hypervisor Guest External Interrupt (HGEI) line management from a global "one-size-fits-all" model to a fully per-HART (per-CPU) model. It also performs the corresponding cleanup in the IMSIC irqchip driver.
The motivation is to properly support heterogeneous RISC-V SoCs (big.LITTLE, multi-vendor core mixes) where different HARTs may expose different numbers of guest interrupt files / HGEIE bits. The old global kvm_riscv_aia_nr_hgei and imsic_global_config::nr_guest_files assumptions are no longer tenable and are removed.
After this series, HGEI allocation, freeing, and interrupt delivery are completely per-HART, the code is simpler and clearer, and there are no remaining global assumptions about guest interrupt file counts.
Changelog:
V2:
- Add per-HART IMSIC guest files to compute final HGEI count.
- Add min(local->nr_guest_files, nr_guest_files);
V1:
https://lore.kernel.org/kvm-riscv/[email protected]/
Guo Ren (Alibaba DAMO Academy) (4):
RISC-V: KVM: AIA: Make HGEI number and management fully per-CPU
RISC-V: KVM: AIA: Replace global HGEI count with simple enabled bool irqchip/riscv-imsic: Move nr_guest_files to per-HART local config
RISC-V: KVM: AIA: Use per-HART IMSIC guest files to compute final HGEI
count
arch/riscv/include/asm/kvm_aia.h | 2 +- arch/riscv/kvm/aia.c | 66 +++++++++++++++---------- arch/riscv/kvm/aia_device.c | 4 +- arch/riscv/kvm/main.c | 3 +- drivers/irqchip/irq-riscv-imsic-state.c | 10 ++-- include/linux/irqchip/riscv-imsic.h | 6 +-- 6 files changed, 53 insertions(+), 38 deletions(-)
Signature · 1 line
2.43.0
PATCH v2 1/4 RISC-V: KVM: AIA: Make HGEI number and management fully per-CPU
From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
Previously the number of Hypervisor Guest External Interrupt (HGEI) lines was stored in a single global variable kvm_riscv_aia_nr_hgei and assumed to be the same for all HARTs. This assumption does not hold on heterogeneous RISC-V SoCs where different cores may expose different HGEIE CSR widths.
Introduce nr_hgei field into the per-CPU struct aia_hgei_control and probe the actual supported HGEI count for the current HART in kvm_riscv_aia_enable() using the standard RISC-V CSR probe technique:
csr_write(CSR_HGEIE, -1UL);
nr = fls_long(csr_read(CSR_HGEIE));
if (nr)
nr--;All HGEI allocation, free and disable paths (kvm_riscv_aia_free_hgei(), kvm_riscv_aia_disable(), etc.) now use the per-CPU value instead of the global one.
The early global kvm_riscv_aia_nr_hgei is kept only for deciding whether SGEI interrupt registration is needed; the real per-HART initialization of lock and free_bitmap is moved to enable time.
This makes KVM AIA robust on big.LITTLE-style and multi-vendor asymmetric platforms.
--- arch/riscv/kvm/aia.c | 40 ++++++++++++++++++++++++---------------- 1 file changed, 24 insertions(+), 16 deletions(-)
arch/riscv/kvm/aia.c +24 -16
@@ -23,6 +23,7 @@ struct aia_hgei_control {raw_spinlock_t lock;unsigned long free_bitmap;struct kvm_vcpu *owners[BITS_PER_LONG ]; +unsigned int nr_hgei; }; static DEFINE_PER_CPU(struct aia_hgei_control, aia_hgei);static int hgei_parent_irq;
@@ -452,7 +453,7 @@ void kvm_riscv_aia_free_hgei(int cpu, int hgei)raw_spin_lock_irqsave (&hgctrl->lock, flags); -if (hgei >0 && hgei <= kvm_riscv_aia_nr_hgei) { +if (hgei >0 && hgei <= hgctrl->nr_hgei) {if (!(hgctrl->free_bitmap &BIT (hgei))) { hgctrl->free_bitmap |=BIT (hgei); hgctrl->owners[hgei] =NULL ;
@@ -486,21 +487,8 @@ static irqreturn_t hgei_interrupt(int irq, void *dev_id)static int aia_hgei_init (void ) { -int cpu, rc; +int rc;struct irq_domain *domain; -struct aia_hgei_control *hgctrl; - -/* Initialize per-CPU guest external interrupt line management */ -for_each_possible_cpu (cpu) { - hgctrl =per_cpu_ptr (&aia_hgei, cpu); -raw_spin_lock_init (&hgctrl->lock); -if (kvm_riscv_aia_nr_hgei) { - hgctrl->free_bitmap = -BIT (kvm_riscv_aia_nr_hgei +1 ) -1 ; - hgctrl->free_bitmap &= ~BIT (0 ); - }else - hgctrl->free_bitmap =0 ; - }/* Skip SGEI interrupt setup for zero guest external interrupts */ if (!kvm_riscv_aia_nr_hgei)
@@ -545,9 +533,29 @@ static void aia_hgei_exit(void)void kvm_riscv_aia_enable (void ) { +struct aia_hgei_control *hgctrl; +if (!kvm_riscv_aia_available ())return ; + hgctrl =this_cpu_ptr (&aia_hgei); + +/* Figure-out number of bits in HGEIE */ +csr_write (CSR_HGEIE ,-1UL ); + hgctrl->nr_hgei =fls_long (csr_read (CSR_HGEIE )); +csr_write (CSR_HGEIE ,0 ); +if (hgctrl->nr_hgei) + hgctrl->nr_hgei--; + +if (hgctrl->nr_hgei) { + hgctrl->free_bitmap =BIT (hgctrl->nr_hgei +1 ) -1 ; + hgctrl->free_bitmap &= ~BIT (0 ); + }else { + hgctrl->free_bitmap =0 ; + } + +raw_spin_lock_init (&hgctrl->lock); +csr_write (CSR_HVICTL ,aia_hvictl_value (false));csr_write (CSR_HVIPRIO1 ,0x0 );csr_write (CSR_HVIPRIO2 ,0x0 );
@@ -588,7 +596,7 @@ void kvm_riscv_aia_disable(void)raw_spin_lock_irqsave (&hgctrl->lock, flags); -for (i =0 ; i <= kvm_riscv_aia_nr_hgei; i++) { +for (i =0 ; i <= hgctrl->nr_hgei; i++) { vcpu = hgctrl->owners[i];if (!vcpu)continue ;
Signature · 1 line
2.43.0
Anup Patel
maintainerRe: [PATCH V2 1/4] RISC-V: KVM: AIA: Make HGEI number and management fully per-CPU
Quoted 19 lines · On Sat, Apr 25, 2026 at 6:29 AM <[email protected]> wrote:
From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
Previously the number of Hypervisor Guest External Interrupt (HGEI) lines was stored in a single global variable
kvm_riscv_aia_nr_hgeiand assumed to be the same for all HARTs. This assumption does not hold on heterogeneous RISC-V SoCs where different cores may expose different HGEIE CSR widths.Introduce
nr_hgeifield into the per-CPUstruct aia_hgei_controland probe the actual supported HGEI count for the current HART inkvm_riscv_aia_enable()using the standard RISC-V CSR probe technique:csr_write(CSR_HGEIE, -1UL); nr = fls_long(csr_read(CSR_HGEIE)); if (nr) nr--;
Quoted 7 lines
All HGEI allocation, free and disable paths (
kvm_riscv_aia_free_hgei(),kvm_riscv_aia_disable(), etc.) now use the per-CPU value instead of the global one.The early global
kvm_riscv_aia_nr_hgeiis kept only for deciding whether SGEI interrupt registration is needed; the real per-HART initialization of lock andfree_bitmapis moved to enable time.
We should keep the kvm_riscv_aia_nr_hgei to imply two things:
1) Minimum number of HGEI lines available across all HARTs
2) Non-zero kvm_riscv_aia_nr_hgei means HGEI is enabled
The PATCH2 is redundant and must be droped. I will take care care of this at the time of merging.
Quoted 60 lines
This makes KVM AIA robust on big.LITTLE-style and multi-vendor asymmetric platforms.
Signed-off-by Guo Ren (Alibaba DAMO Academy) <[email protected]>--- arch/riscv/kvm/aia.c | 40 ++++++++++++++++++++++++---------------- 1 file changed, 24 insertions(+), 16 deletions(-)arch/riscv/kvm/aia.c +5
-15@@ -23,6 +23,7 @@ struct aia_hgei_control {raw_spinlock_t lock;unsigned long free_bitmap;struct kvm_vcpu *owners[BITS_PER_LONG ]; +unsigned int nr_hgei; }; static DEFINE_PER_CPU(struct aia_hgei_control, aia_hgei);static int hgei_parent_irq;@@ -452,7 +453,7 @@ void kvm_riscv_aia_free_hgei(int cpu, int hgei)raw_spin_lock_irqsave (&hgctrl->lock, flags); -if (hgei >0 && hgei <= kvm_riscv_aia_nr_hgei) { +if (hgei >0 && hgei <= hgctrl->nr_hgei) {if (!