DMA pool lock contention

Keith Busch kbusch at kernel.org
Thu Apr 17 11:17:57 PDT 2025


On Thu, Apr 17, 2025 at 10:06:11AM -0700, Caleb Sander Mateos wrote:
> Hi Linux NVMe folks,
> On a 32 KB NVMe passthru read workload, we see 2.4% of CPU time is
> spent in _raw_spin_lock_irqsave called from dma_pool_alloc and
> dma_pool_free. 

Just curious, do you have multiple cpu NUMA nodes here? Or do these CPUs
share a last-level-cache?



More information about the Linux-nvme mailing list