Messages in this thread | | | From | Linus Torvalds <> | Date | Wed, 18 Apr 2012 10:24:13 -0700 | Subject | Re: [v3.4-rc1] ACPI regression bisected |
| |
On Wed, Apr 18, 2012 at 3:29 AM, Thomas Gleixner <tglx@linutronix.de> wrote: > > Linus, can you please apply directly?
Applied. Can we perhaps also make request_threaded_irq() noisily reject things that are not marked IRQF_ONESHOT and have a NULL low-level handler?
That still allows people to do IRQF_SHARED | IRQF_ONESHOT, and that won't be rejected until somebody else actually tries to share it. Then it breaks that somebody else ;(
Sadly, that would tend to break fairly silently.
I wonder if we should make the mismatch error printout be unconditional (and make just the "dump_stack()" part be conditional on CONFIG_DEBUG_SHIRQ)?
Linus
| |