[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH 3/4] cpus: protect TimerState writes with a spin
From: |
Paolo Bonzini |
Subject: |
Re: [Qemu-devel] [PATCH 3/4] cpus: protect TimerState writes with a spinlock |
Date: |
Mon, 10 Sep 2018 01:39:28 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.9.1 |
On 01/09/2018 00:07, Emilio G. Cota wrote:
> On Mon, Aug 20, 2018 at 17:09:02 +0200, Paolo Bonzini wrote:
>> In the next patch, we will need to write cpu_ticks_offset from any
>> thread, even outside the BQL. Currently, it is protected by the BQL
>> just because cpu_enable_ticks and cpu_disable_ticks happen to hold it,
>> but the critical sections are well delimited and it's easy to remove
>> the BQL dependency.
>>
>> Add a spinlock that matches vm_clock_seqlock, and hold it when writing
>> to the TimerState. This also lets us fix cpu_update_icount when 64-bit
>> atomics are not available.
>>
>> Fields of TiemrState are reordered to avoid padding.
>>
>> Signed-off-by: Paolo Bonzini <address@hidden>
>> ---
>> cpus.c | 72 ++++++++++++++++++++++++++++++++++++++--------------------
>> 1 file changed, 47 insertions(+), 25 deletions(-)
>>
>> diff --git a/cpus.c b/cpus.c
>> index 680706aefa..63ddd4fd21 100644
>> --- a/cpus.c
>> +++ b/cpus.c
> (snip)
>> @@ -244,11 +250,15 @@ void cpu_update_icount(CPUState *cpu)
>> int64_t executed = cpu_get_icount_executed(cpu);
>> cpu->icount_budget -= executed;
>>
>> -#ifdef CONFIG_ATOMIC64
>> +#ifndef CONFIG_ATOMIC64
>> + seqlock_write_lock(&timers_state.vm_clock_seqlock,
>> + &timers_state.vm_clock_lock);
>> +#endif
>> atomic_set__nocheck(&timers_state.qemu_icount,
>> timers_state.qemu_icount + executed);
>> -#else /* FIXME: we need 64bit atomics to do this safely */
>> - timers_state.qemu_icount += executed;
>> +#ifndef CONFIG_ATOMIC64
>> + seqlock_write_unlock(&timers_state.vm_clock_seqlock,
>> + &timers_state.vm_clock_lock);
>> #endif
>
>
> I'm puzzled by this hunk. Why are we only adding the seqlock_write
> if !CONFIG_ATOMIC64, if we always read .qemu_icount with seqlock_read?
Yeah, I missed that qemu_icount is read by icount_adjust and so it
indirectly affects qemu_icount_bias. It needs the seqlock always.
Paolo
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- Re: [Qemu-devel] [PATCH 3/4] cpus: protect TimerState writes with a spinlock,
Paolo Bonzini <=