search for: shootdown

On 03/07/2019 18:02, Nadav Amit wrote: >> On Jul 3, 2019, at 7:04 AM, Juergen Gross <jgross at suse.com> wrote: >> >> On 03.07.19 01:51, Nadav Amit wrote: >>> To improve TLB shootdown performance, flush the remote and local TLBs >>> concurrently. Introduce flush_tlb_multi() that does so. Introduce >>> paravirtual versions of flush_tlb_multi() for KVM, Xen and hyper-v (Xen >>> and hyper-v are only compile-tested). >>> While the updated smp infr...

> On Jun 25, 2019, at 8:36 PM, Andy Lutomirski <luto at kernel.org> wrote: > > On Wed, Jun 12, 2019 at 11:49 PM Nadav Amit <namit at vmware.com> wrote: >> To improve TLB shootdown performance, flush the remote and local TLBs >> concurrently. Introduce flush_tlb_multi() that does so. The current >> flush_tlb_others() interface is kept, since paravirtual interfaces need >> to be adapted first before it can be removed. This is left for future >> work. In...

On 03.07.19 01:51, Nadav Amit wrote: > To improve TLB shootdown performance, flush the remote and local TLBs > concurrently. Introduce flush_tlb_multi() that does so. Introduce > paravirtual versions of flush_tlb_multi() for KVM, Xen and hyper-v (Xen > and hyper-v are only compile-tested). > > While the updated smp infrastructure is capable of r...

On 03.07.19 01:51, Nadav Amit wrote: > To improve TLB shootdown performance, flush the remote and local TLBs > concurrently. Introduce flush_tlb_multi() that does so. Introduce > paravirtual versions of flush_tlb_multi() for KVM, Xen and hyper-v (Xen > and hyper-v are only compile-tested). > > While the updated smp infrastructure is capable of r...

This patch implements persistent grants for the xen-blk{front,back} mechanism. The effect of this change is to reduce the number of unmap operations performed, since they cause a (costly) TLB shootdown. This allows the I/O performance to scale better when a large number of VMs are performing I/O. Previously, the blkfront driver was supplied a bvec[] from the request queue. This was granted to dom0; dom0 performed the I/O and wrote directly into the grant-mapped memory and unmapped it; blkfront t...

...ed the two cores to exchange some bus traffic for cache coherency, even though is is completely valid in the C++ abstract machine model for the load of g to return a stale cache value. This is only a real problem on strongly ordered architectures (e.g. x86), but given the relative cost of a cache shootdown and everything else in this test case (with the exception of the thread creation), I wouldn't be surprised if it ended up slowing things down. Especially given that a modern superscalar CPU will speculatively execute the load ANYWAY if it can do so from cache, and if it can't then the perf...

...er interrupts PND: 0 0 0 0 Performance pending work RES: 436674 440302 2195268 1451020 Rescheduling interrupts CAL: 169 265 203 251 Function call interrupts TLB: 43920 44257 50177 52884 TLB shootdowns TRM: 0 0 0 0 Thermal event interrupts THR: 0 0 0 0 Threshold APIC interrupts MCE: 0 0 0 0 Machine check exceptions MCP: 1073 1073 1073 1073 Machine check...

...exchange > some bus traffic for cache coherency, even though is is completely valid in > the C++ abstract machine model for the load of g to return a stale cache > value. This is only a real problem on strongly ordered architectures > (e.g. x86), but given the relative cost of a cache shootdown and everything > else in this test case (with the exception of the thread creation), I > wouldn't be surprised if it ended up slowing things down. Especially given > that a modern superscalar CPU will speculatively execute the load ANYWAY if > it can do so from cache, and if it can...

To improve TLB shootdown performance, flush the remote and local TLBs concurrently. Introduce flush_tlb_multi() that does so. The current flush_tlb_others() interface is kept, since paravirtual interfaces need to be adapted first before it can be removed. This is left for future work. In such PV environments, TLB flushes a...

