Consdiff: use lengths on inputs so they don't need NUL at the end

This is part of #27244, so that we can safely mmap consensus
documents.

5 files changed, 101 insertions, 52 deletions

diff --git a/src/test/bench.c b/src/test/bench.c
index 959d4374b1..9da1b46a1b 100644
--- a/src/test/bench.c
+++ b/src/test/bench.c
@@ -702,11 +702,13 @@ main(int argc, const char **argv)
       perror("X");
       return 1;
     }
+    size_t f1len = strlen(f1);
+    size_t f2len = strlen(f2);
     for (i = 0; i < N; ++i) {
-      char *diff = consensus_diff_generate(f1, f2);
+      char *diff = consensus_diff_generate(f1, f1len, f2, f2len);
       tor_free(diff);
     }
-    char *diff = consensus_diff_generate(f1, f2);
+    char *diff = consensus_diff_generate(f1, f1len, f2, f2len);
     printf("%s", diff);
     tor_free(f1);
     tor_free(f2);
diff --git a/src/test/fuzz/fuzz_diff.c b/src/test/fuzz/fuzz_diff.c
index 1079856fdb..8966856be2 100644
--- a/src/test/fuzz/fuzz_diff.c
+++ b/src/test/fuzz/fuzz_diff.c
@@ -10,9 +10,11 @@
 #include "test/fuzz/fuzzing.h"
 
 static int
-mock_consensus_compute_digest_(const char *c, consensus_digest_t *d)
+mock_consensus_compute_digest_(const char *c, size_t len,
+                               consensus_digest_t *d)
 {
   (void)c;
+  (void)len;
   memset(d->sha3_256, 3, sizeof(d->sha3_256));
   return 0;
 }
@@ -42,14 +44,14 @@ fuzz_main(const uint8_t *stdin_buf, size_t data_size)
   if (! separator)
     return 0;
   size_t c1_len = separator - stdin_buf;
-  char *c1 = tor_memdup_nulterm(stdin_buf, c1_len);
+  const char *c1 = (const char *)stdin_buf;
   size_t c2_len = data_size - c1_len - SEPLEN;
-  char *c2 = tor_memdup_nulterm(separator + SEPLEN, c2_len);
+  const char *c2 = (const char *)separator + SEPLEN;
 
-  char *c3 = consensus_diff_generate(c1, c2);
+  char *c3 = consensus_diff_generate(c1, c1_len, c2, c2_len);
 
   if (c3) {
-    char *c4 = consensus_diff_apply(c1, c3);
+    char *c4 = consensus_diff_apply(c1, c1_len, c3, strlen(c3));
     tor_assert(c4);
     if (strcmp(c2, c4)) {
       printf("%s\n", escaped(c1));
@@ -61,9 +63,6 @@ fuzz_main(const uint8_t *stdin_buf, size_t data_size)
     tor_free(c3);
     tor_free(c4);
   }
-  tor_free(c1);
-  tor_free(c2);
 
   return 0;
 }
-
diff --git a/src/test/fuzz/fuzz_diff_apply.c b/src/test/fuzz/fuzz_diff_apply.c
index 165d0e6126..9b25185225 100644
--- a/src/test/fuzz/fuzz_diff_apply.c
+++ b/src/test/fuzz/fuzz_diff_apply.c
@@ -10,9 +10,11 @@
 #include "test/fuzz/fuzzing.h"
 
 static int
-mock_consensus_compute_digest_(const char *c, consensus_digest_t *d)
+mock_consensus_compute_digest_(const char *c, size_t len,
+                               consensus_digest_t *d)
 {
   (void)c;
+  (void)len;
   memset(d->sha3_256, 3, sizeof(d->sha3_256));
   return 0;
 }
@@ -50,16 +52,13 @@ fuzz_main(const uint8_t *stdin_buf, size_t data_size)
   if (! separator)
     return 0;
   size_t c1_len = separator - stdin_buf;
-  char *c1 = tor_memdup_nulterm(stdin_buf, c1_len);
+  const char *c1 = (const char *)stdin_buf;
   size_t c2_len = data_size - c1_len - SEPLEN;
-  char *c2 = tor_memdup_nulterm(separator + SEPLEN, c2_len);
+  const char *c2 = (const char *)separator + SEPLEN;
 
