chrisccoulson | fta - this is a reproducer for the chromium/breakpad issue btw - http://paste.ubuntu.com/580859/ | 00:11 |
---|---|---|
=== m_conley_away is now known as m_conley | ||
=== asac_ is now known as asac | ||
micahg | fta: chromium seems to work fine on natty | 02:51 |
LLStarks | chrisccoulson, is the search popover supposed to cover the status popover? | 03:04 |
=== m_conley is now known as m_conley_away | ||
LLStarks | *statusbar | 03:05 |
LLStarks | nvm | 03:07 |
=== m_conley_away is now known as m_conley | ||
fta | micahg, [03:51] <micahg> fta: chromium seems to work fine on natty <= uh? did you expect it not to? I need more context.. | 15:14 |
fta | asac, ping | 15:14 |
asac | fta: hoh | 15:24 |
fta | how are you? | 15:24 |
fta | any update on the arm board(s)? | 15:24 |
asac | sorry, that dropped off my busy schedule ... /me hits himself | 15:26 |
asac | let me add that to next week tech lead agenda | 15:26 |
asac | monday i will know more | 15:26 |
asac | done | 15:27 |
asac | will not be forgotten | 15:27 |
asac | https://wiki.linaro.org/EngineeringUnits/Management/Meetings/2011-03-21 | 15:27 |
fta | bug 735877 | 15:36 |
ubot2 | Launchpad bug 735877 in chromium-browser "Illegal instruction in chromium on startup on armel" [Undecided,New] https://launchpad.net/bugs/735877 | 15:36 |
fta | asac, ^^ | 15:36 |
asac | right | 15:39 |
asac | i had folks staring on this today | 15:39 |
asac | fta: is that a patch of mine? | 15:39 |
asac | otherwise he is probably right ... vfpv3-d16 | 15:40 |
fta | nope, it's from upstream | 15:40 |
fta | no idea what that vfpv3-d16 is, and how i should know about it. is it documented somewhere? | 15:40 |
asac | hmm | 15:40 |
davidgiluk | fta: Hi | 15:45 |
asac | fta: meet davidgiluk | 15:45 |
asac | he wants to help you discuss upstream if needed ;) | 15:45 |
fta | hi | 15:45 |
davidgiluk | http://src.chromium.org/viewvc/chrome/trunk/src/build/common.gypi?view=diff&r1=40698&r2=40699 | 15:45 |
asac | so i guess question is if we want to make the point upstream to use -d16 for maximum compatibility | 15:46 |
asac | or if we just want to set this for ubuntu | 15:46 |
asac | https://wiki.ubuntu.com/ARM/Thumb2 | 15:46 |
asac | this is still the default set of flags that we expect | 15:46 |
asac | was decided to be maximum compatible while getting most benefit of most armv7 | 15:46 |
davidgiluk | fta: So my understanding is that there are 2 semi-separate chip features; Neon, and the number of vector registers; some chips have no neon, and have only 16 vector registers - and that's what we default to | 15:47 |
asac | so in ubuntu - unless there is runtime detection for things like neon - -mfpu=vfpv3-d16 is the default we have to use | 15:47 |
fta | davidgiluk, that particular changeset is not the real change, it just moved the flag from a hardcoded value to a variable, which i can change in my package | 15:48 |
davidgiluk | fta: Hmm true | 15:49 |
fta | so it seems i need to pass arm_fpu=vfpv3-d16 too now | 15:49 |
davidgiluk | fta: Yes | 15:49 |
davidgiluk | fta: What concerns me is why has this changed - or has it always been brokwn? | 15:49 |
davidgiluk | e | 15:49 |
fta | i will discuss with upstream. they mostly want arm for chrome-os, not sure what h/w they target | 15:49 |
davidgiluk | fta: The Tegra2 chip is the case it broke for Ramana which is non-neon and as I understand it d16 | 15:50 |
fta | it didn't change in a while, but i used to have a bogus test in my packaging, leading to armv7=0 | 15:50 |
fta | i fixed it in ch10 | 15:50 |
davidgiluk | fta: IMHO it would be best not to pass any flag at all and leave it to the compiler defaults | 15:50 |
fta | no, the chromium build system overrides most system defaults | 15:51 |
davidgiluk | ok | 15:51 |
ramana | fta: well then in that case we (or they) have to be building for the lowest common denominator which is vfpv3-d16 | 15:51 |
ramana | I believe that's the reason why we ended defaulting to vfpv3-d16 | 15:52 |
asac | right | 15:52 |
asac | thats why we do that in ubuntu | 15:52 |
asac | similar to overriding the neon=1 default hat chromium has | 15:52 |
davidgiluk | but has this ever worked on the ac100? | 15:52 |
asac | its interesting ... seems noone cared even though we got bashed that our qt crashed on ac100 ;) | 15:52 |
fta | most probably, it's just that i'm not well aware of those arm specificities. ubuntu vs debian vs chrome-os, all have different targets in mind | 15:53 |
asac | and we had to backout NEON | 15:53 |
ramana | davidgiluk: well michaelh1 recommended I tried chromium-browser on my ac100 over firefox. | 15:53 |
asac | fta: right ... just add vfpv3-d16 for now. then all is fine | 15:53 |
ramana | so I gave that a shot and that's how it showed up | 15:53 |
asac | we are working on runtime detecting neon for chromium/skia | 15:53 |
ramana | so I think he's got it working . | 15:53 |
asac | upstream is not responsive though :( | 15:53 |
davidgiluk | ramana: OK | 15:54 |
fta | ok, i'm fine with arm_fpu=vfpv3-d16 | 15:54 |
asac | fta: thanks!! | 15:54 |
fta | will land that in the next stable update | 15:54 |
ramana | with an older package ofcourse. | 15:54 |
ramana | fta: Thanks. | 15:54 |
davidgiluk | fta: Can you take bug 735877 then? | 15:56 |
ubot2 | Launchpad bug 735877 in chromium-browser "Illegal instruction in chromium on startup on armel" [Undecided,New] https://launchpad.net/bugs/735877 | 15:56 |
fta | so by default, i'll have: arm_thumb=1 arm_neon=0 arm_fpu=vfpv3-d16. and armv7=1 only for maverick onward | 15:56 |
fta | is that correct? | 15:56 |
davidgiluk | yeh I think so - I don't know the pre-maverick world | 15:57 |
=== evilvish is now known as vish | ||
fta | davidgiluk, ramana, asac: here is when the vfpv3 value 1st appeared: http://codereview.chromium.org/660067 | 16:14 |
davidgiluk | fta: Looking at ffmpeg it has a configure flag you can pass that would cause it to pass a -mfpu=vfpv3 but it doesn't look like the default | 16:18 |
fta | (it's an old patch) | 16:19 |
micahg | did i connect twice or once? | 18:17 |
jdstrand | micahg: | 18:19 |
jdstrand | 13:07 -!- micahg [~quassel@ubuntu/member/micahg] has joined #ubuntu-mozillateam | 18:19 |
jdstrand | 13:14 -!- micahg [~quassel@ubuntu/member/micahg] has quit [Remote host closed the connection] | 18:19 |
jdstrand | 13:17 -!- micahg [~micahg@ubuntu/member/micahg] has joined #ubuntu-mozillateam | 18:19 |
fta | micahg, would you be able to help with the chromium bugs? because i'm really bad with this due to lack of time. most should go upstream anyway | 19:51 |
=== magcius is now known as _magcius | ||
micahg | fta, sure, stuff related to regressions I can definitely help with idr if I have an upstream account | 19:55 |
micahg | fta, where's the upstream tracker again? | 19:56 |
fta | http://code.google.com/p/chromium/issues/list | 19:56 |
=== _magcius is now known as magcius | ||
micahg | fta, is #chromium a good place for me to idle/ask questions as well? | 20:05 |
fta | micahg, sure, but it's purely a dev channel, so they hate end user questions, which should go to the support channels, cf /topic | 20:07 |
micahg | fta, ok | 20:07 |
chrisccoulson | heh, this looks a bit like the chrome release schedule: http://people.mozilla.com/~sayrer/2011/temp/process.html | 20:21 |
chrisccoulson | "This proposal makes security updates occur along with Firefox releases, meaning we'll no longer be maintaining old branches." | 20:21 |
chrisccoulson | oh, fudge | 20:21 |
chrisccoulson | i guess we're screwed with all browsers now ;) | 20:22 |
micahg | :( | 20:22 |
* micahg needs to talk to upstream | 20:22 | |
chrisccoulson | fta - you have a separate nightly PPA for each channel of chromium don't you? | 20:23 |
chrisccoulson | and you don't make them parallel installable? | 20:23 |
fta | correct | 20:24 |
fta | http://people.ubuntu.com/~fta/ppa-dashboard/chromium-daily.html | 20:24 |
chrisccoulson | fta - i'm wondering whether that might be a better approach for firefox too, given the proposals for the release process | 20:24 |
fta | i should not have used "daily" in the name | 20:24 |
chrisccoulson | as there's going to what looks like 4 active channels | 20:24 |
fta | chrisccoulson, looking at chromium, most users will end up using either the stable builds or the dailies, not much users in between | 20:30 |
fta | jcastro, is the scroll wheel supposed to work in the dash now? | 20:41 |
fta | i had it working a few days ago at work, but never at home | 20:41 |
fta | similar setups | 20:41 |
jcastro | I am not sure | 20:41 |
jcastro | ah yeah | 20:41 |
jcastro | it works for me | 20:41 |
jcastro | fta: new compiz upload, see if this one helps | 20:44 |
fta | jcastro, also is it possible to filter out the "Recents" in the dash? mine is filled with stuff i don't need in there, like mp3.. i have rhythmbox playing all day long | 20:44 |
jcastro | fta: there's a blacklist thing you can do in zeitgeist | 20:44 |
fta | where? how? | 20:45 |
jcastro | however I've not looked into it, I think you want to search for blacklisting, zeitgeist, and gnome-activity-journal | 20:45 |
jcastro | seiflofty on #ayatana can sort you out probably | 20:45 |
fta | upgrading... | 20:55 |
fta | jcastro, any update for the ch webapps since yesterday? | 20:55 |
jcastro | fta: trevino's going to look at it on friday, he's on holiday right now | 20:55 |
fta | k | 20:58 |
=== m_conley is now known as m_conley_away | ||
fta | jdstrand, [ 9012.547609] type=1400 audit(1300318178.116:20): apparmor="DENIED" operation="file_mmap" parent=4945 profile="/usr/bin/evince-thumbnailer" name="/usr/lib/i386-linux-gnu/gconv/UTF-16.so" pid=4946 comm="evince-thumbnai" requested_mask="m" denied_mask="m" fsuid=1000 ouid=0 | 23:30 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!