performance-metrics: Add QCOW2 async multicluster read micro benchmark

Add micro_bench_qcow_async_multi_cluster_read which reads 8
contiguous clusters (512 KiB) per request through the QcowAsync
io_uring path. With coalesced mappings this can hit the io_uring
fast path for a single Readv SQE.

Signed-off-by: Anatol Belski <anbelski@linux.microsoft.com>
This commit is contained in:
Anatol Belski
2026-03-27 11:06:08 +01:00
committed by Rob Bradford
parent 377c260196
commit 2041ba5a91
2 changed files with 53 additions and 1 deletions

View File

@@ -378,7 +378,7 @@ mod adjuster {
}
}
const TEST_LIST: [PerformanceTest; 88] = [
const TEST_LIST: [PerformanceTest; 90] = [
PerformanceTest {
name: "boot_time_ms",
func_ptr: performance_boot_time,
@@ -1565,6 +1565,30 @@ const TEST_LIST: [PerformanceTest; 88] = [
},
unit_adjuster: adjuster::s_to_us,
},
PerformanceTest {
name: "micro_block_qcow_async_multi_cluster_read_128_us",
func_ptr: micro_bench_block::micro_bench_qcow_async_multi_cluster_read,
control: PerformanceTestControl {
test_timeout: 10,
test_iterations: 20,
warmup_iterations: 5,
num_ops: Some(128),
..PerformanceTestControl::default()
},
unit_adjuster: adjuster::s_to_us,
},
PerformanceTest {
name: "micro_block_qcow_async_multi_cluster_read_256_us",
func_ptr: micro_bench_block::micro_bench_qcow_async_multi_cluster_read,
control: PerformanceTestControl {
test_timeout: 10,
test_iterations: 20,
warmup_iterations: 5,
num_ops: Some(256),
..PerformanceTestControl::default()
},
unit_adjuster: adjuster::s_to_us,
},
];
fn run_test_with_timeout(

View File

@@ -428,3 +428,31 @@ pub fn micro_bench_qcow_async_random_read(control: &PerformanceTestControl) -> f
drain_async_completions(async_io.as_mut(), num_ops);
start.elapsed().as_secs_f64()
}
/// Issue large multi-cluster reads from a prepopulated QCOW2 image
/// through the QcowAsync io_uring path.
///
/// Each read spans 8 contiguous clusters (512 KiB). With coalesced
/// mappings, this can hit the io_uring fast path for a single Readv.
///
/// Returns the total read wall clock time in seconds.
pub fn micro_bench_qcow_async_multi_cluster_read(control: &PerformanceTestControl) -> f64 {
const CLUSTERS_PER_READ: usize = 8;
let num_ops = control.num_ops.expect("num_ops required") as usize;
let (_tmp, disk) = util::qcow_async_tempfile(num_ops);
let mut async_io = disk
.new_async_io(num_ops as u32)
.expect("new_async_io failed");
let read_size = CLUSTERS_PER_READ * QCOW_CLUSTER_SIZE as usize;
let mut buf = vec![0u8; read_size];
let iovec = read_iovec(&mut buf);
let num_reads = num_ops / CLUSTERS_PER_READ;
let start = Instant::now();
submit_reads(async_io.as_mut(), num_reads, read_size as u64, &[iovec]);
drain_async_completions(async_io.as_mut(), num_reads);
start.elapsed().as_secs_f64()
}