[00:02] even now they still have the irish teletext page numbers... but its not on waybackmachine as robots.txt stopped it... http://www.rte.ie/aertel/desktopxhtml/103-1.html [00:02] that's... not teletext [00:02] looks like red button [00:02] no but back when teletext was still in use. the pages looked identical to the teletext pages [00:03] now it's more... digital service [00:03] yeah [00:03] but still the same page numbers [00:03] About ready to lob my new laptop out the window thanks to bug 1522922 [00:03] bug 1522922 in xserver-xorg-video-intel (Ubuntu) "Screen flickering in Intel i915 driver" [High,Triaged] https://launchpad.net/bugs/1522922 [00:03] digital launched without page numbers. everyone hated it [00:03] so they brought back the numbers, but it still sucks [00:03] drink enough coffee to make you phase at same speed as blink? :-P [00:04] tried patching out the bits that people say broke it, and kernel failed to build :( [00:04] if you drop the screen res/colour depth does the flicker go away? [00:05] woo popey pretty much lives on IRC even now :d [00:05] popey: that sounds like that skylake driver issues in older kernels, which ver you dealing with? [00:05] its not an older kernel issue [00:05] oh ok [00:05] it affects these chips if you drive it less than the full capability [00:05] so if it's capable of doing 4k external (e.g.) but has a 1080p you will see it [00:05] i read about flickering happening on skylake without this preliminary support param being added [00:05] as 1080p panel is less than capacity [00:06] yay Linux [00:06] It is allegedly two patches, they have a new patch (two in fact) but those got rejected by Intel QA [00:06] some months ago [00:06] and it's just sat there unfixed since [00:06] i can't remember the last time i had a problem with nvidia :P [00:06] and we ship this kernel on thursday [00:07] popey: Then revert that pair? [00:07] as i said, i tried but kernel biuld failed [00:07] <- not a kernel hacker :) [00:07] popey: Oh, got the error? It's probably pretty easy to fix the build error [00:07] git revert? [00:07] i have the two patch files and -R reverted them [00:08] lemme try the build again, have rebooted since so lost scrollback [00:08] did you commit the changes after? [00:08] no, I'm not using git [00:08] oh [00:08] did it even reverse apply clean? [00:08] lol why not?? [00:08] uh [00:09] because i was building from the source that built the kernel I'm running [00:09] building from the source deb? [00:09] yes [00:09] ew [00:09] so i could shove it in a ppa [00:09] i don't even know how to make that work [00:09] okay, I'll do this on my own, never mind. [00:09] without modifying the source [00:09] popey: Yeh throw us the errors you get, it looks like it should revert OK [00:10] popey: Remember to patch -R them in the opposite order to that they were applied [00:10] ok [00:10] ta [00:10] why are you running the internal LCD at not-it's normal-res ? [00:11] I'm not [00:11] I'm running it at 1080p, native res [00:11] the bug says if you run it less than the chipset capability [00:11] I mean, it's entirely possibly I'm seeing a different bug [00:11] oh less than chipset capacity, I thought it was less than panel capacity [00:11] which is why I want to test reverting these two patches to see [00:11] because others reported it worked [00:12] hm [00:12] http://paste.ubuntu.com/15884141/ [00:13] oh, you've got a machine called gort too :-) [00:13] :) [00:15] popey: OK, without understanding anything about the actual hardware..... [00:15] popey: You see that line where it says train_set_valid = true; ? try making that false. [00:16] popey: and leave the rest as is [00:16] undo the patch revert? [00:16] popey: Yeh, back to clean source [00:16] ok [00:16] ta [00:16] popey: My theory goes that the whole thing is a cache and it only uses the cache if that flag is valid; so if it's never valid then perhaps it'll just do the old path [00:20] popey: Or apply the patch in comment 31 [00:26] that's a really weird patch [00:26] it doesn't actually introduce any new behaviour, it just sometimes avoids doing the old behaviour [00:28] ali1234: What's so weird about that? [00:29] ali1234: It's an optimisation to try and avoid doing that thing sometimes [00:29] yes [00:29] but [00:30] it seems very long winded [00:30] ali1234: I suspect there's a lot of ordering stuff about when you can avoid it and error recovery and stuff - which they've obviously not got right [00:31] yeah that big bit with DRM_DEBUG_KMS [00:31] it will always run if control reaches it [00:31] i suspect that is the problem [00:31] it tries to reuse the stuff, then always hits that bit of code and decides actually it can't reuse it [00:32] that's the only bit that actually does anything [00:33] but the logic of that is weird [00:33] if (thing is valid) { thing is not actually valid}; [00:34] i'd heard from