Date: Thu, 24 May 2018 09:34:49 -0700 From: Matthew Macy <mmacy@freebsd.org> To: Pedro Giffuni <pfg@freebsd.org> Cc: Warner Losh <imp@bsdimp.com>, Michael Tuexen <Michael.Tuexen@macmic.franken.de>, Gleb Smirnoff <glebius@freebsd.org>, src-committers <src-committers@freebsd.org>, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: Re: svn commit: r333860 - head/sys/kern Message-ID: <CAPrugNrEqnY-YQ0z8=x_gdvdk2GSQSLkH23P1tQsTNXvpJanZA@mail.gmail.com> In-Reply-To: <9b7c8c0f-4c33-1a14-f0db-5370c3e5def7@FreeBSD.org> References: <201805190510.w4J5AqfS054367@repo.freebsd.org> <20180523222743.GU71675@FreeBSD.org> <CAPrugNpKxEVx%2Bg0D6qUwq=yw_W-QeuntupwAjteK1yALkJktag@mail.gmail.com> <20180523225729.GV71675@FreeBSD.org> <CAPrugNpsiYQ1MKkgGbkwPEeTcsTCdD_u5GAvYpMJKHzeSbCL=w@mail.gmail.com> <20180524044252.GW71675@FreeBSD.org> <CAPrugNoVPCbB0qTY=Z1dwCZ%2BbCbVbA1d5b0-tw5qaPxWRK0q8g@mail.gmail.com> <5B9EE208-384F-44AD-9B47-059D77FE9B34@macmic.franken.de> <CAPrugNooA9Jfx5=h8fYb5%2BKPUOjSOD_NR8QX=bBkVKCAu=96-A@mail.gmail.com> <ABB8E26D-ABAE-42E6-9A1C-FAD0E8DDA3C6@macmic.franken.de> <CAPrugNq_peOYp-1ufu8d%2BOpmspmT09kUqxV%2BvRYsHqv9tYUTGg@mail.gmail.com> <CANCZdfov%2Bb441dN7shE5JY%2BxgKnT41sVfdwrirxUDUOxkSQrJQ@mail.gmail.com> <CAPrugNqGhGiSyVTORvy%2Bv0U2rgd1SN-%2BOw0m3USp0hiv5wg7vg@mail.gmail.com> <9b7c8c0f-4c33-1a14-f0db-5370c3e5def7@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> > False positives are compiler bugs. No they're not. No more than missed optimization opportunities. They're limitations in the control flow analysis. > > It does happen, with GCC more than with clang, that the compiler has too > many bugs and it's a bad practice to pessimize code to work around them. It doesn't pessimize the compiled output. It may make the code less readable in the opinion of some. I've found that 10% of the warnings were actually legitimate issues. It's unfortunate that the majority are just noise that have to be waded through. > At > very least you should add a comment when adding unnecessary initializations, > something like /* workaround GCC */, but dropping broken warnings is best. That's legitimate.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAPrugNrEqnY-YQ0z8=x_gdvdk2GSQSLkH23P1tQsTNXvpJanZA>