qemu-kvm: Hook cleanups and extended use of upstream code
v2 cover qemu-kvm: Hook cleanups and extended use of upstream code
This version addresses the feedback on v2, namely:
- assert(<vm stopped> || <current thread == env->thread>) on low-level load/save registers
- fixed mpstate initialization
Yet untested is -no-kvm-irqchip with smp due to some bug in unpatched qemu-kvm or the kernel modules. Still investigating.
Pull URL is still
git://git.kiszka.org/qemu-kvm.git queues/vcpu-state
PS: The corresponding upstream queue is now available under queues/kvm-upstream in the same repository. Will send it out later if there are no further remarks on this series to avoid flooding the mailing lists. Early testers are nevertheless welcome.
Jan Kiszka (21):
qemu-kvm: Drop vmport changes
KVM: Make vmport KVM-compatible
qemu-kvm: Clean up register access API
KVM: x86: Fix up misreported CPU features
qemu-kvm: Use upstream kvm_enabled and cpu_synchronize_state
qemu-kvm: Use upstream kvm_setup_guest_memory
qemu-kvm: Use some more upstream prototypes
qemu-kvm: Use upstream kvm_arch_get_supported_cpuid
qemu-kvm: Use upstream kvm_pit_in_kernel
KVM: Move and rename regs_modified
KVM: Rework of guest debug state writing
qemu-kvm: Use upstream kvm_vcpu_dirty
qemu-kvm: Use upstream guest debug code
qemu-kvm: Rework VCPU state writeback API
qemu-kvm: Clean up mpstate synchronization
KVM: x86: Restrict writeback of VCPU state
qemu-kvm: Use VCPU event state for reset and vmsave/load
qemu-kvm: Cleanup/fix TSC and PV clock writeback
qemu-kvm: Clean up KVM's APIC hooks
qemu-kvm: Move kvm_set_boot_cpu_id
qemu-kvm: Bring qemu_init_vcpu back home
cpu-defs.h | 2 +- exec.c | 17 -- hw/apic.c | 47 +----- hw/i8254.c | 6 +- hw/i8259.c | 2 +- hw/ioapic.c | 2 +- hw/msix.c | 3 +- hw/pc.c | 13 +-- hw/pcspk.c | 4 +- hw/piix_pci.c | 2 +- hw/ppc_newworld.c | 3 - hw/ppc_oldworld.c | 3 - hw/s390-virtio.c | 1 - hw/vmport.c | 14 +-- kvm-all.c | 51 +++--- kvm.h | 35 +++-- qemu-kvm-ia64.c | 6 +- qemu-kvm-x86.c | 433 +++++++++++-------------------------------------- qemu-kvm.c | 165 +++---------------- qemu-kvm.h | 161 ++----------------- savevm.c | 4 + sysemu.h | 4 + target-i386/cpu.h | 9 +- target-i386/helper.c | 2 + target-i386/kvm.c | 61 +++++-- target-i386/machine.c | 27 --- target-ia64/machine.c | 5 +- target-ppc/kvm.c | 2 +- target-ppc/machine.c | 4 - target-s390x/kvm.c | 3 +- vl.c | 32 ++++- 31 files changed, 298 insertions(+), 825 deletions(-)
PATCH v2 1/21 qemu-kvm: Drop vmport changes
This attempt to make vmport KVM compatible is half-broken and is scheduled to be replaced by proper upstream support.
--- hw/vmport.c | 13 +------------ 1 files changed, 1 insertions(+), 12 deletions(-)
hw/vmport.c +1 -12
@@ -21,12 +21,10 @@* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN * THE SOFTWARE. */ -#include "hw.h" #include "isa.h" #include "pc.h" #include "sysemu.h" -#include "qemu-kvm.h" //#define VMPORT_DEBUG
@@ -59,10 +57,6 @@ static uint32_t vmport_ioport_read(void *opaque, uint32_t addr)CPUState *env = cpu_single_env;unsigned char command;uint32_t eax; -uint32_t ret; - -if (kvm_enabled ()) -kvm_save_registers (env); eax = env->regs[R_EAX ];if (eax !=VMPORT_MAGIC )
@@ -79,12 +73,7 @@ static uint32_t vmport_ioport_read(void *opaque, uint32_t addr)return eax; } - ret = s->func[command](s->opaque[command], addr); - -if (kvm_enabled ()) -kvm_load_registers (env); - -return ret; +return s->func[command](s->opaque[command], addr); }static void vmport_ioport_write (void *opaque,uint32_t addr,uint32_t val)
Signature · 1 line
1.6.0.2
Re: [PATCH v2 01/21] qemu-kvm: Drop vmport changes
Quoted 3 lines · Jan Kiszka
This attempt to make vmport KVM compatible is half-broken and is scheduled to be replaced by proper upstream support.
Does "scheduled" mean you have patches for adding cpu_synchronize_state() to vmport?
Signature · 1 line
error compiling committee.c: too many arguments to function
Re: [PATCH v2 01/21] qemu-kvm: Drop vmport changes
Quoted 7 lines · Avi Kivity
Quoted 3 lines · Jan Kiszka
This attempt to make vmport KVM compatible is half-broken and is scheduled to be replaced by proper upstream support.
Does "scheduled" mean you have patches for adding
cpu_synchronize_state()to vmport?
See patch 2 in this series, and Marcelo already queued that one for upstream.
Jan
PATCH v2 2/21 KVM: Make vmport KVM-compatible
In reply to Jan KiszkaThe vmport "device" accesses the VCPU registers, so it requires proper cpu_synchronize_state. Add it to vmport_ioport_read, which also synchronizes vmport_ioport_write.
--- hw/vmport.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)
hw/vmport.c +3 -0
@@ -25,6 +25,7 @@#include "isa.h" #include "pc.h" #include "sysemu.h" +#include "kvm.h" //#define VMPORT_DEBUG
@@ -58,6 +59,8 @@ static uint32_t vmport_ioport_read(void *opaque, uint32_t addr)unsigned char command;uint32_t eax; +cpu_synchronize_state (env); + eax = env->regs[R_EAX ];if (eax !=VMPORT_MAGIC )return eax;
Signature · 1 line
1.6.0.2
PATCH v2 3/21 qemu-kvm: Clean up register access API
In reply to Jan Kiszkaqemu-kvm's functios for accessing the VCPU registers are kvm_arch_load/save_regs. Use them directly instead of going through various wrappers. Specifically, we do not need on_vcpu wrapping as all users either already run in the related thread or call while the vm is stopped. This is now also validated during runtime via asserts.
--- qemu-kvm-x86.c | 4 ++++ qemu-kvm.c | 43 ++++++------------------------------------- qemu-kvm.h | 14 +++----------- target-ia64/machine.c | 4 ++-- 4 files changed, 15 insertions(+), 50 deletions(-)
qemu-kvm-x86.c +4 -0
@@ -911,6 +911,8 @@ void kvm_arch_load_regs(CPUState *env)struct kvm_msr_entry msrs[100 ];int rc, n, i; +assert (kvm_cpu_is_stopped (env) || env->thread_id ==kvm_get_thread_id ()); + regs.rax = env->regs[R_EAX ]; regs.rbx = env->regs[R_EBX ]; regs.rcx = env->regs[R_ECX ];
@@ -1072,6 +1074,8 @@ void kvm_arch_save_regs(CPUState *env)uint32_t hflags;uint32_t i, n, rc, bit; +assert (kvm_cpu_is_stopped (env) || env->thread_id ==kvm_get_thread_id ()); +kvm_get_regs (env, ®s); env->regs[R_EAX ] = regs.rax;
qemu-kvm.c +6 -37
89 lines not on this page · raw message
qemu-kvm.h +3 -11
@@ -902,8 +902,6 @@ int kvm_main_loop(void);int kvm_init_ap (void );#ifndef QEMU_KVM_NO_CPU int kvm_vcpu_inited (CPUState *env); -void kvm_load_registers (CPUState *env); -void kvm_save_registers (CPUState *env);void kvm_load_mpstate (CPUState *env);void kvm_save_mpstate (CPUState *env);int kvm_cpu_exec (CPUState *env);
@@ -1068,8 +1066,6 @@ void kvm_load_tsc(CPUState *env);#ifdef TARGET_I386 #define qemu_kvm_has_pit_state2 () (0)#endif -#define kvm_load_registers (env) do {} while(0) -#define kvm_save_registers (env) do {} while(0)#define kvm_save_mpstate (env) do {} while(0)#define qemu_kvm_cpu_stop (env) do {} while(0)static inline void kvm_init_vcpu (CPUState *env)
@@ -1098,13 +1094,6 @@ static inline int kvm_sync_vcpus(void) }#ifndef QEMU_KVM_NO_CPU -void kvm_arch_get_registers (CPUState *env); - -static inline void kvm_arch_put_registers (CPUState *env) -{ -kvm_load_registers (env); -} -void kvm_cpu_synchronize_state (CPUState *env);static inline void cpu_synchronize_state (CPUState *env)
@@ -1164,6 +1153,9 @@ int kvm_check_extension(KVMState *s, unsigned int ext);int kvm_tpr_enable_vapic (CPUState *env); +unsigned long kvm_get_thread_id (void ); +int kvm_cpu_is_stopped (CPUState *env); +#endif #endif
target-ia64/machine.c +2 -2
@@ -9,7 +9,7 @@ void cpu_save(QEMUFile *f, void *opaque)CPUState *env = opaque;if (kvm_enabled ()) { -kvm_save_registers (env); +kvm_arch_save_regs (env);kvm_arch_save_mpstate (env); } }
@@ -19,7 +19,7 @@ int cpu_load(QEMUFile *f, void *opaque, int version_id)CPUState *env = opaque;if (kvm_enabled ()) { -kvm_load_registers (env); +kvm_arch_load_regs (env);kvm_arch_load_mpstate (env); }return 0 ;
Signature · 1 line
1.6.0.2
PATCH v2 4/21 KVM: x86: Fix up misreported CPU features
In reply to Jan KiszkaQuoted 1 line
From qemu-kvm: Kernels before 2.6.30 misreported some essential CPU
features via KVM_GET_SUPPORTED_CPUID. Fix them up.
--- target-i386/kvm.c | 8 +++++++- 1 files changed, 7 insertions(+), 1 deletions(-)
target-i386/kvm.c +7 -1
@@ -101,12 +101,18 @@ uint32_t kvm_arch_get_supported_cpuid(CPUState *env, uint32_t function, int reg)break ;case R_EDX : ret = cpuid->entries[i].edx; -if (function ==0x80000001 ) { +switch (function) { +case 1 : +/* KVM before 2.6.30 misreports the following features */ + ret |=CPUID_MTRR |CPUID_PAT |CPUID_MCE |CPUID_MCA ; +break ; +case 0x80000001 :/* On Intel, kvm returns cpuid according to the Intel spec, * so add missing bits according to the AMD spec: */ cpuid_1_edx =kvm_arch_get_supported_cpuid (env,1 ,R_EDX ); ret |= cpuid_1_edx &0xdfeff7ff ; +break ; }break ; }
Signature · 1 line
1.6.0.2
PATCH v2 5/21 qemu-kvm: Use upstream kvm_enabled and cpu_synchronize_state
In reply to Jan KiszkaThey are identical, no need for private copies. This requires replacing qemu-kvm.h includes with kvm.h, a good thing anyway, and reveals that there is no need for QEMU_KVM_NO_CPU protection.
--- hw/i8254.c | 2 +- hw/i8259.c | 2 +- hw/ioapic.c | 2 +- hw/msix.c | 3 +-- hw/pc.c | 2 +- hw/piix_pci.c | 2 +- kvm.h | 7 +++---- qemu-kvm.h | 41 ----------------------------------------- vl.c | 3 ++- 9 files changed, 11 insertions(+), 53 deletions(-)
hw/i8254.c +1 -1
@@ -25,7 +25,7 @@#include "pc.h" #include "isa.h" #include "qemu-timer.h" -#include "qemu-kvm.h" +#include "kvm.h" #include "i8254.h" //#define DEBUG_PIT
hw/i8259.c +1 -1
@@ -27,7 +27,7 @@#include "monitor.h" #include "qemu-timer.h" -#include "qemu-kvm.h" +#include "kvm.h" /* debug PIC */ //#define DEBUG_PIC
hw/ioapic.c +1 -1
@@ -26,7 +26,7 @@#include "qemu-timer.h" #include "host-utils.h" -#include "qemu-kvm.h" +#include "kvm.h" //#define DEBUG_IOAPIC
hw/msix.c +1 -2
@@ -14,8 +14,7 @@#include "hw.h" #include "msix.h" #include "pci.h" -#define QEMU_KVM_NO_CPU -#include "qemu-kvm.h" +#include "kvm.h" /* Declaration from linux/pci_regs.h */ #define PCI_CAP_ID_MSIX 0x11/* MSI-X */
hw/pc.c +1 -1
@@ -47,7 +47,7 @@#include "multiboot.h" #include "device-assignment.h" -#include "qemu-kvm.h" +#include "kvm.h" /* output Bochs bios info messages */ //#define DEBUG_BIOS
hw/piix_pci.c +1 -1
@@ -28,7 +28,7 @@#include "pci_host.h" #include "isa.h" #include "sysbus.h" -#include "qemu-kvm.h" +#include "kvm.h" /* * I440FX chipset data sheet.
kvm.h +3 -4
@@ -18,8 +18,6 @@#include "qemu-queue.h" #include "qemu-kvm.h" -#ifdef KVM_UPSTREAM -#ifdef CONFIG_KVM extern int kvm_allowed;
@@ -28,6 +26,7 @@ extern int kvm_allowed;#define kvm_enabled () (0)#endif +#ifdef KVM_UPSTREAM struct kvm_run ;/* external API */
@@ -138,6 +137,8 @@ int kvm_check_extension(KVMState *s, unsigned int extension);uint32_t kvm_arch_get_supported_cpuid (CPUState *env,uint32_t function,int reg); +#endif +void kvm_cpu_synchronize_state (CPUState *env);/* generic hooks - to be moved/refactored once there are more users */
@@ -150,5 +151,3 @@ static inline void cpu_synchronize_state(CPUState *env) }#endif - -#endif
qemu-kvm.h +0 -41
156 lines not on this page · raw message
vl.c +2 -1
@@ -157,7 +157,6 @@ int main(int argc, char **argv)#include "qemu-option.h" #include "qemu-config.h" #include "qemu-objects.h" -#include "qemu-kvm.h" #include "hw/device-assignment.h" #include "disas.h"
@@ -5799,8 +5798,10 @@ int main(int argc, char **argv, char **envp)fprintf (stderr,"failed to initialize KVM\n" );exit (1 );#endif +#ifdef CONFIG_KVM fprintf (stderr,"Could not initialize KVM, will disable KVM support\n" ); kvm_allowed =0 ; +#endif } }
Signature · 1 line
1.6.0.2
PATCH v2 6/21 qemu-kvm: Use upstream kvm_setup_guest_memory
In reply to Jan KiszkaNothing missing in upstream kvm_setup_guest_memory, it is even more careful about error handling.
--- kvm-all.c | 3 --- kvm.h | 3 +-- qemu-kvm.c | 15 --------------- qemu-kvm.h | 1 - 4 files changed, 1 insertions(+), 21 deletions(-)
kvm-all.c +0 -3
@@ -886,7 +886,6 @@ int kvm_has_vcpu_events(void)return kvm_state->vcpu_events; } -#ifdef KVM_UPSTREAM void kvm_setup_guest_memory (void *start,size_t size) {if (!kvm_has_sync_mmu ()) {
@@ -905,8 +904,6 @@ void kvm_setup_guest_memory(void *start, size_t size) } } -#endif /* KVM_UPSTREAM */ -#ifdef KVM_CAP_SET_GUEST_DEBUG #ifdef KVM_UPSTREAM
kvm.h +1 -2
@@ -54,10 +54,9 @@ int kvm_has_vcpu_events(void);int kvm_put_vcpu_events (CPUState *env);int kvm_get_vcpu_events (CPUState *env); -#ifdef KVM_UPSTREAM -void kvm_setup_guest_memory (void *start,size_t size); +#ifdef KVM_UPSTREAM int kvm_coalesce_mmio_region (target_phys_addr_t start,ram_addr_t size);int kvm_uncoalesce_mmio_region (target_phys_addr_t start,ram_addr_t size);
qemu-kvm.c +0 -15
@@ -2321,21 +2321,6 @@ void kvm_set_phys_mem(target_phys_addr_t start_addr, ram_addr_t size,return ; } -int kvm_setup_guest_memory (void *area,unsigned long size) -{ -int ret =0 ; - -#ifdef MADV_DONTFORK -if (kvm_enabled () && !kvm_has_sync_mmu ()) - ret =madvise (area, size,MADV_DONTFORK ); -#endif - -if (ret) -perror ("madvise" ); - -return ret; -} -#ifdef KVM_CAP_SET_GUEST_DEBUG struct kvm_set_guest_debug_data {
qemu-kvm.h +0 -1
@@ -923,7 +923,6 @@ void kvm_cpu_destroy_phys_mem(target_phys_addr_t start_addr, unsigned long size);void kvm_qemu_log_memory (target_phys_addr_t start,target_phys_addr_t size,int log); -int kvm_setup_guest_memory (void *area,unsigned long size);int kvm_qemu_create_memory_alias (uint64_t phys_start,uint64_t len,uint64_t target_phys);int kvm_qemu_destroy_memory_alias (uint64_t phys_start);
Signature · 1 line
1.6.0.2
PATCH v2 7/21 qemu-kvm: Use some more upstream prototypes
In reply to Jan KiszkaDrop our private typedef of KVMState and use more identical upstream prototypes.
--- kvm.h | 10 +++++++--- qemu-kvm.c | 4 +++- qemu-kvm.h | 24 +++--------------------- 3 files changed, 13 insertions(+), 25 deletions(-)
kvm.h +7 -3
@@ -32,11 +32,13 @@ struct kvm_run;/* external API */ int kvm_init (int smp_cpus); +#endif /* KVM_UPSTREAM */ int kvm_init_vcpu (CPUState *env);int kvm_cpu_exec (CPUState *env); +#ifdef KVM_UPSTREAM void kvm_set_phys_mem (target_phys_addr_t start_addr,ram_addr_t size,ram_addr_t phys_offset);
@@ -47,19 +49,19 @@ int kvm_physical_sync_dirty_bitmap(target_phys_addr_t start_addr,int kvm_log_start (target_phys_addr_t phys_addr,ram_addr_t size);int kvm_log_stop (target_phys_addr_t phys_addr,ram_addr_t size);int kvm_set_migration_log (int enable); +#endif /* KVM_UPSTREAM */ int kvm_has_sync_mmu (void ); -#endif /* KVM_UPSTREAM */ int kvm_has_vcpu_events (void );int kvm_put_vcpu_events (CPUState *env);int kvm_get_vcpu_events (CPUState *env);void kvm_setup_guest_memory (void *start,size_t size); -#ifdef KVM_UPSTREAM int kvm_coalesce_mmio_region (target_phys_addr_t start,ram_addr_t size);int kvm_uncoalesce_mmio_region (target_phys_addr_t start,ram_addr_t size); +#ifdef KVM_UPSTREAM int kvm_insert_breakpoint (CPUState *current_env,target_ulong addr,target_ulong len,int type); int kvm_remove_breakpoint(CPUState *current_env, target_ulong addr,
@@ -69,6 +71,7 @@ int kvm_update_guest_debug(CPUState *env, unsigned long reinject_trap);int kvm_pit_in_kernel (void );int kvm_irqchip_in_kernel (void ); +#endif /* KVM_UPSTREAM */ /* internal API */
@@ -97,7 +100,6 @@ int kvm_arch_init(KVMState *s, int smp_cpus);int kvm_arch_init_vcpu (CPUState *env); -#endif void kvm_arch_reset_vcpu (CPUState *env);#ifdef KVM_UPSTREAM
@@ -131,9 +133,11 @@ int kvm_arch_remove_hw_breakpoint(target_ulong addr,void kvm_arch_remove_all_hw_breakpoints (void );void kvm_arch_update_guest_debug (CPUState *env,struct kvm_guest_debug *dbg); +#endif int kvm_check_extension (KVMState *s,unsigned int extension); +#ifdef KVM_UPSTREAM uint32_t kvm_arch_get_supported_cpuid (CPUState *env,uint32_t function,int reg);#endif
qemu-kvm.c +3 -1
@@ -1909,12 +1909,14 @@ static void *ap_main_loop(void *_env)return NULL ; } -void kvm_init_vcpu (CPUState *env) +int kvm_init_vcpu (CPUState *env) {pthread_create (&env->kvm_cpu_state.thread,NULL , ap_main_loop, env);while (env->created ==0 )qemu_cond_wait (&qemu_vcpu_cond); + +return 0 ; }int kvm_vcpu_inited (CPUState *env)
qemu-kvm.h +3 -21
@@ -891,7 +891,6 @@ int kvm_init_ap(void);int kvm_vcpu_inited (CPUState *env);void kvm_load_mpstate (CPUState *env);void kvm_save_mpstate (CPUState *env); -int kvm_cpu_exec (CPUState *env);int kvm_insert_breakpoint (CPUState * current_env,target_ulong addr,target_ulong len,int type); int kvm_remove_breakpoint(CPUState * current_env, target_ulong addr,
@@ -933,9 +932,6 @@ void kvm_arch_save_regs(CPUState *env);void kvm_arch_load_regs (CPUState *env);void kvm_arch_load_mpstate (CPUState *env);void kvm_arch_save_mpstate (CPUState *env); -int kvm_arch_init_vcpu (CPUState *cenv); -int kvm_arch_pre_run (CPUState *env,struct kvm_run *run); -int kvm_arch_post_run (CPUState *env,struct kvm_run *run);int kvm_arch_has_work (CPUState *env);void kvm_arch_process_irqchip_events (CPUState *env);int kvm_arch_try_push_interrupts (void *opaque);
@@ -981,8 +977,6 @@ void kvm_tpr_access_report(CPUState *env, uint64_t rip, int is_write);void kvm_tpr_vcpu_start (CPUState *env);int qemu_kvm_get_dirty_pages (unsigned long phys_addr,void *buf); -int kvm_coalesce_mmio_region (target_phys_addr_t start,ram_addr_t size); -int kvm_uncoalesce_mmio_region (target_phys_addr_t start,ram_addr_t size);int kvm_arch_init_irq_routing (void );
@@ -1021,17 +1015,14 @@ void qemu_kvm_cpu_stop(CPUState *env);int kvm_arch_halt (CPUState *env);int handle_tpr_access (void *opaque,CPUState *env,uint64_t rip,int is_write); -int kvm_has_sync_mmu (void );#define qemu_kvm_pit_in_kernel () kvm_pit_in_kernel(kvm_context)#define qemu_kvm_has_gsi_routing () kvm_has_gsi_routing(kvm_context)#ifdef TARGET_I386 #define qemu_kvm_has_pit_state2 () kvm_has_pit_state2(kvm_context)#endif -void kvm_init_vcpu (CPUState *env);void kvm_load_tsc (CPUState *env);#else -#define kvm_has_sync_mmu () (0)#define kvm_nested 0#define qemu_kvm_pit_in_kernel () (0)#define qemu_kvm_has_gsi_routing () (0)
@@ -1040,10 +1031,6 @@ void kvm_load_tsc(CPUState *env);#endif #define kvm_save_mpstate (env) do {} while(0)#define qemu_kvm_cpu_stop (env) do {} while(0) -static inline void kvm_init_vcpu (CPUState *env) -{ -} -static inline void kvm_load_tsc (CPUState *env) { }
@@ -1086,7 +1073,7 @@ typedef struct KVMSlot {typedef struct kvm_dirty_log KVMDirtyLog ; -typedef struct KVMState { +struct KVMState {KVMSlot slots[32 ];int fd;int vmfd;
@@ -1100,14 +1087,9 @@ typedef struct KVMState {int irqchip_in_kernel;struct kvm_context kvm_context; -} KVMState; - -extern KVMState *kvm_state; +}; -int kvm_ioctl (KVMState *s,int type, ...); -int kvm_vm_ioctl (KVMState *s,int type, ...); -int kvm_vcpu_ioctl (CPUState *env,int type, ...); -int kvm_check_extension (KVMState *s,unsigned int ext); +extern struct KVMState *kvm_state;int kvm_tpr_enable_vapic (CPUState *env);
Signature · 1 line
1.6.0.2
PATCH v2 8/21 qemu-kvm: Use upstream kvm_arch_get_supported_cpuid
In reply to Jan KiszkaIt is idential to our version now, so drop the copy.
--- kvm.h | 3 - qemu-kvm-x86.c | 106 ----------------------------------------------------- qemu-kvm.h | 5 -- target-i386/kvm.c | 4 +- 4 files changed, 2 insertions(+), 116 deletions(-)
kvm.h +0 -3
@@ -137,11 +137,8 @@ void kvm_arch_update_guest_debug(CPUState *env, struct kvm_guest_debug *dbg);int kvm_check_extension (KVMState *s,unsigned int extension); -#ifdef KVM_UPSTREAM uint32_t kvm_arch_get_supported_cpuid (CPUState *env,uint32_t function,int reg); -#endif -void kvm_cpu_synchronize_state (CPUState *env);/* generic hooks - to be moved/refactored once there are more users */
qemu-kvm-x86.c +0 -106
118 lines not on this page · raw message
qemu-kvm.h +0 -5
@@ -859,8 +859,6 @@ int kvm_assign_set_msix_entry(kvm_context_t kvm, struct kvm_assigned_msix_entry *entry);#endif -uint32_t kvm_get_supported_cpuid (kvm_context_t kvm,uint32_t function,int reg); -#else /* !CONFIG_KVM */ typedef struct kvm_context *kvm_context_t ;
@@ -1051,9 +1049,6 @@ static inline int kvm_sync_vcpus(void)return 0 ; } -uint32_t kvm_arch_get_supported_cpuid (CPUState *env,uint32_t function, -int reg); -static inline int kvm_set_migration_log (int enable) {return kvm_physical_memory_set_dirty_tracking (enable);
target-i386/kvm.c +2 -2
@@ -25,8 +25,6 @@#include "gdbstub.h" #include "host-utils.h" -#ifdef KVM_UPSTREAM -#ifdef CONFIG_KVM_PARA #include <linux/kvm_para.h> #endif
@@ -133,6 +131,8 @@ uint32_t kvm_arch_get_supported_cpuid(CPUState *env, uint32_t function, int reg)#endif +#ifdef KVM_UPSTREAM +static void kvm_trim_features (uint32_t *features,uint32_t supported) {int i;
Signature · 1 line
1.6.0.2
PATCH v2 9/21 qemu-kvm: Use upstream kvm_pit_in_kernel
In reply to Jan KiszkaDrop private version in favor of recently added upstream service and track it state directly in KVMState.
