Date: Thu, 13 Oct 2011 17:02:58 -0400 From: Arnaud Lacombe <lacombar@gmail.com> To: pyunyh@gmail.com Cc: fodillemlinkarim@gmail.com, freebsd-net@freebsd.org, perryh@pluto.rain.com, kob6558@gmail.com Subject: Re: if_msk.c link negotiation / packet drops Message-ID: <CACqU3MUFXycVwhjeVZ9ykJueOVH%2BcS-zb=iBa-UkqK6FoJVRrg@mail.gmail.com> In-Reply-To: <20111013204747.GA13219@michelle.cdnetworks.com> References: <4E94637A.5090607@gmail.com> <20111011171029.GA5661@michelle.cdnetworks.com> <CAN6yY1tWQZwdqgYdN3uBBdXiGQ2OFDMYbSjhEUeTimHjBnR9iA@mail.gmail.com> <4E959F06.6040906@gmail.com> <20111012170347.GA9138@michelle.cdnetworks.com> <4e969a67.YJyWMt0xI7pFL%2BxJ%perryh@pluto.rain.com> <20111013204747.GA13219@michelle.cdnetworks.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi, On Thu, Oct 13, 2011 at 4:47 PM, YongHyeon PYUN <pyunyh@gmail.com> wrote: > On Thu, Oct 13, 2011 at 12:59:35AM -0700, perryh@pluto.rain.com wrote: >> YongHyeon PYUN <pyunyh@gmail.com> wrote: >> > On Wed, Oct 12, 2011 at 10:07:02AM -0400, Karim wrote: >> > > ... why are we ORing the same call twice isn't the same thing >> > > as calling it once: >> > > >> > > bmsr =3D PHY_READ(sc, E1000_SR) | PHY_READ(sc, E1000_SR); >> > >> > The E1000_SR_LINK_STATUS bit is latched low so it should be read >> > twice. >> >> It might not be a bad idea to check the generated code to be sure >> that the read _is_ being done twice. =A0An optimizer might well come >> to the same conclusion as Karim, and discard the "redundant" second >> instance (unless there's a "volatile" declaration somewhere in the >> expansion of PHY_READ, to explicitly indicate that it has side >> effects). > > Last time I checked it, compiler generated correct code. > Tried again on amd64 and I can still see the code is there. > What about other architecture (especially i386) ? which optimization level did you use ? which compiler version ? About the last question, I know for sure that there has been change in FreeBSD's gcc between 7-STABLE, and FreeBSD -CURRENT. I agree with perryh@ than such hardware requirement _requires_ being explicit in the code, ie proper `volatile' marking. - Arnaud
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CACqU3MUFXycVwhjeVZ9ykJueOVH%2BcS-zb=iBa-UkqK6FoJVRrg>