To improve TLB shootdown performance, flush the remote and local TLBs concurrently. Introduce flush_tlb_multi() that does so. The current flush_tlb_others() interface is kept, since paravirtual interfaces need to be adapted first before it can be removed. This is left for future work. In such PV environments, TLB flushes a...

To improve TLB shootdown performance, flush the remote and local TLBs concurrently. Introduce flush_tlb_multi() that does so. The current flush_tlb_others() interface is kept, since paravirtual interfaces need to be adapted first before it can be removed. This is left for future work. In such PV environments, TLB flushes a...

To improve TLB shootdown performance, flush the remote and local TLBs concurrently. Introduce flush_tlb_multi() that does so. The current flush_tlb_others() interface is kept, since paravirtual interfaces need to be adapted first before it can be removed. This is left for future work. In such PV environments, TLB flushes a...

...o exchange some bus traffic for cache coherency, even though is is > completely valid in the C++ abstract machine model for the load of g > to return a stale cache value. This is only a real problem on > strongly ordered architectures (e.g. x86), but given the relative > cost of a cache shootdown and everything else in this test case > (with the exception of the thread creation), I wouldn't be surprised > if it ended up slowing things down. Especially given that a modern > superscalar CPU will speculatively execute the load ANYWAY if it can > do so from cache, and if it can&...

...b4 219: 64752376 0 PCI-MSI-edge ioc0 NMI: 0 0 Non-maskable interrupts LOC: 993263108 915510467 Local timer interrupts RES: 16326632 9071480 Rescheduling interrupts CAL: 303 529 function call interrupts TLB: 1521462 4810165 TLB shootdowns TRM: 0 0 Thermal event interrupts SPU: 0 0 Spurious interrupts ERR: 0 MIS: 0 Vin?cius Fontes www.asteriskforum.com.br - Informa??es e discuss?o sobre Asterisk e telefonia IP

...apply_paravirt() > completely. (I think Xen can be changed to set CPU feature bits to > specify which ops it needs during early boot, then those ops can be > patched in using early alternatives.) JFYI starting 4.14 Xen PV is not the only user of pv_mmu_ops, Hyper-V uses it for TLB shootdown now. -- Vitaly

On Wed, Jun 12, 2019 at 11:49 PM Nadav Amit <namit at vmware.com> wrote: > > To improve TLB shootdown performance, flush the remote and local TLBs > concurrently. Introduce flush_tlb_multi() that does so. The current > flush_tlb_others() interface is kept, since paravirtual interfaces need > to be adapted first before it can be removed. This is left for future > work. In such PV environ...

I'm looking at adding Xen SMP support, so I'm trying to work out what pv_ops we need, and how to cut into the existing smp stuff. smpboot.c has a mixture of stuff which is generally useful for SMP stuff (the various CPU sets, and presumably the sibling relationships are useful in principle), but also a whole pile of APIC stuff which is irrelevent to Xen. It has these exported

I'm looking at adding Xen SMP support, so I'm trying to work out what pv_ops we need, and how to cut into the existing smp stuff. smpboot.c has a mixture of stuff which is generally useful for SMP stuff (the various CPU sets, and presumably the sibling relationships are useful in principle), but also a whole pile of APIC stuff which is irrelevent to Xen. It has these exported

...219: 2014578 0 PCI-MSI-edge eth3 NMI: 0 0 Non-maskable interrupts LOC: 2180853 1019374 Local timer interrupts RES: 1655 872 Rescheduling interrupts CAL: 49 74 function call interrupts TLB: 5322 5969 TLB shootdowns TRM: 0 0 Thermal event interrupts SPU: 0 0 Spurious interrupts ERR: 0 MIS: 0 lsusb Bus 007 Device 003: ID e4e4:1152 Bus 007 Device 002: ID e4e4:1142 Bus 007 Device 001: ID 0000:0000 Bus 005 Device 001: ID 0000:0000 Bus 006 Devic...

To improve TLB shootdown performance, flush the remote and local TLBs concurrently. Introduce flush_tlb_multi() that does so. The current flush_tlb_others() interface is kept, since paravirtual interfaces need to be adapted first before it can be removed. This is left for future work. In such PV environments, TLB flushes a...