--- hw/i8254.c | 4 ++-- hw/pc.c | 2 +- hw/pcspk.c | 4 ++-- kvm-all.c | 2 +- kvm.h | 2 +- qemu-kvm-x86.c | 12 ++++++------ qemu-kvm.c | 5 ----- qemu-kvm.h | 13 +------------ 8 files changed, 14 insertions(+), 30 deletions(-)
hw/i8254.c +2 -2
@@ -491,7 +491,7 @@ void hpet_disable_pit(void) {PITChannelState *s = &pit_state.channels[0 ]; -if (kvm_enabled () &&qemu_kvm_pit_in_kernel ()) { +if (kvm_enabled () &&kvm_pit_in_kernel ()) {if (qemu_kvm_has_pit_state2 ()) {kvm_hpet_disable_kpit (); }else {
@@ -515,7 +515,7 @@ void hpet_enable_pit(void)PITState *pit = &pit_state;PITChannelState *s = &pit->channels[0 ]; -if (kvm_enabled () &&qemu_kvm_pit_in_kernel ()) { +if (kvm_enabled () &&kvm_pit_in_kernel ()) {if (qemu_kvm_has_pit_state2 ()) {kvm_hpet_enable_kpit (); }else {
hw/pc.c +1 -1
@@ -951,7 +951,7 @@ static void pc_init1(ram_addr_t ram_size, ioapic_irq_hack = isa_irq; }#ifdef CONFIG_KVM_PIT -if (kvm_enabled () &&qemu_kvm_pit_in_kernel ()) +if (kvm_enabled () &&kvm_pit_in_kernel ()) pit =kvm_pit_init (0x40 ,isa_reserve_irq (0 )); else#endif
hw/pcspk.c +2 -2
@@ -56,7 +56,7 @@ static void kvm_get_pit_ch2(PITState *pit, {struct kvm_pit_state pit_state; -if (kvm_enabled () &&qemu_kvm_pit_in_kernel ()) { +if (kvm_enabled () &&kvm_pit_in_kernel ()) {kvm_get_pit (kvm_context, &pit_state); pit->channels[2 ].mode = pit_state.channels[2 ].mode; pit->channels[2 ].count = pit_state.channels[2 ].count;
@@ -71,7 +71,7 @@ static void kvm_get_pit_ch2(PITState *pit,static void kvm_set_pit_ch2 (PITState *pit,struct kvm_pit_state *inkernel_state) { -if (kvm_enabled () &&qemu_kvm_pit_in_kernel ()) { +if (kvm_enabled () &&kvm_pit_in_kernel ()) { inkernel_state->channels[2 ].mode = pit->channels[2 ].mode; inkernel_state->channels[2 ].count = pit->channels[2 ].count; inkernel_state->channels[2].count_load_time =
kvm-all.c +1 -1
@@ -164,13 +164,13 @@ int kvm_irqchip_in_kernel(void)return kvm_state->irqchip_in_kernel; } -#ifdef KVM_UPSTREAM int kvm_pit_in_kernel (void ) {return kvm_state->pit_in_kernel; } +#ifdef KVM_UPSTREAM int kvm_init_vcpu (CPUState *env) {KVMState *s = kvm_state;
kvm.h +1 -1
@@ -68,10 +68,10 @@ int kvm_remove_breakpoint(CPUState *current_env, target_ulong addr, target_ulong len, int type);void kvm_remove_all_breakpoints (CPUState *current_env);int kvm_update_guest_debug (CPUState *env,unsigned long reinject_trap); +#endif /* KVM_UPSTREAM */ int kvm_pit_in_kernel (void );int kvm_irqchip_in_kernel (void ); -#endif /* KVM_UPSTREAM */ /* internal API */
qemu-kvm-x86.c +6 -6
@@ -119,13 +119,13 @@ static int kvm_create_pit(kvm_context_t kvm)#ifdef KVM_CAP_PIT int r; - kvm->pit_in_kernel =0 ; + kvm_state->pit_in_kernel =0 ;if (!kvm->no_pit_creation) { r =kvm_ioctl (kvm_state,KVM_CHECK_EXTENSION ,KVM_CAP_PIT );if (r >0 ) { r =kvm_vm_ioctl (kvm_state,KVM_CREATE_PIT );if (r >=0 ) - kvm->pit_in_kernel =1 ; + kvm_state->pit_in_kernel =1 ;else {fprintf (stderr,"Create kernel PIC irqchip failed\n" );return r;
@@ -311,14 +311,14 @@ int kvm_set_lapic(CPUState *env, struct kvm_lapic_state *s)int kvm_get_pit (kvm_context_t kvm,struct kvm_pit_state *s) { -if (!kvm->pit_in_kernel) +if (!kvm_pit_in_kernel ())return 0 ;return kvm_vm_ioctl (kvm_state,KVM_GET_PIT , s); }int kvm_set_pit (kvm_context_t kvm,struct kvm_pit_state *s) { -if (!kvm->pit_in_kernel) +if (!kvm_pit_in_kernel ())return 0 ;return kvm_vm_ioctl (kvm_state,KVM_SET_PIT , s); }
@@ -326,14 +326,14 @@ int kvm_set_pit(kvm_context_t kvm, struct kvm_pit_state *s)#ifdef KVM_CAP_PIT_STATE2 int kvm_get_pit2 (kvm_context_t kvm,struct kvm_pit_state2 *ps2) { -if (!kvm->pit_in_kernel) +if (!kvm_pit_in_kernel ())return 0 ;return kvm_vm_ioctl (kvm_state,KVM_GET_PIT2 , ps2); }int kvm_set_pit2 (kvm_context_t kvm,struct kvm_pit_state2 *ps2) { -if (!kvm->pit_in_kernel) +if (!kvm_pit_in_kernel ())return 0 ;return kvm_vm_ioctl (kvm_state,KVM_SET_PIT2 , ps2); }
qemu-kvm.c +0 -5
@@ -1040,11 +1040,6 @@ int kvm_set_signal_mask(CPUState *env, const sigset_t *sigset)return r; } -int kvm_pit_in_kernel (kvm_context_t kvm) -{ -return kvm->pit_in_kernel; -} -int kvm_inject_nmi (CPUState *env) {#ifdef KVM_CAP_USER_NMI
qemu-kvm.h +1 -12
@@ -64,8 +64,6 @@ struct kvm_context {int irqchip_inject_ioctl;/// do not create in-kernel pit if set int no_pit_creation; -/// in-kernel pit status -int pit_in_kernel;#ifdef KVM_CAP_IRQ_ROUTING struct kvm_irq_routing *irq_routes;int nr_allocated_irq_routes;
@@ -563,13 +561,6 @@ void kvm_inject_x86_mce(CPUState *cenv, int bank, uint64_t status, int abort_on_error);/*! -* \brief Query wheather in kernel pit is used -* -* \param kvm Pointer to the current kvm_context -*/ -int kvm_pit_in_kernel (kvm_context_t kvm); - -/*! * \brief Initialize coalesced MMIO * * Check for coalesced MMIO capability and store in context
@@ -1014,7 +1005,6 @@ int kvm_arch_halt(CPUState *env);int handle_tpr_access (void *opaque,CPUState *env,uint64_t rip,int is_write); -#define qemu_kvm_pit_in_kernel () kvm_pit_in_kernel(kvm_context)#define qemu_kvm_has_gsi_routing () kvm_has_gsi_routing(kvm_context)#ifdef TARGET_I386 #define qemu_kvm_has_pit_state2 () kvm_has_pit_state2(kvm_context)
@@ -1022,7 +1012,6 @@ int handle_tpr_access(void *opaque, CPUState *env, uint64_t rip,void kvm_load_tsc (CPUState *env);#else #define kvm_nested 0 -#define qemu_kvm_pit_in_kernel () (0)#define qemu_kvm_has_gsi_routing () (0)#ifdef TARGET_I386 #define qemu_kvm_has_pit_state2 () (0)
@@ -1055,7 +1044,6 @@ static inline int kvm_set_migration_log(int enable) } -int kvm_irqchip_in_kernel (void ); #ifdef CONFIG_KVM typedef struct KVMSlot {
@@ -1080,6 +1068,7 @@ struct KVMState { QTAILQ_HEAD(, kvm_sw_breakpoint) kvm_sw_breakpoints;#endif int irqchip_in_kernel; +int pit_in_kernel;struct kvm_context kvm_context; };
Signature · 1 line
1.6.0.2
PATCH v2 10/21 KVM: Move and rename regs_modified
In reply to Jan KiszkaTouching the user space representation of KVM's VCPU state is - naturally - a per-VCPU thing. So move the dirty flag into KVM_CPU_COMMON and rename it at this chance to reflect its true meaning.
--- cpu-defs.h | 1 + kvm-all.c | 12 ++++++------ 2 files changed, 7 insertions(+), 6 deletions(-)
cpu-defs.h +1 -0
@@ -208,6 +208,7 @@ struct KVMCPUState {struct KVMState *kvm_state; \struct kvm_run *kvm_run; \int kvm_fd; \ +int kvm_vcpu_dirty; \uint32_t stop;/* Stop request */ \uint32_t stopped;/* Artificially stopped */ \struct KVMCPUState kvm_cpu_state;
kvm-all.c +6 -6
@@ -573,9 +573,9 @@ static void kvm_run_coalesced_mmio(CPUState *env, struct kvm_run *run)void kvm_cpu_synchronize_state (CPUState *env) { -if (!env->kvm_state->regs_modified) { +if (!env->kvm_vcpu_dirty) {kvm_arch_get_registers (env); - env->kvm_state->regs_modified =1 ; + env->kvm_vcpu_dirty =1 ; } }
@@ -593,9 +593,9 @@ int kvm_cpu_exec(CPUState *env)break ; } -if (env->kvm_state->regs_modified) { +if (env->kvm_vcpu_dirty) {kvm_arch_put_registers (env); - env->kvm_state->regs_modified =0 ; + env->kvm_vcpu_dirty =0 ; }kvm_arch_pre_run (env, run);
@@ -951,9 +951,9 @@ static void kvm_invoke_set_guest_debug(void *data)struct kvm_set_guest_debug_data *dbg_data = data;CPUState *env = dbg_data->env; -if (env->kvm_state->regs_modified) { +if (env->kvm_vcpu_dirty) {kvm_arch_put_registers (env); - env->kvm_state->regs_modified =0 ; + env->kvm_vcpu_dirty =0 ; } dbg_data->err =kvm_vcpu_ioctl (env,KVM_SET_GUEST_DEBUG , &dbg_data->dbg); }
Signature · 1 line
1.6.0.2
PATCH v2 11/21 KVM: Rework of guest debug state writing
In reply to Jan KiszkaSo far we synchronized any dirty VCPU state back into the kernel before updating the guest debug state. This was a tribute to a deficite in x86 kernels before 2.6.33. But as this is an arch-dependent issue, it is better handle in the x86 part of KVM and remove the writeback point for generic code.
--- kvm-all.c | 12 ++++-------- target-i386/cpu.h | 9 ++++++++- target-i386/kvm.c | 11 +++++++++++ 3 files changed, 23 insertions(+), 9 deletions(-)
kvm-all.c +4 -8
@@ -951,10 +951,6 @@ static void kvm_invoke_set_guest_debug(void *data)struct kvm_set_guest_debug_data *dbg_data = data;CPUState *env = dbg_data->env; -if (env->kvm_vcpu_dirty) { -kvm_arch_put_registers (env); - env->kvm_vcpu_dirty =0 ; - } dbg_data->err =kvm_vcpu_ioctl (env,KVM_SET_GUEST_DEBUG , &dbg_data->dbg); }
@@ -962,12 +958,12 @@ int kvm_update_guest_debug(CPUState *env, unsigned long reinject_trap) {struct kvm_set_guest_debug_data data; - data.dbg.control =0 ; -if (env->singlestep_enabled) - data.dbg.control =KVM_GUESTDBG_ENABLE |KVM_GUESTDBG_SINGLESTEP ; + data.dbg.control = reinject_trap; +if (env->singlestep_enabled) { + data.dbg.control |=KVM_GUESTDBG_ENABLE |KVM_GUESTDBG_SINGLESTEP ; + }kvm_arch_update_guest_debug (env, &data.dbg); - data.dbg.control |= reinject_trap; data.env = env;on_vcpu (env, kvm_invoke_set_guest_debug, &data);
target-i386/cpu.h +8 -1
@@ -21,6 +21,10 @@#include "config.h" +#ifdef CONFIG_KVM +#include <linux/kvm.h> /* for kvm_guest_debug */ +#endif +#ifdef TARGET_X86_64 #define TARGET_LONG_BITS 64#else
@@ -718,7 +722,10 @@ typedef struct CPUX86State {uint8_t has_error_code;uint32_t sipi_vector;uint32_t cpuid_kvm_features; - +#if defined(CONFIG_KVM ) && defined(KVM_CAP_SET_GUEST_DEBUG ) +struct kvm_guest_debug kvm_guest_debug; +#endif +/* in order to simplify APIC support, we leave this pointer to the user */ struct APICState *apic_state;
target-i386/kvm.c +11 -0
@@ -865,6 +865,15 @@ int kvm_arch_put_registers(CPUState *env)if (ret <0 )return ret; +/* +* Kernels before 2.6.33 overwrote flags.TF injected via SET_GUEST_DEBUG +* while updating GP regs. Work around this by updating the debug state +* once again. +*/ + ret =kvm_vcpu_ioctl (env,KVM_SET_GUEST_DEBUG , &env->kvm_guest_debug); +if (ret <0 ) +return ret; + ret =kvm_put_fpu (env);if (ret <0 )return ret;
@@ -1163,6 +1172,8 @@ void kvm_arch_update_guest_debug(CPUState *env, struct kvm_guest_debug *dbg) (len_code[hw_breakpoint[n].len] << (18 + n*4 )); } } +/* Keep a copy for the writeback workaround in kvm_arch_put_registers */ +memcpy (&env->kvm_guest_debug, dbg,sizeof (env->kvm_guest_debug)); }#endif /* KVM_CAP_SET_GUEST_DEBUG */ #endif
Signature · 1 line
1.6.0.2
PATCH v2 12/21 qemu-kvm: Use upstream kvm_vcpu_dirty
In reply to Jan KiszkaDrop regs_modified in favor of upstream's equivalent and clean up kvm_cpu_synchronize_state at this chance.
--- cpu-defs.h | 1 - hw/pc.c | 2 +- qemu-kvm.c | 18 +++++++++--------- 3 files changed, 10 insertions(+), 11 deletions(-)
cpu-defs.h +0 -1
@@ -142,7 +142,6 @@ struct KVMCPUState {pthread_t thread;int signalled;struct qemu_work_item *queued_work_first, *queued_work_last; -int regs_modified; };#define CPU_TEMP_BUF_NLONGS 128
hw/pc.c +1 -1
@@ -744,7 +744,7 @@ CPUState *pc_new_cpu(const char *cpu_model)fprintf (stderr,"Unable to find x86 CPU definition\n" );exit (1 ); } - env->kvm_cpu_state.regs_modified =1 ; + env->kvm_vcpu_dirty =1 ;if ((env->cpuid_features &CPUID_APIC ) || smp_cpus >1 ) { env->cpuid_apic_id = env->cpu_index;/* APIC reset callback resets cpu */
qemu-kvm.c +9 -9
@@ -861,9 +861,9 @@ int pre_kvm_run(kvm_context_t kvm, CPUState *env) {kvm_arch_pre_run (env, env->kvm_run); -if (env->kvm_cpu_state.regs_modified) { +if (env->kvm_vcpu_dirty) {kvm_arch_load_regs (env); - env->kvm_cpu_state.regs_modified =0 ; + env->kvm_vcpu_dirty =0 ; }pthread_mutex_unlock (&qemu_mutex);
@@ -1530,16 +1530,16 @@ static void on_vcpu(CPUState *env, void (*func)(void *data), void *data)static void do_kvm_cpu_synchronize_state (void *_env) {CPUState *env = _env; -if (!env->kvm_cpu_state.regs_modified) { -kvm_arch_save_regs (env); - env->kvm_cpu_state.regs_modified =1 ; - } + +kvm_arch_save_regs (env); }void kvm_cpu_synchronize_state (CPUState *env) { -if (!env->kvm_cpu_state.regs_modified) +if (!env->kvm_vcpu_dirty) {on_vcpu (env, do_kvm_cpu_synchronize_state, env); + env->kvm_vcpu_dirty =1 ; + } }static void inject_interrupt (void *data)
@@ -2329,9 +2329,9 @@ static void kvm_invoke_set_guest_debug(void *data) {struct kvm_set_guest_debug_data *dbg_data = data; -if (cpu_single_env->kvm_cpu_state.regs_modified) { +if (cpu_single_env->kvm_vcpu_dirty) {kvm_arch_save_regs (cpu_single_env); - cpu_single_env->kvm_cpu_state.regs_modified =0 ; + cpu_single_env->kvm_vcpu_dirty =0 ; } dbg_data->err = kvm_set_guest_debug(cpu_single_env,
Signature · 1 line
1.6.0.2
PATCH v2 13/21 qemu-kvm: Use upstream guest debug code
In reply to Jan KiszkaCode was absolute identical except for previous cleanup in upstream.
--- kvm-all.c | 7 +- kvm.h | 4 - qemu-kvm-x86.c | 178 ++-------------------------------------------------- qemu-kvm.c | 44 ------------- qemu-kvm.h | 37 ----------- target-i386/kvm.c | 2 +- 6 files changed, 11 insertions(+), 261 deletions(-)
kvm-all.c +3 -4
@@ -919,7 +919,9 @@ static void on_vcpu(CPUState *env, void (*func)(void *data), void *data)func (data); #endif } -#endif /* KVM_UPSTREAM */ +#else /* !KVM_UPSTREAM */ +static void on_vcpu (CPUState *env,void (*func)(void *data),void *data); +#endif /* !KVM_UPSTREAM */ struct kvm_sw_breakpoint *kvm_find_sw_breakpoint (CPUState *env,target_ulong pc)
@@ -938,8 +940,6 @@ int kvm_sw_breakpoints_active(CPUState *env)return !QTAILQ_EMPTY (&env->kvm_state->kvm_sw_breakpoints); } -#ifdef KVM_UPSTREAM -struct kvm_set_guest_debug_data {struct kvm_guest_debug dbg;CPUState *env;
@@ -969,7 +969,6 @@ int kvm_update_guest_debug(CPUState *env, unsigned long reinject_trap)on_vcpu (env, kvm_invoke_set_guest_debug, &data);return data.err; } -#endif int kvm_insert_breakpoint (CPUState *current_env,target_ulong addr,target_ulong len,int type)
kvm.h +0 -4
@@ -61,14 +61,12 @@ void kvm_setup_guest_memory(void *start, size_t size);int kvm_coalesce_mmio_region (target_phys_addr_t start,ram_addr_t size);int kvm_uncoalesce_mmio_region (target_phys_addr_t start,ram_addr_t size); -#ifdef KVM_UPSTREAM int kvm_insert_breakpoint (CPUState *current_env,target_ulong addr,target_ulong len,int type);int kvm_remove_breakpoint (CPUState *current_env,target_ulong addr,target_ulong len,int type);void kvm_remove_all_breakpoints (CPUState *current_env);int kvm_update_guest_debug (CPUState *env,unsigned long reinject_trap); -#endif /* KVM_UPSTREAM */ int kvm_pit_in_kernel (void );int kvm_irqchip_in_kernel (void );
@@ -101,7 +99,6 @@ int kvm_arch_init(KVMState *s, int smp_cpus);int kvm_arch_init_vcpu (CPUState *env);void kvm_arch_reset_vcpu (CPUState *env); -#ifdef KVM_UPSTREAM struct kvm_guest_debug ;struct kvm_debug_exit_arch ;
@@ -133,7 +130,6 @@ int kvm_arch_remove_hw_breakpoint(target_ulong addr,void kvm_arch_remove_all_hw_breakpoints (void );void kvm_arch_update_guest_debug (CPUState *env,struct kvm_guest_debug *dbg); -#endif int kvm_check_extension (KVMState *s,unsigned int extension);
qemu-kvm-x86.c +7 -171
190 lines not on this page · raw message
qemu-kvm.c +0 -44
56 lines not on this page · raw message
qemu-kvm.h +0 -37
@@ -323,10 +323,6 @@ static inline int kvm_reset_mpstate(CPUState *env)*/ int kvm_inject_irq (CPUState *env,unsigned irq); -#ifdef KVM_CAP_SET_GUEST_DEBUG -int kvm_set_guest_debug (CPUState *env,struct kvm_guest_debug *dbg); -#endif -#if defined(__i386__) || defined(__x86_64__)/*! * \brief Setup a vcpu's cpuid instruction emulation
@@ -880,12 +876,6 @@ int kvm_init_ap(void);int kvm_vcpu_inited (CPUState *env);void kvm_load_mpstate (CPUState *env);void kvm_save_mpstate (CPUState *env); -int kvm_insert_breakpoint (CPUState * current_env,target_ulong addr, -target_ulong len,int type); -int kvm_remove_breakpoint (CPUState * current_env,target_ulong addr, -target_ulong len,int type); -void kvm_remove_all_breakpoints (CPUState * current_env); -int kvm_update_guest_debug (CPUState *env,unsigned long reinject_trap);void kvm_apic_init (CPUState *env);/* called from vcpu initialization */ void qemu_kvm_load_lapic (CPUState *env);
@@ -928,33 +918,6 @@ void kvm_arch_push_nmi(void *opaque);void kvm_arch_cpu_reset (CPUState *env);int kvm_set_boot_cpu_id (uint32_t id); -struct kvm_guest_debug ; -struct kvm_debug_exit_arch ; - -struct kvm_sw_breakpoint { -target_ulong pc; -target_ulong saved_insn; -int use_count; -QTAILQ_ENTRY (kvm_sw_breakpoint) entry; -}; - -QTAILQ_HEAD (kvm_sw_breakpoint_head, kvm_sw_breakpoint); - -int kvm_arch_debug (struct kvm_debug_exit_arch *arch_info); -int kvm_sw_breakpoints_active (CPUState *env); -struct kvm_sw_breakpoint *kvm_find_sw_breakpoint (CPUState *env, -target_ulong pc); -int kvm_arch_insert_sw_breakpoint (CPUState * current_env, -struct kvm_sw_breakpoint *bp); -int kvm_arch_remove_sw_breakpoint (CPUState * current_env, -struct kvm_sw_breakpoint *bp); -int kvm_arch_insert_hw_breakpoint (target_ulong addr,target_ulong len, -int type); -int kvm_arch_remove_hw_breakpoint (target_ulong addr,target_ulong len, -int type); -void kvm_arch_remove_all_hw_breakpoints (void ); -void kvm_arch_update_guest_debug (CPUState *env,struct kvm_guest_debug *dbg); -void qemu_kvm_aio_wait_start (void );void qemu_kvm_aio_wait (void );void qemu_kvm_aio_wait_end (void );
target-i386/kvm.c +1 -1
@@ -1003,6 +1003,7 @@ int kvm_arch_handle_exit(CPUState *env, struct kvm_run *run) return ret; } +#endif #ifdef KVM_CAP_SET_GUEST_DEBUG int kvm_arch_insert_sw_breakpoint (CPUState *env,struct kvm_sw_breakpoint *bp)
@@ -1176,6 +1177,5 @@ void kvm_arch_update_guest_debug(CPUState *env, struct kvm_guest_debug *dbg) memcpy(&env->kvm_guest_debug, dbg, sizeof(env->kvm_guest_debug)); }#endif /* KVM_CAP_SET_GUEST_DEBUG */ -#endif #include "qemu-kvm-x86.c"
Signature · 1 line
1.6.0.2
PATCH v2 14/21 qemu-kvm: Rework VCPU state writeback API
In reply to Jan KiszkaThis grand cleanup drops all reset and vmsave/load related synchronization points in favor of four(!) generic hooks:
- cpu_synchronize_all_states in qemu_savevm_state_complete (initial sync from kernel before vmsave)
- cpu_synchronize_all_post_init in qemu_loadvm_state (writeback after vmload)
- cpu_synchronize_all_post_init in main after machine init
- cpu_synchronize_all_post_reset in qemu_system_reset (writeback after system reset)
These writeback points + the existing one of VCPU exec after cpu_synchronize_state map on three levels of writeback:
- KVM_PUT_ASYNC_STATE (during runtime, other VCPUs continue to run)
- KVM_PUT_RESET_STATE (on synchronous system reset, all VCPUs stopped)
- KVM_PUT_FULL_STATE (on init or vmload, all VCPUs stopped as well)
This level is passed to the arch-specific VCPU state writing function that will decide which concrete substates need to be written. That way, no writer of load, save or reset functions that interact with in-kernel KVM states will ever have to worry about synchronization again. That also means that a lot of reasons for races, segfaults and deadlocks are eliminated.
cpu_synchronize_state remains untouched, just as Anthony suggested. We continue to need it before reading or writing of VCPU states that are also tracked by in-kernel KVM subsystems.
Consequently, this patch removes many cpu_synchronize_state calls that are now redundant, just like remaining explicit register syncs. It does not touch qemu-kvm's special hooks for mpstate, vcpu_events, or tsc loading. They will be cleaned up by individual patches.