-  char *c3 = consensus_diff_apply(c1, c2);
+  char *c3 = consensus_diff_apply(c1, c1_len, c2, c2_len);
 
-  tor_free(c1);
-  tor_free(c2);
   tor_free(c3);
 
   return 0;
 }
-
diff --git a/src/test/test_consdiff.c b/src/test/test_consdiff.c
index b836befd22..23e8f7167e 100644
--- a/src/test/test_consdiff.c
+++ b/src/test/test_consdiff.c
@@ -14,6 +14,39 @@
 #define tt_str_eq_line(a,b) \
   tt_assert(line_str_eq((b),(a)))
 
+static int
+consensus_split_lines_(smartlist_t *out, const char *s, memarea_t *area)
+{
+  size_t len = strlen(s);
+  return consensus_split_lines(out, s, len, area);
+}
+
+static int
+consensus_compute_digest_(const char *cons,
+                          consensus_digest_t *digest_out)
+{
+  size_t len = strlen(cons);
+  char *tmp = tor_memdup(cons, len);
+  // We use memdup here to ensure that the input is NOT nul-terminated.
+  // This makes it likelier for us to spot bugs.
+  int r = consensus_compute_digest(tmp, len, digest_out);
+  tor_free(tmp);
+  return r;
+}
+
+static int
+consensus_compute_digest_as_signed_(const char *cons,
+                                    consensus_digest_t *digest_out)
+{
+  size_t len = strlen(cons);
+  char *tmp = tor_memdup(cons, len);
+  // We use memdup here to ensure that the input is NOT nul-terminated.
+  // This makes it likelier for us to spot bugs.
+  int r = consensus_compute_digest_as_signed(tmp, len, digest_out);
+  tor_free(tmp);
+  return r;
+}
+
 static void
 test_consdiff_smartlist_slice(void *arg)
 {
@@ -58,7 +91,7 @@ test_consdiff_smartlist_slice_string_pos(void *arg)
 
   /* Create a regular smartlist. */
   (void)arg;
-  consensus_split_lines(sl, "a\nd\nc\na\nb\n", area);
+  consensus_split_lines_(sl, "a\nd\nc\na\nb\n", area);
 
   /* See that smartlist_slice_string_pos respects the bounds of the slice. */
   sls = smartlist_slice(sl, 2, 5);
@@ -87,8 +120,8 @@ test_consdiff_lcs_lengths(void *arg)
   int e_lengths2[] = { 0, 1, 1, 2, 3, 4 };
 
   (void)arg;
-  consensus_split_lines(sl1, "a\nb\nc\nd\ne\n", area);
-  consensus_split_lines(sl2, "a\nc\nd\ni\ne\n", area);
+  consensus_split_lines_(sl1, "a\nb\nc\nd\ne\n", area);
+  consensus_split_lines_(sl2, "a\nc\nd\ni\ne\n", area);
 
   sls1 = smartlist_slice(sl1, 0, -1);
   sls2 = smartlist_slice(sl2, 0, -1);
@@ -119,10 +152,10 @@ test_consdiff_trim_slices(void *arg)
   memarea_t *area = memarea_new();
 
   (void)arg;
-  consensus_split_lines(sl1, "a\nb\nb\nb\nd\n", area);
-  consensus_split_lines(sl2, "a\nc\nc\nc\nd\n", area);
-  consensus_split_lines(sl3, "a\nb\nb\nb\na\n", area);
-  consensus_split_lines(sl4, "c\nb\nb\nb\nc\n", area);
+  consensus_split_lines_(sl1, "a\nb\nb\nb\nd\n", area);
+  consensus_split_lines_(sl2, "a\nc\nc\nc\nd\n", area);
+  consensus_split_lines_(sl3, "a\nb\nb\nb\na\n", area);
+  consensus_split_lines_(sl4, "c\nb\nb\nb\nc\n", area);
   sls1 = smartlist_slice(sl1, 0, -1);
   sls2 = smartlist_slice(sl2, 0, -1);
   sls3 = smartlist_slice(sl3, 0, -1);
@@ -165,8 +198,8 @@ test_consdiff_set_changed(void *arg)
   memarea_t *area = memarea_new();
 