(hgctrl->free_bitmap &BIT (hgei))) { hgctrl->free_bitmap |=BIT (hgei); hgctrl->owners[hgei] =NULL ;@@ -486,21 +487,8 @@ static irqreturn_t hgei_interrupt(int irq, void *dev_id)static int aia_hgei_init (void ) { -int cpu, rc; +int rc;struct irq_domain *domain; -struct aia_hgei_control *hgctrl; - -/* Initialize per-CPU guest external interrupt line management */ -for_each_possible_cpu (cpu) { - hgctrl =per_cpu_ptr (&aia_hgei, cpu); -raw_spin_lock_init (&hgctrl->lock); -if (kvm_riscv_aia_nr_hgei) { - hgctrl->free_bitmap = -BIT (kvm_riscv_aia_nr_hgei +1 ) -1 ; - hgctrl->free_bitmap &= ~BIT (0 ); - }else - hgctrl->free_bitmap =0 ; - }/* Skip SGEI interrupt setup for zero guest external interrupts */ if (!kvm_riscv_aia_nr_hgei)@@ -545,9 +533,29 @@ static void aia_hgei_exit(void)void kvm_riscv_aia_enable (void ) { +struct aia_hgei_control *hgctrl; +if (!kvm_riscv_aia_available ())
arch/riscv/kvm/aia.c static void aia_hgei_exit(void) · line 548
Quoted 10 lines
arch/riscv/kvm/aia.c +8
-0return ; + hgctrl =this_cpu_ptr (&aia_hgei); + +/* Figure-out number of bits in HGEIE */ +csr_write (CSR_HGEIE ,-1UL ); + hgctrl->nr_hgei =fls_long (csr_read (CSR_HGEIE )); +csr_write (CSR_HGEIE ,0 ); +if (hgctrl->nr_hgei) + hgctrl->nr_hgei--;
For completness of this patch, we should also have the following:
if (gc)
hgctrl->nr_hgei= min((ulong)hgctrl->nr_hgei, gc->nr_guest_files);
else
hgctrl->nr_hgei = 0;Also, over here we should check and update kvm_riscv_aia_nr_hgei to keep it minimum accross HARTs:
if (hgctrl->nr_hgei && hgctrl->nr_hgei < kvm_riscv_aia_nr_hgei)
kvm_riscv_aia_nr_hgei = hgctrl->nr_hgei;Quoted 23 lines
arch/riscv/kvm/aia.c +11
-2+ +if (hgctrl->nr_hgei) { + hgctrl->free_bitmap =BIT (hgctrl->nr_hgei +1 ) -1 ; + hgctrl->free_bitmap &= ~BIT (0 ); + }else { + hgctrl->free_bitmap =0 ; + } + +raw_spin_lock_init (&hgctrl->lock); +csr_write (CSR_HVICTL ,aia_hvictl_value (false));csr_write (CSR_HVIPRIO1 ,0x0 );csr_write (CSR_HVIPRIO2 ,0x0 );@@ -588,7 +596,7 @@ void kvm_riscv_aia_disable(void)raw_spin_lock_irqsave (&hgctrl->lock, flags); -for (i =0 ; i <= kvm_riscv_aia_nr_hgei; i++) { +for (i =0 ; i <= hgctrl->nr_hgei; i++) { vcpu = hgctrl->owners[i];if (!vcpu)continue ; --
Quoted 2 lines
2.43.0
Regards,
Anup
Re: [PATCH V2 1/4] RISC-V: KVM: AIA: Make HGEI number and management fully per-CPU
Quoted 29 lines · Anup Patel
Quoted 19 lines · On Sat, Apr 25, 2026 at 6:29 AM <[email protected]> wrote:
From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
Previously the number of Hypervisor Guest External Interrupt (HGEI) lines was stored in a single global variable
kvm_riscv_aia_nr_hgeiand assumed to be the same for all HARTs. This assumption does not hold on heterogeneous RISC-V SoCs where different cores may expose different HGEIE CSR widths.Introduce
nr_hgeifield into the per-CPUstruct aia_hgei_controland probe the actual supported HGEI count for the current HART inkvm_riscv_aia_enable()using the standard RISC-V CSR probe technique:csr_write(CSR_HGEIE, -1UL); nr = fls_long(csr_read(CSR_HGEIE)); if (nr) nr--;Quoted 7 lines
All HGEI allocation, free and disable paths (
kvm_riscv_aia_free_hgei(),kvm_riscv_aia_disable(), etc.) now use the per-CPU value instead of the global one.The early global
kvm_riscv_aia_nr_hgeiis kept only for deciding whether SGEI interrupt registration is needed; the real per-HART initialization of lock andfree_bitmapis moved to enable time.
Quoted 6 lines
We should keep the
kvm_riscv_aia_nr_hgeito imply two things:
1) Minimum number of HGEI lines available across all HARTs
2) Non-zerokvm_riscv_aia_nr_hgeimeans HGEI is enabledThe PATCH2 is redundant and must be droped. I will take care care of this at the time of merging.
Okay.
Quoted 74 lines
Quoted 61 lines
This makes KVM AIA robust on big.LITTLE-style and multi-vendor asymmetric platforms.
Signed-off-by Guo Ren (Alibaba DAMO Academy) <[email protected]>--- arch/riscv/kvm/aia.c | 40 ++++++++++++++++++++++++---------------- 1 file changed, 24 insertions(+), 16 deletions(-)arch/riscv/kvm/aia.c +5
-15@@ -23,6 +23,7 @@ struct aia_hgei_control {raw_spinlock_t lock;unsigned long free_bitmap;struct kvm_vcpu *owners[BITS_PER_LONG ]; +unsigned int nr_hgei; }; static DEFINE_PER_CPU(struct aia_hgei_control, aia_hgei);static int hgei_parent_irq;@@ -452,7 +453,7 @@ void kvm_riscv_aia_free_hgei(int cpu, int hgei)raw_spin_lock_irqsave (&hgctrl->lock, flags); -if (hgei >0 && hgei <= kvm_riscv_aia_nr_hgei) { +if (hgei >0 && hgei <= hgctrl->nr_hgei) {if (!(hgctrl->free_bitmap &BIT (hgei))) { hgctrl->free_bitmap |=BIT (hgei); hgctrl->owners[hgei] =NULL ;@@ -486,21 +487,8 @@ static irqreturn_t hgei_interrupt(int irq, void *dev_id)static int aia_hgei_init (void ) { -int cpu, rc; +int rc;struct irq_domain *domain; -struct aia_hgei_control *hgctrl; - -/* Initialize per-CPU guest external interrupt line management */ -for_each_possible_cpu (cpu) { - hgctrl =per_cpu_ptr (&aia_hgei, cpu); -raw_spin_lock_init (&hgctrl->lock); -if (kvm_riscv_aia_nr_hgei) { - hgctrl->free_bitmap = -BIT (kvm_riscv_aia_nr_hgei +1 ) -1 ; - hgctrl->free_bitmap &= ~BIT (0 ); - }else - hgctrl->free_bitmap =0 ; - }/* Skip SGEI interrupt setup for zero guest external interrupts */ if (!kvm_riscv_aia_nr_hgei)@@ -545,9 +533,29 @@ static void aia_hgei_exit(void)void kvm_riscv_aia_enable (void ) { +struct aia_hgei_control *hgctrl; +if (!kvm_riscv_aia_available ())arch/riscv/kvm/aia.c static void aia_hgei_exit(void) · line 548
Quoted 10 lines
arch/riscv/kvm/aia.c +8
-0return ; + hgctrl =this_cpu_ptr (&aia_hgei); + +/* Figure-out number of bits in HGEIE */ +csr_write (CSR_HGEIE ,-1UL ); + hgctrl->nr_hgei =fls_long (csr_read (CSR_HGEIE )); +csr_write (CSR_HGEIE ,0 ); +if (hgctrl->nr_hgei) + hgctrl->nr_hgei--;For completness of this patch, we should also have the following:
Quoted 2 lines
if (gc)
hgctrl->nr_hgei= min((ulong)hgctrl->nr_hgei, gc->nr_guest_files);Thanks for the review!
On keeping kvm_riscv_aia_nr_hgei as the minimum across all HARTs:
I agree to keep kvm_riscv_aia_nr_hgei as a non-zero indicator for HGEI enablement. However, I have a concern about making it track the per-HART minimum. The purpose of this patch is precisely to let each HART use its own nr_hgei independently, so heterogeneous HARTs with more HGEI lines are not artificially limited by weaker ones. Collapsing that back into a global minimum would partially defeat the goal.
ref: global->nr_guest_files = min(global->nr_guest_files, local->nr_guest_files);
Quoted 37 lines
else hgctrl->nr_hgei = 0;Also, over here we should check and update
kvm_riscv_aia_nr_hgeito keep it minimum accross HARTs:if (hgctrl->nr_hgei && hgctrl->nr_hgei < kvm_riscv_aia_nr_hgei) kvm_riscv_aia_nr_hgei = hgctrl->nr_hgei;Quoted 23 lines
arch/riscv/kvm/aia.c +11
-2+ +if (hgctrl->nr_hgei) { + hgctrl->free_bitmap =BIT (hgctrl->nr_hgei +1 ) -1 ; + hgctrl->free_bitmap &= ~BIT (0 ); + }else { + hgctrl->free_bitmap =0 ; + } + +raw_spin_lock_init (&hgctrl->lock); +csr_write (CSR_HVICTL ,aia_hvictl_value (false));csr_write (CSR_HVIPRIO1 ,0x0 );csr_write (CSR_HVIPRIO2 ,0x0 );@@ -588,7 +596,7 @@ void kvm_riscv_aia_disable(void)raw_spin_lock_irqsave (&hgctrl->lock, flags); -for (i =0 ; i <= kvm_riscv_aia_nr_hgei; i++) { +for (i =0 ; i <= hgctrl->nr_hgei; i++) { vcpu = hgctrl->owners[i];if (!vcpu)continue ; --Quoted 2 lines
2.43.0
Regards,
Anup
Signature · 2 lines
Best Regards Guo Ren
PATCH v2 2/4 RISC-V: KVM: AIA: Replace global HGEI count with simple enabled bool
In reply to [email protected]From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
Now that HGEI line management is fully per-CPU (via struct aia_hgei_control::nr_hgei), the global kvm_riscv_aia_nr_hgei is no longer needed. Replace it with a simple bool kvm_riscv_aia_hgei_enabled that only indicates whether HGEI support is available at all.