--- exec.c | 17 ----------------- hw/apic.c | 3 --- hw/pc.c | 1 - hw/ppc_newworld.c | 3 --- hw/ppc_oldworld.c | 3 --- hw/s390-virtio.c | 1 - kvm-all.c | 19 +++++++++++++------ kvm.h | 22 +++++++++++++++++++++- qemu-kvm-ia64.c | 2 +- qemu-kvm-x86.c | 3 +-- qemu-kvm.c | 16 +++++++++++++--- qemu-kvm.h | 2 +- savevm.c | 4 ++++ sysemu.h | 4 ++++ target-i386/kvm.c | 2 +- target-i386/machine.c | 10 ---------- target-ia64/machine.c | 2 -- target-ppc/kvm.c | 2 +- target-ppc/machine.c | 4 ---- target-s390x/kvm.c | 3 +-- vl.c | 29 +++++++++++++++++++++++++++++ 21 files changed, 90 insertions(+), 62 deletions(-)
exec.c +0 -17
@@ -529,21 +529,6 @@ void cpu_exec_init_all(unsigned long tb_size)#if defined(CPU_SAVE_VERSION ) && !defined(CONFIG_USER_ONLY ) -static void cpu_common_pre_save (void *opaque) -{ -CPUState *env = opaque; - -cpu_synchronize_state (env); -} - -static int cpu_common_pre_load (void *opaque) -{ -CPUState *env = opaque; - -cpu_synchronize_state (env); -return 0 ; -} -static int cpu_common_post_load (void *opaque,int version_id) {CPUState *env = opaque;
@@ -561,8 +546,6 @@ static const VMStateDescription vmstate_cpu_common = { .version_id = 1, .minimum_version_id = 1, .minimum_version_id_old = 1, - .pre_save = cpu_common_pre_save, - .pre_load = cpu_common_pre_load, .post_load = cpu_common_post_load, .fields = (VMStateField []) {VMSTATE_UINT32 (halted, CPUState),
hw/apic.c +0 -3
@@ -488,7 +488,6 @@ void apic_init_reset(CPUState *env)if (!s)return ; -cpu_synchronize_state (env); s->tpr =0 ; s->spurious_vec =0xff ; s->log_dest =0 ;
@@ -1070,8 +1069,6 @@ static void apic_reset(void *opaque)APICState *s = opaque;int bsp; -cpu_synchronize_state (s->cpu_env); - bsp =cpu_is_bsp (s->cpu_env); s->apicbase =0xfee00000 | (bsp ?MSR_IA32_APICBASE_BSP :0 ) |MSR_IA32_APICBASE_ENABLE ;
hw/pc.c +0 -1
@@ -744,7 +744,6 @@ CPUState *pc_new_cpu(const char *cpu_model)fprintf (stderr,"Unable to find x86 CPU definition\n" );exit (1 ); } - env->kvm_vcpu_dirty =1 ;if ((env->cpuid_features &CPUID_APIC ) || smp_cpus >1 ) { env->cpuid_apic_id = env->cpu_index;/* APIC reset callback resets cpu */
hw/ppc_newworld.c +0 -3
@@ -139,9 +139,6 @@ static void ppc_core99_init (ram_addr_t ram_size, envs[i] = env; } -/* Make sure all register sets take effect */ -cpu_synchronize_state (env); -/* allocate RAM */ ram_offset =qemu_ram_alloc (ram_size);cpu_register_physical_memory (0 , ram_size, ram_offset);
hw/ppc_oldworld.c +0 -3
@@ -164,9 +164,6 @@ static void ppc_heathrow_init (ram_addr_t ram_size, envs[i] = env; } -/* Make sure all register sets take effect */ -cpu_synchronize_state (env); -/* allocate RAM */ if (ram_size > (2047 <<20 )) { fprintf(stderr,
hw/s390-virtio.c +0 -1
@@ -185,7 +185,6 @@ static void s390_init(ram_addr_t ram_size,exit (1 ); } -cpu_synchronize_state (env); env->psw.addr =KERN_IMAGE_START ; env->psw.mask =0x0000000180000000ULL ; }
kvm-all.c +13 -6
@@ -152,10 +152,6 @@ static void kvm_reset_vcpu(void *opaque)CPUState *env = opaque; kvm_arch_reset_vcpu(env); -if (kvm_arch_put_registers (env)) { -fprintf (stderr,"Fatal: kvm vcpu reset failed\n" ); -abort (); - } }#endif
@@ -206,7 +202,6 @@ int kvm_init_vcpu(CPUState *env)if (ret ==0 ) {qemu_register_reset (kvm_reset_vcpu, env);kvm_arch_reset_vcpu (env); - ret =kvm_arch_put_registers (env); } err:return ret;
@@ -579,6 +574,18 @@ void kvm_cpu_synchronize_state(CPUState *env) } } +void kvm_cpu_synchronize_post_reset (CPUState *env) +{ +kvm_arch_put_registers (env,KVM_PUT_RESET_STATE ); + env->kvm_vcpu_dirty =0 ; +} + +void kvm_cpu_synchronize_post_init (CPUState *env) +{ +kvm_arch_put_registers (env,KVM_PUT_FULL_STATE ); + env->kvm_vcpu_dirty =0 ; +} +int kvm_cpu_exec (CPUState *env) {struct kvm_run *run = env->kvm_run;
@@ -594,7 +601,7 @@ int kvm_cpu_exec(CPUState *env) }if (env->kvm_vcpu_dirty) { -kvm_arch_put_registers (env); +kvm_arch_put_registers (env,KVM_PUT_ASYNC_STATE ); env->kvm_vcpu_dirty =0 ; }
kvm.h +21 -1
@@ -92,7 +92,11 @@ int kvm_arch_pre_run(CPUState *env, struct kvm_run *run);int kvm_arch_get_registers (CPUState *env); -int kvm_arch_put_registers (CPUState *env); +#define KVM_PUT_ASYNC_STATE 1 +#define KVM_PUT_RESET_STATE 2 +#define KVM_PUT_FULL_STATE 3 + +int kvm_arch_put_registers (CPUState *env,int level);int kvm_arch_init (KVMState *s,int smp_cpus);
@@ -136,6 +140,8 @@ int kvm_check_extension(KVMState *s, unsigned int extension);uint32_t kvm_arch_get_supported_cpuid (CPUState *env,uint32_t function,int reg);void kvm_cpu_synchronize_state (CPUState *env); +void kvm_cpu_synchronize_post_reset (CPUState *env); +void kvm_cpu_synchronize_post_init (CPUState *env);/* generic hooks - to be moved/refactored once there are more users */
@@ -146,4 +152,18 @@ static inline void cpu_synchronize_state(CPUState *env) } } +static inline void cpu_synchronize_post_reset (CPUState *env) +{ +if (kvm_enabled ()) { +kvm_cpu_synchronize_post_reset (env); + } +} + +static inline void cpu_synchronize_post_init (CPUState *env) +{ +if (kvm_enabled ()) { +kvm_cpu_synchronize_post_init (env); + } +} +#endif
qemu-kvm-ia64.c +1 -1
@@ -16,7 +16,7 @@ int kvm_arch_qemu_create_context(void)return 0 ; } -void kvm_arch_load_regs (CPUState *env) +void kvm_arch_load_regs (CPUState *env,int level) { }
qemu-kvm-x86.c +1 -2
@@ -803,7 +803,7 @@ static void get_seg(SegmentCache *lhs, const struct kvm_segment *rhs) | (rhs->avl * DESC_AVL_MASK); } -void kvm_arch_load_regs (CPUState *env) +void kvm_arch_load_regs (CPUState *env,int level) {struct kvm_regs regs;struct kvm_fpu fpu;
@@ -1369,7 +1369,6 @@ void kvm_arch_push_nmi(void *opaque)void kvm_arch_cpu_reset (CPUState *env) {kvm_arch_reset_vcpu (env); -kvm_arch_load_regs (env);kvm_put_vcpu_events (env);if (!cpu_is_bsp (env)) {if (kvm_irqchip_in_kernel ()) {
qemu-kvm.c +13 -3
@@ -862,7 +862,7 @@ int pre_kvm_run(kvm_context_t kvm, CPUState *env)kvm_arch_pre_run (env, env->kvm_run);if (env->kvm_vcpu_dirty) { -kvm_arch_load_regs (env); +kvm_arch_load_regs (env,KVM_PUT_ASYNC_STATE ); env->kvm_vcpu_dirty =0 ; }
@@ -1535,6 +1535,18 @@ void kvm_cpu_synchronize_state(CPUState *env) } } +void kvm_cpu_synchronize_post_reset (CPUState *env) +{ +kvm_arch_load_regs (env,KVM_PUT_RESET_STATE ); + env->kvm_vcpu_dirty =0 ; +} + +void kvm_cpu_synchronize_post_init (CPUState *env) +{ +kvm_arch_load_regs (env,KVM_PUT_FULL_STATE ); + env->kvm_vcpu_dirty =0 ; +} +static void inject_interrupt (void *data) {cpu_interrupt (current_env, (long ) data);
@@ -1880,8 +1892,6 @@ static void *ap_main_loop(void *_env)kvm_arch_init_vcpu (env); -kvm_arch_load_regs (env); -/* signal VCPU creation */ current_env->created =1 ;pthread_cond_signal (&qemu_vcpu_cond);
qemu-kvm.h +1 -1
@@ -908,7 +908,7 @@ int kvm_qemu_destroy_memory_alias(uint64_t phys_start);int kvm_arch_qemu_create_context (void );void kvm_arch_save_regs (CPUState *env); -void kvm_arch_load_regs (CPUState *env); +void kvm_arch_load_regs (CPUState *env,int level);void kvm_arch_load_mpstate (CPUState *env);void kvm_arch_save_mpstate (CPUState *env);int kvm_arch_has_work (CPUState *env);
savevm.c +4 -0
@@ -1368,6 +1368,8 @@ int qemu_savevm_state_complete(Monitor *mon, QEMUFile *f) {SaveStateEntry *se; +cpu_synchronize_all_states (); + QTAILQ_FOREACH(se, &savevm_handlers, entry) {if (se->save_live_state ==NULL )continue ;
@@ -1568,6 +1570,8 @@ int qemu_loadvm_state(QEMUFile *f) } } +cpu_synchronize_all_post_init (); + ret =0 ; out:
sysemu.h +4 -0
@@ -59,6 +59,10 @@ int load_vmstate(Monitor *mon, const char *name);void do_delvm (Monitor *mon,const QDict *qdict);void do_info_snapshots (Monitor *mon); +void cpu_synchronize_all_states (void ); +void cpu_synchronize_all_post_reset (void ); +void cpu_synchronize_all_post_init (void ); +void qemu_announce_self (void );void main_loop_wait (int timeout);
target-i386/kvm.c +1 -1
@@ -857,7 +857,7 @@ int kvm_get_vcpu_events(CPUState *env) }#ifdef KVM_UPSTREAM -int kvm_arch_put_registers (CPUState *env) +int kvm_arch_put_registers (CPUState *env,int level) {int ret;
target-i386/machine.c +0 -10
@@ -322,7 +322,6 @@ static void cpu_pre_save(void *opaque)CPUState *env = opaque;int i; -cpu_synchronize_state (env);if (kvm_enabled ()) {kvm_save_mpstate (env);kvm_get_vcpu_events (env);
@@ -342,14 +341,6 @@ static void cpu_pre_save(void *opaque)#endif } -static int cpu_pre_load (void *opaque) -{ -CPUState *env = opaque; - -cpu_synchronize_state (env); -return 0 ; -} -static int cpu_post_load (void *opaque,int version_id) {CPUState *env = opaque;
@@ -389,7 +380,6 @@ static const VMStateDescription vmstate_cpu = { .minimum_version_id = 3, .minimum_version_id_old = 3, .pre_save = cpu_pre_save, - .pre_load = cpu_pre_load, .post_load = cpu_post_load, .fields = (VMStateField []) {VMSTATE_UINTTL_ARRAY (regs, CPUState,CPU_NB_REGS ),
target-ia64/machine.c +0 -2
@@ -9,7 +9,6 @@ void cpu_save(QEMUFile *f, void *opaque)CPUState *env = opaque;if (kvm_enabled ()) { -kvm_arch_save_regs (env);kvm_arch_save_mpstate (env); } }
@@ -19,7 +18,6 @@ int cpu_load(QEMUFile *f, void *opaque, int version_id)CPUState *env = opaque;if (kvm_enabled ()) { -kvm_arch_load_regs (env);kvm_arch_load_mpstate (env); }return 0 ;
target-ppc/kvm.c +1 -1
@@ -57,7 +57,7 @@ void kvm_arch_reset_vcpu(CPUState *env) { } -int kvm_arch_put_registers (CPUState *env) +int kvm_arch_put_registers (CPUState *env,int level) {struct kvm_regs regs;int ret;
target-ppc/machine.c +0 -4
@@ -8,8 +8,6 @@ void cpu_save(QEMUFile *f, void *opaque)CPUState *env = (CPUState *)opaque;unsigned int i, j; -cpu_synchronize_state (env); -for (i =0 ; i <32 ; i++)qemu_put_betls (f, &env->gpr[i]);#if !defined(TARGET_PPC64 )
@@ -97,8 +95,6 @@ int cpu_load(QEMUFile *f, void *opaque, int version_id)CPUState *env = (CPUState *)opaque;unsigned int i, j; -cpu_synchronize_state (env); -for (i =0 ; i <32 ; i++)qemu_get_betls (f, &env->gpr[i]);#if !defined(TARGET_PPC64 )
target-s390x/kvm.c +1 -2
@@ -91,7 +91,7 @@ void kvm_arch_reset_vcpu(CPUState *env)/* FIXME: add code to reset vcpu. */ } -int kvm_arch_put_registers (CPUState *env) +int kvm_arch_put_registers (CPUState *env,int level) {struct kvm_regs regs;int ret;
@@ -296,7 +296,6 @@ static int handle_hypercall(CPUState *env, struct kvm_run *run)cpu_synchronize_state (env); r =s390_virtio_hypercall (env); -kvm_arch_put_registers (env);return r; }
vl.c +29 -0
@@ -3047,6 +3047,33 @@ static void nographic_update(void *opaque)qemu_mod_timer (nographic_timer, interval +qemu_get_clock (rt_clock)); } +void cpu_synchronize_all_states (void ) +{ +CPUState *cpu; + +for (cpu = first_cpu; cpu; cpu = cpu->next_cpu) { +cpu_synchronize_state (cpu); + } +} + +void cpu_synchronize_all_post_reset (void ) +{ +CPUState *cpu; + +for (cpu = first_cpu; cpu; cpu = cpu->next_cpu) { +cpu_synchronize_post_reset (cpu); + } +} + +void cpu_synchronize_all_post_init (void ) +{ +CPUState *cpu; + +for (cpu = first_cpu; cpu; cpu = cpu->next_cpu) { +cpu_synchronize_post_init (cpu); + } +} +struct vm_change_state_entry {VMChangeStateHandler *cb;void *opaque;
@@ -3195,6 +3222,7 @@ void qemu_system_reset(void) QTAILQ_FOREACH_SAFE(re, &reset_handlers, entry, nre) { re->func (re->opaque); } +cpu_synchronize_all_post_reset (); }void qemu_system_reset_request (void )
@@ -5956,6 +5984,7 @@ int main(int argc, char **argv, char **envp) machine->init (ram_size, boot_devices, kernel_filename, kernel_cmdline, initrd_filename, cpu_model); +cpu_synchronize_all_post_init ();#ifndef _WIN32/* must be after terminal init, SDL library changes signal handlers */
Signature · 1 line
1.6.0.2
Re: [PATCH v2 14/21] qemu-kvm: Rework VCPU state writeback API
Quoted 11 lines · Jan Kiszka
This grand cleanup drops all reset and vmsave/load related synchronization points in favor of four(!) generic hooks:
-
cpu_synchronize_all_statesinqemu_savevm_state_complete(initial sync from kernel before vmsave)
-cpu_synchronize_all_post_initinqemu_loadvm_state(writeback after vmload)
-cpu_synchronize_all_post_initin main after machine init
-cpu_synchronize_all_post_resetinqemu_system_reset(writeback after system reset)
Quoted 5 lines
These writeback points + the existing one of VCPU exec after
cpu_synchronize_statemap on three levels of writeback:-
KVM_PUT_ASYNC_STATE(during runtime, other VCPUs continue to run)
Wouldn't that be SYNC_STATE (state that is modified by the current vcpu only)?
Quoted 10 lines
-
KVM_PUT_RESET_STATE(on synchronous system reset, all VCPUs stopped)
-KVM_PUT_FULL_STATE(on init or vmload, all VCPUs stopped as well)This level is passed to the arch-specific VCPU state writing function that will decide which concrete substates need to be written. That way, no writer of load, save or reset functions that interact with in-kernel KVM states will ever have to worry about synchronization again. That also means that a lot of reasons for races, segfaults and deadlocks are eliminated.
Quoted 10 lines
cpu_synchronize_stateremains untouched, just as Anthony suggested. We continue to need it before reading or writing of VCPU states that are also tracked by in-kernel KVM subsystems.Consequently, this patch removes many
cpu_synchronize_statecalls that are now redundant, just like remaining explicit register syncs. It does not touch qemu-kvm's special hooks for mpstate,vcpu_events, or tsc loading. They will be cleaned up by individual patches.
I'm uneasy about this. What are the rules for putting cpu_synchronize_state() now?
Signature · 1 line
error compiling committee.c: too many arguments to function
Re: [PATCH v2 14/21] qemu-kvm: Rework VCPU state writeback API
Quoted 12 lines · Avi Kivity
Quoted 11 lines · Jan Kiszka
This grand cleanup drops all reset and vmsave/load related synchronization points in favor of four(!) generic hooks:
-
cpu_synchronize_all_statesinqemu_savevm_state_complete(initial sync from kernel before vmsave)
-cpu_synchronize_all_post_initinqemu_loadvm_state(writeback after vmload)
-cpu_synchronize_all_post_initin main after machine init
-cpu_synchronize_all_post_resetinqemu_system_reset(writeback after system reset)
Quoted 8 lines
Quoted 5 lines
These writeback points + the existing one of VCPU exec after
cpu_synchronize_statemap on three levels of writeback:-
KVM_PUT_ASYNC_STATE(during runtime, other VCPUs continue to run)Wouldn't that be
SYNC_STATE(state that is modified by the current vcpu only)?
It's async /wrt other VCPUs. They continue to run and may interact with this VCPU while updating its state.
Quoted 22 lines
Quoted 11 lines
-
KVM_PUT_RESET_STATE(on synchronous system reset, all VCPUs stopped)
-KVM_PUT_FULL_STATE(on init or vmload, all VCPUs stopped as well)This level is passed to the arch-specific VCPU state writing function that will decide which concrete substates need to be written. That way, no writer of load, save or reset functions that interact with in-kernel KVM states will ever have to worry about synchronization again. That also means that a lot of reasons for races, segfaults and deadlocks are eliminated.
Quoted 10 lines
cpu_synchronize_stateremains untouched, just as Anthony suggested. We continue to need it before reading or writing of VCPU states that are also tracked by in-kernel KVM subsystems.Consequently, this patch removes many
cpu_synchronize_statecalls that are now redundant, just like remaining explicit register syncs. It does not touch qemu-kvm's special hooks for mpstate,vcpu_events, or tsc loading. They will be cleaned up by individual patches.
Quoted 2 lines
I'm uneasy about this. What are the rules for putting
cpu_synchronize_state()now?
As before for code that accesses the state during runtime: Before you read or write some bit of it, call cpu_synchronize_state().
Only reset and save/restore handlers do not have to worry about synchronization anymore. It makes no sense to overload them with arch-specific KVM knowledge about what shall be written and when.
Jan
Re: [PATCH v2 14/21] qemu-kvm: Rework VCPU state writeback API
Quoted 18 lines · Jan Kiszka
Quoted 16 lines · Avi Kivity
Quoted 14 lines · Jan Kiszka
This grand cleanup drops all reset and vmsave/load related synchronization points in favor of four(!) generic hooks:
-
cpu_synchronize_all_statesinqemu_savevm_state_complete(initial sync from kernel before vmsave)
-cpu_synchronize_all_post_initinqemu_loadvm_state(writeback after vmload)
-cpu_synchronize_all_post_initin main after machine init
-cpu_synchronize_all_post_resetinqemu_system_reset(writeback after system reset)These writeback points + the existing one of VCPU exec after
cpu_synchronize_statemap on three levels of writeback:
Quoted 9 lines
Quoted 6 lines
Quoted 3 lines
-
KVM_PUT_ASYNC_STATE(during runtime, other VCPUs continue to run)Wouldn't that be
SYNC_STATE(state that is modified by the current vcpu only)?It's async /wrt other VCPUs. They continue to run and may interact with this VCPU while updating its state.
Well, to me it makes more sense to name them from the point of view of the vcpu that is doing the update.
Quoted 10 lines
Quoted 3 lines
I'm uneasy about this. What are the rules for putting
cpu_synchronize_state()now?As before for code that accesses the state during runtime: Before you read or write some bit of it, call
cpu_synchronize_state().Only reset and save/restore handlers do not have to worry about synchronization anymore. It makes no sense to overload them with arch-specific KVM knowledge about what shall be written and when.
OK.
Signature · 1 line
error compiling committee.c: too many arguments to function
Re: [PATCH v2 14/21] qemu-kvm: Rework VCPU state writeback API
Quoted 25 lines · Avi Kivity
Quoted 24 lines · Jan Kiszka
Quoted 22 lines · Avi Kivity
Quoted 11 lines · Jan Kiszka
This grand cleanup drops all reset and vmsave/load related synchronization points in favor of four(!) generic hooks:
-
cpu_synchronize_all_statesinqemu_savevm_state_complete(initial sync from kernel before vmsave)
-cpu_synchronize_all_post_initinqemu_loadvm_state(writeback after vmload)
-cpu_synchronize_all_post_initin main after machine init
-cpu_synchronize_all_post_resetinqemu_system_reset(writeback after system reset)Quoted 6 lines
These writeback points + the existing one of VCPU exec after
cpu_synchronize_statemap on three levels of writeback:-
KVM_PUT_ASYNC_STATE(during runtime, other VCPUs continue to run)Wouldn't that be
SYNC_STATE(state that is modified by the current vcpu only)?
Quoted 6 lines
Quoted 3 lines
It's async /wrt other VCPUs. They continue to run and may interact with this VCPU while updating its state.
Well, to me it makes more sense to name them from the point of view of the vcpu that is doing the update.
I'm open for a better name - except for "sync" as writebacks are always synchronous from the POV of the modified VCPU. Is KVM_PUT_RUNTIME_STATE clearer?
Jan
Re: [PATCH v2 14/21] qemu-kvm: Rework VCPU state writeback API
Quoted 6 lines · Jan Kiszka
Quoted 4 lines
Well, to me it makes more sense to name them from the point of view of the vcpu that is doing the update.
I'm open for a better name - except for "sync" as writebacks are always synchronous from the POV of the modified VCPU.
I meant that the state itself is synchronous from the POV of the vcpu. That is, rax can only be changed by the core, but INIT state may be changed by the other cores.
Quoted 3 lines
Is
KVM_PUT_RUNTIME_STATE
clearer?
They're all runtime. I guess this issue will always be confusing, so best to slap a comment on top of the define to explain it.
Signature · 1 line
error compiling committee.c: too many arguments to function
PATCH v2 15/21 qemu-kvm: Clean up mpstate synchronization
In reply to Jan KiszkaPush mpstate reading/writing into kvm_arch_load/save_regs and, on x86, properly synchronize with halted in the accessor functions.
--- hw/apic.c | 7 --- qemu-kvm-ia64.c | 4 +- qemu-kvm-x86.c | 102 ++++++++++++++++++++++++++++++------------------- qemu-kvm.c | 30 -------------- qemu-kvm.h | 15 ------- target-i386/machine.c | 6 --- target-ia64/machine.c | 3 + 7 files changed, 69 insertions(+), 98 deletions(-)
hw/apic.c +0 -7
@@ -507,13 +507,6 @@ void apic_init_reset(CPUState *env) s->wait_for_sipi =1 ; env->halted = !(s->apicbase &MSR_IA32_APICBASE_BSP ); -#ifdef KVM_CAP_MP_STATE -if (kvm_enabled () &&kvm_irqchip_in_kernel ()) { - env->mp_state - = env->halted ?KVM_MP_STATE_UNINITIALIZED :KVM_MP_STATE_RUNNABLE ; -kvm_load_mpstate (env); - } -#endif }static void apic_startup (APICState *s,int vector_num)
qemu-kvm-ia64.c +3 -1
@@ -124,7 +124,9 @@ void kvm_arch_cpu_reset(CPUState *env) {if (kvm_irqchip_in_kernel (kvm_context)) {#ifdef KVM_CAP_MP_STATE -kvm_reset_mpstate (env->kvm_cpu_state.vcpu_ctx); +struct kvm_mp_state mp_state = {.mp_state =KVM_MP_STATE_UNINITIALIZED + }; +kvm_set_mpstate (env, &mp_state);#endif }else { env->interrupt_request &= ~CPU_INTERRUPT_HARD ;
qemu-kvm-x86.c +63 -39
132 lines not on this page · raw message
qemu-kvm.c +0 -30
@@ -1579,36 +1579,6 @@ void kvm_update_interrupt_request(CPUState *env) } } -static void kvm_do_load_mpstate (void *_env) -{ -CPUState *env = _env; - -kvm_arch_load_mpstate (env); -} - -void kvm_load_mpstate (CPUState *env) -{ -if (kvm_enabled () && qemu_system_ready &&kvm_vcpu_inited (env)) -on_vcpu (env, kvm_do_load_mpstate, env); -} - -static void kvm_do_save_mpstate (void *_env) -{ -CPUState *env = _env; - -kvm_arch_save_mpstate (env); -#ifdef KVM_CAP_MP_STATE -if (kvm_irqchip_in_kernel ()) - env->halted = (env->mp_state ==KVM_MP_STATE_HALTED ); -#endif -} - -void kvm_save_mpstate (CPUState *env) -{ -if (kvm_enabled ()) -on_vcpu (env, kvm_do_save_mpstate, env); -} -int kvm_cpu_exec (CPUState *env) {int r;
qemu-kvm.h +0 -15
@@ -299,16 +299,6 @@ int kvm_get_mpstate(CPUState *env, struct kvm_mp_state *mp_state);* */ int kvm_set_mpstate (CPUState *env,struct kvm_mp_state *mp_state); -/*! -* * \brief Reset VCPU MP state -* -*/ -static inline int kvm_reset_mpstate (CPUState *env) -{ -struct kvm_mp_state mp_state = {.mp_state =KVM_MP_STATE_UNINITIALIZED - }; -return kvm_set_mpstate (env, &mp_state); -}#endif /*!
@@ -874,8 +864,6 @@ static inline void kvm_inject_x86_mce(CPUState *cenv, int bank,int kvm_main_loop (void );int kvm_init_ap (void );int kvm_vcpu_inited (CPUState *env); -void kvm_load_mpstate (CPUState *env); -void kvm_save_mpstate (CPUState *env);void kvm_apic_init (CPUState *env);/* called from vcpu initialization */ void qemu_kvm_load_lapic (CPUState *env);
@@ -909,8 +897,6 @@ int kvm_arch_qemu_create_context(void);void kvm_arch_save_regs (CPUState *env);void kvm_arch_load_regs (CPUState *env,int level); -void kvm_arch_load_mpstate (CPUState *env); -void kvm_arch_save_mpstate (CPUState *env);int kvm_arch_has_work (CPUState *env);void kvm_arch_process_irqchip_events (CPUState *env);int kvm_arch_try_push_interrupts (void *opaque);
@@ -979,7 +965,6 @@ void kvm_load_tsc(CPUState *env);#ifdef TARGET_I386 #define qemu_kvm_has_pit_state2 () (0)#endif -#define kvm_save_mpstate (env) do {} while(0)#define qemu_kvm_cpu_stop (env) do {} while(0)static inline void kvm_load_tsc (CPUState *env) {
target-i386/machine.c +0 -6
@@ -323,7 +323,6 @@ static void cpu_pre_save(void *opaque)int i;if (kvm_enabled ()) { -kvm_save_mpstate (env);kvm_get_vcpu_events (env); }
@@ -362,12 +361,7 @@ static int cpu_post_load(void *opaque, int version_id)tlb_flush (env,1 );if (kvm_enabled ()) { -/* when in-kernel irqchip is used, env->halted causes deadlock -because no userspace IRQs will ever clear this flag */ - env->halted =0 ; -kvm_load_tsc (env); -kvm_load_mpstate (env);kvm_put_vcpu_events (env); }
target-ia64/machine.c +3 -0
@@ -4,6 +4,9 @@#include "exec-all.h" #include "qemu-kvm.h" +void kvm_arch_save_mpstate (CPUState *env); +void kvm_arch_load_mpstate (CPUState *env); +void cpu_save (QEMUFile *f,void *opaque) {CPUState *env = opaque;
Signature · 1 line
1.6.0.2
PATCH v2 16/21 KVM: x86: Restrict writeback of VCPU state
In reply to Jan KiszkaDo not write nmi_pending, sipi_vector, and mpstate unless we at least go through a reset. And TSC as well as KVM wallclocks should only be written on full sync, otherwise we risk to drop some time on during state read-modify-write.
