Signed overflow optimization hazards in the kernel [LWN.net]

Signed overflow optimization hazards in the kernel

Posted Aug 16, 2012 9:02 UTC (Thu) by ken (subscriber, #625)
In reply to: Signed overflow optimization hazards in the kernel by hobocken
Parent article: Signed overflow optimization hazards in the kernel

Exactly I noticed that to. It's been -O2 as far as I can remember.

to post comments

Signed overflow optimization hazards in the kernel

Posted Aug 16, 2012 13:19 UTC (Thu) by PaulMcKenney (✭ supporter ✭, #9624) [Link] (2 responses)

And you are absolutely right: -O2 and no sign of -fwrapv. Perhaps this is more urgent than I was thinking...

Signed overflow optimization hazards in the kernel

Posted Aug 16, 2012 13:58 UTC (Thu) by abatters (✭ supporter ✭, #6932) [Link] (1 responses)

http://marc.info/?t=124806309500001&r=1&w=2

The kernel currently uses -fno-strict-overflow because -fwrapv was buggy in some versions of gcc. Unfortunately -fno-strict-overflow is also buggy in some other versions of gcc. Depending on the version of gcc I compile with, I have to patch the makefile to use one or the other.

Signed overflow optimization hazards in the kernel

Posted Aug 16, 2012 22:49 UTC (Thu) by PaulMcKenney (✭ supporter ✭, #9624) [Link]

Whew! It goes back to being non-urgent. ;-)