code wiki / _hdl_build / nx_sha1_extvec_gate.nx
`nx_sha1_extvec_gate.nx` source

↩ module page · 270 lines · 12988 B
1// nx_sha1_extvec_gate.nx -- validated against RFC3174-resultarray0, read from the pinned+corroborated RFC3174 document.
2// ⚠HEADER CORRECTED 2026-08-01: this file was CLONED from nx_hmac_extvec_gate.nx and inherited its
3// header verbatim, so it claimed to be that gate validating HMAC-SHA-256 against RFC 4231. Nine of
4// twenty-three extvec gates carried the same wrong self-description. ★★★★★A CLONED FILE INHERITS ITS
5// PARENT'S CLAIMS, AND ON AN EVIDENCE ARTIFACT THE HEADER IS A PROVENANCE CLAIM, NOT A COMMENT --
6// an auditor reading headers would have concluded RFC 4231 validated all nine subjects.
7//
8// SUPERSEDES the single-case version. Going from 1 vector to 7 matters because the cases are deliberately
9// chosen by the authority to hit DIFFERENT code paths, and the ones I was NOT running are the interesting
10// ones: case 3 uses a 50-byte repeated data block, cases 6 and 7 use a 131-BYTE KEY (longer than the 64-byte
11// SHA-256 block, so the key must be HASHED first), and case 5 publishes a TRUNCATED 128-bit MAC.
12// ★A gate that ran only case 1 would never touch the key-longer-than-block branch -- the single most
13// commonly botched part of HMAC. Running one vector from a seven-vector suite is not "validated against
14// RFC 4231"; it is validated against one line of it.
15//
16// ⚠VARIABLE-LENGTH FIELDS, HANDLED BY TERMINATOR NOT BY LENGTH. Key/Data lengths differ per case and their
17// annotations are inconsistent -- "(20 bytes)" for keys but ("Hi There") for data -- so a length cannot be
18// read uniformly. Instead the hex run is read until the first `(`, which terminates both forms.
19// ⚠THE MAC HAS NO `(` TERMINATOR and case 5's is TRUNCATED to 16 bytes, so it is read as hex pairs until a
20// pair is not both-hex, capped at 32. That correctly stops at the section heading that follows -- note
21// "4.3." begins with '4', a HEX DIGIT, and is only rejected because '.' is not: the PAIR rule saves this,
22// a single-nibble rule would have swallowed it.
23//
24// Construction unchanged: no expected value in this source, document pinned to a socket-time digest, every
25// key/data/MAC read from that pinned document, and a completeness check that refuses GREEN below 7.
26// license_tier: ORIGINAL  expect_exit: 0
27import "nx_syscalls.nx"
28import "nx_sha256_wasm.nx"
29import "nx_sha1.nx"
30import "nx_gate_verdict.nx"
31
32func w(s: *u8) -> i64 { var n: i64 = 0; while s[n] != (0 as u8) { n = n + 1 } sys_write(1, s, n); return 0 }
33func wb(b: *u8, n: i64) -> i64 { sys_write(1, b, n); return 0 }
34
35func nn(v: i64) -> i64 {
  var m: i64 = v
  if m < 0 { w("-" as *u8); m = 0 - m }
  let t: *u8 = sys_mmap(32)
  var k: i64 = 0
  if m == 0 { t[0] = 48 as u8; k = 1 }
  while m > 0 { t[k] = (48 + (m % 10)) as u8; m = m / 10; k = k + 1 }
  let b: *u8 = sys_mmap(32)
  var j: i64 = 0
  while j < k { b[j] = t[k - 1 - j]; j = j + 1 }
  sys_write(1, b, k)
  return 0
47}
48
49func hexnib(v: i64) -> i64 { if v < 10 { return 48 + v } return 87 + v }
50
51func hexval(c: i64) -> i64 {
  if c >= 48 { if c <= 57 { return c - 48 } }
  if c >= 97 { if c <= 102 { return c - 87 } }
  if c >= 65 { if c <= 70 { return c - 55 } }
  return 0 - 1
56}
57
58func isws(c: i64) -> i64 {
  if c == 32 { return 1 }
  if c == 10 { return 1 }
  if c == 13 { return 1 }
  if c == 9 { return 1 }
  return 0
64}
65
66func starts(b: *u8, n: i64, at: i64, s: *u8) -> i64 {
  var i: i64 = 0
  while s[i] != (0 as u8) {
      if at + i >= n { return 0 }
      if b[at + i] != s[i] { return 0 }
      i = i + 1
  }
  return 1
74}
75
76func findfrom(b: *u8, n: i64, s: *u8, from: i64) -> i64 {
  var p: i64 = from
  while p < n {
      if starts(b, n, p, s) == 1 { return p }
      p = p + 1
  }
  return 0 - 1
83}
84
85// Read hex PAIRS (whitespace between pairs is skipped) until a pair is not both-hex, or `(` is reached,
86// or cap is hit. Returns the byte count.
