https://bugs.kde.org/show_bug.cgi?id=374940
--- Comment #6 from Julian Seward <jsew...@acm.org> --- (In reply to Matthias Urlichs from comment #4) > * the kernel needs to handle negative %esp offsets all the time because when > you grow the stack the kernel "sees" the old contents of %esp (necessarily, > otherwise it couldn't restart the faulting instruction). Page fault based stack extension only applies to the main thread's stack. Why does libev have to be non-conforming in this way? There are plenty of other packages that do high-performance thread stuff and I've never heard of any of them doing anything like this. -- You are receiving this mail because: You are watching all bug changes.