somewhere that intel engineers are finding their own hardware a bit of a challenge even under Windows these days [00:34] i would have explicitly set the flag to not valid, reversed the if, and then set the flag back if it was successfully made valid [00:34] The intel video driver has been awful on linux for a long time now [00:35] mmm [00:35] daftykins: The damn stuff is all getting very complicated - VGA was just nice and simple; this things having to do protocol negotiations with the LCD panel, and people care about really really low power and there's firmware blobs and who knows what [00:36] oh totally [00:36] (That probably means there's firmware on the panel as well? That's a scary thought) [00:36] well VGA seems to be a constant hassle with EDID woes these days, judging from my time in #ubuntu giving support [00:37] 9 times out of 10, switching to a digital interface would at least let you see what was going wrong more clearly, if not just solve it immediately [00:37] the fact it's still delving into xorg.conf after all these years though is such cringe [00:38] yeh fighting X seems to have got more common again [00:38] a really gifted guy that comes in #ubuntu sometimes has an amazing setup of 5+ displays from a business line Dell laptop which he has an external graphics card connected into via expresscard [00:39] amazing amount of fiddling to get his display setup going :D [00:40] oh heck, is that one display at each point of the pentacle to get that to work? [00:40] hmm i think i see the problem [00:40] here is the context: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/i915/intel_dp.c?h=v4.4#n3735 [00:42] ali1234: That return is unusual [00:42] immediately above that, intel_dp_reset_link_train was called [00:42] then if the flag is marked as valid it is called again in the loop [00:42] and then the return [00:43] but presumably that doesn't happen if it didn't happen the first time [00:43] then below it we have some "voltage_tries" stuff [00:43] so that loop is designed to run a few times, maybe, before it works [00:45] i suspect deleting lines 3724-3730 might actually be the correct fix [00:45] maybe someone forgot to delete it [00:46] or possibly only call it if (!intel_dp->train_set_valid) [00:48] ali1234: I'm not sure - I don't understand how much it's trying; my reading is it's resetting that clock recovery stuff, if that comes up ok then it's done; only if that fails does it then go into the stuff where it tests the training set stuff? [00:49] i read it like this [00:49] before the patch: call intel_dp_reset_link_train, then make sure everything came up okay [00:50] if not, call intel_dp_reset_link_train and loop [00:50] up to 5 times [00:51] so what's the point of the new bit of code? [00:51] it only makes sense if they are trying to avoid the reset entirely [00:51] in which case... the reset before the loop shouldn't happen [00:51] ali1234: I'm not sure if it's one reset or a few; is the link training the same thing as the clock recovery? [00:52] lines 3725, 3757, 3776 [00:52] all the same call [00:53] hmm true [00:54] ali1234: Yeh except the 2nd patch in the series changes the intel_dp_reset_link_train [00:55] i'm looking at the 4.4 flat code [00:55] ali1234: It avoids reseting the data if the flag is true [00:55] second patch? [00:55] you mean the memset? [00:55] yeh [00:56] i dont think that is relevant [00:56] also that's the first patch [00:56] ali1234: so in both cases it wiggles the training stuff, but after these patches it starts the training stuff from the previously cached settings [00:56] you can tell because it adds the bool, the second one just uses it [00:57] yeh, but I'm nots ure it's bad to call that dp_reset_link_train [00:57] does it actually look at the flag inside that function? [00:57] if it's just writing the same data back [00:57] yeh [00:57] oh is that the function where the memset happens? [00:57] i see [00:57] ali1234: https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/patch/?id=4e96c97742f4201edf1b0f8e1b1b6b2ac6ff33e7 [00:57] ali1234: See the first change in there [00:57] yeah [00:58] Heh, i love just mentioning something in here and you guys having fun with understanding the code :) [00:58] popey: *trying* understanding the code [00:58] * penguin42 knows very little about eDP, the intel training stuff or any of that [00:59] but I think most serial systems have training phases these days [00:59] so... [01:00] easiest way to revert this is to set the flag to false right before the line 3725 [01:00] then it should behave exactly as before [01:01] hence my suggestion before of just removing the thing that ever set it to true [01:01] ..still building [01:02] yeah that should also work [01:02] although... where does it get initialized the first time? [01:02] popey: Still? [01:02] popey: you're using -j right? [01:06] um [01:06] train_set_valid is