CODEWITHSUNDEEP
assemblyx86
Shubhang Desai
Shubhang Desai

Reputation: 319

Is this jns ever false? eflag in x86

I am reading some assembly, and I'm fairly sure that I have some logically impossible/dead code in it. Here it is:

shr       %eax
test      %eax, %eax
jns       0x[something]
[if body]
[something]

Is it ever, in any case, possible for the if-body to get executed? I feel like the answer is no, but then again, wouldn't the compiler optimize this out?

Apologize for the vagueness in the code, didn't want to add more code than was necessary to give the question context. Let me know if more information would be useful.

Upvotes: 0

Views: 1603

Answers (1)

Sean Werkema
Sean Werkema

Reputation: 6135

Unless there is a branch directly to the if-body from somewhere else, it cannot possibly be reached through that jns. The shr (logical shift right) instruction will always result in a high bit (sign bit) of zero, so that test and jns will always have the same behavior no matter what value eax started out with.

That said, it would be worthwhile to search elsewhere in the code and see if anything jumps to the test, or to the jnz, or to the [if body], since those instructions might still not be dead code if they're reachable another way.

Upvotes: 3

Related Questions