You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I started fuzzing slices again after the recent fix to #2147 (hurray). Unfortunately, there seem to be more nasty edge cases.
In this action we assign 2 (0000 0010) to h.h.a while also assigning 1 to the least significant bit. The final value of h.h.a should be 3 (0000 0011). However, in 35_SimplifyDefUse, the compiler removes the assignment to h.h.a. Copy-out should only copy back the least significant bit, but it looks like the pass assumes that the entire header is copied back. At least that is how I understand it.
I started fuzzing slices again after the recent fix to #2147 (hurray). Unfortunately, there seem to be more nasty edge cases.
In this action we assign 2 (0000 0010) to
h.h.a
while also assigning 1 to the least significant bit. The final value ofh.h.a
should be 3 (0000 0011). However, in35_SimplifyDefUse
, the compiler removes the assignment toh.h.a
. Copy-out should only copy back the least significant bit, but it looks like the pass assumes that the entire header is copied back. At least that is how I understand it.to
slice_assignment.p4.txt
slice_assignment.stf.txt
The text was updated successfully, but these errors were encountered: