Prev: SATA NCQ - blessing for throughput, curse for latency?
Next: [PATCH]cfq-iosched: make nonrot check logic consistent
From: Jupyung Lee on 10 Nov 2009 23:40
In its current preempt-rt kernel, ksoftirq() includes a series of primitives
related with kernel preemption and irq on/off, in the following order:

preempt_disable() ... (1)
local_irq_disable() ... (2)
__preempt_enable_no_resched() ... (3)
local_irq_enable() ... (4)

A problem arises if a process is awakened between (1) and (2) because
it is not given a chance to preempt the currently running process
for a while since (4): since (4), the kernel is preemptible, but there
is no explicit reschedule point.

A simple suggestion to resolve the problem is to add a reschedule point,
preempt_check_resched(), just after (4).

Signed-off-by: Jupyung Lee <jupyung(a)gmail.com>
---
kernel/softirq.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/kernel/softirq.c b/kernel/softirq.c
index 590049c..7c1b041 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -1037,6 +1037,7 @@ sleep_more:
set_softirq_pending(local_softirq_pending() & ~softirq_mask);
local_bh_disable();
local_irq_enable();
+ preempt_check_resched();

h = &softirq_vec[data->nr];
if (h)
--
1.5.4.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 | 
Pages: 1
Prev: SATA NCQ - blessing for throughput, curse for latency?
Next: [PATCH]cfq-iosched: make nonrot check logic consistent