Gave a talk y'day trying to convince people that using ASAN is easy - yet the linux kernel just broke it... https://bugzilla.kernel.org/show_bug.cgi?id=196537 …
Quick read shows this is clearly ASAN's fault. You can't assume availability of a fixed address range. Must assign dynamically at start/load
-
-
They said in github they don't want dynamic cause it needs more code/cpu lol
-
Yeah I figured. They're wrong. The whole ASAN library-side stuff is unusably bad code, UB/vuln detection full of UB/vulns itself.
-
And since it's incompatible with musl and probably always will be, there's been talk of rewriting the ASAN lib side to be clean & portable..
-
can you and
@kayseesee discuss this and try to find a viable solution everyone's happy with? -
-
I'm not sure if I have anything of immediate value to offer there except "assuming you can use a fixed address is a bug".
-
I interpreted
@hanno's suggestion more as an idea about major project direction & making ASAN safe for production use (& maybe musl-compat).
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.