--- kvm.h | 2 +- qemu-kvm-x86.c | 2 +- target-i386/kvm.c | 32 ++++++++++++++++++++------------ target-i386/machine.c | 2 +- 4 files changed, 23 insertions(+), 15 deletions(-)
kvm.h +1 -1
@@ -53,7 +53,7 @@ int kvm_set_migration_log(int enable);int kvm_has_sync_mmu (void );int kvm_has_vcpu_events (void ); -int kvm_put_vcpu_events (CPUState *env); +int kvm_put_vcpu_events (CPUState *env,int level);int kvm_get_vcpu_events (CPUState *env);void kvm_setup_guest_memory (void *start,size_t size);
qemu-kvm-x86.c +1 -1
@@ -1398,7 +1398,7 @@ void kvm_arch_push_nmi(void *opaque)void kvm_arch_cpu_reset (CPUState *env) {kvm_arch_reset_vcpu (env); -kvm_put_vcpu_events (env); +kvm_put_vcpu_events (env,KVM_PUT_RESET_STATE );kvm_reset_mpstate (env);if (!cpu_is_bsp (env) && !kvm_irqchip_in_kernel ()) { env->interrupt_request &= ~CPU_INTERRUPT_HARD ;
target-i386/kvm.c +20 -12
@@ -544,7 +544,7 @@ static void kvm_msr_entry_set(struct kvm_msr_entry *entry, entry->data = value; } -static int kvm_put_msrs (CPUState *env) +static int kvm_put_msrs (CPUState *env,int level) {struct {struct kvm_msrs info;
@@ -558,7 +558,6 @@ static int kvm_put_msrs(CPUState *env)kvm_msr_entry_set (&msrs[n++],MSR_IA32_SYSENTER_EIP , env->sysenter_eip);if (kvm_has_msr_star (env))kvm_msr_entry_set (&msrs[n++],MSR_STAR , env->star); -kvm_msr_entry_set (&msrs[n++],MSR_IA32_TSC , env->tsc);kvm_msr_entry_set (&msrs[n++],MSR_VM_HSAVE_PA , env->vm_hsave);#ifdef TARGET_X86_64 /* FIXME if lm capable */
@@ -567,8 +566,12 @@ static int kvm_put_msrs(CPUState *env)kvm_msr_entry_set (&msrs[n++],MSR_FMASK , env->fmask);kvm_msr_entry_set (&msrs[n++],MSR_LSTAR , env->lstar);#endif -kvm_msr_entry_set (&msrs[n++],MSR_KVM_SYSTEM_TIME , env->system_time_msr); -kvm_msr_entry_set (&msrs[n++],MSR_KVM_WALL_CLOCK , env->wall_clock_msr); +if (level ==KVM_PUT_FULL_STATE ) { +kvm_msr_entry_set (&msrs[n++],MSR_IA32_TSC , env->tsc); +kvm_msr_entry_set (&msrs[n++],MSR_KVM_SYSTEM_TIME , + env->system_time_msr); +kvm_msr_entry_set (&msrs[n++],MSR_KVM_WALL_CLOCK , env->wall_clock_msr); + } msr_data.info.nmsrs = n;
@@ -786,7 +789,7 @@ static int kvm_get_mp_state(CPUState *env) }#endif -int kvm_put_vcpu_events (CPUState *env) +int kvm_put_vcpu_events (CPUState *env,int level) {#ifdef KVM_CAP_VCPU_EVENTS struct kvm_vcpu_events events;
@@ -810,8 +813,11 @@ int kvm_put_vcpu_events(CPUState *env) events.sipi_vector = env->sipi_vector; - events.flags = -KVM_VCPUEVENT_VALID_NMI_PENDING |KVM_VCPUEVENT_VALID_SIPI_VECTOR ; + events.flags =0 ; +if (level >=KVM_PUT_RESET_STATE ) { + events.flags |= +KVM_VCPUEVENT_VALID_NMI_PENDING |KVM_VCPUEVENT_VALID_SIPI_VECTOR ; + }return kvm_vcpu_ioctl (env,KVM_SET_VCPU_EVENTS , &events);#else
@@ -882,15 +888,17 @@ int kvm_arch_put_registers(CPUState *env, int level)if (ret <0 )return ret; - ret =kvm_put_msrs (env); + ret =kvm_put_msrs (env, level);if (ret <0 )return ret; - ret =kvm_put_mp_state (env); -if (ret <0 ) -return ret; +if (level >=KVM_PUT_RESET_STATE ) { + ret =kvm_put_mp_state (env); +if (ret <0 ) +return ret; + } - ret =kvm_put_vcpu_events (env); + ret =kvm_put_vcpu_events (env, level);if (ret <0 )return ret;
target-i386/machine.c +1 -1
@@ -362,7 +362,7 @@ static int cpu_post_load(void *opaque, int version_id)if (kvm_enabled ()) {kvm_load_tsc (env); -kvm_put_vcpu_events (env); +kvm_put_vcpu_events (env,KVM_PUT_FULL_STATE ); }return 0 ;
Signature · 1 line
1.6.0.2
PATCH v2 17/21 qemu-kvm: Use VCPU event state for reset and vmsave/load
In reply to Jan KiszkaPush reading/writing of vcpu_events into kvm_arch_load/save_regs to avoid KVM-specific hooks in generic code.
--- kvm.h | 2 -- qemu-kvm-x86.c | 6 ++++-- target-i386/kvm.c | 4 ++-- target-i386/machine.c | 6 ------ 4 files changed, 6 insertions(+), 12 deletions(-)
kvm.h +0 -2
@@ -53,8 +53,6 @@ int kvm_set_migration_log(int enable);int kvm_has_sync_mmu (void );int kvm_has_vcpu_events (void ); -int kvm_put_vcpu_events (CPUState *env,int level); -int kvm_get_vcpu_events (CPUState *env);void kvm_setup_guest_memory (void *start,size_t size);
qemu-kvm-x86.c +4 -2
@@ -986,6 +986,8 @@ void kvm_arch_load_regs(CPUState *env, int level)/* Avoid deadlock: no user space IRQ will ever clear it. */ env->halted =0 ; } + +kvm_put_vcpu_events (env, level); }void kvm_load_tsc (CPUState *env)
@@ -1157,6 +1159,7 @@ void kvm_arch_save_regs(CPUState *env) } }kvm_arch_save_mpstate (env); +kvm_get_vcpu_events (env); } static void do_cpuid_ent(struct kvm_cpuid_entry2 *e, uint32_t function,
@@ -1231,7 +1234,7 @@ int kvm_arch_init_vcpu(CPUState *cenv)qemu_kvm_load_lapic (cenv); - cenv->interrupt_injected =-1 ; +kvm_arch_reset_vcpu (cenv);#ifdef KVM_CPUID_SIGNATURE /* Paravirtualization CPUIDs */
@@ -1398,7 +1401,6 @@ void kvm_arch_push_nmi(void *opaque)void kvm_arch_cpu_reset (CPUState *env) {kvm_arch_reset_vcpu (env); -kvm_put_vcpu_events (env,KVM_PUT_RESET_STATE );kvm_reset_mpstate (env);if (!cpu_is_bsp (env) && !kvm_irqchip_in_kernel ()) { env->interrupt_request &= ~CPU_INTERRUPT_HARD ;
target-i386/kvm.c +2 -2
@@ -789,7 +789,7 @@ static int kvm_get_mp_state(CPUState *env) }#endif -int kvm_put_vcpu_events (CPUState *env,int level) +static int kvm_put_vcpu_events (CPUState *env,int level) {#ifdef KVM_CAP_VCPU_EVENTS struct kvm_vcpu_events events;
@@ -825,7 +825,7 @@ int kvm_put_vcpu_events(CPUState *env, int level)#endif } -int kvm_get_vcpu_events (CPUState *env) +static int kvm_get_vcpu_events (CPUState *env) {#ifdef KVM_CAP_VCPU_EVENTS struct kvm_vcpu_events events;
target-i386/machine.c +0 -6
@@ -5,7 +5,6 @@#include "exec-all.h" #include "kvm.h" -#include "qemu-kvm.h" static const VMStateDescription vmstate_segment = { .name ="segment" ,
@@ -322,10 +321,6 @@ static void cpu_pre_save(void *opaque)CPUState *env = opaque;int i; -if (kvm_enabled ()) { -kvm_get_vcpu_events (env); - } -/* FPU */ env->fpus_vmstate = (env->fpus & ~0x3800 ) | (env->fpstt &0x7 ) <<11 ; env->fptag_vmstate =0 ;
@@ -362,7 +357,6 @@ static int cpu_post_load(void *opaque, int version_id)if (kvm_enabled ()) {kvm_load_tsc (env); -kvm_put_vcpu_events (env,KVM_PUT_FULL_STATE ); }return 0 ;
Signature · 1 line
1.6.0.2
PATCH v2 18/21 qemu-kvm: Cleanup/fix TSC and PV clock writeback
In reply to Jan KiszkaDrop kvm_load_tsc in favor of level-dependent writeback in kvm_arch_load_regs. KVM's PV clock MSRs fall in the same category and should therefore only be written back on full sync.
--- qemu-kvm-x86.c | 19 +++++-------------- qemu-kvm.h | 4 ---- target-i386/machine.c | 5 ----- 3 files changed, 5 insertions(+), 23 deletions(-)
qemu-kvm-x86.c +5 -14
@@ -972,8 +972,11 @@ void kvm_arch_load_regs(CPUState *env, int level) set_msr_entry(&msrs[n++], MSR_LSTAR , env->lstar); }#endif -set_msr_entry (&msrs[n++],MSR_KVM_SYSTEM_TIME , env->system_time_msr); -set_msr_entry (&msrs[n++],MSR_KVM_WALL_CLOCK , env->wall_clock_msr); +if (level ==KVM_PUT_FULL_STATE ) { +set_msr_entry (&msrs[n++],MSR_IA32_TSC , env->tsc); +set_msr_entry (&msrs[n++],MSR_KVM_SYSTEM_TIME , env->system_time_msr); +set_msr_entry (&msrs[n++],MSR_KVM_WALL_CLOCK , env->wall_clock_msr); + } rc =kvm_set_msrs (env, msrs, n);if (rc ==-1 )
@@ -990,18 +993,6 @@ void kvm_arch_load_regs(CPUState *env, int level)kvm_put_vcpu_events (env, level); } -void kvm_load_tsc (CPUState *env) -{ -int rc; -struct kvm_msr_entry msr; - -set_msr_entry (&msr,MSR_IA32_TSC , env->tsc); - - rc =kvm_set_msrs (env, &msr,1 ); -if (rc ==-1 ) -perror ("kvm_set_tsc FAILED.\n" ); -} -void kvm_arch_save_regs (CPUState *env) {struct kvm_regs regs;
qemu-kvm.h +0 -4
@@ -958,7 +958,6 @@ int handle_tpr_access(void *opaque, CPUState *env, uint64_t rip,#ifdef TARGET_I386 #define qemu_kvm_has_pit_state2 () kvm_has_pit_state2(kvm_context)#endif -void kvm_load_tsc (CPUState *env);#else #define kvm_nested 0#define qemu_kvm_has_gsi_routing () (0)
@@ -966,9 +965,6 @@ void kvm_load_tsc(CPUState *env);#define qemu_kvm_has_pit_state2 () (0)#endif #define qemu_kvm_cpu_stop (env) do {} while(0) -static inline void kvm_load_tsc (CPUState *env) -{ -}#endif void kvm_mutex_unlock (void );
target-i386/machine.c +0 -5
@@ -354,11 +354,6 @@ static int cpu_post_load(void *opaque, int version_id)hw_breakpoint_insert (env, i);tlb_flush (env,1 ); - -if (kvm_enabled ()) { -kvm_load_tsc (env); - } -return 0 ; }
Signature · 1 line
1.6.0.2
PATCH v2 19/21 qemu-kvm: Clean up KVM's APIC hooks
In reply to Jan KiszkaThe APIC is part of the VCPU state, so trigger its readout and writeback from kvm_arch_save/load_regs. Thanks to the transparent sync on reset and vmsave/load, we can also drop explicit sync code, reducing the diff to upstream.
