This bug was originally fixed in commit 35a4933a8959 ("time:
Avoid signed overflow in timekeeping_get_ns()"). When the code was
refactored in commit 6bd58f09e1d8 ("time: Add cycles to nanoseconds
translation") the signed overflow fix was lost. Re-introduce it
in a less subtle way by changing the type of "nsec" to unsigned
and adding a comment explaining why.
Signed-off-by: Chris Metcalf <cmetc...@mellanox.com>
---
v2: just use "u64 nsec" to fix the signed shift problem
kernel/time/timekeeping.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 37dec7e3db43..8c06a2aa9b5f 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -302,7 +302,7 @@ static inline u32 arch_gettimeoffset(void) { return 0; }
static inline s64 timekeeping_delta_to_ns(struct tk_read_base *tkr,
cycle_t delta)
{
- s64 nsec;
+ u64 nsec; /* Avoid possibility of a negative right shift. */
nsec = delta * tkr->mult + tkr->xtime_nsec;
nsec >>= tkr->shift;
--
2.7.2
