https://sourceware.org/bugzilla/show_bug.cgi?id=32713
--- Comment #4 from Brian Mak <makb at juniper dot net> --- (In reply to Mark Wielaard from comment #2) > See also this kernel thread: > https://lore.kernel.org/all/39fc2866-dff3-43c9-9d40-e8ff30a21...@juniper.net/ > Looks like the kernel people believe this in "in spec" so doesn't really > break user space handling. So we'll have to figure out to work around it > somehow. Per that reply, we are actually out of spec. I'm working on a patch to bring us back into spec, while still maintaining a sorted VMA size ordering. Meanwhile, a separate patch is being worked up to gate the VMA sorting behind a sysctl. The part that is out of spec is the fact that the program headers are also written in VMA size order. Loadable segments must be specified in virtual memory ordering in the program header table, per the ELF specs: "Loadable segment entries in the program header table appear in ascending order, sorted on the p_vaddr member." Hope this clears things up. -- You are receiving this mail because: You are on the CC list for the bug.
