fix(virtqueue): order descriptor publication against the notification check

[stlankes]

Both rings decide whether to notify the device by reading state the device writes, right after publishing a descriptor, with nothing ordering the two. The preceding barrier orders the descriptor contents against the field that publishes them, not that field against this later load, so a weakly ordered machine can read the flag before our publication is visible.

This causes deadlocks in polling mode, which are prevented by adding memory barriers.

[github-actions]

Benchmark Results

Details

Benchmark	Current: 7184aa0	Previous: 2e23902	Performance Ratio
startup_benchmark Build Time	94.18 s	80.34 s	1.17 ❗
startup_benchmark File Size	0.78 MB	0.80 MB	0.98 ❗
Startup Time - 1 core	0.75 s (±0.02 s)	0.75 s (±0.02 s)	1.00
Startup Time - 2 cores	0.75 s (±0.02 s)	0.74 s (±0.02 s)	1.02
Startup Time - 4 cores	0.77 s (±0.01 s)	0.74 s (±0.02 s)	1.03 ❗
multithreaded_benchmark Build Time	91.99 s	82.11 s	1.12 ❗
multithreaded_benchmark File Size	0.88 MB	0.86 MB	1.03 ❗
Multithreaded Pi Efficiency - 2 Threads	65.57 % (±7.98 %)	85.89 % (±6.61 %)	0.76 ❗
Multithreaded Pi Efficiency - 4 Threads	41.56 % (±2.91 %)	43.43 % (±2.56 %)	0.96
Multithreaded Pi Efficiency - 8 Threads	20.00 % (±1.57 %)	25.76 % (±1.53 %)	0.78 ❗
micro_benchmarks Build Time	216.18 s	80.40 s	2.69 ❗
micro_benchmarks File Size	0.88 MB	0.86 MB	1.03 ❗
Scheduling time - 1 thread	197.73 ticks (±50.00 ticks)	62.65 ticks (±4.06 ticks)	3.16 ❗
Scheduling time - 2 threads	107.44 ticks (±32.65 ticks)	34.08 ticks (±4.10 ticks)	3.15 ❗
Micro - Time for syscall (getpid)	9.36 ticks (±4.41 ticks)	3.45 ticks (±0.58 ticks)	2.71 ❗
Memcpy speed - (built_in) block size 4096	56802.00 MByte/s (±40001.94 MByte/s)	82448.38 MByte/s (±56997.13 MByte/s)	0.69
Memcpy speed - (built_in) block size 1048576	14874.16 MByte/s (±12761.27 MByte/s)	30585.98 MByte/s (±24707.84 MByte/s)	0.49
Memcpy speed - (built_in) block size 16777216	12680.41 MByte/s (±10804.21 MByte/s)	26340.06 MByte/s (±21720.96 MByte/s)	0.48
Memset speed - (built_in) block size 4096	57037.56 MByte/s (±40149.36 MByte/s)	82292.76 MByte/s (±56891.50 MByte/s)	0.69
Memset speed - (built_in) block size 1048576	15231.27 MByte/s (±12965.56 MByte/s)	31323.85 MByte/s (±25145.86 MByte/s)	0.49
Memset speed - (built_in) block size 16777216	13106.68 MByte/s (±11094.44 MByte/s)	27104.68 MByte/s (±22209.94 MByte/s)	0.48
Memcpy speed - (rust) block size 4096	54882.81 MByte/s (±39121.98 MByte/s)	74097.96 MByte/s (±51811.44 MByte/s)	0.74
Memcpy speed - (rust) block size 1048576	14427.31 MByte/s (±12693.24 MByte/s)	30361.60 MByte/s (±24602.37 MByte/s)	0.48
Memcpy speed - (rust) block size 16777216	13138.89 MByte/s (±11445.15 MByte/s)	27625.34 MByte/s (±22806.88 MByte/s)	0.48
Memset speed - (rust) block size 4096	55373.89 MByte/s (±39464.05 MByte/s)	74373.47 MByte/s (±51976.48 MByte/s)	0.74
Memset speed - (rust) block size 1048576	14726.38 MByte/s (±12831.83 MByte/s)	31110.89 MByte/s (±25033.24 MByte/s)	0.47
Memset speed - (rust) block size 16777216	13410.92 MByte/s (±11572.30 MByte/s)	28386.93 MByte/s (±23265.03 MByte/s)	0.47
alloc_benchmarks Build Time	213.34 s	74.76 s	2.85 ❗
alloc_benchmarks File Size	0.86 MB	0.87 MB	0.98 ❗
Allocations - Allocation success	91.38 %	91.31 %	1.00 ❗
Allocations - Deallocation success	100.00 %	100.00 %	1
Allocations - Pre-fail Allocations	61.60 %	61.44 %	1.00 ❗
Allocations - Average Allocation time	23456.36 Ticks (±1601.07 Ticks)	5860.58 Ticks (±98.43 Ticks)	4.00 ❗
Allocations - Average Allocation time (no fail)	24362.93 Ticks (±2025.26 Ticks)	6554.81 Ticks (±92.86 Ticks)	3.72 ❗
Allocations - Average Deallocation time	6864.14 Ticks (±1641.54 Ticks)	1805.01 Ticks (±250.35 Ticks)	3.80 ❗
mutex_benchmark Build Time	222.70 s	79.82 s	2.79 ❗
mutex_benchmark File Size	0.88 MB	0.86 MB	1.03 ❗
Mutex Stress Test Average Time per Iteration - 1 Threads	36.84 ns (±6.97 ns)	12.10 ns (±0.41 ns)	3.04 ❗
Mutex Stress Test Average Time per Iteration - 2 Threads	32.94 ns (±8.80 ns)	40.26 ns (±1.68 ns)	0.82

This comment was automatically generated by workflow using github-action-benchmark.

[mkroening]

Looks good to me! 👍