[PATCH 3/3] ARM: msm: Wait for timer clear to complete
Stephen Boyd
sboyd at codeaurora.org
Thu Mar 14 23:31:39 EDT 2013
Without looping on the status bit, there is no way to guarantee
that a clear of the timer has actually completed. This can cause
us to enable the timer before the count has cleared and miss a
timer interrupt. To simplify this patch, remove the timer
register setup done during timer init, since it's duplicate work
that is eventually done in the set_next_event() callback.
Signed-off-by: Stephen Boyd <sboyd at codeaurora.org>
---
arch/arm/mach-msm/timer.c | 41 +++++++++++++++++++++++------------------
1 file changed, 23 insertions(+), 18 deletions(-)
diff --git a/arch/arm/mach-msm/timer.c b/arch/arm/mach-msm/timer.c
index 9f033c3..284313f 100644
--- a/arch/arm/mach-msm/timer.c
+++ b/arch/arm/mach-msm/timer.c
@@ -30,20 +30,22 @@
#include "common.h"
-#define TIMER_MATCH_VAL 0x0000
-#define TIMER_COUNT_VAL 0x0004
-#define TIMER_ENABLE 0x0008
-#define TIMER_ENABLE_CLR_ON_MATCH_EN BIT(1)
-#define TIMER_ENABLE_EN BIT(0)
-#define TIMER_CLEAR 0x000C
-#define DGT_CLK_CTL 0x10
-#define DGT_CLK_CTL_DIV_4 0x3
+#define TIMER_MATCH_VAL 0x0000
+#define TIMER_COUNT_VAL 0x0004
+#define TIMER_ENABLE 0x0008
+#define TIMER_ENABLE_CLR_ON_MATCH_EN BIT(1)
+#define TIMER_ENABLE_EN BIT(0)
+#define TIMER_CLEAR 0x000C
+#define DGT_CLK_CTL 0x10
+#define DGT_CLK_CTL_DIV_4 0x3
+#define TIMER_STS_GPT0_CLR_PEND BIT(10)
#define GPT_HZ 32768
#define MSM_DGT_SHIFT 5
static void __iomem *event_base;
+static void __iomem *sts_base;
static irqreturn_t msm_timer_interrupt(int irq, void *dev_id)
{
@@ -68,6 +70,11 @@ static int msm_timer_set_next_event(unsigned long cycles,
writel_relaxed(ctrl, event_base + TIMER_CLEAR);
writel_relaxed(cycles, event_base + TIMER_MATCH_VAL);
+
+ if (sts_base)
+ while (readl_relaxed(sts_base) & TIMER_STS_GPT0_CLR_PEND)
+ cpu_relax();
+
writel_relaxed(ctrl | TIMER_ENABLE_EN, event_base + TIMER_ENABLE);
return 0;
}
@@ -138,9 +145,6 @@ static int __cpuinit msm_local_timer_setup(struct clock_event_device *evt)
if (!smp_processor_id())
return 0;
- writel_relaxed(0, event_base + TIMER_ENABLE);
- writel_relaxed(0, event_base + TIMER_CLEAR);
- writel_relaxed(~0, event_base + TIMER_MATCH_VAL);
evt->irq = msm_clockevent.irq;
evt->name = "local_timer";
evt->features = msm_clockevent.features;
@@ -178,9 +182,6 @@ static void __init msm_timer_init(u32 dgt_hz, int sched_bits, int irq,
struct clocksource *cs = &msm_clocksource;
int res;
- writel_relaxed(0, event_base + TIMER_ENABLE);
- writel_relaxed(0, event_base + TIMER_CLEAR);
- writel_relaxed(~0, event_base + TIMER_MATCH_VAL);
ce->cpumask = cpumask_of(0);
ce->irq = irq;
@@ -275,6 +276,7 @@ void __init msm_dt_timer_init(void)
of_node_put(np);
event_base = base + 0x4;
+ sts_base = base + 0x88;
source_base = cpu0_base + 0x24;
freq /= 4;
writel_relaxed(DGT_CLK_CTL_DIV_4, source_base + DGT_CLK_CTL);
@@ -283,7 +285,8 @@ void __init msm_dt_timer_init(void)
}
#endif
-static int __init msm_timer_map(phys_addr_t addr, u32 event, u32 source)
+static int __init msm_timer_map(phys_addr_t addr, u32 event, u32 source,
+ u32 sts)
{
void __iomem *base;
@@ -294,6 +297,8 @@ static int __init msm_timer_map(phys_addr_t addr, u32 event, u32 source)
}
event_base = base + event;
source_base = base + source;
+ if (sts)
+ sts_base = base + sts;
return 0;
}
@@ -302,7 +307,7 @@ void __init msm7x01_timer_init(void)
{
struct clocksource *cs = &msm_clocksource;
- if (msm_timer_map(0xc0100000, 0x0, 0x10))
+ if (msm_timer_map(0xc0100000, 0x0, 0x10, 0x0))
return;
cs->read = msm_read_timer_count_shift;
cs->mask = CLOCKSOURCE_MASK((32 - MSM_DGT_SHIFT));
@@ -313,14 +318,14 @@ void __init msm7x01_timer_init(void)
void __init msm7x30_timer_init(void)
{
- if (msm_timer_map(0xc0100000, 0x4, 0x24))
+ if (msm_timer_map(0xc0100000, 0x4, 0x24, 0x80))
return;
msm_timer_init(24576000 / 4, 32, 1, false);
}
void __init qsd8x50_timer_init(void)
{
- if (msm_timer_map(0xAC100000, 0x0, 0x10))
+ if (msm_timer_map(0xAC100000, 0x0, 0x10, 0x34))
return;
msm_timer_init(19200000 / 4, 32, 7, false);
}
--
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
hosted by The Linux Foundation
More information about the linux-arm-kernel
mailing list