=== BruceMa is now known as BruceMa-afk | ||
amitk | cking: just reading about stress-ng. In CPU compute, can it generate %loading that is frequency invariant? IOW, 50% load at 2GHz != 50% load at 1GHz | 10:37 |
---|---|---|
cking | amitk, it's not intelligent like that, it just breaks the load into time slices of idle and loaded, so it ignores the clock freq | 10:40 |
amitk | cking: ok, we've modified cyclictest for that behaviour and are current modifying rt-app to do the same | 10:41 |
amitk | cking: but the memory load might be interesting, I'll check it out | 10:42 |
cking | amitk, if you see any niches where I can add more stressors let me know, I'd like to make stress-ng more useful over time | 10:42 |
amitk | cking: sure, I'll play around a bit. Our focus hasn't been finding the boundaries (stressing) but more about making it easy for sched maintainers to be able to simulate typical mobile (android) behaviour on a std. linux distro - e.g. mp3 playback, web browsing. | 10:45 |
amitk | cking: but i can see stress-ng being useful to setup thermal constraints, for example | 10:46 |
cking | amitk, indeed, stress-ng was more like "how can I fully load a machine and see how much I can make the CPU and cache toasty" | 10:46 |
cking | and the old stress tool wasn't really useful enough for what I was playing with | 10:46 |
cking | mainly because I was also tinkering with thermald and I wanted to see if I could make that kick on a hot intel box | 10:47 |
rbansal1 | Hi | 10:51 |
rbansal1 | Dose someone know, when I am trying to compile Linux 3.14 kernel I get the following error : ERROR: modinfo: could not open /lib/modules/3.14.0/modules.dep | 10:51 |
rbansal1 | Why this error is coming ? I tried on Ubuntu and CentOS and on both I get almost similar error | 10:51 |
rbansal1 | Any clue on this issue will be helpful. | 10:52 |
=== ara is now known as Guest7746 | ||
rtg | apw, I pushed the rebase to 3.16-rc2. I guess we ought to start doing some testing this week. | 13:27 |
apw | rtg, yeah, it must be getting close, we should get it in the CKT PPA soon so jibel can test against it | 13:38 |
rtg | apw, ah, right. I'll do that today. | 13:38 |
=== shadeslayer_ is now known as shadeslayer | ||
rtg | BenC, don't forget to give some love to powerpc64-emb | 13:47 |
BenC | rtg: Right, on it, thanks | 13:47 |
rtg | BenC, do you think you'll get it figured out this morning ? I'll wait on the first upload if so. | 13:47 |
BenC | rtg: I’m hoping so. I’ll know more withing about 15-30 mins | 13:48 |
rtg | BenC, cool, will wait for a bit then | 13:48 |
BenC | rtg: This is a compile error, correct? | 13:50 |
rtg | BenC, yup. | 13:51 |
rtg | BenC, link error IIRC | 13:51 |
BenC | rtg: Any chance you already have a log of the error? | 13:51 |
rtg | BenC, hmm, probably not. all of my local build tests are fairly ephemeral. it wouldn't take but 10 minutes or so to recreate it. | 13:52 |
BenC | rtg: I have a build started… | 13:53 |
BenC | rtg: I have the fix…one-liner and only affects 85xx (freescale) | 15:11 |
=== hatch__ is now known as hatch | ||
rtg | BenC, cool, thats the best kind. I'll probably look at it and think, "Why couldn't I have figured that out ?" | 15:13 |
BenC | rtg: The error you saw related to smp_(give|take)_timebase, correct? | 15:15 |
BenC | *smp_generic_(give|take)_timebase | 15:15 |
rtg | BenC, yup, thats the one | 15:17 |
BenC | rtg: Patch sent to kernel-team | 15:21 |
apw | smb, hey ... we pulled a lot of VIRT* things =y is that still appropriate | 15:33 |
smb | apw, I would assume yes. At least as long as we want to have boot essentials in for virt and bare-metal at the same time. | 15:35 |
smb | apw, Of course you could ask me to do that part of the conf review. :) | 15:35 |
apw | smb, heh no as long as they are needed i am happy | 15:36 |
smb | apw, At least I am under the impression they were. Though things can silently change, so I would not want to bet money on it | 15:38 |
infinity | smb: I'm willing to bet a lot of that stuff can be modular, but that requires more than just IRC handwaving. :P | 15:39 |
infinity | Like, actual testing. On multiple arches. Ick. | 15:39 |
smb | infinity, A lot _can_ be modular. The problem with them was that (again at least at some point) there was nothing that would trigger them auto-load | 15:40 |
infinity | smb: Well, being auto-loaded is implied in my "can be modular" statement. If something needs to be manually loaded, that's a no-go. | 15:40 |
smb | infinity, Ah ok. So yeah. Some may have changed but agreed, I neither would want things to change without very well tested well. | 15:42 |
infinity | "very well tested well". | 15:43 |
smb | textual deja-vu | 15:43 |
smb | Or not remembering the word one typed a second ago | 15:43 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!