Since a clocksource change post 3.2-rc1, tasks on my DB1500 board hang after random amounts of time (from a few minutes to a few hours), regardless of load. Debugging showed that the compare-match register value is a few seconds lower than the current counter value. The minimum value of 8 was initialy determined by a trial-and-error approach. Currently it is sufficient for all Alchemys (without PCI apparently), independent of CPU clock; only the DB1500 and DB1550 boards experience these timer-related tasks hangs now. This patch increases the minimum timeout by 1 (to 9 counter ticks) which seems sufficient since the systems are still working perfectly fine after over 24 hours. Signed-off-by: Manuel Lauss <manuel.lauss@xxxxxxxxxxxxxx> --- Please consider for 3.2 arch/mips/alchemy/common/time.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/arch/mips/alchemy/common/time.c b/arch/mips/alchemy/common/time.c index a594a85..63ba51f 100644 --- a/arch/mips/alchemy/common/time.c +++ b/arch/mips/alchemy/common/time.c @@ -146,7 +146,7 @@ static int __init alchemy_time_init(unsigned int m2int) cd->shift = 32; cd->mult = div_sc(32768, NSEC_PER_SEC, cd->shift); cd->max_delta_ns = clockevent_delta2ns(0xffffffff, cd); - cd->min_delta_ns = clockevent_delta2ns(8, cd); /* ~0.25ms */ + cd->min_delta_ns = clockevent_delta2ns(9, cd); /* ~0.28ms */ clockevents_register_device(cd); setup_irq(m2int, &au1x_rtcmatch2_irqaction); -- 1.7.8
