code wiki / _hdl_build / _dxg_probe.nx
_dxg_probe.nx
buildroot/runtime/_hdl_build/_dxg_probe.nx
about
_dxg_probe.nx -- SOVEREIGN-GPU foundation RECON probe (no libvulkan/libcuda/libdxcore,
no libc): raw openat("/dev/dxg") + raw ioctl on the real dxgkrnl thunk for the RTX 5080.
This is a SCOPE/RECON probe, NOT the gate. It proves three things from a REAL run:
(A) /dev/dxg opens with a raw openat() -> prints the fd (>=0 = real device handle).
(B) a TAMPER ioctl with a non-dxg magic returns a real device errno (NOT 0) -> the
fd is the real device, not a no-op; a bogus code does NOT falsely succeed.
(C) a magic-0x47 LX_DX* ioctl returns the kernel's REAL response code (0 or -errno),
distinguishable from the tamper -> the 0x47 ioctl path reaches dxgkrnl.
The dxgkrnl uapi (WSL2-Linux-Kernel drivers/hv/dxgkrnl, include/uapi/misc/d3dkmthk-style):
ioctl numbers are _IOWR(0x47, nr, struct ...) -> 32-bit code =
dir(2 bits, _IOWR=3=0xC0000000) | size(14 bits, <<16) | magic(8 bits 0x47, <<8) | nr(8 bits).
LX_DXENUMADAPTERS3 nr is in the LX_DX* enum; this probe issues a representative 0x47 code and a
stub arg to read the kernel's response (a real call needs the exact struct -- see the gate plan).
license_tier: ORIGINAL
dependencies 1 imports · 0 importers
imports: nx_syscalls.nx
imported by: nobody (leaf or entry point)
call flow from main pre-order; caps 40 nodes / depth 6 declared; ↻ = already shown
structs
| none |
consts
| none |
functions
| 19 | func dp_p(s: *u8) -> i64 { var n: i64=0; while s[n]!=(0 as u8){n=n+1} sys_write(1,s,n); return 0 } |
| 20 | func dp_n(v: i64) -> i64 { let bb: *u8=sys_mmap(28); var m: i64=v; if m<0{m=0-m;sys_write(1,"-" as *u8,1)}; let t: *u8=sys_mmap(28); var k: i64=0; if m==0{t[0]=48;k=1}; while m>0{t[k]=(48+(m%10)) as u8;m=m/10;k=k+1}; var i: i64=0; while i<k{bb[i]=t[k-1-i];i=i+1}; sys_write(1,bb,k); return 0 } |
| 22 | func main() -> i64 |