Re:Writing a Self-Mutating x86_64 C Program (2013)
by jakobdabo on Friday January 11, 2019 @01:32PM
(#18885642)
Modern x86s may be Harvard architecture internally (with separate code and data L1 caches), but they still present themselves to the developer as classic Von Neuman, which is easier to program.
[ Parent
| on HN ]
Re:Writing a Self-Mutating x86_64 C Program (2013)
by simias on Friday January 11, 2019 @01:51PM
(#18885821)
So did they keep it that way purely for historical reason and back-compat? Few people write self-modifying code (or even code loaders) these days, it seems like a tricky feature to implement in hardware for relatively little gain.
Re:Writing a Self-Mutating x86_64 C Program (2013)
by jandrese on Friday January 11, 2019 @02:26PM
(#18886089)
Depends if the self-modifying code in question is stuff like kernel drivers for popular hardware on widely used operating systems. The x86 manufacturers do bend over backwards for compatibility. You can run 16 bit code on the latest i9 processor if you like, including MSDOS if your motherboard still supports legacy boot modes. That is a remarkable level of backwards compatibility.