Re: [Xen-devel] [PATCH v5 8/8] microcode: update microcode on cores in parallel

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 13.02.19 at 03:30, <chao.gao@xxxxxxxxx> wrote:
> On Tue, Feb 12, 2019 at 06:55:20AM -0700, Jan Beulich wrote:
>>>>> On 12.02.19 at 14:25, <roger.pau@xxxxxxxxxx> wrote:
>>> On Tue, Feb 12, 2019 at 05:51:41AM -0700, Jan Beulich wrote:
>>>> >>> On 28.01.19 at 08:06, <chao.gao@xxxxxxxxx> wrote:
>>>> > @@ -314,9 +310,7 @@ static int apply_microcode(unsigned int cpu)
>>>> >  
>>>> >      mc_intel = patch->data;
>>>> >      BUG_ON(!mc_intel);
>>>> > -
>>>> > -    /* serialize access to the physical write to MSR 0x79 */
>>>> > -    spin_lock_irqsave(&microcode_update_lock, flags);
>>>> > +    BUG_ON(local_irq_is_enabled());
>>>> >  
>>>> >      /* write microcode via MSR 0x79 */
>>>> >      wrmsrl(MSR_IA32_UCODE_WRITE, (unsigned long)mc_intel->bits);
>>>> > @@ -329,7 +323,6 @@ static int apply_microcode(unsigned int cpu)
>>>> >      rdmsrl(MSR_IA32_UCODE_REV, msr_content);
>>>> >      val[1] = (uint32_t)(msr_content >> 32);
>>>> >  
>>>> > -    spin_unlock_irqrestore(&microcode_update_lock, flags);
>>>> >      if ( val[1] != mc_intel->hdr.rev )
>>>> >      {
>>>> >          printk(KERN_ERR "microcode: CPU%d update from revision "
>>>> 
>>>> Am I understanding right that you now rely on upper layers in the
>>>> call tree to avoid calling into here in parallel for two hyperthreads
>>>> of the same core? I can't see how you avoid this situation during
>>>> AP bringup, for example. Did I overlook anything in this regard?
>>> 
>>> IIRC microcode update is done in the serialized part of AP bringup,
>>> before the call to smp_callin, which guarantees serialization.
>>
>>Hmm, yes, right now it is. But I'd call this "happens to be that way"
>>rather than "is guaranteed to be that way" - prior to commit
>>f97838bbd9 it did happen later.
> 
> How about employing another lock, "early_ucode_update_lock", to
> guarantee serialization.
> 
> In particular, early_microcode_update_cpu() and microcode_resume_cpu()
> will acquire this lock before ucode update.

That's a (temporary) option, but would over-serialize things again.
I was rather trying to hint towards a per-core lock.

Jan



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel