On Wed, Feb 13, 2019 at 07:56:26AM +0100, Martin Husemann wrote: > On Tue, Feb 12, 2019 at 09:20:23PM -0000, Christos Zoulas wrote: > > Well, regardless of what the right permissions of .eh_frame are, > > we could just nuke the code and default to the "new" behavior... > > We can then put back the varasm.c change... > > This of course needs to be evaluated carefully. > > Why would anyone modify the eh frame list at runtime? Or is any other > writable data placed there on purpose? I didn't see any, and if - wouldn't > it better go into a separate section?
.eh_frame was historically writeable on MIPS because they f**ked up the relocations: inter-section relocations were not allowed. The "fix" for that was to use indirect pointer encodings and allow ld to optimize the .eh_frame encoding. Joerg
