GHES: Elliminate double-loop in the NMI handler

There's no real need to iterate twice over the HW error sources in the
NMI handler. With the previous cleanups, elliminating the second loop is
almost trivial.

Signed-off-by: Borislav Petkov <bp@suse.de>
This commit is contained in:
Borislav Petkov 2015-03-18 10:12:35 +01:00
parent 6169ddf846
commit 2383844d48

View file

@ -851,25 +851,18 @@ static int ghes_notify_nmi(unsigned int cmd, struct pt_regs *regs)
if (sev >= GHES_SEV_PANIC)
__ghes_panic(ghes);
ret = NMI_HANDLED;
}
if (ret == NMI_DONE)
goto out;
list_for_each_entry_rcu(ghes, &ghes_nmi, list) {
if (!(ghes->flags & GHES_TO_CLEAR))
continue;
__process_error(ghes);
ghes_clear_estatus(ghes);
ret = NMI_HANDLED;
}
#ifdef CONFIG_ARCH_HAVE_NMI_SAFE_CMPXCHG
irq_work_queue(&ghes_proc_irq_work);
#endif
out:
raw_spin_unlock(&ghes_nmi_lock);
return ret;
}