87func parserun(b: *u8, n: i64, from: i64, out: *u8, cap: i64) -> i64 {
  var p: i64 = from
  var got: i64 = 0
  var done: i64 = 0
  while done == 0 {
      if got >= cap { done = 1 }
      else {
          // skip whitespace
          var d1: i64 = 0
          while d1 == 0 {
              if p >= n { d1 = 1 }
              else { if isws(b[p] as i64) == 1 { p = p + 1 } else { d1 = 1 } }
          }
          if p + 1 >= n { done = 1 }
          else {
              // `(` ends a LINE SEGMENT, not the value: RFC 4231 annotates EVERY wrapped line, e.g.
              //   Data = 7768...6e7420  ("what do ya want ")
              //          666f...693f    ("for nothing?")
              // Treating `(` as the value terminator stopped case 2 at 16 of its 28 bytes. On `(`, skip
              // to the next line and continue ONLY if it resumes with a hex pair; otherwise stop.
              if b[p] == (40 as u8) {
                  var dl: i64 = 0
                  while dl == 0 {
                      if p >= n { dl = 1 }
                      else { if b[p] == (10 as u8) { p = p + 1; dl = 1 } else { p = p + 1 } }
                  }
                  var dw: i64 = 0
                  while dw == 0 {
                      if p >= n { dw = 1 }
                      else { if isws(b[p] as i64) == 1 { p = p + 1 } else { dw = 1 } }
                  }
                  // ⚠"Da" IS VALID HEX. The next line may be `   Data =  7768...`, and testing only
                  // "do the first two chars parse as hex" accepted D,a and read the LABEL as byte 0xDA --
                  // adding one phantom byte to every key (case 1 read 21 of 20, case 2 read 5 of 4).
                  // Third form of the same hazard today: an ASCII gutter, then English prose, now a
                  // FIELD LABEL that happens to spell hex.
                  // STRUCTURAL DISCRIMINATOR: a continuation line has NO '='; every label line has one.
                  var eqfound: i64 = 0
                  var sc: i64 = p
                  var de: i64 = 0
                  while de == 0 {
                      if sc >= n { de = 1 }
                      else { if b[sc] == (10 as u8) { de = 1 }
                      else { if b[sc] == (61 as u8) { eqfound = 1; de = 1 } else { sc = sc + 1 } } }
                  }
                  if eqfound == 1 { done = 1 }
                  else {
                      if p + 1 >= n { done = 1 }
                      else {
                          if hexval(b[p] as i64) < 0 { done = 1 }
                          else { if hexval(b[p + 1] as i64) < 0 { done = 1 } }
                      }
                  }
              }
              else {
                  let h1: i64 = hexval(b[p] as i64)
                  let h2: i64 = hexval(b[p + 1] as i64)
                  if h1 < 0 { done = 1 }
                  else { if h2 < 0 { done = 1 }
                  else {
                      out[got] = ((h1 * 16) + h2) as u8
                      got = got + 1
                      p = p + 2
                  } }
              }
          }
      }
  }
  return got
156}
157
158// Find the next "Key" FIELD LABEL at line start. Verified against the raw bytes of RFC 4231:
159//   line 192 "   Key =          0b0b..."   <- cases 1,2,4,5,6,7
160//   line 251 "   Key            aaaa..."   <- CASE 3: no '=' at all
161//   line 177 "   Keys, data, and digests..." <- PROSE. "Key" is a PREFIX of "Keys".
162// Anchoring on "\n   Key" alone matched that prose line and drove the parse to ZERO cases. So the label is
163// accepted only when the character AFTER "Key" is a SPACE or '=' -- which "Keys" fails on 's'.
164// ★A prefix match is not a token match. Every anchor in this file is now checked against the byte AFTER it.
165func find_key_label(b: *u8, n: i64, from: i64) -> i64 {
  var p: i64 = from
  var done: i64 = 0
  while done == 0 {
      let h: i64 = findfrom(b, n, "\n   Key" as *u8, p)
      if h < 0 { return 0 - 1 }
      let c: i64 = b[h + 7] as i64
      if c == 32 { return h }
      if c == 61 { return h }
      p = h + 7
  }
  return 0 - 1
177}
178func main() -> i64 {
  w("nx_sha1_extvec_gate -- SHA-1 vs RFC 3174 resultarray, READ FROM THE FETCHED DOCUMENT\n" as *u8)
  w("  PURPOSE: ADJUDICATE the unresolved RFC 2202 HMAC-SHA-1 RED. This gate shares NOTHING with the\n" as *u8)
  w("  HMAC layer, so it can tell a broken hash core from a broken HMAC wrapper or reader.\n" as *u8)
  let lp: *i64 = sys_mmap(16) as *i64
  lp[0] = 0
  let b: *u8 = sys_read_file("knowledge/extvec/rfc3174.txt\x00" as *u8, lp)
  if lp[0] <= 0 { w("RED: fetched vector file absent.\n" as *u8); return 1 }
  let ctx: *u8 = sys_mmap(1024)
  let dg: *u8 = sys_mmap(64)
  nx_sha256_one_shot(b, lp[0], ctx, dg)
  let hx: *u8 = sys_mmap(80)
  var i: i64 = 0
  while i < 32 { hx[i*2] = hexnib(((dg[i] as i64)/16)&15) as u8; hx[i*2+1] = hexnib((dg[i] as i64)&15) as u8; i = i + 1 }
  let wnt: *u8 = "aa0ca4ef21609e64897980488c5555109266c29cd814fe97680c916566d29332\x00" as *u8
  var pin: i64 = 1
  i = 0
  while i < 64 { if hx[i] != wnt[i] { pin = 0 } i = i + 1 }
  w("  acquisition digest: " as *u8); wb(hx, 64); w("\n" as *u8)
  if pin == 0 { w("RED: PIN FAILED.\n" as *u8); return 1 }
  w("  PIN OK -- bytes match the digest computed in-process at the socket\n" as *u8)
199
  // The message: #define TEST1   "abc"   -- read from the document, not typed.
  let t1: i64 = findfrom(b, lp[0], "#define TEST1" as *u8, 0)
  if t1 < 0 { w("RED: no '#define TEST1'\n" as *u8); return 1 }
  let q1: i64 = findfrom(b, lp[0], "\"" as *u8, t1)
  if q1 < 0 { w("RED: no quote after TEST1\n" as *u8); return 1 }
  let msg: *u8 = sys_mmap(64)
  var mn: i64 = 0
  var p: i64 = q1 + 1
  var d: i64 = 0
  while d == 0 {
      if p >= lp[0] { d = 1 }
      else { if b[p] == (34 as u8) { d = 1 } else { msg[mn] = b[p]; mn = mn + 1; p = p + 1 } }
  }
  if mn != 3 { w("RED: TEST1 parsed " as *u8); nn(mn); w(" bytes, expected 3 -- the READER.\n" as *u8); return 1 }
214
  // The answer: the FIRST entry of char *resultarray[4]. Anchored on the array declaration so no
  // intermediate or unrelated hex block can be mistaken for it.
  let ra: i64 = findfrom(b, lp[0], "char *resultarray" as *u8, 0)
  if ra < 0 { w("RED: no 'char *resultarray'\n" as *u8); return 1 }
  let rq: i64 = findfrom(b, lp[0], "\"" as *u8, ra)
  if rq < 0 { w("RED: no quoted result\n" as *u8); return 1 }
  let exp: *u8 = sys_mmap(64)
  let en: i64 = parserun(b, lp[0], rq + 1, exp, 20)
  if en != 20 { w("RED: result parsed " as *u8); nn(en); w(" bytes, expected 20 -- the READER.\n" as *u8); return 1 }
  w("  parsed TEST1 (3 bytes) and resultarray[0] (20 bytes) from the document\n" as *u8)
225
  let got: *u8 = sys_mmap(64)
  sha1(msg, mn, got)
228
  var pass: i64 = 0
  var fail: i64 = 0
  var same: i64 = 1
  i = 0
  while i < 20 { if got[i] != exp[i] { same = 0 } i = i + 1 }
  if same == 1 {
      pass = pass + 1
      w("  PASS: sha1(TEST1) == resultarray[0]\n" as *u8)
      w("  ==> THE HASH CORE IS CORRECT. The RFC 2202 HMAC-SHA-1 RED is therefore NOT the hash;\n" as *u8)
      w("      it is the HMAC wrapper or my RFC 2202 reader. Suspect list narrowed by elimination.\n" as *u8)
  } else {
      fail = fail + 1
      w("  FAIL: sha1(TEST1) != resultarray[0]\n" as *u8)
      w("  ==> THE HASH CORE ITSELF DISAGREES WITH RFC 3174. Both HMAC-SHA-1 paths are downstream\n" as *u8)
      w("      victims, and the defect is in sha1(), not in the HMAC layer.\n" as *u8)
      w("  expected[0..3]=" as *u8); nn(exp[0] as i64); w(" " as *u8); nn(exp[1] as i64)
      w("  got[0..3]=" as *u8); nn(got[0] as i64); w(" " as *u8); nn(got[1] as i64); w("\n" as *u8)
  }
247
  // NEGATIVE CONTROL
  let m2: *u8 = "abd\x00" as *u8
  sha1(m2, 3, got)
  var same2: i64 = 1
  i = 0
  while i < 20 { if got[i] != exp[i] { same2 = 0 } i = i + 1 }
  if same2 == 0 { pass = pass + 1; w("  PASS NEG: \"abd\" differs (the check can fail)\n" as *u8) }
  else { fail = fail + 1; w("  FAIL NEG\n" as *u8) }
256
  w("\n  refsrc=https://www.rfc-editor.org/rfc/rfc3174.txt\n" as *u8)
  w("  refsrcdig=" as *u8); wb(hx, 64); w("\n" as *u8)
  w("  ref=RFC3174-resultarray0   gate=nx_sha1_extvec_gate\n" as *u8)
  w("nx_sha1_extvec_gate: pass=" as *u8); nn(pass); w(" fail=" as *u8); nn(fail)
  // MIGRATED onto nx_gate_verdict by nx_gate_dry_apply (D001, minimal form): every check
  // row above is untouched, so the PASS/FAIL vector cannot change; only the hand-rolled
  // verdict emission is replaced by the ONE shared base class. Proven by nx_gate_migrate verify.
  let ctr__dry: *i64 = gv_ctr()
  ctr__dry[0] = pass
  ctr__dry[1] = pass + fail
  let rc__dry: i64 = gv_verdict("SHA1-EXTVEC-GATE" as *u8, ctr__dry, "teeth unchanged; verdict emission migrated onto the shared base class" as *u8)
  sys_exit(rc__dry)
  return rc__dry
270}
nx_sha1_extvec_gate.nx source

`nx_sha1_extvec_gate.nx` source