--- hw/apic.c | 37 +++++-------------------------------- qemu-kvm-x86.c | 4 ++-- qemu-kvm.h | 5 ++--- 3 files changed, 9 insertions(+), 37 deletions(-)
hw/apic.c +5 -32
@@ -24,8 +24,6 @@#include "host-utils.h" #include "kvm.h" -#include "qemu-kvm.h" -//#define DEBUG_APIC /* APIC Local Vector Table */
@@ -951,36 +949,22 @@ static void kvm_kernel_lapic_load_from_user(APICState *s)#endif -void qemu_kvm_load_lapic (CPUState *env) +void kvm_load_lapic (CPUState *env) {#ifdef KVM_CAP_IRQCHIP -if (kvm_enabled () &&kvm_vcpu_inited (env) &&kvm_irqchip_in_kernel ()) { -kvm_kernel_lapic_load_from_user (env->apic_state); - } -#endif -} - -static void apic_pre_save (void *opaque) -{ -#ifdef KVM_CAP_IRQCHIP -APICState *s = (void *)opaque; -if (kvm_enabled () &&kvm_irqchip_in_kernel ()) { -kvm_kernel_lapic_save_to_user (s); +kvm_kernel_lapic_load_from_user (env->apic_state); }#endif } -static int apic_post_load (void *opaque,int version_id) +void kvm_save_lapic (CPUState *env) {#ifdef KVM_CAP_IRQCHIP -APICState *s = opaque; -if (kvm_enabled () &&kvm_irqchip_in_kernel ()) { -kvm_kernel_lapic_load_from_user (s); +kvm_kernel_lapic_save_to_user (env->apic_state); }#endif -return 0 ; }/* This function is only used for old state version 1 and 2 */
@@ -1019,9 +1003,6 @@ static int apic_load_old(QEMUFile *f, void *opaque, int version_id)if (version_id >=2 )qemu_get_timer (f, s->timer); - -qemu_kvm_load_lapic (s->cpu_env); -return 0 ; }
@@ -1052,9 +1033,7 @@ static const VMStateDescription vmstate_apic = {VMSTATE_INT64 (next_time, APICState),VMSTATE_TIMER (timer, APICState), VMSTATE_END_OF_LIST() - }, - .pre_save = apic_pre_save, - .post_load = apic_post_load, + } };static void apic_reset (void *opaque)
@@ -1077,7 +1056,6 @@ static void apic_reset(void *opaque)*/ s->lvt[APIC_LVT_LINT0 ] =0x700 ; } -qemu_kvm_load_lapic (s->cpu_env); }static CPUReadMemoryFunc *const apic_mem_read[3 ] = {
@@ -1121,11 +1099,6 @@ int apic_init(CPUState *env)vmstate_register (s->idx, &vmstate_apic, s);qemu_register_reset (apic_reset, s); -/* apic_reset must be called before the vcpu threads are initialized and load -* registers, in qemu-kvm. -*/ -apic_reset (s); - local_apics[s->idx] = s;return 0 ; }
qemu-kvm-x86.c +2 -2
@@ -984,6 +984,7 @@ void kvm_arch_load_regs(CPUState *env, int level)if (level >=KVM_PUT_RESET_STATE ) {kvm_arch_load_mpstate (env); +kvm_load_lapic (env); }if (kvm_irqchip_in_kernel ()) {/* Avoid deadlock: no user space IRQ will ever clear it. */
@@ -1150,6 +1151,7 @@ void kvm_arch_save_regs(CPUState *env) } }kvm_arch_save_mpstate (env); +kvm_save_lapic (env);kvm_get_vcpu_events (env); }
@@ -1223,8 +1225,6 @@ int kvm_arch_init_vcpu(CPUState *cenv)CPUState copy;uint32_t i, j, limit; -qemu_kvm_load_lapic (cenv); -kvm_arch_reset_vcpu (cenv);#ifdef KVM_CPUID_SIGNATURE
qemu-kvm.h +2 -3
@@ -864,9 +864,8 @@ static inline void kvm_inject_x86_mce(CPUState *cenv, int bank,int kvm_main_loop (void );int kvm_init_ap (void );int kvm_vcpu_inited (CPUState *env); -void kvm_apic_init (CPUState *env); -/* called from vcpu initialization */ -void qemu_kvm_load_lapic (CPUState *env); +void kvm_save_lapic (CPUState *env); +void kvm_load_lapic (CPUState *env);void kvm_hpet_enable_kpit (void );void kvm_hpet_disable_kpit (void );
Signature · 1 line
1.6.0.2
PATCH v2 20/21 qemu-kvm: Move kvm_set_boot_cpu_id
In reply to Jan KiszkaSetting the boot CPU ID is arch-specific KVM stuff. So push it where it belongs to.
--- hw/pc.c | 3 --- qemu-kvm-x86.c | 3 ++- 2 files changed, 2 insertions(+), 4 deletions(-)
hw/pc.c +0 -3
@@ -803,9 +803,6 @@ static void pc_init1(ram_addr_t ram_size,#endif } -if (kvm_enabled ()) { -kvm_set_boot_cpu_id (0 ); - }for (i =0 ; i < smp_cpus; i++) { env =pc_new_cpu (cpu_model); }
qemu-kvm-x86.c +2 -1
@@ -695,7 +695,8 @@ int kvm_arch_qemu_create_context(void)if (kvm_check_extension (kvm_state,KVM_CAP_ADJUST_CLOCK ))vmstate_register (0 , &vmstate_kvmclock, &kvmclock_data);#endif -return 0 ; + +return kvm_set_boot_cpu_id (0 ); } static void set_msr_entry(struct kvm_msr_entry *entry, uint32_t index,
Signature · 1 line
1.6.0.2
PATCH v2 21/21 qemu-kvm: Bring qemu_init_vcpu back home
In reply to Jan KiszkaThere is no need for the this hack anymore, initialization is now robust against reordering as it doesn't try to write the VCPU state on its own.
--- hw/pc.c | 5 ----- target-i386/helper.c | 2 ++ 2 files changed, 2 insertions(+), 5 deletions(-)
hw/pc.c +0 -5
@@ -751,11 +751,6 @@ CPUState *pc_new_cpu(const char *cpu_model) }else {qemu_register_reset ((QEMUResetHandler *)cpu_reset, env); } - -/* kvm needs this to run after the apic is initialized. Otherwise, -* it can access invalid state and crash. -*/ -qemu_init_vcpu (env);return env; }
target-i386/helper.c +2 -0
@@ -1953,6 +1953,8 @@ CPUX86State *cpu_x86_init(const char *cpu_model) }mce_init (env); +qemu_init_vcpu (env); +return env; }
Signature · 1 line
1.6.0.2
Quoted 7 lines · Jan Kiszka
This version addresses the feedback on v2, namely:
-assert(<vm stopped> || <current thread == env->thread>)on low-level load/save registers
- fixed mpstate initializationYet untested is -no-kvm-irqchip with smp due to some bug in unpatched qemu-kvm or the kernel modules. Still investigating.
Don't recall it ever working properly. Guess that IPI emulation with signals is problematic.
Quoted 9 lines
Pull URL is still
git://git.kiszka.org/qemu-kvm.gitqueues/vcpu-statePS: The corresponding upstream queue is now available under queues/kvm-upstream in the same repository. Will send it out later if there are no further remarks on this series to avoid flooding the mailing lists. Early testers are nevertheless welcome.
Can the necessary changes for cleanups go in qemu.git first, so after they're merged in qemu-kvm, we can apply the unification patches.
And later do the same with synchronization rework (API looks fine to me).
Nice work!
Quoted 11 lines · Marcelo Tosatti
Quoted 7 lines · Jan Kiszka
This version addresses the feedback on v2, namely:
-assert(<vm stopped> || <current thread == env->thread>)on low-level load/save registers
- fixed mpstate initializationYet untested is -no-kvm-irqchip with smp due to some bug in unpatched qemu-kvm or the kernel modules. Still investigating.
Don't recall it ever working properly. Guess that IPI emulation with signals is problematic.
OK, then it only works by chance on some boxes. Specifically, forcing qemu onto a single host core "solves" the issue - suspicious.
Quoted 10 lines
Quoted 9 lines
Pull URL is still
git://git.kiszka.org/qemu-kvm.gitqueues/vcpu-statePS: The corresponding upstream queue is now available under queues/kvm-upstream in the same repository. Will send it out later if there are no further remarks on this series to avoid flooding the mailing lists. Early testers are nevertheless welcome.
Quoted 5 lines
Can the necessary changes for cleanups go in qemu.git first, so after they're merged in qemu-kvm, we can apply the unification patches.
And later do the same with synchronization rework (API looks fine to me).
Fine with me. Will send out my upstream queue - against uq/master or directly against qemu.git then?
Quoted 3 lines
Nice work!
Thanks,
Jan
Signature · 2 lines
Siemens AG, Corporate Technology, CT T DE IT 1 Corporate Competence Center Embedded Linux
Quoted 26 lines · Jan Kiszka
Quoted 11 lines · Marcelo Tosatti
Quoted 7 lines · Jan Kiszka
This version addresses the feedback on v2, namely:
-assert(<vm stopped> || <current thread == env->thread>)on low-level load/save registers
- fixed mpstate initializationYet untested is -no-kvm-irqchip with smp due to some bug in unpatched qemu-kvm or the kernel modules. Still investigating.
Don't recall it ever working properly. Guess that IPI emulation with signals is problematic.
OK, then it only works by chance on some boxes. Specifically, forcing qemu onto a single host core "solves" the issue - suspicious.
Quoted 10 lines
Quoted 9 lines
Pull URL is still
git://git.kiszka.org/qemu-kvm.gitqueues/vcpu-statePS: The corresponding upstream queue is now available under queues/kvm-upstream in the same repository. Will send it out later if there are no further remarks on this series to avoid flooding the mailing lists. Early testers are nevertheless welcome.
Quoted 8 lines
Quoted 5 lines
Can the necessary changes for cleanups go in qemu.git first, so after they're merged in qemu-kvm, we can apply the unification patches.
And later do the same with synchronization rework (API looks fine to me).
Fine with me. Will send out my upstream queue - against uq/master or directly against qemu.git then?
Please send only the cleanups (up to 0160a98b9415603f583aec49b21dbd98b46ac969) for qemu.git for now, since Avi/others might have comments on the synchronization changes.
Quoted 6 lines · Marcelo Tosatti
Quoted 5 lines · Jan Kiszka
This version addresses the feedback on v2, namely:
-assert(<vm stopped> || <current thread == env->thread>)on low-level load/save registers
- fixed mpstate initialization
Quoted 6 lines
Quoted 2 lines
Yet untested is -no-kvm-irqchip with smp due to some bug in unpatched qemu-kvm or the kernel modules. Still investigating.
Don't recall it ever working properly. Guess that IPI emulation with signals is problematic.
It did for me.
--
Gleb.
[Qemu-devel] Need QEMU source and documentation
Hi all,
I need QEMU source and documentation but qemu.org is down. Is there any other place from where I can get both source and documentation?
Regards,
Taimoor
_________________________________________________________________
Hotmail: Free, trusted and rich email service.
https://signup.live.com/signup.aspx?id=60969
Re: [Qemu-devel] Need QEMU source and documentation
2010/2/4 Taimoor Mirza <[email protected]>:
Quoted 6 lines
Hi all,
I need QEMU source and documentation but qemu.org is down. Is there any other place from where I can get both source and documentation?
You can get it from savannah.
http://savannah.nongnu.org/projects/qemu
Quoted 6 lines
Regards,
Taimoor________________________________
Hotmail: Free, trusted and rich email service. Get it now.
Re: [Qemu-devel] Need QEMU source and documentation
In reply to Taimoor MirzaQuoted 5 lines · Taimoor Mirza
Hi all,
I need QEMU source and documentation but qemu.org is down. Is there any other place from where I can get both source and documentation?
It's back again.
It wasn't actually down. qemu.org runs in a qemu virtual machine with kvm acceleration. In this case, I had top running on the serial port and had disconnected from the serial console.
top continuously generates output which filled the socket buffer and ultimately caused write() to block freezing the entire virtual machine until I reconnected to the serial console.
We desperately need to rewrite the char device handling.
Regards,
Anthony Liguori
Quoted 5 lines
Regards,
Taimoor------------------------------------------------------------------------
Hotmail: Free, trusted and rich email service. Get it now. <https://signup.live.com/signup.aspx?id=60969>
Quoted 5 lines · Jan Kiszka
This version addresses the feedback on v2, namely:
-assert(<vm stopped> || <current thread == env->thread>)on low-level load/save registers
- fixed mpstate initialization
With those patched, doing "info cpus" in monitor kill the guest, which, unfortunately, means that reworked VCPU state writeback is still not so easy to use correctly.
--
Gleb.
Quoted 10 lines · Gleb Natapov
Quoted 5 lines · Jan Kiszka
This version addresses the feedback on v2, namely:
-assert(<vm stopped> || <current thread == env->thread>)on low-level load/save registers
- fixed mpstate initializationWith those patched, doing "info cpus" in monitor kill the guest, which, unfortunately, means that reworked VCPU state writeback is still not so easy to use correctly.
Regression of patch 12 ("qemu-kvm: Use upstream kvm_vcpu_dirty"), will be fixed in v3.
Thanks,
Jan
---
qemu-kvm.c +1 -1
@@ -1532,13 +1532,13 @@ static void do_kvm_cpu_synchronize_state(void *_env)CPUState *env = _env;kvm_arch_save_regs (env); + env->kvm_vcpu_dirty =1 ; }void kvm_cpu_synchronize_state (CPUState *env) {if (!env->kvm_vcpu_dirty) {on_vcpu (env, do_kvm_cpu_synchronize_state, env); - env->kvm_vcpu_dirty =1 ; } }
Quoted 7 lines · Jan Kiszka
Quoted 6 lines · Gleb Natapov
Quoted 5 lines · Jan Kiszka
This version addresses the feedback on v2, namely:
-assert(<vm stopped> || <current thread == env->thread>)on low-level load/save registers
- fixed mpstate initialization
Quoted 8 lines
Quoted 4 lines
With those patched, doing "info cpus" in monitor kill the guest, which, unfortunately, means that reworked VCPU state writeback is still not so easy to use correctly.
Regression of patch 12 ("qemu-kvm: Use upstream
kvm_vcpu_dirty"), will be fixed in v3.
Looks better now indeed. Thanks.
--
Gleb.