Date: Mon, 14 Jan 2013 14:08:06 -0800 From: Evan Martin <evan@chromium.org> To: Yuri <yuri@rawbw.com> Cc: freebsd-chromium@freebsd.org, Jonathan Chen <jonc@chen.org.nz> Subject: Re: Oddness with latest chromium Message-ID: <CAFzwtj1-3-zwVn4bKsCT-o0bbKuOm74N_K_c6CiMGJDuyNiEzA@mail.gmail.com> In-Reply-To: <50F47E0F.9070900@rawbw.com> References: <50F3B4EF.6090206@chen.org.nz> <CANcjpOD0_Sc=ktAKEYie30UzGraNUDbfkxr4rEtFwhZHTj=pXg@mail.gmail.com> <CAJuc1zPH_BSkJoNZ2T53bS=fM7PgU4=XH8f0FU=dhrFeJ5F%2Bqg@mail.gmail.com> <CANcjpOCEEu%2B=wnDyEROczWw-rePcCO4ah1q_9S1yqfLPL6hm3g@mail.gmail.com> <CAJuc1zM2d5s1vDH=OrgFXvW%2BGz1R7raNzdAG2fq%2BV4bryCr%2BVw@mail.gmail.com> <50F47E0F.9070900@rawbw.com>
next in thread | previous in thread | raw e-mail | index | archive | help
(I wrote some of the code in this area for Chrome, but all my code has likely been removed by this point. It's used for better graphics performance -- reducing copies of pixels rendered by WebKit as they make their way to the X server.) It looks like there's some explicit attempt by Chrome to detect FreeBSD and work around this area. It wouldn't surprise me if the fallback codepath was wrong in some way. http://code.google.com/searchframe#OAMlx_jo-ck/src/ui/base/x/x11_util.cc&exact_package=chromium&q=shmget&type=cs&l=391 It appears that this code attempts to detect if you have this flag set and if so it falls back to a slower codepath. It wouldn't surprise me at all if this slower codepath had bitrotted in some way. It appears the Opera browser relies on it as well: http://forums.pcbsd.org/showthread.php?t=8018 One ChangeLog of theirs says "Note: On FreeBSD shared memory doesn't work by FreeBSD design." which seems a bit sad. Here's the Chrome code that relies on the detachment behavior, which has a comment describing why it does it: http://dxr.mozilla.org/mozilla-central/ipc/chromium/src/chrome/common/transport_dib_linux.cc.html#l39 >From some discussion in other bug trackers, it appears the behavior Chrome (and other tools like VMWare) rely upon isn't allowed by POSIX, which is why you can toggle the strict behavior with a flag on FreeBSD. But that's just my impression from reading some random forum posts. On Mon, Jan 14, 2013 at 1:52 PM, Yuri <yuri@rawbw.com> wrote: > > On 01/14/2013 13:30, Jonathan Chen wrote: >> >> Ok, that appears to be the key tunable that starts to make everything work >> again. I've been using chromium for about an hour now and haven't >> encountered any unexpected behaviour so far. > > > This also means that some error condition due to the lack of this tunable setting went unnoticed and chrome was malfunctioning instead of warning about the issue. > > Yuri > > _______________________________________________ > freebsd-chromium@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-chromium > To unsubscribe, send any mail to "freebsd-chromium-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAFzwtj1-3-zwVn4bKsCT-o0bbKuOm74N_K_c6CiMGJDuyNiEzA>