https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106763
--- Comment #4 from George Pee <georgepee at gmail dot com> --- Yes, it's possible that this isn't a compiler bug. I thought that it might be because the problem started showing up after upgrading the toolchain. I wasn't sure if the compiler was failing to emit some kind of alignment or fp setting code. I running on linux 4.9.118, I enabled CONFIG_DEBUG_USER=y and set user_debug=31 in the kernel cmdline. This is what the kernel reports. I'm starting to look through it. [ 51.337524] a.out (3638): undefined instruction: pc=00010778 [ 51.337536] CPU: 3 PID: 3638 Comm: a.out Tainted: P O 4.9.118 #2 [ 51.337547] task: 8572b000 task.stack: 8a002000 [ 51.337555] PC is at 0x10778 [ 51.337560] LR is at 0x60dc51b4 [ 51.337567] pc : [<00010778>] lr : [<60dc51b4>] psr: 60000010 [ 51.337567] sp : 72909c50 ip : 60dc51c0 fp : 72909c54 [ 51.337572] r10: 60ff5000 r9 : 00000000 r8 : 00000000 [ 51.337578] r7 : 00000000 r6 : 00010668 r5 : 00000000 r4 : 00003346 [ 51.337583] r3 : 00000000 r2 : 00000001 r1 : 00000000 r0 : 6ff59dd5 [ 51.337589] Flags: nZCv IRQs on FIQs on Mode USER_32 ISA ARM Segment user [ 51.337595] Control: 50c0383d Table: 3e0c406a DAC: 00000015 [ 51.337605] Code: eef81be7 eddf2b05 eec10ba2 eef37b60 (ee173990)
