MD5与SHA256填充步骤是否一致？C语言实现问题咨询

Hey there! Let's break down why your MD5 works but SHA256 is misbehaving—your hunch about padding differences is spot-on. Here's the key breakdown and fix:

Core Differences Between MD5 and SHA256 Padding #

Both algorithms follow a similar initial padding pattern, but the length field at the end is drastically different:

• Shared Step: First append a single 0x80 byte (binary 10000000) to the input message, then fill with 0x00 bytes until the total length is aligned for the final block. Both use 512-bit (64-byte) blocks.
• Critical Difference: Length Field
  • MD5: Adds an 8-byte (64-bit) little-endian value representing the original message length in bits.
  • SHA256: Adds a 16-byte (128-bit) big-endian value representing the original message length in bits.

Your "toto" Test Case Analysis #

For the input "toto" (4 bytes = 32 bits):

• MD5 Padding (Correct):
  You appended 1 byte 0x80, 51 bytes 0x00, then the 8-byte little-endian length (0x20 0x00 0x00 0x00 0x00 0x00 0x00 0x00). This totals 64 bytes, which fits MD5's requirements perfectly—hence your working result.
• SHA256 Padding (Incorrect):
  You reused the exact same padding as MD5, which means:
  1. You only added an 8-byte length field instead of the required 16 bytes.
  2. The length field is in little-endian instead of big-endian.
     This throws off the entire block structure SHA256 expects, leading to your invalid hash result.

Fix Steps #

1. Adjust Padding Calculation:
   For SHA256, after adding 0x80, fill with 0x00 bytes until the total length is 64 bytes minus 16 bytes (to leave space for the 128-bit length field). For "toto", that means 43 bytes of 0x00 (4 original + 1 0x80 + 43 0x00 = 48 bytes; 48 + 16 = 64).
2. Implement the 128-bit Big-Endian Length Field:
   For 32 bits of input length, the SHA256 length field is 12 bytes of 0x00 followed by 4 bytes 0x00 0x00 0x00 0x20 (big-endian representation of 32).
3. Validate the Correct Hash:
   The standard SHA256 hash for "toto" is c7911af3adbd12a035b289556d96470a75d486555177e0e2e67d71f7a0a26660—your corrected implementation should match this.

内容的提问来源于stack exchange，提问作者Hugo Eynard