On 11.06.19 17:16, Ralf Ramsauer wrote:
On 6/11/19 5:07 PM, Jan Kiszka wrote:
On 11.06.19 01:03, Ralf Ramsauer wrote:
If we should ever compile timing.c for 32 bit inmates, then unsigned
long is
the wrong type. A 32 bit type (aka unsigned long on 32 bit) is too
small to
hold the result.
This ensures that we won't hit a bug when enabling on 32 bit. Once we
will
enable timing.c on 32 bit, we will have to implement 64 bit division
intrinsics.
Signed-off-by: Ralf Ramsauer <[email protected]>
---
inmates/lib/x86/include/inmate.h | 2 +-
inmates/lib/x86/timing.c | 8 ++++----
2 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/inmates/lib/x86/include/inmate.h
b/inmates/lib/x86/include/inmate.h
index 7664570d..e4f75206 100644
--- a/inmates/lib/x86/include/inmate.h
+++ b/inmates/lib/x86/include/inmate.h
@@ -241,7 +241,7 @@ void ioapic_pin_set_vector(unsigned int pin,
enum ioapic_trigger_mode trigger_mode,
unsigned int vector);
-unsigned long pm_timer_read(void);
+unsigned long long pm_timer_read(void);
unsigned long tsc_read_ns(void);
unsigned long tsc_init(void);
diff --git a/inmates/lib/x86/timing.c b/inmates/lib/x86/timing.c
index de41fc35..d179da0f 100644
--- a/inmates/lib/x86/timing.c
+++ b/inmates/lib/x86/timing.c
@@ -92,13 +92,13 @@ unsigned long tsc_init(void)
return tsc_freq;
}
-unsigned long pm_timer_read(void)
+unsigned long long pm_timer_read(void)
{
unsigned int cpu = cpu_id();
- unsigned long tmr;
+ unsigned long long tmr;
- tmr = ((inl(comm_region->pm_timer_address) & 0x00ffffff) *
NS_PER_SEC)
- / PM_TIMER_HZ;
+ tmr = ((unsigned long long)(inl(comm_region->pm_timer_address)
+ & 0x00ffffff) * NS_PER_SEC) / PM_TIMER_HZ;
if (tmr < pm_timer_last[cpu])
pm_timer_overflows[cpu] += PM_TIMER_OVERFLOW;
pm_timer_last[cpu] = tmr;
This is not yet consistent: pm_timer_last is still unsigned long.
Do we need the conversion in this series already?
Nope. It's just a 'reminder' that sth. is odd there. But we can ignore
it. Same for 2/10.
Just some assorted stuff that I hit when (unsuccessfully) trying to
enable SMP on 32 bit.
Both patches make sense. I would just address this typing consistently, rather
then hiding the issue again.
Jan
--
You received this message because you are subscribed to the Google Groups
"Jailhouse" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/jailhouse-dev/ca4ba28e-157e-0576-49e8-f2ef0b4689e1%40web.de.
For more options, visit https://groups.google.com/d/optout.
