RFC x86: allocate up to 32 tlb invalidate vectors
Get series with b4
b4 am 1288766655.23014.113.camel@sli10-conroeVersions newest first
- v1 RFC Latest4 patches 15 repliesIn review
Patches in v1 No applies recorded yet
| # | Subject and files | Size | Replies | Applied |
|---|---|---|---|---|
| 1/4 | x86: cleanup vector usagearch/ | +21 | 3 replies | |
| 2/4 | x86: allocate 32 tlb_invalidate_interrupt handlersarch/ arch/ arch/ +1 more | +103 | 1 reply | |
| 3/4 | x86: make tlb invalidate vector number configurablearch/ | +9 | 0 replies | |
| 4/4 | x86: avoid tlbstate lock if no enough cpusarch/ | +7 | 9 replies |
Reviews trailers per patch
1 of 4 patches has a Reviewed-by or Acked-by.
| Patch | Eric Dumazet |
|---|---|
| 1/4 | |
| 2/4 | |
| 3/4 | |
| 4/4 | A-b |
Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.
Cover letter [RFC 0/4]x86: allocate up to 32 tlb invalidate vectors
Hi,
In workload with heavy page reclaim, flush_tlb_page() is frequently used. We currently have 8 vectors for tlb flush, which is fine for small machines. But for big machines with a lot of CPUs, the 8 vectors are shared by all CPUs and we need lock to protect them. This will cause a lot of lock contentions. please see the patch 3 for detailed number of the lock contention.
Andi Kleen suggests we can use 32 vectors for tlb flush, which should be fine for even 8 socket machines. Test shows this reduces lock contention dramatically (see patch 3 for number).
One might argue if this will waste too many vectors and leave less vectors for devices. This could be a problem. But even we use 32 vectors, we still leave 78 vectors for devices. And we now have per-cpu vector, vector isn't scarce any more, but I'm open if anybody has objections.
Thanks,
Shaohua