13:26 <NaviFullroot> Hi ppl, I have a q:

13:26 <NaviFullroot> What is the difference between exynos-iommu and smmu on armv7(exynos54xx)?

13:26 <NaviFullroot> Is it possible to use smmu driver instead of exynos-iommu?

13:49 <krzk> NaviFullroot: iommu is sysmmu

13:49 <krzk> NaviFullroot: what do you mean by smmu?

13:52 <krzk> NaviFullroot: actually the opposite - exynos sysmmu is an iommu :)

13:58 <javier___> krzk: speaking about exynos iommu, I just answered your question in the ML about why it was not enabled by default in exynos_defconfig

13:59 <krzk> Thomas Pietrowski also replied some time ago but I didn't quite get what he had in mind

14:01 <javier___> krzk: I see, I missed his answer but now reading I never saw the issue he mentions

14:02 <gordan> What is the deal with audio support on the Peach Pi? I've been finding recently that more often than not the Peach Pi audio devices just outright disappear and show up greyed out in KDE system configuration. Only the dummy output shows up not greyed out. Not that audio ever worked on it, but at least the hardware devices were visible.

14:02 <gordan> This is with 4.3.0 kernel.

14:02 <gordan> Is there a set of out-of-tree patches somewhere that makes audio output work?

14:03 <javier___> krzk: I would had thought that v4.3 would be in good shape if that already contain Marek's patches that fixed the issue I mentioned when iommu was disabled

14:03 <krzk> everything is messy :)

14:04 <javier___> gordan: audio is the only remaining peripheral that is not working reliable on Peach Pi/Pi (and Snow FWIW)

14:05 <javier___> gordan: I've been that in my TODO list for a long time but I never find the time to look at it and I'm quite ignorant in ASoC / codecs to start with

14:08 <javier___> krzk: but that's the reason why I suggested to give a try in kernelci before posting / landing a patch to enable exynos iommu since there could be issues in other platforms

14:08 <krzk> javier___: I can't find your reply... when did you send it?

14:09 <javier___> krzk: subject "[PATCH] ARM: dts: exynos5422-odroid*: remove fimd node"

14:10 <javier___> krzk: is in the archives now http://permalink.gmane.org/gmane.linux.kernel.samsung-soc/50140

14:13 <krzk> krzk: that's strange... I don't have it in my archives... it didn't go to spam neither...

14:13 <javier___> weird... I posted like 20 mins ago

14:13 <krzk> anyway, I agree overall, but a testing also for Exynso4 and Exynos5250 board would be nice

14:14 <krzk> btw, I set up my testing server - currently only with Odroid XU3

14:14 <krzk> http://www.krzk.eu/tgrid?length=30

14:15 <krzk> I need to get some more boards... from Samsung... and at least one more build machine :)

14:16 <javier___> krzk: yes, kernelci would at least do a boot test on exynos4 and 5250 but also some manual test would be nice to see that display is actually working

14:17 <krzk> javier___: yeah, display test :)

14:17 <gordan> javier___: Shouldn't it be possible to get the correct audio driver pulled straight out of the ChromeOS kernel and either drop it into a mainline kernel, or at least bisect it for differences?

14:18 <javier___> gordan: yes, the downstream kernel can be used as a reference for sure. Is just not trivial because the drivers in downstream and mainline are very different

14:19 <javier___> gordan: also the downstream driver does some layering violations and do more stuff that they should into the ASoC machine driver

14:19 <gordan> :(

14:20 <javier___> gordan: as I said, I've that on my TODO literally for months, so if you feel brave to look at it or know more about audio, feel free to give a try :)

14:20 <javier___> patches are always welcomed ;)

14:22 <gordan> javier___: Unfortunately my kernel level development experience is extremely thin. Not completely non-existant, but pretty close to. It is unlikely I'd be able to do it without a sufficiently large amount of support from someone who is at expert level that the guidance overhead would mean it'd be quicker if they just did it directly. :(

14:37 <daniels> javier___: i need to kick sjoerd about getting extremely basic display tests (basically, 'is it a peach pi? right, then ensure it has a 1080p display on eDP-1, and also a HDMI-1 connector') triggered from kernelci

14:37 <daniels> javier___: they're still in the lava test defs :P

14:38 <sjoerd> daniels: me?

14:40 <javier___> daniels: iirc it was you that never pushed in the corect repo :P

14:40 <javier___> *correct

14:40 <javier___> daniels: but seriously, getting those tests running would be awesome

14:41 <javier___> I was just telling sjoerd that found a "Division by zero in kernel" in Exynos DRM driver for v4.4-rc2

14:44 <daniels> javier___: eek, where?

14:45 <daniels> javier___: and, hm, i thought padovan pushed to the right repo but i guess not

14:45 <daniels> sjoerd: i didn't think we had test-definitions run on kernelci/next?

14:45 <sjoerd> ah no that part we don't indeed

14:46 <javier___> daniels: I meant that I didn't find it yet in https://git.collabora.com/cgit/singularity/qa/test-definitions.git/tree/kernel

14:46 <daniels> javier___: yeah indeed

14:49 <javier___> daniels: but I know the tests and are awesome :)

14:50 <daniels> javier___: any day now ... ;)

14:51 <NaviFullroot> krzk: well, i'm looking at drivers/iommu and found exynos-iommu.c and arm-smmu.c driver's, they both implements iommu_ops.

14:52 <NaviFullroot> Also, is there possible to boot exynos 5422(odroid xu3) on kernel 4+ with HYP mode?

14:56 <javier___> NaviFullroot: krzk already left

14:56 <javier___> NaviFullroot: it depends on whether your firmware boots the kernel in HYP or SVC mode AFAIU

14:57 <javier___> the odroid XU3 firmware is really bad, it doesn't even allow proper MCPM so things like CPUidle don't work

14:58 <javier___> so I wouldn't be surprised if KVM wouldn't work too

14:59 <NaviFullroot> javier___: well, u-boot boots it in HYP mode, and kernel 3.10 works(including kvm). But same u-boot and kernel 4.2 doesn't work.

15:01 <javier___> NaviFullroot: interesting, I should give a try to do that as well

15:01 <javier___> probably I never will though :P

15:02 <javier___> my list just keeps growing...