On 22 Sep 2015, at 16:23, Ian Lepore <i...@freebsd.org> wrote: > > On Tue, 2015-09-22 at 11:53 +0200, Dimitry Andric wrote: >> On 22 Sep 2015, at 11:43, Slawa Olhovchenkov <s...@zxy.spb.ru> wrote: >>> >>> On Tue, Sep 22, 2015 at 09:35:35AM +0000, Dimitry Andric wrote: >> ... >>>> - imm |= (-1 << 7); >>>> + imm |= -(1 << 7); >>> >>> May be (~0 << 7) is more simple to understund? >> >> This will give the same warning. ~0 will implicitly convert to -1. >> >> (It would be better to convert all these masks to unsigned, where such >> shifting is always defined, but that gives a lot more churn.) >> >> -Dimitry >> > > How is (-1U << 7) more churn?
For this one line, it isn't, of course. Writing -(1 << 7) is simply my personal preference, and it matches the type of 'imm'. But what I meant was changing all mask-containing int variables to unsigned, and suffixing all mask literals U, and so on. This is what should have been done by GNU in the first place. :) -Dimitry
signature.asc
Description: Message signed with OpenPGP using GPGMail