   (void)arg;
-  consensus_split_lines(sl1, "a\nb\na\na\n", area);
-  consensus_split_lines(sl2, "a\na\na\na\n", area);
+  consensus_split_lines_(sl1, "a\nb\na\na\n", area);
+  consensus_split_lines_(sl2, "a\na\na\na\n", area);
 
   /* Length of sls1 is 0. */
   sls1 = smartlist_slice(sl1, 0, 0);
@@ -240,8 +273,8 @@ test_consdiff_calc_changes(void *arg)
   memarea_t *area = memarea_new();
 
   (void)arg;
-  consensus_split_lines(sl1, "a\na\na\na\n", area);
-  consensus_split_lines(sl2, "a\na\na\na\n", area);
+  consensus_split_lines_(sl1, "a\na\na\na\n", area);
+  consensus_split_lines_(sl2, "a\na\na\na\n", area);
 
   sls1 = smartlist_slice(sl1, 0, -1);
   sls2 = smartlist_slice(sl2, 0, -1);
@@ -259,7 +292,7 @@ test_consdiff_calc_changes(void *arg)
   tt_assert(!bitarray_is_set(changed2, 3));
 
   smartlist_clear(sl2);
-  consensus_split_lines(sl2, "a\nb\na\nb\n", area);
+  consensus_split_lines_(sl2, "a\nb\na\nb\n", area);
   tor_free(sls1);
   tor_free(sls2);
   sls1 = smartlist_slice(sl1, 0, -1);
@@ -282,7 +315,7 @@ test_consdiff_calc_changes(void *arg)
   bitarray_clear(changed1, 3);
 
   smartlist_clear(sl2);
-  consensus_split_lines(sl2, "b\nb\nb\nb\n", area);
+  consensus_split_lines_(sl2, "b\nb\nb\nb\n", area);
   tor_free(sls1);
   tor_free(sls2);
   sls1 = smartlist_slice(sl1, 0, -1);
@@ -610,8 +643,8 @@ test_consdiff_gen_ed_diff(void *arg)
   /* Test 'a', 'c' and 'd' together. See that it is done in reverse order. */
   smartlist_clear(cons1);
   smartlist_clear(cons2);
-  consensus_split_lines(cons1, "A\nB\nC\nD\nE\n", area);
-  consensus_split_lines(cons2, "A\nC\nO\nE\nU\n", area);
+  consensus_split_lines_(cons1, "A\nB\nC\nD\nE\n", area);
+  consensus_split_lines_(cons2, "A\nC\nO\nE\nU\n", area);
   diff = gen_ed_diff(cons1, cons2, area);
   tt_ptr_op(NULL, OP_NE, diff);
   tt_int_op(7, OP_EQ, smartlist_len(diff));
@@ -627,8 +660,8 @@ test_consdiff_gen_ed_diff(void *arg)
 
   smartlist_clear(cons1);
   smartlist_clear(cons2);
-  consensus_split_lines(cons1, "B\n", area);
-  consensus_split_lines(cons2, "A\nB\n", area);
+  consensus_split_lines_(cons1, "B\n", area);
+  consensus_split_lines_(cons2, "A\nB\n", area);
   diff = gen_ed_diff(cons1, cons2, area);
   tt_ptr_op(NULL, OP_NE, diff);
   tt_int_op(3, OP_EQ, smartlist_len(diff));
@@ -656,7 +689,7 @@ test_consdiff_apply_ed_diff(void *arg)
   diff = smartlist_new();
   setup_capture_of_logs(LOG_WARN);
 
-  consensus_split_lines(cons1, "A\nB\nC\nD\nE\n", area);
+  consensus_split_lines_(cons1, "A\nB\nC\nD\nE\n", area);
 
