Overly broad PVS calculation when a compound expression depends on one input variable in multipe sub-expressions

[xusheng6]

rcx_2 is only dependent on the value of rcx_1, which has a PVS of {0, 8}, so we should be able to track its value

Binary: quick bird dances boldly

[hongyu-dev]

Hi @xusheng6, can you provide me more information about what is

Binary: quick bird dances boldly

Also if you can point me which directory or files I should target on for fixing this issue? Thank you!

[hongyu-dev]

This has been fixed

[xusheng6]

Nope, this has not really been fixed -- we currently calculates a PVS for it, but it is incorrect. rcx_1 can only take two possible values, so rcx_2 should also have only two possible values. But we are reporting four possible values for it, i.e., 0x0, 0x8, 0x10, 0x18. That is because we tracked the PVS of the two sub-expressions, and each of them can have two possible values, so adding them together leads to four possible values