Skip to content

Fix unsigned underflow in SimpleString::subString() on empty strings - #1889

Merged
basvodde merged 1 commit into
cpputest:masterfrom
94xhn:fix/substring-empty-string-underflow
Jul 30, 2026
Merged

Fix unsigned underflow in SimpleString::subString() on empty strings#1889
basvodde merged 1 commit into
cpputest:masterfrom
94xhn:fix/substring-empty-string-underflow

Conversation

@94xhn

@94xhn 94xhn commented Jul 14, 2026

Copy link
Copy Markdown
Contributor

SimpleString::subString(size_t beginPos, size_t amount) rejects an out-of-range beginPos with:

if (beginPos > size()-1) return "";

size() returns size_t (unsigned). When called on an empty string, size() is 0, so size()-1 underflows to SIZE_MAX. The check beginPos > SIZE_MAX is then false for essentially any beginPos, so the bounds check is silently defeated and the function falls through to:

SimpleString newString = getBuffer() + beginPos;

which reads out of bounds of the (1-byte, null-terminator-only) buffer of an empty SimpleString.

Repro (confirmed with a guard-page allocator that places an empty SimpleString's buffer directly against a PAGE_NOACCESS page):

SimpleString empty("");
SimpleString result = empty.subString(1, 1); // reads 1 byte past the buffer

This reliably crashes before the fix and returns "" after it.

Fix: change the check to beginPos >= size(), which is mathematically identical to the original for every size() >= 1 (so every existing non-empty-string call site is byte-for-byte unaffected) but no longer underflows when size() == 0.

Verified:

  • Compiled a minimal repro harness against the unmodified sources: empty.subString(1, 1) segfaults before the fix.
  • Recompiled the same harness against the fixed source: no crash, subString(1, 1) on an empty string correctly returns "".
  • Rebuilt and ran the existing SimpleString test group (AllTests -g SimpleString, 130 tests / 238 checks) against both the original and fixed source — identical OK results before and after, so this is a pure bug fix with no behavior change for any existing test or non-empty-string caller.

subString(beginPos, amount) rejected out-of-range beginPos with
`if (beginPos > size()-1) return "";`. Since size() returns size_t,
calling this on an empty string (size() == 0) makes size()-1 wrap
around to SIZE_MAX, so the bounds check is defeated for any beginPos
and the function falls through to an out-of-bounds read of the
string's internal buffer.

Use `beginPos >= size()` instead, which is mathematically equivalent
to the original check for every size() >= 1 and additionally handles
size() == 0 correctly.
@basvodde

Copy link
Copy Markdown
Member

Thanks!

@basvodde
basvodde merged commit a5270d6 into cpputest:master Jul 30, 2026
30 of 65 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants