[00:54] good morning [00:59] hello [00:59] hey MrCollins [00:59] hows it [01:00] early woken, still bit blurry [01:00] coffee to the rescue [01:00] I hear ya [01:00] You must be in Europe or Asia? :) [01:00] Middle Europe IIRC [01:01] belgium [01:01] cool. [01:02] Always wanted to visit Europe etc. [01:02] MrCollins: you should visit topyli [01:02] not trying to make this offtopic. [01:03] Croatia? im lost lol [01:03] MrCollins: Finland [01:03] I would like to! [01:04] I am American but of mainly Irish descent. Maybe thats why, when the rare occasion presents itself, I drink half the bar! [01:04] scandinavian countrys are nice, great gov and highspeed internet [01:05] My grandfather visited France in 1944 :) Thats the last of us to have been across the pond. [01:05] I digress [01:05] sacré bleu [01:05] to the rules hehe [07:36] good morning === akem__ is now known as akem [17:41] ouch! just discovered a remotely exploitable kernel/DRI/GEM bug with a CVE from 2015 that hasn't been fixed [17:41] sorry, from 2013 ! [17:41] :o [17:41] uh [17:41] CVE-2013-7445 [17:42] fossa kernel? [17:42] https://ubuntu.com/security/CVE-2013-7445 which, via links, eventually leads via https://bugs.freedesktop.org/show_bug.cgi?id=106136 to https://gitlab.freedesktop.org/drm/intel/-/issues/110 [17:42] Freedesktop bug 106136 in DRM/Intel "per-process/context memory usage accounting for i915" [Enhancement,Resolved: moved] [17:42] 7 months ago: "As there is no activity, closing this issue." [17:43] basically; a crafted web page can cause a browser to consume all memory via allocating multiple CANVAS elements which consume GEM objects which are not accounted for [17:49] CVE's shouldn't be allowed to "As there is no activity, closing this issue." without a fix being released [17:50] I only came across by tracing some vulnerabilities reported by debsecan, due to BugHunter1000's comments in #ubuntu earlier [17:51] this must be the first time bugzie did something useful. [17:53] i'm glad you spotted this, and am wondering how many other cve's are in deferred state because upstream moved bugtrackers or considered bugs stale. [17:53] the SUSE security team say it's too complex to fix, so left alone [17:53] but upstream should really be on something like this like flies on a ...! [17:53] if this is 'just' a dos issue on desktops then maybe it's not that bad. [17:54] yes :-/ [17:54] it's not - there's a comment about servers running video pipelines being affected too [17:54] I think I'll bring it up in -hardened [17:55] tomreyn: is that really bugzie? [17:55] leftyfb: i do not know, i just notice behavorial similarities. ;) [17:57] leftyfb: actually, no, probably not. [17:59] this person has a much higher level of understanding of what they're talking about (not ubuntu specifically, but security, linux in general). so i jumped to conclusions, sorry. [18:00] BugHunter1000 was wrong about there being lots of outstanding vulns - they didn't do even basic background checks of the CVEs/packages reported by debscan as I have [18:04] yes, that's what made me think it may be him initally [18:06] I don't have time to develop a simple proof/exploit right now - working towards an important deadline tomorrow on something else. Anyone here fancy trying to create one? Looks like it *might* only need a simlpe HTML page with a looping Javascript creating and drawing into multiple canvas elements (so the backing pages are dirty) and not releasing/freeing them [18:10] my javascript skills are lacking. maybe this can be a start: https://codepen.io/2toria/pen/BipvF [18:35] anyone with less than 32GB RAM can test my demo (ensure you've not got anything important running!) [18:37] TJ-: where can i find that? i've got an intel grapjhics laptop running ubuntu 18.04 (i think, haven't used it in a while) with 8 ? GB RAM. [18:37] I've got it created 1920x1080 canvas every second; for up to 1000 iterations - might need to do more and reduce the delay to trigger it earlier [18:37] tomreyn: I'll send a link privately [18:40] thanks, got it. i'll need to update it first of all, will take a while. [18:40] tomreyn: I need to head home now for dinner; will be back on later [18:40] ok [18:41] i'll be around for some more hours [18:41] tomreyn: reduce the sleep to 10 and increase the loop iterations to 99999999 - and in a terminal do "watch -n 1 free" [18:42] I'm feeling like posting a Hacker News item on this to ensure it gets eyeballs on it! [18:42] * TJ- zooms off [18:49] geez, 2 GB upgrades [18:51] what for? [19:14] 18.04, i just didn't have this system running for a good while ;) [19:19] and back! [19:24] TJ-: wb. just started it up [19:24] ram is going down [19:25] thanks. wondering if I should adapt so the current iteration number is drawn inside the new canvas so we can see how far its got especially id/when it starts to die [19:25] i should probably have diusabled swap [19:26] a larger memory allocation poer cycle would be good [19:26] "sudo swapoff" ? [19:27] yes, next run i'll do that [19:28] also may need to randomise the colour and alpha