--- arch/riscv/include/asm/kvm_aia.h | 2 +- arch/riscv/kvm/aia.c | 18 ++++++++---------- arch/riscv/kvm/aia_device.c | 4 ++-- arch/riscv/kvm/main.c | 3 +-- 4 files changed, 12 insertions(+), 15 deletions(-)
arch/riscv/include/asm/kvm_aia.h +1 -1
@@ -79,7 +79,7 @@ struct kvm_vcpu_aia {#define irqchip_in_kernel (k) ((k)->arch.aia.in_kernel) -extern unsigned int kvm_riscv_aia_nr_hgei; +extern bool kvm_riscv_aia_hgei_enabled;extern unsigned int kvm_riscv_aia_max_ids;DECLARE_STATIC_KEY_FALSE (kvm_riscv_aia_available);#define kvm_riscv_aia_available () \
arch/riscv/kvm/aia.c +8 -10
@@ -28,7 +28,7 @@ struct aia_hgei_control { static DEFINE_PER_CPU(struct aia_hgei_control, aia_hgei);static int hgei_parent_irq; -unsigned int kvm_riscv_aia_nr_hgei; +bool kvm_riscv_aia_hgei_enabled;unsigned int kvm_riscv_aia_max_ids;DEFINE_STATIC_KEY_FALSE (kvm_riscv_aia_available);
@@ -491,7 +491,7 @@ static int aia_hgei_init(void)struct irq_domain *domain;/* Skip SGEI interrupt setup for zero guest external interrupts */ -if (!kvm_riscv_aia_nr_hgei) +if (!kvm_riscv_aia_hgei_enabled) goto skip_sgei_interrupt;/* Find INTC irq domain */
@@ -524,7 +524,7 @@ static int aia_hgei_init(void)static void aia_hgei_exit (void ) {/* Do nothing for zero guest external interrupts */ -if (!kvm_riscv_aia_nr_hgei) +if (!kvm_riscv_aia_hgei_enabled)return ;/* Free per-CPU SGEI interrupt */
@@ -631,6 +631,7 @@ int kvm_riscv_aia_init(void) {int rc;const struct imsic_global_config *gc; +unsigned int kvm_riscv_aia_nr_hgei;if (!riscv_isa_extension_available (NULL , SxAIA))return -ENODEV ;
@@ -641,21 +642,18 @@ int kvm_riscv_aia_init(void) kvm_riscv_aia_nr_hgei =fls_long (csr_read (CSR_HGEIE ));csr_write (CSR_HGEIE ,0 );if (kvm_riscv_aia_nr_hgei) - kvm_riscv_aia_nr_hgei--; + kvm_riscv_aia_hgei_enabled = true;/* * Number of usable HGEI lines should be minimum of per-HART * IMSIC guest files and number of bits in HGEIE */ -if (gc) - kvm_riscv_aia_nr_hgei =min ((ulong )kvm_riscv_aia_nr_hgei, - gc->nr_guest_files); -else - kvm_riscv_aia_nr_hgei =0 ; +if (!gc) + kvm_riscv_aia_hgei_enabled =0 ;/* Find number of guest MSI IDs */ kvm_riscv_aia_max_ids =IMSIC_MAX_ID ; -if (gc && kvm_riscv_aia_nr_hgei) +if (gc && kvm_riscv_aia_hgei_enabled) kvm_riscv_aia_max_ids = gc->nr_guest_ids +1 ;/* Initialize guest external interrupt line management */
arch/riscv/kvm/aia_device.c +2 -2
@@ -71,7 +71,7 @@ static int aia_config(struct kvm *kvm, unsigned long type,* external interrupts (i.e. non-zero * VS-level IMSIC pages). */ -if (!kvm_riscv_aia_nr_hgei) +if (!kvm_riscv_aia_hgei_enabled)return -EINVAL ;break ;default :
@@ -628,7 +628,7 @@ void kvm_riscv_aia_init_vm(struct kvm *kvm)*/ /* Initialize default values in AIA global context */ - aia->mode = (kvm_riscv_aia_nr_hgei) ? + aia->mode = (kvm_riscv_aia_hgei_enabled) ?KVM_DEV_RISCV_AIA_MODE_AUTO :KVM_DEV_RISCV_AIA_MODE_EMUL ; aia->nr_ids = kvm_riscv_aia_max_ids -1 ; aia->nr_sources =0 ;
arch/riscv/kvm/main.c +1 -2
@@ -169,8 +169,7 @@ static int __init riscv_kvm_init(void)kvm_info ("VMID %ld bits available\n" ,kvm_riscv_gstage_vmid_bits ());if (kvm_riscv_aia_available ()) -kvm_info ("AIA available with %d guest external interrupts\n" , - kvm_riscv_aia_nr_hgei); +kvm_info ("AIA available with guest external interrupts\n" );kvm_riscv_setup_vendor_features ();
Signature · 1 line
2.43.0
Anup Patel
maintainerRe: [PATCH V2 2/4] RISC-V: KVM: AIA: Replace global HGEI count with simple enabled bool
Quoted 8 lines · On Sat, Apr 25, 2026 at 6:29 AM <[email protected]> wrote:
From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
Now that HGEI line management is fully per-CPU (via struct
aia_hgei_control::nr_hgei), the globalkvm_riscv_aia_nr_hgeiis no longer needed. Replace it with a simplebool kvm_riscv_aia_hgei_enabledthat only indicates whether HGEI support is available at all.
As mentioned in PATCH1, this PATCH is not required.
