On 09.12.20 15:02, Mark Rutland wrote:

On Wed, Dec 09, 2020 at 01:27:10PM +0000, Mark Rutland wrote:

quoted

On Sun, Nov 22, 2020 at 01:44:53PM -0800, Andy Lutomirski wrote:

quoted

On Sat, Nov 21, 2020 at 10:55 PM Jürgen Groß [off-list ref] wrote:

quoted

On 20.11.20 12:59, Peter Zijlstra wrote:

quoted

If someone were to write horrible code like:

       local_irq_disable();
       local_irq_save(flags);
       local_irq_enable();
       local_irq_restore(flags);

we'd be up some creek without a paddle... now I don't _think_ we have
genius code like that, but I'd feel saver if we can haz an assertion in
there somewhere...

quoted

I was just talking to Peter on IRC about implementing the same thing for
arm64, so could we put this in the generic irqflags code? IIUC we can
use raw_irqs_disabled() to do the check.

As this isn't really entry specific (and IIUC the cases this should
catch would break lockdep today), maybe we should add a new
DEBUG_IRQFLAGS for this, that DEBUG_LOCKDEP can also select?

Something like:

#define local_irq_restore(flags)                               \
        do {                                                    \
                if (!raw_irqs_disabled_flags(flags)) {          \
                        trace_hardirqs_on();                    \
                } else if (IS_ENABLED(CONFIG_DEBUG_IRQFLAGS) {  \
                        if (unlikely(raw_irqs_disabled())       \

Whoops; that should be !raw_irqs_disabled().

quoted

                                warn_bogus_irqrestore();        \
                }                                               \
                raw_local_irq_restore(flags);                   \
         } while (0)

... perhaps? (ignoring however we deal with once-ness).

If no-one shouts in the next day or two I'll spin this as its own patch.

Fine with me. So I'll just ignore a potential error case in my patch.

Thanks,


Juergen