never initialized? [01:06] hmmmm [01:06] (possibly) [01:06] so maybe it's just trying to use garbage? [01:06] that would certainly explain the flickering [01:07] that structure might be safely 0 initialised as a whole [01:09] ali1234: I think it might be part of the intel_dig_port that they use kzalloc to allocate [01:10] it's only doing j2 i think [01:10] comes from enc_to_intel_dp which is converting some other thing [01:11] oh, -j4 [01:11] popey: My normal thing is 2x the number of cores I have [01:11] popey: The big problem with distro configs is they enable *everything* [01:12] popey: There are probably no users of USB lego mindstorm drivers on the same machine as fibre channel storage and 100Gbps infiniband connections [01:12] probably [01:13] yeah, I'm trying to change as little as possible [01:13] because in the past I have changed too much and knackered some other part of my daily workflow [01:15] * penguin42 did about 15 kernel builds during the week trying to bisect down a screwup that was breaking something I was using; slowly got more annoyed removing random USB drivers and the like to speed it u [01:15] p [01:16] yeah, it's in intel_digital_port which is kzalloc'd [01:16] i wanted to try a really old kernel, turns out it didn't have the aes bit to decrypt my luks hdd [01:16] really? Hasn't that stuff been there for ever? [01:17] sure, but we switched to a new cypher [01:17] from ~4.x [01:17] oh [01:17] so my default install can only be unlocked with a 4.x kernel [01:17] which is a useful datapoint [01:18] heh, building modules for TV cards I will never own [01:19] I wonder how many of these drivers are in use by like one guy somewhere [01:20] popey: Often the case some of this stuff is now none [01:20] i know for a fact that some of those DVB drivers don't work any more [01:20] what, kernel changed and they weren't updated? [01:21] popey: I do have an unfortunate problem with one of my dads old laptops, the VGA/DRM driver for it was removed because someone noticed it was hopelessly insecure and no one wanted to spend the time to fix it; so the laptop is now using the generic drivers on 14.04 and is slow as hell [01:21] i don't know why they don't work... they just don't [01:21] popey: Welcome to bit rot [01:21] ew [01:21] vesa [01:21] yeh [01:22] ooh, it's building zfs [01:23] which is separate it seems [01:23] checking zfs license... CDDL [01:23] heh [01:26] whilst i'd be the first to say people should no longer be using them, those that insist on keeping using netbooks have basically got a non-workable system with a lot of modern stuff now because the intel driver seems to have regressed for GM965 since 12.04 [01:26] there's not that much choice on the low power x86 [01:27] The T43 I got given this week turned out to work quite well with Ubuntu MATE 16.04 :) [01:27] Not bad for a >10 year old computer [01:28] yeah anything but netbooks really [01:28] that sounds like a possible core 2 duo? [01:28] i _think_ so [01:28] sometimes i just think there are some folk who are enabling bad decisions too much in #ubuntu for claiming lubuntu will resurrect some of the museum pieces out there [01:28] Pentium M 2.0GHz [01:29] Was hilarious getting the factory XP install on it [01:29] ah the M [01:29] :D [01:29] watching all the IBM nonsense self-install was fun [01:29] popey: Did you have to pass the force PAE during the install? [01:29] no [01:29] interesting; we had to do that for someone with a mate install the other week [01:30] not all pentium M' are equal? [01:30] i seem to recall the lubuntu installs had to be picked at some point as they came with a forcepae config as default or some such [01:30] yeah i think later ones may've changed something [01:30] maybe it's default yeah, dunno [01:30] popey: Apparently not, and I don't really understand why - some of them have PAE, some of them don't; and apparently some of them have it but say they dont [01:30] it's a cracking little laptop [01:30] We are a 5 Thinkpad household now [01:31] the T43 was apparently the last one made under IBM name [01:31] and has no windows key [01:31] I tend to use my Model M keyboard that's similarly windows-key-less [01:31] there's a wiki page I found with a list of niche features people look for in thinkpads, with the list of models that fulfil the requirements [01:32] i use super too much to go without [01:32] I do find it odd that my T450 shipped with all the VT stuff switched off in the BIOS [01:34] yeh this day and age [01:34] that's pretty common across all manufacturers, i once read a claim that it's because 'malware could take over your host OS and turn it into a VM, inserting programs you couldn't see running' [01:34] ahh [01:35] actually the gifted guy i mentioned earlier helped me patch the BIOS on a Sony TZ i have which doesn't have the VT option in its' BIOS and was disabled [01:35] speaking of which, i should play with coreboot one day [01:35] quick edit after showing him a copy of the NVRAM i think for him to oonfirm the options position and hey presto - KVM got enabled :> [01:35] now i have machines that can use it [01:35] yeh I've heard that idea of rootkitting like that (I can't remember the name someone came up with); I find it unlikely but hmm [01:36] right, it's apparently now -0c out there; which is very annoying for mid April [01:36] :O [01:37] lots of LDing now [01:37] feels close to the end [01:40] maybe not :) [01:40] i was surprised to see VT-d and VT-x were off on my new skylake build as default, too === WillMoogle_ is now known as WillMoogle [01:48] bah, gonna leave it running and sleep [01:48] thanks for the help chaps [01:48] I remember sleep. [01:48] yeh sleep sounds a good plan [01:53] and then they were gone [01:57] daftykins: Any sign of your package yet? [02:01] nah still stuck in Germany i think on the last update, lets see if anything else has changed [02:02] i'm not sure how it'd make it here from there, probably via London [02:08] amazon reckon Wednesday! [02:08] still, i think waiting is the best part of new things :) [02:08] That's why I still shoot on film. ;-) [02:08] hehehe [02:09] That and I can't afford a Pentax K3 [02:09] ...and it's nice to have at least one hobby that's not computerised. [02:10] one of my clients is hugely into his cameras, constantly buying new models that once again rename the RAW file extension they use and needing yet another program to open them [02:10] often has these quite fancy super compact Leica models with huge displays on the back [02:11] Sounds dismal. How do his prints look though? [02:11] full size SLRs too, often goes on about all the ISO # stuff to me [02:11] well in the true modern style nothing ever gets printed :) [02:11] I use Ilford 400 film. 400 is my ISO. ;-) [02:11] just dumped to disk and maybe viewed on screen [02:14] I suppose if you're taking photographs for Web sites or something that might be ok. [02:14] ...or to email to distant relatives. [02:14] which way would be ok? [02:15] images should never be emailed :) [02:15] daftykins: Why's that? [02:15] Because they double in size when you attach them? ;-) [02:16] wasn't designed for it, just bad netiquette as well [02:16] daftykins: I understand but in practice very few people know or care. [02:16] oh yeah, they need some education :> [02:16] ...but you can email a link to a download. [02:17] it's quite funny seeing a client getting links to picasa from a guy that's managing his property refurbishment down in Spain right now [02:17] the links aren't working :D [02:17] "this one works for me" the guy says, he showed me it and i had to point out that well, of course it will if you're logged in as the owner... :D [02:19] I've never used Picasa [02:19] i thought i'd read Google were shuttering it [02:19] yeah... they are [02:20] oh it's become the main Google Photos instead, ok [02:22] Ah, I've used that one. [02:23] * ball thinks about buying a new hard disk drive [02:24] out of space? [03:25] daftykins: No, I'm getting disk errors from the one I installed. [03:26] It /might/ be because the case wants it installed upside down or it might just be a crummy old disk drive. [03:26] If I buy a 2.5" SATA disk I can mount it the right way up. [03:26] checked the ol' SMART table with smartctl? [03:27] Yes. Let me check it again now. [03:27] I get green lights all the way down and 0 reallocated sectors. [03:27] Temp is +27C [03:28] Write error rate is 1805 [03:28] ...which is odd. [03:28] power-on hours 12,798 [03:29] ...that's less than two years. [03:29] Perhaps I just need new cables or it doesn't like being upside down. [03:30] nothing for current, pending or remapped sectors? [03:30] Current pending sector 539. [03:30] Offline uncorrectable 160 [03:30] oh dear, she's toast [03:30] 0 remapped. [03:31] yip that one needs to go to the great electronic graveyard [03:31] daftykins: Oh good. I'll wipe it and throw it away then. [03:31] I have a 7200 RPM 2.5" 500G disk picked out. [03:31] I suppose I could order that tomorrow. [03:31] brb [03:31] so small! [03:32] daftykins: No real benefit to a 1.8" in this case ;-) [03:34] hehe [03:34] i actually have been given a Sony VAIO TT21 by a client for decommissioning, it contains a 1.2" SSD [03:34] If I go with 2.5" Vs. 3.5", I can at least put the disk the right way up. [03:35] i can't argue with that, i feel weird mounting drives in odd positions - a lot of Dells i've put in for folk have them sideways o0 [03:37] It used to be that the manufacturers had approved orientations: upright or on one side but not upside down or on end. [03:37] ...I date back to MFM and RLL drives though so that may be old information. [03:37] makes sense to me [03:58] hello knightwise [04:17] Goodnight everyone. === funkyHat1 is now known as funkyHat === Chrisfu- is now known as Chrisfu === Matrixiumn is now known as Xack [08:41] morning boys and girls. === xack is now known as Guest81809 === Matrixiumn is now known as Xack === zmoylan-1i is now known as zmoylan-pi === m0nkey__ is now known as m0nkey_ [12:19] !isitoutyet [12:19] No, it's not out yet! [14:43] penguin42: that one line fix has stopped the flicker [14:43] thank you [14:44] popey: which fix? [14:44] popey: Wahey! [14:44] popey: I'm an expert voodoo programmer [14:45] !isitcompiledyet [14:50] MartijnVdS: uh, train_set_valid = true; set to false [14:50] rebuilt kernel, installed no blanking flicker [14:51] popey: ah :) For me the enable_fbc thing worked well enough (on last year's XPS 13 Developer edition) [14:52] what kind of flickering were you getting [14:52] I have two separate issues [14:52] random black screen blanking periodically, fixed by above [14:52] random flickering in chrome/chromium [14:53] popey: Is the flicker something just within the drawing area or the screen as a whole? [14:53] doesn't chrome have hidden options to change how it handles video display? [14:53] I had random full screen flickers, sort of "out of sync" for a fraction of a second [14:53] penguin42: which one? :) [14:53] popey: The chrome one [14:53] MartijnVdS: yeah, same here [14:53] penguin42: blocks of the screen [14:53] but since 4.4/4.5 it's fine for me with default settings [14:53] like a tile here/there [14:53] MartijnVdS: tried 4.6, still broken here [14:53] (Debian testing) [14:53] popey: which model do you have? [14:53] T450 [14:54] Ah not a Dell [14:55] There are so many weird small issues with new Intel chips lately [14:55] yup [14:55] MartijnVdS: popey needs to have a nipple to keep him happy [14:55] :) [14:55] penguin42: xD [14:56] I really love my XPS [14:58] the t450 also has 3 real pad buttons as well I think which is a big plus [14:58] I just use 1/2/3 finger taps [14:58] And I've disabled the touch screen too because .. dirty screen [15:00] the other day I saw someone using a touchscreen in a text editor surprisingly effectively [16:25] So : now snappy is a part of Wily : what should I use it for? Currently still using apt for most things. [16:25] Well, everything AFAIA === zmoylan-1i is now known as zmoylan-pi === funkyHat2 is now known as funkyHat [19:19] ali1234: did we find out what's new in the pi3? [19:19] the pi zero? no [19:19] er, new pi zero [19:19] yeah [19:19] two weeks [19:19] new information: connectors won't move [19:20] unless you have a hammer... [19:55] hrmm my favourite choice kornbluth seems to be having issues today [19:58] https://twitter.com/swiftonsecurity/status/721786968795820033 [20:00] still part of the path then they laugh at you, then you win... :-) [20:01] https://twitter.com/swiftonsecurity/status/721790151026810880 [20:02] that clip is so apt XD [20:03] you mean, git apply; ./configure; make; make install! [20:03] what was it that xubuntu was doing for me on a laptop, something that really goes against the basic use of a portable machine [20:03] it worked? :-P [20:03] nope! [20:04] it wasn’t hibernating? [20:04] see i can't abide using suspend or hibernate. for me a laptop is either on or off and nary another mode is tolerated... :-) [20:06] zmoylan-pi: couldn't agree more, don't do all these silly power states [20:06] what movie is this clip from? https://twitter.com/swiftonsecurity/status/721791105474297857 [20:06] if it's suspend and it's in my bag there's a small chance the fan will come on and suck some dust or crap in my bag into the laptop. and hibernate just took forever when i tried it on windows and the few times on linux... [20:06] you haven't owned an SSD yet have you? :) [20:07] either way it's not worth the disk writes [20:07] diddledan: got the replacement Surface Pro 4 back, turned it on to find it had the 10240 Win10 build on - ugh, fully reinstalled and still suffers from battery drain when in sleep [20:08] totally useless devices until they properly fix that [20:08] * zmoylan-pi misses my psions... now they did power right... [20:09] press on, psion ready, press off psion goes asleep and waits [20:09] palm worked as well... [20:09] quite liked my 3c :) got it as a kid wanting a laptop but having no money, though [20:09] the reason i got the psion 3a was that i couldn't afford a laptop [20:10] and it fitted in my pocket [20:10] and had a decent battery life, keyboard, software, community... [20:12] solid state software! [20:12] os, apps, development in rom running multitasking in 512k used as ram AND storage [20:21] back later perhaps, if kornbluth sorts itself out :) [21:03] I don't know what kornbluth is [21:04] a webservice of some kind? === Chrisfu- is now known as Chrisfu === andyjones2001_ is now known as andyjones2001 [23:17] http://www.geek.com/microsoft/microsoft-was-serious-about-linux-apps-running-natively-on-windows-10-1652205/