so as to avoid any possible samepage merging [19:31] TJ-: it's surviving, though [19:32] available mem never gets entirely depleted, i guess it does GC still [19:32] i'll try without swap [19:32] tomreyn: not sure if this is the way to trigger it; didn't see any exploit examples so may need to experiment somewhat [19:37] firefox's about:performance says the POC tab only consumes 2.1 MB RAM, so obviously the memory is consumed elsewhere [19:38] top says reserved memory allocation on the firefox container the tab runs in is constantly growing though [19:42] "Gah, your tab just crashed." :) [19:44] check slabtop while running it [19:47] too late, system sully loaded, can't do anything [19:47] *fully [19:47] oops [19:47] well, *next* time you run it.. :) [19:48] geez this thing went hot [19:48] i guess the fans must be dusty [19:49] anyways, i guess it works then, tj [19:49] now i got to have food, bbl [19:53] tomreyn: there's a new version available at the original URL; with this you can set the iteration and delay_ms at top of the HTML [19:53] sarnold: you want to test this too? [19:53] TJ-: nope :) [19:54] TJ-: I already had my fill of i915 memory allocation problems this year [19:54] I've brought my 32GB Ryzen to a stop! [19:54] sarnold: this is across all GPUs [19:56] oh fun [19:58] around 200 iterations and it was making this Zen2 Ryzen laptop with 32GB sluggish [19:58] People still use firefox ? [19:59] Ussat: that is a snide remark [20:00] i'll always consider it my primary [20:00] Ussat: this isn't the first time that you've made unhelpful remarks [20:01] 1) it was a question, and 2) was an honest question [20:01] so....keep YOUR snide remarks out [20:01] a remark is different from a question [20:02] Just like me saying that intelligent people use Firefox [20:02] That is a remark, I asked a question [20:02] it's quite obvious that you're being dishonest - and trying to stir up conflict with that reply even, on ignore you go [20:03] Now to test and measure this on a range of browsers [20:03] OH noes....ignore [20:03] Ussat: oh heck yeah, chromium-browser never felt like 'linux'. granted firefox is getting further and further away from that feeling :( [20:04] I switched all my browsers to Chrome [20:04] Ussat: it used to be that firefox could be made to handle ^W and ^U like vim / emacs / bash etc but they took that away from me a while ago.. [20:04] middle-click paste was a big one [20:04] does chromium-browser navigate to an url on middle-click paste? [20:05] chromium/chrome/blink are becoming the new Internet Explorer [20:05] Not sure what you mean by chromium-browser......I guess thats the chrome upstream ? and yes [20:05] its configureable [20:06] hmm I may need to give it another look [20:07] pentadactyl used to be reason enough to stick with firefox, but the webmumble things that replaced the old plugin interface just aren't as good [20:08] I like chrome because it seemlessly syncs across all my systems, devices [20:09] most browsers do I think [20:09] certainly Firefox does [20:10] Firefox does not sync browsing history, or open tabs [20:11] Yes it does [20:12] Not from what I have seen, but its doesnt matter, I prefer Chrome, one of the reasons is the many usefull plugins [20:13] https://support.mozilla.org/en-US/kb/view-synced-tabs-other-devices [20:13] Again, the main reason is the extensions I use [20:19] heh, that sounds like my firefox use.. tridactyl, noscript, privacybadger, open in browser [20:20] Eventually, I imagine I will migrate away from chrome to the new MS Edge, since its so much better on my battery on my Surface Pro 7 [20:21] lol [20:22] OH...I thought you had me on ignore..... [20:22] I was so upset when you said that [20:23] I mean the new Edge is basically Chrome, so...why not ? [22:43] hey small-data, how are you? [22:44] tomreyn: never better! you?? [22:45] good, thanks. but i really just wondered whether you're human. ;) [22:45] haha, that depends on who you talk to [22:46] i hadn't seen you talk before, just joingn more ubuntu channels, was wondering [22:46] and then there was the nickname. ;-) ok, time to dig a hole in the ground to dig this conspiracy theory in. [22:47] was looking for help with a problem, tried a couple other channels just to see what was up. [22:47] ha, no worries. ttyl, need to reboot now and see if I can make any progress. [22:47] good luck. [22:52] Should ask them what Distro they use? [22:53] he's here because the recent grub update to ubuntu pointed out problems in his firmware, so it's probably ubuntu [22:53] sarnold: I saw them post the same thing on #linuxmint-help yesterday [22:53] jeremy31: lol [22:53] I thought mint didn't bother passing along updates? [22:54] sarnold: That changed, they used to hold back kernel and firmware updates [22:55] i thought there was doubt over the timely release of security updates? [22:57] Mint still uses Ubuntu repos for 90+% of packages [22:59] aldcor was on #linuxmint-help asking the same question as in #ubuntu. I gave him some advise on #linuxmint-help about questions in #ubuntu