Oh cool, allocations made by apps launched from the Finder on macOS are *never freed* due to a launchd bug. Yes, it’s as bad as it sounds. https://bugs.chromium.org/p/chromium/issues/detail?id=861939#c7 …
-
-
Hmm, maybe it’s intended, because Xcode forcibly disables the nano allocator. But that sure seems like questionable behavior: no pages ever returned to the OS unless you add an undocumented flag to your Info.plist?
Show this threadThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
libsystem_malloc is scary. Tried malloc interposition on macOS. There is an official (essentially broken) API. Googled many failed attempts at osx malloc interposition. Before interposer is run libsystem has allocated and there are several hardwired zones. ASan uses dyld hacks.
-
… was trying to control libsystem_malloc and mmap address space. I successfully interposed several of the mach vmmap APIs, enough to get libc apps working with control over the address of the first allocation; source (extracted from my app) is here https://github.com/michaeljclark/libSystem-mmap …
End of conversation
New conversation -
-
-
huh is it possible something like this could happen on linux? I've seen similar unexplainable behavior with firefox and (to a lesser extent) chrome on linux, where the processes are unaware of their own memory usage, unlike the OS
-
I think we have complete control over malloc/free/etc. on Linux unless we're preempted with LD_PRELOAD? But also, Firefox will use MADV_FREE instead of MADV_DONTNEED if it's defined on the build host (so, not Mozilla's builds, but possibly downstream builds).
- 1 more reply
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.