   /* Command without range. */
   smartlist_add_linecpy(diff, area, "a");
@@ -829,7 +862,7 @@ test_consdiff_apply_ed_diff(void *arg)
   smartlist_clear(diff);
 
   /* Test appending text, 'a'. */
-  consensus_split_lines(diff, "3a\nU\nO\n.\n0a\nV\n.\n", area);
+  consensus_split_lines_(diff, "3a\nU\nO\n.\n0a\nV\n.\n", area);
   cons2 = apply_ed_diff(cons1, diff, 0);
   tt_ptr_op(NULL, OP_NE, cons2);
   tt_int_op(8, OP_EQ, smartlist_len(cons2));
@@ -846,7 +879,7 @@ test_consdiff_apply_ed_diff(void *arg)
   smartlist_free(cons2);
 
   /* Test deleting text, 'd'. */
-  consensus_split_lines(diff, "4d\n1,2d\n", area);
+  consensus_split_lines_(diff, "4d\n1,2d\n", area);
   cons2 = apply_ed_diff(cons1, diff, 0);
   tt_ptr_op(NULL, OP_NE, cons2);
   tt_int_op(2, OP_EQ, smartlist_len(cons2));
@@ -857,7 +890,7 @@ test_consdiff_apply_ed_diff(void *arg)
   smartlist_free(cons2);
 
   /* Test changing text, 'c'. */
-  consensus_split_lines(diff, "4c\nT\nX\n.\n1,2c\nM\n.\n", area);
+  consensus_split_lines_(diff, "4c\nT\nX\n.\n1,2c\nM\n.\n", area);
   cons2 = apply_ed_diff(cons1, diff, 0);
   tt_ptr_op(NULL, OP_NE, cons2);
   tt_int_op(5, OP_EQ, smartlist_len(cons2));
@@ -871,7 +904,7 @@ test_consdiff_apply_ed_diff(void *arg)
   smartlist_free(cons2);
 
   /* Test 'a', 'd' and 'c' together. */
-  consensus_split_lines(diff, "4c\nT\nX\n.\n2d\n0a\nM\n.\n", area);
+  consensus_split_lines_(diff, "4c\nT\nX\n.\n2d\n0a\nM\n.\n", area);
   cons2 = apply_ed_diff(cons1, diff, 0);
   tt_ptr_op(NULL, OP_NE, cons2);
   tt_int_op(6, OP_EQ, smartlist_len(cons2));
@@ -918,12 +951,12 @@ test_consdiff_gen_diff(void *arg)
       );
 
   tt_int_op(0, OP_EQ,
-      consensus_compute_digest_as_signed(cons1_str, &digests1));
+      consensus_compute_digest_as_signed_(cons1_str, &digests1));
   tt_int_op(0, OP_EQ,
-      consensus_compute_digest(cons2_str, &digests2));
+      consensus_compute_digest_(cons2_str, &digests2));
 
-  consensus_split_lines(cons1, cons1_str, area);
-  consensus_split_lines(cons2, cons2_str, area);
+  consensus_split_lines_(cons1, cons1_str, area);
+  consensus_split_lines_(cons2, cons2_str, area);
 
   diff = consdiff_gen_diff(cons1, cons2, &digests1, &digests2, area);
   tt_ptr_op(NULL, OP_EQ, diff);
@@ -937,9 +970,9 @@ test_consdiff_gen_diff(void *arg)
       "directory-signature foo bar\nbar\n"
       );
   tt_int_op(0, OP_EQ,
-      consensus_compute_digest_as_signed(cons1_str, &digests1));
+      consensus_compute_digest_as_signed_(cons1_str, &digests1));
   smartlist_clear(cons1);
-  consensus_split_lines(cons1, cons1_str, area);
+  consensus_split_lines_(cons1, cons1_str, area);
   diff = consdiff_gen_diff(cons1, cons2, &digests1, &digests2, area);
   tt_ptr_op(NULL, OP_NE, diff);
   tt_int_op(11, OP_EQ, smartlist_len(diff));
@@ -991,8 +1024,8 @@ test_consdiff_apply_diff(void *arg)
       "directory-signature foo bar\nbar\n"
       );
   tt_int_op(0, OP_EQ,
-      consensus_compute_digest(cons1_str, &digests1));
-  consensus_split_lines(cons1, cons1_str, area);
+      consensus_compute_digest_(cons1_str, &digests1));
+  consensus_split_lines_(cons1, cons1_str, area);
 
