Discussion
I remember when the *context functions in posix were depercated and many folks had were crazy mad they were. And the funny thing is the *context functions are even more security issues than the setjmp/longjmp functions will ever be.
Just goes to show there is no keeping folks happy.
This was over 15 years ago now when I remember these complaints. Yes for new stuff pointing them at libucontext is the correct approach.
I was just pointing out the whole PHP with setjmp/longjmp is being used as an excuse to diss C now and when folks did remove functionality in the past of security holes like setjmp/longjmp people complained hard.
@pinskia @ariadne I am so not-happy with the narrative of "longjmp is insecure" being pushed by ignorant C-haters who don't even know what it does (and who often think it's like a goto) in response to this.
The exact same thing would have happened if PHP had bubbled-up the error condition but not actually handled it properly.
I have not looked in detail but I'm pretty sure they could have fixed the problem without the flaky 4kloc patch or whatever it is just by making the code at the longjmp target *actually do the right thing*. And gotten a patch we could actually be confident in.
But instead, "longjmp is insecure" ideology got us this unauditable mess.
@lanodan Yes.
Looks like they were actually removed in POSIX-2008 in the end.
Install Fedizen
Get the full app experience
@pinskia just point them at libucontext and tell them to move on :p