x86/entry: Avoid very early RET
authorPeter Zijlstra <peterz@infradead.org>
Tue, 14 Jun 2022 21:15:46 +0000 (23:15 +0200)
committerBorislav Petkov <bp@suse.de>
Mon, 27 Jun 2022 08:33:58 +0000 (10:33 +0200)
commit7c81c0c9210c9bfab2bae76aab2999de5bad27db
tree04f9f4c9c7289fb9b8fbe3ab9927c01712f74c2b
parentaa3d480315ba6c3025a60958e1981072ea37c3df
x86/entry: Avoid very early RET

Commit

  ee774dac0da1 ("x86/entry: Move PUSH_AND_CLEAR_REGS out of error_entry()")

manages to introduce a CALL/RET pair that is before SWITCH_TO_KERNEL_CR3,
which means it is before RETBleed can be mitigated.

Revert to an earlier version of the commit in Fixes. Down side is that
this will bloat .text size somewhat. The alternative is fully reverting
it.

The purpose of this patch was to allow migrating error_entry() to C,
including the whole of kPTI. Much care needs to be taken moving that
forward to not re-introduce this problem of early RETs.

Fixes: ee774dac0da1 ("x86/entry: Move PUSH_AND_CLEAR_REGS out of error_entry()")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Borislav Petkov <bp@suse.de>
arch/x86/entry/entry_64.S