On 2012-06-23 00:59, Anthony Liguori wrote:
> On 06/22/2012 05:45 PM, Jan Kiszka wrote:
>> This sketches a possible path to get rid of the iothread lock on vmexits
>> in KVM mode. On x86, the the in-kernel irqchips has to be used because
>> we otherwise need to synchronize APIC and other per-cpu state accesses
>> that could be changed concurrently. Not yet fully analyzed is the NMI
>> injection path in the absence of an APIC.
>>
>> s390x should be fine without specific locking as their pre/post-run
>> callbacks are empty. Power requires locking for the pre-run callback.
>>
>> This patch is untested, but a similar version was successfully used in
>> a x86 setup with a network I/O path that needed no central iothread
>> locking anymore (required special MMIO exit handling).
>> ---
>> kvm-all.c | 18 ++++++++++++++++--
>> target-i386/kvm.c | 7 +++++++
>> target-ppc/kvm.c | 4 ++++
>> 3 files changed, 27 insertions(+), 2 deletions(-)
>>
>> diff --git a/kvm-all.c b/kvm-all.c
>> index f8e4328..9c3e26f 100644
>> --- a/kvm-all.c
>> +++ b/kvm-all.c
>> @@ -1460,6 +1460,8 @@ int kvm_cpu_exec(CPUArchState *env)
>> return EXCP_HLT;
>> }
>>
>> + qemu_mutex_unlock_iothread();
>> +
>> do {
>> if (env->kvm_vcpu_dirty) {
>> kvm_arch_put_registers(env, KVM_PUT_RUNTIME_STATE);
>> @@ -1476,14 +1478,16 @@ int kvm_cpu_exec(CPUArchState *env)
>> */
>> qemu_cpu_kick_self();
>> }
>> - qemu_mutex_unlock_iothread();
>>
>> run_ret = kvm_vcpu_ioctl(env, KVM_RUN, 0);
>>
>> - qemu_mutex_lock_iothread();
>> kvm_arch_post_run(env, run);
>>
>> + /* TODO: push coalesced mmio flushing to the point where we
>> access
>> + * devices that are using it (currently VGA and E1000). */
>> + qemu_mutex_lock_iothread();
>> kvm_flush_coalesced_mmio_buffer();
>> + qemu_mutex_unlock_iothread();
>>
>> if (run_ret< 0) {
>> if (run_ret == -EINTR || run_ret == -EAGAIN) {
>> @@ -1499,19 +1503,23 @@ int kvm_cpu_exec(CPUArchState *env)
>> switch (run->exit_reason) {
>> case KVM_EXIT_IO:
>> DPRINTF("handle_io\n");
>> + qemu_mutex_lock_iothread();
>> kvm_handle_io(run->io.port,
>> (uint8_t *)run + run->io.data_offset,
>> run->io.direction,
>> run->io.size,
>> run->io.count);
>> + qemu_mutex_unlock_iothread();
>> ret = 0;
>> break;
>> case KVM_EXIT_MMIO:
>> DPRINTF("handle_mmio\n");
>> + qemu_mutex_lock_iothread();
>> cpu_physical_memory_rw(run->mmio.phys_addr,
>> run->mmio.data,
>> run->mmio.len,
>> run->mmio.is_write);
>> + qemu_mutex_unlock_iothread();
>> ret = 0;
>> break;
>> case KVM_EXIT_IRQ_WINDOW_OPEN:
>> @@ -1520,7 +1528,9 @@ int kvm_cpu_exec(CPUArchState *env)
>> break;
>> case KVM_EXIT_SHUTDOWN:
>> DPRINTF("shutdown\n");
>> + qemu_mutex_lock_iothread();
>> qemu_system_reset_request();
>> + qemu_mutex_unlock_iothread();
>> ret = EXCP_INTERRUPT;
>> break;
>> case KVM_EXIT_UNKNOWN:
>> @@ -1533,11 +1543,15 @@ int kvm_cpu_exec(CPUArchState *env)
>> break;
>> default:
>> DPRINTF("kvm_arch_handle_exit\n");
>> + qemu_mutex_lock_iothread();
>> ret = kvm_arch_handle_exit(env, run);
>> + qemu_mutex_unlock_iothread();
>> break;
>> }
>> } while (ret == 0);
>>
>> + qemu_mutex_lock_iothread();
>> +
>> if (ret< 0) {
>> cpu_dump_state(env, stderr, fprintf, CPU_DUMP_CODE);
>> vm_stop(RUN_STATE_INTERNAL_ERROR);
>> diff --git a/target-i386/kvm.c b/target-i386/kvm.c
>> index 0d0d8f6..0ad64d1 100644
>> --- a/target-i386/kvm.c
>> +++ b/target-i386/kvm.c
>> @@ -1631,7 +1631,10 @@ void kvm_arch_pre_run(CPUX86State *env, struct
>> kvm_run *run)
>>
>> /* Inject NMI */
>> if (env->interrupt_request& CPU_INTERRUPT_NMI) {
>
> Strictly speaking, wouldn't we need to use testbit() and setbit()? I
> would expect in the very least a barrier would be needed.I need to think about this as well. We ignored it so far, just saw it when hacking up this patch. > > Looks pretty nice overall. I'll need to apply and spend some time > carefully walking through the code. Without getting the coalesced mmio flushing out of the way, it does not buy us that much yet. But I have some idea... Jan
signature.asc
Description: OpenPGP digital signature
