When is segmentation a good idea?

Many people would say never. This is not true. Segmentation is a good idea for large assembly-based projects. It's not a good idea for C projects, since C wasn't constructed with segmentation in mind.

Here is a list of the advantages (compared to flat memory model)

• Major data structures can be protected (high level of protection)
• 16-bit offsets can be used instead of 32-bit in many cases (smaller executables)
• 16-bit near calls can be used instead of 32-bit near calls (smaller executables)
• Data segments cannot be executed (security)
• Stack corruption is likely to protection fault in the module causing the error (easier to debug)
• Returns to invalid code is caught at return instruction (easier to debug)

Here is a list of disadvantages (compared to flat memory model)

• Loading selectors takes time. In a assembler project, this isn't very common, so it doesn't affect performance too much.

Objections

None registered this far ;-)

Disagree?

Send a mail. If it's relevant, I will put it under objections.