On Mon, Mar 10, 2025 at 07:49:56AM -0700, H. Peter Anvin wrote: > >> Alternatively, you can co-opt the gcc BR I already filed on this and > >> argue there that there are new reasons to support the alternate > >> construct. > > I should probably clarify that this wasn't flippant, but a serious > request. > > If this works by accident on existing gcc, and works on clang, that is > a very good reason for making it the supported way of doing this going > forward for both compilers. Per-compiler hacks are nasty, and although > we are pretty good about coping with them in the kernel, some user > space app developer is guaranteed to get it wrong. > > Frame pointers are actually more relevant in user space because user > space tends to be compiled with a wider range of debug and > architecture options, and of course there is simply way more user > space code out there. I opened a gcc bug: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119279 -- Josh
|