The x86 architecture is the weirdo, part 2

Re:The x86 architecture is the weirdo, part 2 by rpasquay on Friday April 22, 2022 @07:54AM (#31120496)

I did not understand why on x86 Windows/MSVC cannot implement exceptions the same way as it was done for other architectures. I read that on x64 a different approach is used ("table based" - sounds to me like the "unwind code" approach mentioned in the article). Is this not possible no x86 for some reason ?

[ Parent | on HN ]

Re:The x86 architecture is the weirdo, part 2 by thedeadfish on Friday April 22, 2022 @08:07AM (#31120588)

Table based exceptions only work if everyone plays ball. Its not practical to introduce them after the fact because they will not work with any existing code including OS callbacks. AMD64 can use them because they were there from the beginning.

[ Parent | on HN ]

Re:The x86 architecture is the weirdo, part 2 by gpderetta on Friday April 22, 2022 @08:50AM (#31120982)

On linux they are also used for x86-32. The previous sjlj ABI was broken about 20 years ago when the Itanium ABI was introduced.

[ Parent | on HN ]

Re:The x86 architecture is the weirdo, part 2 by nikbackm on Friday April 22, 2022 @08:03AM (#31120559)

The answer seems to be in the follow-up comments to the post.

[ Parent | on HN ]