Quoted 124 lines
Signed-off-by Guo Ren (Alibaba DAMO Academy) <[email protected]>--- arch/riscv/include/asm/kvm_aia.h | 2 +- arch/riscv/kvm/aia.c | 18 ++++++++---------- arch/riscv/kvm/aia_device.c | 4 ++-- arch/riscv/kvm/main.c | 3 +-- 4 files changed, 12 insertions(+), 15 deletions(-)arch/riscv/include/asm/kvm_aia.h +1
-1@@ -79,7 +79,7 @@ struct kvm_vcpu_aia {#define irqchip_in_kernel (k) ((k)->arch.aia.in_kernel) -extern unsigned int kvm_riscv_aia_nr_hgei; +extern bool kvm_riscv_aia_hgei_enabled;extern unsigned int kvm_riscv_aia_max_ids;DECLARE_STATIC_KEY_FALSE (kvm_riscv_aia_available);#define kvm_riscv_aia_available () \arch/riscv/kvm/aia.c +8
-10@@ -28,7 +28,7 @@ struct aia_hgei_control { static DEFINE_PER_CPU(struct aia_hgei_control, aia_hgei);static int hgei_parent_irq; -unsigned int kvm_riscv_aia_nr_hgei; +bool kvm_riscv_aia_hgei_enabled;unsigned int kvm_riscv_aia_max_ids;DEFINE_STATIC_KEY_FALSE (kvm_riscv_aia_available);@@ -491,7 +491,7 @@ static int aia_hgei_init(void)struct irq_domain *domain;/* Skip SGEI interrupt setup for zero guest external interrupts */ -if (!kvm_riscv_aia_nr_hgei) +if (!kvm_riscv_aia_hgei_enabled) goto skip_sgei_interrupt;/* Find INTC irq domain */ @@ -524,7 +524,7 @@ static int aia_hgei_init(void)static void aia_hgei_exit (void ) {/* Do nothing for zero guest external interrupts */ -if (!kvm_riscv_aia_nr_hgei) +if (!kvm_riscv_aia_hgei_enabled)return ;/* Free per-CPU SGEI interrupt */ @@ -631,6 +631,7 @@ int kvm_riscv_aia_init(void) {int rc;const struct imsic_global_config *gc; +unsigned int kvm_riscv_aia_nr_hgei;if (!riscv_isa_extension_available (NULL , SxAIA))return -ENODEV ;@@ -641,21 +642,18 @@ int kvm_riscv_aia_init(void) kvm_riscv_aia_nr_hgei =fls_long (csr_read (CSR_HGEIE ));csr_write (CSR_HGEIE ,0 );if (kvm_riscv_aia_nr_hgei) - kvm_riscv_aia_nr_hgei--; + kvm_riscv_aia_hgei_enabled = true;/* * Number of usable HGEI lines should be minimum of per-HART * IMSIC guest files and number of bits in HGEIE */ -if (gc) - kvm_riscv_aia_nr_hgei =min ((ulong )kvm_riscv_aia_nr_hgei, - gc->nr_guest_files); -else - kvm_riscv_aia_nr_hgei =0 ; +if (!gc) + kvm_riscv_aia_hgei_enabled =0 ;/* Find number of guest MSI IDs */ kvm_riscv_aia_max_ids =IMSIC_MAX_ID ; -if (gc && kvm_riscv_aia_nr_hgei) +if (gc && kvm_riscv_aia_hgei_enabled) kvm_riscv_aia_max_ids = gc->nr_guest_ids +1 ;/* Initialize guest external interrupt line management */ arch/riscv/kvm/aia_device.c +2
-2@@ -71,7 +71,7 @@ static int aia_config(struct kvm *kvm, unsigned long type,* external interrupts (i.e. non-zero * VS-level IMSIC pages). */ -if (!kvm_riscv_aia_nr_hgei) +if (!kvm_riscv_aia_hgei_enabled)return -EINVAL ;break ;default :@@ -628,7 +628,7 @@ void kvm_riscv_aia_init_vm(struct kvm *kvm)*/ /* Initialize default values in AIA global context */ - aia->mode = (kvm_riscv_aia_nr_hgei) ? + aia->mode = (kvm_riscv_aia_hgei_enabled) ?KVM_DEV_RISCV_AIA_MODE_AUTO :KVM_DEV_RISCV_AIA_MODE_EMUL ; aia->nr_ids = kvm_riscv_aia_max_ids -1 ; aia->nr_sources =0 ;arch/riscv/kvm/main.c +1
-3@@ -169,8 +169,7 @@ static int __init riscv_kvm_init(void)kvm_info ("VMID %ld bits available\n" ,kvm_riscv_gstage_vmid_bits ());if (kvm_riscv_aia_available ()) -kvm_info ("AIA available with %d guest external interrupts\n" , - kvm_riscv_aia_nr_hgei); +kvm_info ("AIA available with guest external interrupts\n" );kvm_riscv_setup_vendor_features (); --
Quoted 2 lines
2.43.0
Regards,
Anup
PATCH v2 3/4 irqchip/riscv-imsic: Move nr_guest_files to per-HART local config
In reply to [email protected]From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
With the recent KVM AIA per-HART HGEI conversion, the global nr_guest_files is no longer appropriate. Different HARTs in heterogeneous SoCs may have different numbers of guest interrupt files.
Move nr_guest_files from struct imsic_global_config to struct imsic_local_config, and compute it per-CPU in imsic_setup_state() based on the actual MMIO guest file region size.
Update the related comment to reflect that KVM now uses the per-HART value.
This eliminates the last global assumption about guest files and completes the per-HART conversion series for RISC-V AIA/IMSIC.
