Warning for clang's shifting bare integer values default to 32 bit

Question

So, in the upstream sources, we have this line

rc_c_max = allmem - (3 << 30);

Upstream which typically uses gcc, does things "as expected". allmem is uint64_t. But with clang, it appears to assume "3" is 32 bit (or maybe signed) and the results are not "as expected". An easy fix for it is to change it to

rc_c_max = allmem - (3ULL << 30);

and everything works. Changing all the bare-interger shifts, by working out/guessing which needs to be corrected is tedious, is there a warning I can enable for clang to point this out? -Wall -Wextra does not complain about this line.

Output:

1/19/16 9:25:30.000 AM kernel[0]: allmem - (3 << 30) : 0x373333000 : 14817636352
1/19/16 9:25:30.000 AM kernel[0]: allmem - (3ULL << 30) : 0x273333000 : 10522669056

3 has type signed int. You should always be thinking about types when writing code containing shifts . — M.M
– M.M, Commented Jan 20, 2016 at 8:12

Trevor Hickey · Accepted Answer · 2016-01-21 01:05:32Z

2

The flag you are looking for is:

-Wshift-sign-overflow

Although, unlike gcc, I recommend using clang's compiler flag:

-Weverything

That would enable -Wshift-sign-overflow as well.

edited Jan 21, 2016 at 1:05

answered Jan 20, 2016 at 7:00

Trevor Hickey

38.4k37 gold badges180 silver badges287 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

lundman Over a year ago

Perfect, showed one more in my code, and one in XNU sources which would be harder to fix :)

Collectives™ on Stack Overflow

Warning for clang's shifting bare integer values default to 32 bit

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related