feat/PCIe: verify the base address of PCI devices

[zyuiop]

This ensures that a malicious hypervisor will not provide PCI devices that map to system memory

[github-actions]

Benchmark Results

Details

Benchmark	Current: fe42a4a	Previous: 30cb3f9	Performance Ratio
startup_benchmark Build Time	80.77 s	78.59 s	1.03 ❗
startup_benchmark File Size	0.76 MB	0.76 MB	1.01 ❗
Startup Time - 1 core	0.76 s (±0.01 s)	0.73 s (±0.02 s)	1.04 ❗
Startup Time - 2 cores	0.76 s (±0.03 s)	0.75 s (±0.02 s)	1.01
Startup Time - 4 cores	0.75 s (±0.02 s)	0.76 s (±0.02 s)	1.00
multithreaded_benchmark Build Time	81.29 s	80.39 s	1.01 ❗
multithreaded_benchmark File Size	0.86 MB	0.82 MB	1.06 ❗
Multithreaded Pi Efficiency - 2 Threads	90.21 % (±7.65 %)	89.59 % (±5.95 %)	1.01
Multithreaded Pi Efficiency - 4 Threads	43.84 % (±2.63 %)	43.86 % (±2.44 %)	1.00
Multithreaded Pi Efficiency - 8 Threads	25.94 % (±1.27 %)	25.65 % (±1.39 %)	1.01
micro_benchmarks Build Time	82.87 s	87.27 s	0.95 ❗
micro_benchmarks File Size	0.87 MB	0.82 MB	1.06 ❗
Scheduling time - 1 thread	64.87 ticks (±2.06 ticks)	64.58 ticks (±2.95 ticks)	1.00
Scheduling time - 2 threads	37.30 ticks (±4.47 ticks)	35.27 ticks (±2.44 ticks)	1.06
Micro - Time for syscall (getpid)	2.83 ticks (±0.25 ticks)	2.72 ticks (±0.18 ticks)	1.04
Memcpy speed - (built_in) block size 4096	86593.76 MByte/s (±59674.46 MByte/s)	84336.36 MByte/s (±58124.47 MByte/s)	1.03
Memcpy speed - (built_in) block size 1048576	30829.60 MByte/s (±24922.40 MByte/s)	30954.90 MByte/s (±25149.11 MByte/s)	1.00
Memcpy speed - (built_in) block size 16777216	29920.90 MByte/s (±24643.32 MByte/s)	27618.34 MByte/s (±22854.11 MByte/s)	1.08
Memset speed - (built_in) block size 4096	86908.05 MByte/s (±59878.23 MByte/s)	84961.36 MByte/s (±58506.83 MByte/s)	1.02
Memset speed - (built_in) block size 1048576	31574.01 MByte/s (±25348.59 MByte/s)	31697.81 MByte/s (±25573.78 MByte/s)	1.00
Memset speed - (built_in) block size 16777216	30691.13 MByte/s (±25079.90 MByte/s)	28408.60 MByte/s (±23353.48 MByte/s)	1.08
Memcpy speed - (rust) block size 4096	76411.31 MByte/s (±53221.33 MByte/s)	74740.31 MByte/s (±52172.06 MByte/s)	1.02
Memcpy speed - (rust) block size 1048576	30758.96 MByte/s (±24893.83 MByte/s)	30989.85 MByte/s (±25131.37 MByte/s)	0.99
Memcpy speed - (rust) block size 16777216	29933.38 MByte/s (±24647.12 MByte/s)	27766.19 MByte/s (±22932.85 MByte/s)	1.08
Memset speed - (rust) block size 4096	76907.00 MByte/s (±53570.50 MByte/s)	75196.89 MByte/s (±52449.92 MByte/s)	1.02
Memset speed - (rust) block size 1048576	31500.78 MByte/s (±25322.18 MByte/s)	31748.80 MByte/s (±25574.00 MByte/s)	0.99
Memset speed - (rust) block size 16777216	30681.86 MByte/s (±25062.56 MByte/s)	28552.59 MByte/s (±23427.10 MByte/s)	1.07
alloc_benchmarks Build Time	81.36 s	81.63 s	1.00 ❗
alloc_benchmarks File Size	0.84 MB	0.84 MB	1.01 ❗
Allocations - Allocation success	100.00 %	100.00 %	1
Allocations - Deallocation success	100.00 %	100.00 %	1
Allocations - Pre-fail Allocations	100.00 %	100.00 %	1
Allocations - Average Allocation time	5366.07 Ticks (±96.57 Ticks)	5722.73 Ticks (±63.96 Ticks)	0.94 ❗
Allocations - Average Allocation time (no fail)	5366.07 Ticks (±96.57 Ticks)	5722.73 Ticks (±63.96 Ticks)	0.94 ❗
Allocations - Average Deallocation time	666.03 Ticks (±111.08 Ticks)	1530.42 Ticks (±212.44 Ticks)	0.44 ❗
mutex_benchmark Build Time	81.69 s	80.99 s	1.01 ❗
mutex_benchmark File Size	0.87 MB	0.82 MB	1.06 ❗
Mutex Stress Test Average Time per Iteration - 1 Threads	12.02 ns (±0.24 ns)	12.10 ns (±0.36 ns)	0.99
Mutex Stress Test Average Time per Iteration - 2 Threads	14.76 ns (±0.43 ns)	17.00 ns (±3.01 ns)	0.87

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

[mkroening]

Thanks, this looks good to me. :)

Just one suggestion.

[mkroening]

Can't this be simplified?

Suggested change

	(pci_start >= region_start && pci_start <= region_end) // PCI region starts within the memory region
	|| (pci_end >= region_start && pci_end <= region_end) // PCI region ends within the memory region
	|| (pci_start <= region_start && pci_end >= region_end) // PCI region contains the memory region
	pci_start > region_end && pci_end < region_start

At least that is how I did it in PageRange::overlaps.

[zyuiop]

Isn't it more pci_start < region_end && pci_end > region_start ?

I ended up using range intersections, which I think make the intent clearer

[mkroening]

You are right; I mistyped. Would you be okay with that? I'd like to avoid new nightly features if possible. :)

[zyuiop]

Okay, I was supposing this may be the case :)
Pushing a new version