objtool: Fix stack-swizzle for FRAME_POINTER=y
authorPeter Zijlstra <peterz@infradead.org>
Thu, 18 Feb 2021 16:14:10 +0000 (17:14 +0100)
committerThomas Gleixner <tglx@linutronix.de>
Mon, 22 Feb 2021 18:54:09 +0000 (19:54 +0100)
commit724c8a23d589d8a002d2e39633c2f9a5a429616f
tree623c789a1ed733adbc2d11b50449f08a7d95404b
parent3aac798a917be3b8f2f647b834bb06bf2f8df4f1
objtool: Fix stack-swizzle for FRAME_POINTER=y

When objtool encounters the stack-swizzle:

mov %rsp, (%[tos])
mov %[tos], %rsp
...
pop %rsp

Inside a FRAME_POINTER=y build, things go a little screwy because
clearly we're not adjusting the cfa->base. This then results in the
pop %rsp not being detected as a restore of cfa->base so it will turn
into a regular POP and offset the stack, resulting in:

  kernel/softirq.o: warning: objtool: do_softirq()+0xdb: return with modified stack frame

Therefore, have "mov %[tos], %rsp" act like a PUSH (it sorta is
anyway) to balance the things out. We're not too concerned with the
actual stack_size for frame-pointer builds, since we don't generate
ORC data for them anyway.

Fixes: aafeb14e9da2 ("objtool: Support stack-swizzle")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Link: https://lkml.kernel.org/r/YC6UC+rc9KKmQrkd@hirez.programming.kicks-ass.net
tools/objtool/check.c