Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1495552

Re: [RFC] arm64: Enforce observed order for spinlock and data

Show all headers | View raw

On 2016-10-04 13:53, bdegraaf@codeaurora.org wrote:
> On 2016-10-04 06:12, Mark Rutland wrote:
>> On Mon, Oct 03, 2016 at 03:20:57PM -0400, bdegraaf@codeaurora.org 
>> wrote:
>>> On 2016-10-01 14:11, Mark Rutland wrote:
>>> >Hi Brent,
>>> >
>>> >Evidently my questions weren't sufficiently clear; even with your
>>> >answers it's not clear to me what precise issue you're attempting to
>>> >solve.  I've tried to be more specific this time.
>>> >
>>> >At a high-level, can you clarify whether you're attempting to solve is:
>>> >
>>> >(a) a functional correctness issue (e.g. data corruption)
>>> >(b) a performance issue
>>> >
>>> >And whether this was seen in practice, or found through code
>>> >inspection?
>> 
>>> Thinking about this, as the reader/writer code has no known "abuse"
>>> case, I'll remove it from the patchset, then provide a v2 patchset
>>> with a detailed explanation for the lockref problem using the commits
>>> you provided as an example, as well as performance consideration.
>> 
>> If there's a functional problem, let's consider that in isolation 
>> first.
>> Once we understand that, then we can consider doing what is optimal.
>> 
>> As should be obvious from the above, I'm confused because this patch
>> conflates functional details with performance optimisations which (to
>> me) sound architecturally dubious.
>> 
>> I completely agree with Peter that if the problem lies with lockref, 
>> it
>> should be solved in the lockref code.
>> 
>> Thanks,
>> Mark.
> 
> After looking at this, the problem is not with the lockref code per se: 
> it is
> a problem with arch_spin_value_unlocked().  In the out-of-order case,
> arch_spin_value_unlocked() can return TRUE for a spinlock that is in 
> fact
> locked but the lock is not observable yet via an ordinary load.  Other 
> than
> ensuring order on the locking side (as the prior patch did), there is a 
> way
> to make arch_spin_value_unlock's TRUE return value deterministic, but 
> it
> requires that it does a write-back to the lock to ensure we didn't 
> observe
> the unlocked value while another agent was in process of writing back a
> locked value.
> 
> Brent

Scratch that--things get complicated as the lock itself gets "cloned," 
which
could happen during the out-of-order window.  I'll post back later after 
I've
analyzed it fully.

Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

[RFC] arm64: Enforce observed order for spinlock and data Brent DeGraaf <bdegraaf@codeaurora.org> - 2016-09-30 19:50 +0200
  Re: [RFC] arm64: Enforce observed order for spinlock and data Robin Murphy <robin.murphy@arm.com> - 2016-09-30 20:50 +0200
    Re: [RFC] arm64: Enforce observed order for spinlock and data bdegraaf@codeaurora.org - 2016-10-01 17:50 +0200
  Re: [RFC] arm64: Enforce observed order for spinlock and data Peter Zijlstra <peterz@infradead.org> - 2016-09-30 21:00 +0200
  Re: [RFC] arm64: Enforce observed order for spinlock and data Peter Zijlstra <peterz@infradead.org> - 2016-09-30 21:10 +0200
    Re: [RFC] arm64: Enforce observed order for spinlock and data bdegraaf@codeaurora.org - 2016-10-01 18:10 +0200
  Re: [RFC] arm64: Enforce observed order for spinlock and data Mark Rutland <mark.rutland@arm.com> - 2016-09-30 21:40 +0200
    Re: [RFC] arm64: Enforce observed order for spinlock and data bdegraaf@codeaurora.org - 2016-10-01 18:20 +0200
      Re: [RFC] arm64: Enforce observed order for spinlock and data Mark Rutland <mark.rutland@arm.com> - 2016-10-01 20:20 +0200
        Re: [RFC] arm64: Enforce observed order for spinlock and data bdegraaf@codeaurora.org - 2016-10-03 21:30 +0200
          Re: [RFC] arm64: Enforce observed order for spinlock and data Peter Zijlstra <peterz@infradead.org> - 2016-10-04 09:00 +0200
          Re: [RFC] arm64: Enforce observed order for spinlock and data Mark Rutland <mark.rutland@arm.com> - 2016-10-04 12:20 +0200
            Re: [RFC] arm64: Enforce observed order for spinlock and data bdegraaf@codeaurora.org - 2016-10-04 20:00 +0200
              Re: [RFC] arm64: Enforce observed order for spinlock and data bdegraaf@codeaurora.org - 2016-10-04 20:30 +0200
              Re: [RFC] arm64: Enforce observed order for spinlock and data Mark Rutland <mark.rutland@arm.com> - 2016-10-04 21:20 +0200
                Re: [RFC] arm64: Enforce observed order for spinlock and data bdegraaf@codeaurora.org - 2016-10-05 17:00 +0200
                Re: [RFC] arm64: Enforce observed order for spinlock and data bdegraaf@codeaurora.org - 2016-10-05 17:20 +0200
                Re: [RFC] arm64: Enforce observed order for spinlock and data Peter Zijlstra <peterz@infradead.org> - 2016-10-05 17:20 +0200
                Re: [RFC] arm64: Enforce observed order for spinlock and data bdegraaf@codeaurora.org - 2016-10-05 17:40 +0200
                Re: [RFC] arm64: Enforce observed order for spinlock and data bdegraaf@codeaurora.org - 2016-10-12 22:10 +0200

csiph-web