   /* diff doesn't have enough lines. */
   cons2 = consdiff_apply_diff(cons1, diff, &digests1);
@@ -1182,4 +1215,3 @@ struct testcase_t consdiff_tests[] = {
   CONSDIFF_LEGACY(apply_diff),
   END_OF_TESTCASES
 };
-
diff --git a/src/test/test_consdiffmgr.c b/src/test/test_consdiffmgr.c
index 6c0601b504..dc4fea7f6f 100644
--- a/src/test/test_consdiffmgr.c
+++ b/src/test/test_consdiffmgr.c
@@ -21,6 +21,21 @@
 #include "test/test.h"
 #include "test/log_test_helpers.h"
 
+static char *
+consensus_diff_apply_(const char *c, const char *d)
+{
+  size_t c_len = strlen(c);
+  size_t d_len = strlen(d);
+  // We use memdup here to ensure that the input is NOT nul-terminated.
+  // This makes it likelier for us to spot bugs.
+  char *c_tmp = tor_memdup(c, c_len);
+  char *d_tmp = tor_memdup(d, d_len);
+  char *result = consensus_diff_apply(c_tmp, c_len, d_tmp, d_len);
+  tor_free(c_tmp);
+  tor_free(d_tmp);
+  return result;
+}
+
 // ============================== Setup/teardown the consdiffmgr
 // These functions get run before/after each test in this module
 
@@ -153,7 +168,8 @@ lookup_diff_from(consensus_cache_entry_t **out,
                  const char *str1)
 {
   uint8_t digest[DIGEST256_LEN];
-  if (router_get_networkstatus_v3_sha3_as_signed(digest, str1)<0) {
+  if (router_get_networkstatus_v3_sha3_as_signed(digest,
+                                                 str1, strlen(str1))<0) {
     TT_FAIL(("Unable to compute sha3-as-signed"));
     return CONSDIFF_NOT_FOUND;
   }
@@ -181,7 +197,7 @@ lookup_apply_and_verify_diff(consensus_flavor_t flav,
   if (diff_string == NULL || r < 0)
     return -1;
 
-  char *applied = consensus_diff_apply(str1, diff_string);
+  char *applied = consensus_diff_apply_(str1, diff_string);
   tor_free(diff_string);
   if (applied == NULL)
     return -1;
@@ -370,7 +386,8 @@ test_consdiffmgr_make_diffs(void *arg)
   ns = fake_ns_new(FLAV_MICRODESC, now-3600);
   md_ns_body = fake_ns_body_new(FLAV_MICRODESC, now-3600);
   r = consdiffmgr_add_consensus(md_ns_body, ns);
-  router_get_networkstatus_v3_sha3_as_signed(md_ns_sha3, md_ns_body);
+  router_get_networkstatus_v3_sha3_as_signed(md_ns_sha3, md_ns_body,
+                                             strlen(md_ns_body));
   networkstatus_vote_free(ns);
   tt_int_op(r, OP_EQ, 0);
 
@@ -414,7 +431,7 @@ test_consdiffmgr_make_diffs(void *arg)
   r = consensus_cache_entry_get_body(diff, &diff_body, &diff_size);
   tt_int_op(r, OP_EQ, 0);
   diff_text = tor_memdup_nulterm(diff_body, diff_size);
-  applied = consensus_diff_apply(md_ns_body, diff_text);
+  applied = consensus_diff_apply_(md_ns_body, diff_text);
   tt_assert(applied);
   tt_str_op(applied, OP_EQ, md_ns_body_2);