--- drivers/irqchip/irq-riscv-imsic-state.c | 10 ++++++---- include/linux/irqchip/riscv-imsic.h | 6 +++--- 2 files changed, 9 insertions(+), 7 deletions(-)
drivers/irqchip/irq-riscv-imsic-state.c +6 -4
@@ -878,7 +878,6 @@ int __init imsic_setup_state(struct fwnode_handle *fwnode, void *opaque) }/* Configure handlers for target CPUs */ - global->nr_guest_files =BIT (global->guest_index_bits) -1 ;for (i =0 ; i < nr_parent_irqs; i++) { rc =imsic_get_parent_hartid (fwnode, i, &hartid);if (rc) {
@@ -910,23 +909,26 @@ int __init imsic_setup_state(struct fwnode_handle *fwnode, void *opaque) reloff -=ALIGN (resource_size (&mmios[j]),BIT (global->guest_index_bits) *IMSIC_MMIO_PAGE_SZ ); } + + local =per_cpu_ptr (global->local, cpu); + local->nr_guest_files =BIT (global->guest_index_bits) -1 ; +if (index >= nr_mmios) {pr_warn ("%pfwP: MMIO not found for parent irq%d\n" , fwnode, i);continue ; } - local =per_cpu_ptr (global->local, cpu); local->msi_pa = mmios[index].start + reloff; local->msi_va = mmios_va[index] + reloff;/* -* KVM uses global->nr_guest_files to determine the available guest +* KVM uses local->nr_guest_files to determine the available guest * interrupt files on each CPU. Take the minimum number of guest * interrupt files across all CPUs to avoid KVM incorrectly allocating * an unexisted or unmapped guest interrupt file on some CPUs. */ nr_guest_files = (resource_size (&mmios[index]) - reloff) /IMSIC_MMIO_PAGE_SZ -1 ; - global->nr_guest_files =min (global->nr_guest_files, nr_guest_files); + local->nr_guest_files =min (local->nr_guest_files, nr_guest_files); nr_handlers++; }
include/linux/irqchip/riscv-imsic.h +3 -3
@@ -40,6 +40,9 @@struct imsic_local_config {phys_addr_t msi_pa;void __iomem *msi_va; + +/* Number of guest interrupt files per core */ +u32 nr_guest_files; };struct imsic_global_config {
@@ -68,9 +71,6 @@ struct imsic_global_config {/* Number of guest interrupt identities */ u32 nr_guest_ids; -/* Number of guest interrupt files per core */ -u32 nr_guest_files; -/* Per-CPU IMSIC addresses */ struct imsic_local_config __percpu *local; };
Signature · 1 line
2.43.0
Re: [PATCH V2 3/4] irqchip/riscv-imsic: Move nr_guest_files to per-HART local config
Quoted 11 lines · On Sat, Apr 25 2026 at 00:59, [email protected] wrote:
From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
With the recent KVM AIA per-HART HGEI conversion, the global
nr_guest_filesis no longer appropriate. Different HARTs in heterogeneous SoCs may have different numbers of guest interrupt files.Move
nr_guest_filesfromstruct imsic_global_configtostruct imsic_local_config, and compute it per-CPU inimsic_setup_state()based on the actual MMIO guest file region size.
Quoted 7 lines
Update the related comment to reflect that KVM now uses the per-HART value.
This eliminates the last global assumption about guest files and completes the per-HART conversion series for RISC-V AIA/IMSIC.
Signed-off-by Guo Ren (Alibaba DAMO Academy) <[email protected]>
Assuming this goes through the risc-v tree:
Re: [PATCH V2 3/4] irqchip/riscv-imsic: Move nr_guest_files to per-HART local config
Quoted 16 lines · Thomas Gleixner
Quoted 14 lines · On Sat, Apr 25 2026 at 00:59, [email protected] wrote:
From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
With the recent KVM AIA per-HART HGEI conversion, the global
nr_guest_filesis no longer appropriate. Different HARTs in heterogeneous SoCs may have different numbers of guest interrupt files.Move
nr_guest_filesfromstruct imsic_global_configtostruct imsic_local_config, and compute it per-CPU inimsic_setup_state()based on the actual MMIO guest file region size.Update the related comment to reflect that KVM now uses the per-HART value.
Quoted 8 lines
Quoted 4 lines
This eliminates the last global assumption about guest files and completes the per-HART conversion series for RISC-V AIA/IMSIC.
Signed-off-by Guo Ren (Alibaba DAMO Academy) <[email protected]>Assuming this goes through the risc-v tree:
Acked-by Thomas Gleixner <[email protected]>
Got it & Thx Thomas.
I think it'll go through the kvm-riscv tree because it's all about kvm improvements.
Signature · 2 lines
Best Regards Guo Ren
Anup Patel
maintainerRe: [PATCH V2 3/4] irqchip/riscv-imsic: Move nr_guest_files to per-HART local config
In reply to Thomas GleixnerHi Thomas,
Quoted 16 lines · Thomas Gleixner
Quoted 14 lines · On Sat, Apr 25 2026 at 00:59, [email protected] wrote:
From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
With the recent KVM AIA per-HART HGEI conversion, the global
nr_guest_filesis no longer appropriate. Different HARTs in heterogeneous SoCs may have different numbers of guest interrupt files.Move
nr_guest_filesfromstruct imsic_global_configtostruct imsic_local_config, and compute it per-CPU inimsic_setup_state()based on the actual MMIO guest file region size.Update the related comment to reflect that KVM now uses the per-HART value.
