fix(virtio-net): free up send capacity before checking it

[cagatay-y]

Before checking the send capacity to issue a send token, we need to poll to get the capacity back from completed sends.

[github-actions]

Benchmark Results

Benchmark	Current: 36b177f	Previous: ee6e5d2	Performance Ratio
startup_benchmark Build Time	135.67 s	135.29 s	1.00
startup_benchmark File Size	0.90 MB	0.90 MB	1.00
Startup Time - 1 core	0.93 s (±0.01 s)	0.93 s (±0.02 s)	1.00
Startup Time - 2 cores	0.92 s (±0.02 s)	0.94 s (±0.02 s)	0.99
Startup Time - 4 cores	0.92 s (±0.01 s)	0.97 s (±0.02 s)	0.95
multithreaded_benchmark Build Time	141.66 s	141.88 s	1.00
multithreaded_benchmark File Size	1.00 MB	1.00 MB	1.00
Multithreaded Pi Efficiency - 2 Threads	2.64 % (±12.67 %)	2.83 % (±13.59 %)	0.93
Multithreaded Pi Efficiency - 4 Threads	1.51 % (±7.25 %)	1.60 % (±7.68 %)	0.94
Multithreaded Pi Efficiency - 8 Threads	0.71 % (±3.43 %)	0.78 % (±3.74 %)	0.92
micro_benchmarks Build Time	225.51 s	195.06 s	1.16
micro_benchmarks File Size	1.01 MB	1.01 MB	1.00
Scheduling time - 1 thread	3.28 ticks (±15.75 ticks)	3.06 ticks (±14.68 ticks)	1.07
Scheduling time - 2 threads	3.34 ticks (±16.05 ticks)	1.58 ticks (±7.57 ticks)	2.12
Micro - Time for syscall (getpid)	0.18 ticks (±0.85 ticks)	0.16 ticks (±0.79 ticks)	1.07
Memcpy speed - (built_in) block size 4096	1474.06 MByte/s (±7075.47 MByte/s)	1680.11 MByte/s (±8064.52 MByte/s)	0.88
Memcpy speed - (built_in) block size 1048576	545.63 MByte/s (±2619.02 MByte/s)	559.39 MByte/s (±2685.09 MByte/s)	0.98
Memcpy speed - (built_in) block size 16777216	178.33 MByte/s (±856.01 MByte/s)	201.86 MByte/s (±968.91 MByte/s)	0.88
Memset speed - (built_in) block size 4096	1481.48 MByte/s (±7111.11 MByte/s)	1518.99 MByte/s (±7291.14 MByte/s)	0.98
Memset speed - (built_in) block size 1048576	1006.97 MByte/s (±4833.47 MByte/s)	1283.86 MByte/s (±6162.54 MByte/s)	0.78
Memset speed - (built_in) block size 16777216	515.00 MByte/s (±2472.02 MByte/s)	869.39 MByte/s (±4173.09 MByte/s)	0.59
Memcpy speed - (rust) block size 4096	991.74 MByte/s (±4760.33 MByte/s)	685.71 MByte/s (±3291.43 MByte/s)	1.45
Memcpy speed - (rust) block size 1048576	549.83 MByte/s (±2639.18 MByte/s)	547.47 MByte/s (±2627.87 MByte/s)	1.00
Memcpy speed - (rust) block size 16777216	181.38 MByte/s (±870.61 MByte/s)	206.17 MByte/s (±989.64 MByte/s)	0.88
Memset speed - (rust) block size 4096	1200.00 MByte/s (±5760.00 MByte/s)	1304.35 MByte/s (±6260.87 MByte/s)	0.92
Memset speed - (rust) block size 1048576	1313.93 MByte/s (±6306.87 MByte/s)	1280.41 MByte/s (±6145.97 MByte/s)	1.03
Memset speed - (rust) block size 16777216	475.92 MByte/s (±2284.42 MByte/s)	881.11 MByte/s (±4229.33 MByte/s)	0.54
alloc_benchmarks Build Time	197.75 s	186.50 s	1.06
alloc_benchmarks File Size	0.97 MB	0.97 MB	1.00
Allocations - Allocation success	2.00 % (±13.86 %)	2.00 % (±13.86 %)	1
Allocations - Deallocation success	1.40 % (±9.68 %)	1.39 % (±9.66 %)	1.00
Allocations - Pre-fail Allocations	2.00 % (±13.86 %)	2.00 % (±13.86 %)	1
Allocations - Average Allocation time	260.74 Ticks (±1806.85 Ticks)	241.20 Ticks (±1671.42 Ticks)	1.08
Allocations - Average Allocation time (no fail)	260.74 Ticks (±1806.85 Ticks)	241.20 Ticks (±1671.42 Ticks)	1.08
Allocations - Average Deallocation time	16.79 Ticks (±116.36 Ticks)	16.83 Ticks (±116.64 Ticks)	1.00
mutex_benchmark Build Time	204.01 s	186.85 s	1.09
mutex_benchmark File Size	1.01 MB	1.01 MB	1.00
Mutex Stress Test Average Time per Iteration - 1 Threads	0.34 ns (±2.36 ns)	0.34 ns (±2.36 ns)	1
Mutex Stress Test Average Time per Iteration - 2 Threads	0.40 ns (±2.77 ns)	0.36 ns (±2.49 ns)	1.11

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

[mkroening]

@zyuiop, please review too :)

[zyuiop]

The network can get stuck without this patch. LGTM!

[stlankes]

Any idea, why the CI failed?

[mkroening]

Any idea, why the CI failed?

Might be spurious. Retrying.