Paul E. McKenney
Posts
102Following
50Followers
212
Paul E. McKenney
paulmckrcu
@jwseigh As always, much depends on your exact implementation. But what if the RSEQ reader picked up the pointer, took an unrelated long-running interrupt, took (and responded to) the membarrier IPI, then took another unrelated long-running interrupt? I do not believe that this would cause RSEQ to retry (though I might be missing something).
In that scenario, how is the updater to know that it should avoid reclaiming the object whose pointer the RSEQ reader will eventually store into its hazard pointer?
(I get the blog post: If you can uninterruptibly load and store the pointer, then membarrier() will either force the reader to see the new pointer or force the updater to see the updated hazard pointer.
I do not understand how your RSEQ trick works. Not yet, anyway.)
In that scenario, how is the updater to know that it should avoid reclaiming the object whose pointer the RSEQ reader will eventually store into its hazard pointer?
(I get the blog post: If you can uninterruptibly load and store the pointer, then membarrier() will either force the reader to see the new pointer or force the updater to see the updated hazard pointer.
I do not understand how your RSEQ trick works. Not yet, anyway.)
1
0
0
Paul E. McKenney
paulmckrcu
@jwseigh If the vCPU preemption also blocks the guest-OS-level membarrier() call, perhaps you are OK?
1
0
0
Paul E. McKenney
paulmckrcu
@jwseigh Hmmm, interesting... Does RSEQ now fail in response to vCPU preemption?
1
0
0
Paul E. McKenney
paulmckrcu
@jwseigh I thought that membarrier only guaranteed ordering, but perhaps you have a trick that I missed.
1
0
0
Paul E. McKenney
paulmckrcu
@jwseigh Nice! But if I may ask, how does this deal with the possibility that a updater on some other thread concurrently changed the pointer that you are trying to protect?
1
0
0
Paul E. McKenney
paulmckrcu
Today I learned that gawk has a "--debug" command-line argument, and has had for more than ten years. In my defense, I using awk for about 30 years prior to that...
1
3
11
Paul E. McKenney
paulmckrcu
@Louson Never forget: It is easier to circle the square than to square the circle.
0
0
0
Paul E. McKenney
paulmckrcu
@jwseigh Rightly or wrongly, this site says tau is 2π. https://en.wikipedia.org/wiki/Tau_(mathematics)
Me, I glanced too quickly at the first occurrence of 𝜏 (tau) in this thread and mistook it for the Roman-alphabet letter "r". ;-)
Me, I glanced too quickly at the first occurrence of 𝜏 (tau) in this thread and mistook it for the Roman-alphabet letter "r". ;-)
0
0
0
Paul E. McKenney
paulmckrcu
@tommythorn Indeed, 2π day and 𝜏 day are one and the same! But I suspect that more people would prefer to eat two pies than would prefer to eat a tau. ;-)
0
0
1
Paul E. McKenney
paulmckrcu
@morgan I would attempt a witty response, but I would probably end up sounding like a square. ;-)
0
0
1
Paul E. McKenney
paulmckrcu
@brauner Fair enough! But I might nor might not remember that you forgot. ;-)
0
0
1
Paul E. McKenney
paulmckrcu
@brauner RCU uses many mutexes so that RCU users don't have to? ;-)
1
0
3
Paul E. McKenney
paulmckrcu
1
0
2
Paul E. McKenney
paulmckrcu
@ljs It is amazing how destructive unanticipated use cases can be. Destructive, but necessary...
0
0
2
Paul E. McKenney
paulmckrcu
0
0
2
Paul E. McKenney
paulmckrcu
@jann @brauner Or to recover from the next timeout being unduly delayed, for that matter. That said, in my experience, most timeout handlers unconditionally do their work. On the other hand, I could easily believe that unconditional timeout handlers are much less likely to come to your attention. ;-)
0
0
2