Quoted 8 lines
Quoted 4 lines
This eliminates the last global assumption about guest files and completes the per-HART conversion series for RISC-V AIA/IMSIC.
Signed-off-by Guo Ren (Alibaba DAMO Academy) <[email protected]>Assuming this goes through the risc-v tree:
Acked-by Thomas Gleixner <[email protected]>
Thanks, I will take this series through KVM RISC-V tree.
Regards,
Anup
PATCH v2 4/4 RISC-V: KVM: AIA: Use per-HART IMSIC guest files to compute final HGEI count
In reply to [email protected]From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
Now that nr_guest_files has been moved to struct imsic_local_config and is computed per-HART, KVM must respect the actual number of guest interrupt files available on each HART when setting up HGEI.
In kvm_riscv_aia_enable():
- Retrieve the per-CPU IMSIC local config
- Take min(hgctrl->nr_hgei, lc->nr_guest_files) as the final usable HGEI count for this HART
- Use the result to initialize free_bitmap
This ensures correct HGEI allocation on heterogeneous RISC-V SoCs where different cores may have different IMSIC guest file counts, and completes the per-HART conversion series.
--- arch/riscv/kvm/aia.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-)
arch/riscv/kvm/aia.c +10 -2
@@ -534,6 +534,9 @@ static void aia_hgei_exit(void)void kvm_riscv_aia_enable (void ) {struct aia_hgei_control *hgctrl; +const struct imsic_global_config *gc; +const struct imsic_local_config *lc; +unsigned int nr_hgei;if (!kvm_riscv_aia_available ())return ;
@@ -547,8 +550,13 @@ void kvm_riscv_aia_enable(void)if (hgctrl->nr_hgei) hgctrl->nr_hgei--; -if (hgctrl->nr_hgei) { - hgctrl->free_bitmap =BIT (hgctrl->nr_hgei +1 ) -1 ; + gc =imsic_get_global_config (); + lc = (gc) ?this_cpu_ptr (gc->local) :NULL ; +if (lc) + nr_hgei =min (hgctrl->nr_hgei, lc->nr_guest_files); + +if (nr_hgei) { + hgctrl->free_bitmap =BIT (nr_hgei +1 ) -1 ; hgctrl->free_bitmap &= ~BIT (0 ); }else { hgctrl->free_bitmap =0 ;
Signature · 1 line
2.43.0
Re: Re: [PATCH 3/3] irqchip/riscv-imsic: Remove global nr_guest_files after KVM AIA per-HART conversion
In reply to [email protected]Hi Anup,
Quoted 9 lines · Guo Ren
Quoted 8 lines
Hi all,
This patch simply removed
nr_guest_files, which is suboptimal. For v2, I will keepnr_guest_filesinsidestruct imsic_local_config(per-HART) and compute the effective HGEI count inkvm_riscv_aia_enable()as:
Quoted 4 lines
Quoted 1 line
min(hgctrl->nr_hgei, lc->nr_guest_files)Yes, I was going to comment the exact same thing. Good that you realized it early.
Thank you for the feedback. This is V2:
https://lore.kernel.org/kvm-riscv/[email protected]/
Best Regards
GUO Ren
Anup Patel
maintainerQuoted 43 lines · On Sat, Apr 25, 2026 at 6:29 AM <[email protected]> wrote:
From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>
This short series converts RISC-V KVM AIA's Hypervisor Guest External Interrupt (HGEI) line management from a global "one-size-fits-all" model to a fully per-HART (per-CPU) model. It also performs the corresponding cleanup in the IMSIC irqchip driver.
The motivation is to properly support heterogeneous RISC-V SoCs (big.LITTLE, multi-vendor core mixes) where different HARTs may expose different numbers of guest interrupt files / HGEIE bits. The old global
kvm_riscv_aia_nr_hgeiandimsic_global_config::nr_guest_filesassumptions are no longer tenable and are removed.After this series, HGEI allocation, freeing, and interrupt delivery are completely per-HART, the code is simpler and clearer, and there are no remaining global assumptions about guest interrupt file counts.
Changelog:
V2:
- Add per-HART IMSIC guest files to compute final HGEI count.
- Addmin(local->nr_guest_files, nr_guest_files);V1:
https://lore.kernel.org/kvm-riscv/[email protected]/Guo Ren (Alibaba DAMO Academy) (4):
RISC-V: KVM: AIA: Make HGEI number and management fully per-CPU
RISC-V: KVM: AIA: Replace global HGEI count with simple enabled bool irqchip/riscv-imsic: Movenr_guest_filesto per-HART local config
RISC-V: KVM: AIA: Use per-HART IMSIC guest files to compute final HGEI
countarch/riscv/include/asm/kvm_aia.h | 2 +- arch/riscv/kvm/aia.c | 66 +++++++++++++++---------- arch/riscv/kvm/aia_device.c | 4 +- arch/riscv/kvm/main.c | 3 +- drivers/irqchip/irq-riscv-imsic-state.c | 10 ++-- include/linux/irqchip/riscv-imsic.h | 6 +-- 6 files changed, 53 insertions(+), 38 deletions(-)
Quoted 3 lines
--
2.43.0
I will send out v3 series to avoid some back-n-forth.
Regards,
Anup