rtc: ds1685: actually spin forever in poweroff error path

objtool reports the following warnings:

  drivers/rtc/rtc-ds1685.o: warning: objtool: ds1685_rtc_work_queue()+0x0: duplicate frame pointer save
  drivers/rtc/rtc-ds1685.o: warning: objtool: ds1685_rtc_work_queue()+0x3: duplicate frame pointer setup
  drivers/rtc/rtc-ds1685.o: warning: objtool: ds1685_rtc_work_queue()+0x0: frame pointer state mismatch

The warning message needs to be improved, but what it really means in
this case is that ds1685_rtc_poweroff() has a possible code path where
it can actually fall through to the next function in the object code,
ds1685_rtc_work_queue().

The bug is caused by the use of the unreachable() macro in a place which
is actually reachable.  That causes gcc to assume that the printk()
immediately before the unreachable() macro never returns, when in fact
it does.  So gcc places the printk() at the very end of the function's
object code.  When the printk() returns, the next function starts
executing.

The surrounding comment and printk message state that the code should
spin forever, which explains the unreachable() statement.  However the
actual spin code is missing.

Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
This commit is contained in:
Josh Poimboeuf 2016-03-07 09:03:02 -06:00 committed by Alexandre Belloni
parent d586126221
commit 361c6ed6b1

View file

@ -2161,6 +2161,7 @@ ds1685_rtc_poweroff(struct platform_device *pdev)
/* Check for valid RTC data, else, spin forever. */
if (unlikely(!pdev)) {
pr_emerg("platform device data not available, spinning forever ...\n");
while(1);
unreachable();
} else {
/* Get the rtc data. */