Replace buf.Read with io.ReadFull to prevent partial read failures

[oskarszoon]

What Changed

Fix critical deserialization bug in go-subtree where buf.Read() calls could return partial reads, causing "unable to read conflicting node X" errors.

Problem

The subtree deserialization code used buf.Read() in multiple places, checking if the returned byte count matched the expected amount:

if n, err := buf.Read(st.ConflictingNodes[i][:]); err != nil || n != 32 {
    return fmt.Errorf("unable to read conflicting node %d: %w", i, err)
}

However, bufio.Reader.Read() is allowed to return fewer bytes than requested even when more data is available. When the internal buffer had only 8 bytes remaining (common at 32KB buffer boundaries), it would return those 8
bytes immediately instead of the requested 32, causing the error check to fail.

This manifested as consistent failures at the same conflicting node index (e.g., 1376) across multiple servers because they all had the same buffer size and file layout.

Solution

Replaced all buf.Read() calls with io.ReadFull(), which guarantees reading the full requested amount or returning an error. Also:

• Removed the unnecessary ReadBytes() helper function that was a poorly-implemented reinvention of io.ReadFull()
• Cleaned up commented-out io.ReadFull calls that showed previous developers recognized the issue
• Removed unused variables

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud