| === chihchun_afk is now known as chihchun | ||
| === chihchun is now known as chihchun_afk | ||
| === chihchun_afk is now known as chihchun | ||
| === davmor2_ is now known as davmor2 | ||
| === greyback is now known as greyback|food | ||
| === chihchun is now known as chihchun_afk | ||
| racarr | Morning! | 12:26 |
|---|---|---|
| racarr | greyback|food: I found the crash! | 12:26 |
| greyback|food | racarr: yeah! | 12:27 |
| greyback|food | have a branch I can try? | 12:27 |
| racarr | greyback|food: Yeah, of unity-mir actually, the final solution isn't totally evident | 12:27 |
| racarr | juts a sec | 12:27 |
| greyback|food | racarr: oh, what have I done wrong? | 12:28 |
| racarr | greyback|food: Eh, no not really your fault which is why I say the final solution | 12:28 |
| racarr | isn't evident | 12:28 |
| greyback|food | ok | 12:28 |
| racarr | greyback|food: http://pastebin.ubuntu.com/6145326/ | 12:28 |
| racarr | basically snapshot crashes if default_surface is null | 12:29 |
| greyback|food | oh interesting | 12:29 |
| racarr | i.e. the application is open but not surface yet | 12:29 |
| racarr | so it could throw an exception | 12:29 |
| racarr | or perhaps return an empty image or | 12:29 |
| racarr | I dunno | 12:29 |
| greyback|food | well I should not be asking for a snapshot if the surface hasn't been created yet | 12:30 |
| racarr | the good news is it seems really hard to crash with this applied | 12:30 |
| racarr | havent seen any | 12:30 |
| greyback|food | cool | 12:30 |
| === chihchun_afk is now known as chihchun | ||
| kgunn | racarr: nice | 12:33 |
| greyback|food | racarr: wanna propose that change as a MR, and I can review? | 12:33 |
| kgunn | alf_: can you propose an mr to bump the so name of the mir server? | 12:34 |
| racarr | greyback|food: Ok | 12:35 |
| racarr | Morning kgunn :) | 12:35 |
| kgunn | mornin | 12:35 |
| kgunn | gonna grab toast and coffee.... | 12:35 |
| racarr | :) | 12:35 |
| racarr | Oh also im not really up early :p (re: other channel) | 12:36 |
| racarr | im in Pennsylvania | 12:36 |
| racarr | greyback|food: https://code.launchpad.net/~robertcarr/unity-mir/check-default-surface-before-snapshot/+merge/187012 | 12:37 |
| greyback|food | racarr: thanks | 12:38 |
| === greyback|food is now known as greyback | ||
| racarr | greyback: What's remaining to flip mir in the image. | 12:45 |
| racarr | DPMS landed btw and alexandros is working on galaxy nexus support via sysfs stuff | 12:45 |
| racarr | Is it just flicker? | 12:45 |
| greyback | racarr: yes, just flicker | 12:46 |
| racarr | great | 12:48 |
| racarr | greyback: There is a lot of reason to believe that has to do with our HWC module and that upgrading the appropriate bionic side bits | 12:48 |
| racarr | will fix it :) | 12:48 |
| racarr | kdub is on it last I heard | 12:48 |
| greyback | cool | 12:49 |
| greyback | would be great if we can flip soon | 12:49 |
| === mzanetti is now known as mzanetti|food | ||
| racarr | Good afternoon alan :) | 13:10 |
| === mzanetti|food is now known as mzanetti | ||
| kgunn | alf_: scratch my last ping | 14:08 |
| racarr | My focus-tell-dont-ask branch is generating the strangest test failures. ApplicationMEdiatorReport session_create_surface and surface_create_buffer failing with broken pipe | 14:10 |
| racarr | looks like it's exposing something weird... | 14:10 |
| === chihchun is now known as chihchun_afk | ||
| kgunn | didrocks: for libmirclient ffe https://bugs.launchpad.net/ubuntu/+source/mir/+bug/1229212 | 14:23 |
| ubot5 | Ubuntu bug 1229212 in mir (Ubuntu) "[FFe] mir api bump post sprint for libmirclient" [Undecided,New] | 14:23 |
| didrocks | kgunn: you should add that it doesn't have any impact on the default ubuntu experience and the risk of regression is low | 14:23 |
| didrocks | kgunn: also, as per wiki page, you need to subscribe the release team | 14:23 |
| kgunn | didrocks: ack | 14:25 |
| didrocks | kgunn: then, let's join #ubuntu-release and get people looking at it ;) | 14:25 |
| didrocks | (please ping me once there) | 14:25 |
| smartboyhw | No symbols? Ow........ | 14:26 |
| === dandrader is now known as dandrader|afk | ||
| racarr | ...my laptop power cord is failing me so it seems like I am goign to dissapear for a while soon | 14:50 |
| racarr | while I reinstall on an old macbook I have in the closet | 14:50 |
| racarr | its depressing to watch your laptop battery go down and know there is nothing you can do about it | 14:54 |
| racarr | greyback_: kgunn: I guess we need resizing for october? (phone rotation) | 14:56 |
| racarr | so thinking of switching full steam to that? | 14:56 |
| kgunn | racarr: greyback_ actually...cleaning up (controlling) the api for server should come first | 14:57 |
| kgunn | at least something better than we have atm | 14:58 |
| kgunn | i thnk resize is actually something we can use post V1....but i'll defer to greyback_ | 14:58 |
| === dandrader|afk is now known as dandrader | ||
| racarr | kgunn: It's needed for rotation | 15:00 |
| racarr | because of the panel | 15:00 |
| racarr | or maybe we could use a hack in qtubuntu with artificially inflated surface sizes to work around it in the interim | 15:00 |
| kgunn | greyback_ racarr ....panel didn't rotate before (not arguing its not needed...), so was assuming parity ok | 15:01 |
| kgunn | lemme check | 15:01 |
| greyback_ | kgunn: panel does not rotate right now. | 15:02 |
| racarr | kgunn: Well, parity is ok for landing right, but not ok for october final? | 15:04 |
| kgunn | racarr: i believe its ok for oct...but lemme check | 15:05 |
| racarr | Ok | 15:06 |
| racarr | I can start working on stabilizzing the server APIs too | 15:06 |
| racarr | I have some ideas for reducing the width of the interface between unity-mir and mir | 15:06 |
| racarr | or rather, the number of interfaces used | 15:06 |
| racarr | but I'm not sure I have a full picture solution | 15:07 |
| racarr | ok laptop is about to kill itself. if all goes well should be back in an hour or two. Can reach me on | 15:07 |
| racarr | google talk via phone in the mean time | 15:08 |
| kdub | hello folks | 15:12 |
| === chihchun_afk is now known as chihchun | ||
| kgunn | greyback_: can you change the build dep for unity-mir | 15:39 |
| kgunn | ricmm: can you change it for platform-api ? | 15:39 |
| greyback_ | kgunn: can do. | 15:39 |
| kgunn | thanks...didrocks is gonna try to land mir | 15:40 |
| kgunn | latest | 15:40 |
| didrocks | don't forget u-s-c | 15:40 |
| kgunn | didrocks: sorry if i am slow...but why are we making this change ? | 15:42 |
| kgunn | https://code.launchpad.net/~didrocks/mir/remove-hack/+merge/187029 | 15:42 |
| didrocks | kgunn: because we don't force reverse depdencies anymore to use the exact mir version they were built against | 15:43 |
| kgunn | didrocks: ah...backward compat | 15:44 |
| didrocks | yeah ;) | 15:44 |
| greyback_ | kgunn: it's just a client api bump? | 15:48 |
| kgunn | greyback_: ricmm both client & server | 15:49 |
| kdub | ricmm, rsalveti how easy (or difficult) would it be to try a cm-10.2 based phablet build? | 15:49 |
| === dandrader is now known as dandrader|lunch | ||
| kgunn | alf_ racarr kdub read your mail....let me know if you think we should just ressurect the integration branch ? | 16:22 |
| kgunn | is alan_g having connectivity issues ? | 16:22 |
| ricmm | kgunn: didrocks what exact versions should we start depping on in unity-mir and platform-api? | 16:23 |
| ricmm | for the new process, wahts the first dep to dep on | 16:23 |
| didrocks | ricmm: 0.0.11 seems to be the version they will bump Mir to | 16:23 |
| mterry | kgunn, hi | 16:24 |
| kgunn | didrocks: i thot you said since it bumped once | 16:24 |
| kgunn | that we shouldn't bump again | 16:24 |
| kgunn | or do you want us to ? | 16:24 |
| didrocks | kgunn: there are 2 bumps | 16:25 |
| didrocks | the packaging ABI name bump -> not needed to change (in debian/control) | 16:25 |
| didrocks | the upstream version bump (the version you are releasing), you need to bump it to 0.0.11 in debian/changelog | 16:26 |
| ricmm | greyback_: ^ we need to dep client/server on .11 then | 16:29 |
| ricmm | so that we benefit of a rebuild | 16:29 |
| kgunn | kdub: moving here...do you think its worthwhile to chase a workaround of forced second layer for a bit ? | 16:59 |
| kgunn | i'm not sure the feasibility of cm10.2 attempt | 16:59 |
| kgunn | didrocks: you gonna be on for a few more ? | 16:59 |
| didrocks | kgunn: ~30 minutes, I doubt Mir will have the time to finish building by then (even if we start now) | 17:00 |
| didrocks | kgunn: just push everything, I'll staged to proposed an have RAOF tomorrow morning doing the final xmir uploads | 17:00 |
| didrocks | sounds right? | 17:00 |
| kdub | kgunn, unsure. i don't understand how that forces anything to be right though | 17:00 |
| kdub | so it might just be shuffling the problem around | 17:00 |
| RAOF | didrocks: I'm at XDC at the moment; I can do uploads, but not necessarily when you want them :) | 17:03 |
| didrocks | RAOF: can you just email me a link with the 3 xmir packages source I need to upload then? | 17:04 |
| didrocks | I'll do that for you in that case | 17:04 |
| RAOF | didrocks: Sure | 17:04 |
| didrocks | RAOF: thanks! I'll upload that my tomorrow morning then | 17:04 |
| didrocks | kgunn: ^ | 17:04 |
| kgunn | cool | 17:04 |
| RAOF | didrocks: Oh, it's already in the experimental-prevalidation PPA. Are you not going to be copying that over? | 17:04 |
| greyback_ | didrocks: kgunn: unity-mir version bump branch https://code.launchpad.net/~gerboland/unity-mir/bump-mirclient-to-3/+merge/187071 | 17:06 |
| === chihchun is now known as chihchun_afk | ||
| didrocks | RAOF: we can't copy from that, just send me the link of the name of the source then, I'll just resign and uploda then | 17:06 |
| didrocks | RAOF: I prefer to have an email with the name to ensure we don't forget anything ;) | 17:06 |
| RAOF | didrocks: Ok. I'll do that over lunch. | 17:06 |
| didrocks | greyback_: approved (not top approved yet, see the comment) | 17:07 |
| didrocks | RAOF: thanks! | 17:07 |
| greyback_ | didrocks: ack | 17:07 |
| kgunn | didrocks: greyback_ ricmm mterry ... please do so https://code.launchpad.net/~kgunn72/mir/changlog-update-0.0.11-for-mirserver/+merge/187070 | 17:08 |
| didrocks | kgunn: commented | 17:11 |
| didrocks | https://code.launchpad.net/~kgunn72/mir/changlog-update-0.0.11-for-mirserver/+merge/187070/comments/426686 | 17:11 |
| kgunn | didrocks: ack all that...but for server, i thot ffe was for touch ? hence blanket...only client has ffe right ? | 17:16 |
| mterry | ah, that'll teach me | 17:16 |
| didrocks | kgunn: oh, it's just to list it, so that we don't get any issue with the release team :) | 17:17 |
| kgunn | didrocks: ok...will include | 17:17 |
| didrocks | thx | 17:17 |
| kgunn | didrocks: but going fwd...if we break server again....then what ? | 17:17 |
| didrocks | kgunn: no need to list a bug | 17:18 |
| didrocks | kgunn: but otherwise, same procedure | 17:18 |
| kgunn | cool | 17:18 |
| didrocks | mterry: I feel sad that you approved :p | 17:18 |
| didrocks | mterry: I can do now 5 NEW without looking at the copyright file, relying on you catching up during the MIR ;) | 17:19 |
| didrocks | that will be your punishment! | 17:19 |
| mterry | didrocks, Now I feel sad too | 17:19 |
| didrocks | ahah | 17:19 |
| kgunn | didrocks: ok...think its right now... | 17:24 |
| kgunn | greyback_: thanks for the list of headers | 17:25 |
| didrocks | kgunn: the syntax is (LP: #1229212), not (LP#1229212) | 17:25 |
| ubot5 | Launchpad bug 1229212 in mir (Ubuntu) "[FFe] mir api bump post sprint for libmirclient" [Undecided,In progress] https://launchpad.net/bugs/1229212 | 17:25 |
| kgunn | didrocks: ok...one more time | 17:25 |
| didrocks | kgunn: otherwise, almost good (well, you have -0ubuntu1, but daily release will strip and fix that anyway) | 17:25 |
| kgunn | didrocks: ok...just the : and that's it? | 17:28 |
| didrocks | kgunn: : + space | 17:28 |
| kgunn | didrocks: ok...for real for real...it should be done :) | 17:31 |
| * didrocks checking :) | 17:31 | |
| didrocks | kgunn: globally approved ;) | 17:32 |
| kgunn | racarr: you still around or flying ? | 17:34 |
| kgunn | greyback_: so weird...i must have had a bug in my bash...those were there....just moved them, it now seems to work | 17:44 |
| * kgunn so lucky | 17:44 | |
| didrocks | kgunn: I didn't point them out on purpose, not that important (you will remain with "kg" as the signer) | 17:46 |
| === dandrader|lunch is now known as dandrader | ||
| kgunn | greyback_: ok...lucky me has updated again | 17:48 |
| greyback_ | kgunn: did you push? | 17:49 |
| kgunn | greyback_: yes but just... | 17:49 |
| greyback_ | kgunn: ok, approving here | 17:50 |
| kgunn | kdub: back to the forcing 2nd overlay...i thot you were thinking it'd force diff code paths? or...maybe that wasn't your idea & someone else was puting it fwd ? | 17:53 |
| kgunn | kdub: looking at the code, only thing i saw was an extra invalidate | 17:53 |
| kdub | kgunn, i've patched around the obvious problem with the code path, which didn't make much difference | 17:54 |
| kgunn | kdub: so, atm we're back to needing to try CM 10.2 | 17:54 |
| kdub | kgunn, well, at the moment, i'm trying to compile the kernel so I can dig right down to the hardware and figure out what's going on | 17:55 |
| kgunn | kdub: need help? or are you able to build & boot ? | 17:55 |
| kdub | the kernel guys were helpful in pointing me to how to build, now its just a matter of me getting my chroot in order | 17:56 |
| rf9020 | ps aux | grep unity-system-compositor | 19:14 |
| rf9020 | patricia 2833 0.0 0.0 4464 808 pts/0 S+ 21:12 0:00 grep --color=auto unity-system-compositor what is it ? | 19:14 |
| ogra_ | the process entry for the grep you are running | 19:15 |
| ogra_ | ps aux | grep unity-system-compositor | grep -v grep | 19:15 |
| ogra_ | that suppresses finding yourself while searching for something | 19:16 |
| rf9020 | my problem on xorg kernel 3.11.0.8 reso 800x600 with nvidia 319.32 monitor 24" , mir ok if driver no proprietaire , , if driver Nouveaux screen blank !! | 19:18 |
| rf9020 | how to have 1920x1200 | 19:18 |
| === mterry__ is now known as mterry | ||
| === dandrader is now known as dandrader|afk | ||
| racarr | Return of working computer! | 19:37 |
| racarr | kgunn: I am still around. I am flying on wednesday | 19:37 |
| racarr | I was briefly out due to a failure of my laptop power cable | 19:37 |
| rf9020 | what video card supports 100% mir | 19:39 |
| racarr | is amd64 ci still | 19:44 |
| racarr | problematic | 19:44 |
| racarr | it seems to be :( | 19:44 |
| RAOF | rf9020: Your choice of non-ancient intel, nouveau, radeon. | 19:46 |
| === dandrader|afk is now known as dandrader | ||
| rf9020 | ?? radeon ok and nvidia ?? | 19:48 |
| racarr | nvidia is ok as long as the nouveau supports the card | 19:48 |
| rf9020 | i have 8500 gt nvidia | 19:49 |
| racarr | http://nouveau.freedesktop.org/wiki/FeatureMatrix/ | 19:49 |
| rf9020 | nouveau with 8500 gt screen blank with cross | 19:51 |
| rf9020 | no click right , no terminal , but ctrl+alt f1 ok | 19:53 |
| RAOF | Oooh, sweet! There's a new new DRI2Authenticate mechanism that'll let us async that bastartd. | 19:53 |
| kgunn | racarr: hey, was just checking on what you decided to attack - my preference is api :)...basically, pat sent a mail saying all ui changes should be deferred anyway... | 20:22 |
| kgunn | almost viewing it as risk now | 20:22 |
| racarr | kgunn: API it is :) | 20:25 |
| racarr | that and trying to get anything to pass jenkins | 20:26 |
| kgunn | racarr: no doubt on making jenkins happy....jenkins is so fickle | 20:27 |
| racarr | [ FAILED ] DisplayConfigurationTest.display_change_request_for_unauthorized_client_fails (103032 ms) | 20:32 |
| racarr | fhqwfqwhfqwf | 20:33 |
| ricmm | kdub: kgunn can we talk here | 21:06 |
| ricmm | the other channel is noisy | 21:07 |
| ricmm | kdub: can you explain more about what was failing? | 21:07 |
| ricmm | we know SF does the 2 layer rendering with the same kernel you are trying | 21:07 |
| ricmm | and same HWC | 21:07 |
| kdub | ricmm, but i'm not convinced they don't have rendering glitches either | 21:08 |
| ricmm | just use the phone with surface flinger then | 21:08 |
| kdub | ricmm, i haven't gotten the "hwc's force-2 layers" to work | 21:08 |
| ricmm | you wont see the issues | 21:08 |
| ricmm | they might still have other issues, but its not the same issue for sure | 21:09 |
| ricmm | vsync fencing is happening the right way | 21:09 |
| kdub | i won't see the flashing in the same way, i do see problems though | 21:09 |
| ricmm | and im not talking android, im talking the default image with SF + unity8 | 21:09 |
| ricmm | what problems? | 21:09 |
| ricmm | well I mean but the difference is majestic | 21:09 |
| ricmm | its clearly obvious that something is failing with the Mir one | 21:09 |
| ricmm | thats triggering this | 21:09 |
| ricmm | the 2-layer rendering being the most obvious cause for the missed fencing on vsync | 21:10 |
| kdub | but the problems i see with surfaceflinger are similar to vsync timing issues | 21:10 |
| ricmm | porting to 10.2 is not really an option in our current timeframe | 21:10 |
| kdub | right | 21:10 |
| ricmm | what I want is exactly what surface flinger does to circumvent the problem, in our code | 21:10 |
| ricmm | at least as a PoC or a hack until we can move forward with 10.2 | 21:11 |
| ricmm | as it wont happen before 13.10 | 21:11 |
| kdub | right, i'm trying to patch what we have for that | 21:11 |
| kdub | and, i haven't been able to prove that surfaceflinger is doing the right thing either | 21:11 |
| kdub | so i'm currently in the process of trying to patch things | 21:11 |
| kdub | preferably hwc | 21:11 |
| ricmm | I understand, what I'm saying is that even if SF is also having vsync issues... they are not as apparent to the naked eye as the Mir issue | 21:11 |
| ricmm | so you cant put them on the same plane | 21:12 |
| ricmm | as the SF flinger one can be considered "flawless" as we've been shipping it for a while, even if there are underlying hwc issues | 21:12 |
| ricmm | they clearly dont generate the same outcome as the current tearing with Mir | 21:12 |
| ricmm | so, I'd say please keep trying to patch/hack to do what SF is doing re: 2lyr rendering | 21:12 |
| ricmm | until we can trigger the right fencing | 21:13 |
| ricmm | because it feels like you are right on the money with the missed syncs analysis, we just havent been able to get the hack to work to prove it | 21:13 |
| kdub | i can still pursue, i'm just worried that chasing the hack won't be sufficient | 21:14 |
| kgunn | kdub: maybe we can enlist duflu here | 21:15 |
| ricmm | kdub: unless another better option comes along... | 21:16 |
| ricmm | and they pretty much can only come from you as you are the one who knows the issue at hand | 21:16 |
| ricmm | 10.2 is not a realistic option for shipping | 21:16 |
| kdub | sure, understand that | 21:18 |
| kgunn | ricmm: sorry...got distracted, i still think if we could gen up a CM10.2 not for releasing but just to test...it would be very very useful | 21:28 |
| kgunn | ricmm: just confirming that we could get some love from rsalveti to give it a shot at least | 21:28 |
| kdub | i'll try to stabilize the 2 layer hwc hack i have right now | 21:29 |
| === jhodapp is now known as jhodapp|afk | ||
| ricmm | kgunn: it might not even be possible to get that love | 21:34 |
| ricmm | as both salveti and I are buried with stuff | 21:35 |
| kgunn | ricmm: is flicker not important ? :) | 21:35 |
| ricmm | yes, but I prefer if it can be sorted internally in Mir | 21:35 |
| ricmm | as, if SF can do it, Mir can do it | 21:35 |
| ricmm | thats our motto after all ;) | 21:35 |
| kgunn | ricmm: yeah...but i thot we were cool to get some help fom rsalveti ? ...no ? | 21:36 |
| olli | kgunn, isn't that an asac q? | 21:36 |
| kgunn | ricmm: i don't even know how much work i'm asking for....can you give a swag ? | 21:37 |
| kgunn | olli: yeah..i suppose so | 21:38 |
| kgunn | ricmm: it doesn't have to be rsalveti if someone else knows how | 21:38 |
| ricmm | guys, if the hack works for SF it will work for us | 21:39 |
| ricmm | its just not complete yet | 21:39 |
| ricmm | lets not sign off on a huge task as testing 10.2 can be | 21:39 |
| kdub | the way sf and mir composition is triggered is a bit different... i guess right now, the first focus is testing the hack and seeing if it does help the flicker | 21:40 |
| ricmm | kdub: lets exhaust that route as I asked on friday first | 21:41 |
| ricmm | and then we'll see what we can do if all that fails | 21:41 |
| ricmm | rsalveti isnt back until tomorrow and he is the one who can tell you how much work it will be to test 10.2 | 21:41 |
| kgunn | ricmm: thanks... | 21:41 |
| ricmm | but I can assure you than more things will be broken, and that we dont have the manpower to get it in place for 13.10 | 21:41 |
| ricmm | so it is not really a viable solution | 21:41 |
| ricmm | therefore whatever cycles we spend there, will be wasted time | 21:41 |
| kdub | yeah, it seems like it might be a lot of code to move | 21:42 |
| kgunn | but what if that is the right answer? | 21:42 |
| ChickenCutlass | kgunn, getting 10.2 up and running is really a big task. | 21:42 |
| ricmm | it is not, and that is proven | 21:42 |
| kgunn | rock/hardplace | 21:42 |
| ricmm | how? because I dont see the same degree of flickering in SF | 21:42 |
| ricmm | so, that is not really the only solution | 21:42 |
| ricmm | we need to contain the fix in the least amount of code | 21:43 |
| ricmm | and that is either Mir or hwcomposer itself | 21:43 |
| ricmm | but not moving the whole base to 10.2, that will break some serious havoc :) | 21:43 |
| kgunn | or FB | 21:43 |
| ricmm | or fb yes | 21:44 |
| kgunn | i'm not asking for CM10.2 to be released | 21:44 |
| kgunn | and i got the message | 21:44 |
| kgunn | its too big for you to attempt | 21:44 |
| kgunn | so new requirement is "flicker less than or equal to SF" | 21:44 |
| kgunn | ..not no flicker | 21:45 |
| ricmm | the req is feature/performance parity with SF | 21:45 |
| ricmm | if we get that, we ship Mir | 21:45 |
| ricmm | and as not shipping Mir isnt an option, we will get that for sure ;) | 21:45 |
| kgunn | ricmm: i used to do phablet-dev-bootstrap but...did it and couldn't build...did it change to phablet-devenv-bootstrap | 21:57 |
| ChickenCutlass | kgunn, phablet-dev is still the way | 21:59 |
| kgunn | ChickenCutlass: great...its just me | 22:00 |
| kgunn | racarr: any thots on the amd ci test ? | 22:08 |
| kgunn | racarr: i think francis is going to try to build/ci on a dedicated/non-shared host to see if that helps | 22:18 |
| ice9 | what is the benefit of Mir over Wayland? | 22:36 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!