cnxsoft has joined #imx6-dev
ragesoss has quit [Remote host closed the connection]
ragesoss has joined #imx6-dev
<ragesoss> are there any special steps necessary to update the kernel on a gk802 (on debian, from projectgus's installer)?
<ragesoss> I built a new kernel package with a few config changes, and then installed it with dpgk -i.
<ragesoss> when I restarted it, it still used the old kernel, and I noticed that /boot/initrd was still symlinked to the old kernel.
<ragesoss> I tried changing that to the new kernel's initrd.img, and then I tried update-initramfs.
<ragesoss> still, I just get the old kernel when I boot it. could that be because the new kernel is failing? or because I need to update uboot or something else?
<projectgus> ragesoss: the symlink to zImage needs to point to the new kernel as well, that isn't updated automatically
<projectgus> and the initrd symlink as well. 'update-initramfs -u -k all' (from memory) should rebuild a new initrd for each installed kernel
<projectgus> then just update the symlink to point ot the correct one
<projectgus> it's kind of hacky but it's the best I could do without, you know, actually doing it properly ;)
atiti_ has joined #imx6-dev
atiti_ has quit [Client Quit]
atiti_ has joined #imx6-dev
atiti_ has quit [Read error: Connection reset by peer]
diego_r has joined #imx6-dev
jawcraig has quit [Read error: Operation timed out]
jawcraig has joined #imx6-dev
ragesoss has quit [Read error: Operation timed out]
ragesoss has joined #imx6-dev
eebrah_ has joined #imx6-dev
eebrah_ has quit [Ping timeout: 248 seconds]
rz2k has joined #imx6-dev
dshankar has quit [Quit: dshankar]
dshankar has joined #imx6-dev
dshankar has quit [Quit: dshankar]
ballock has joined #imx6-dev
ballock has left #imx6-dev [#imx6-dev]
ballock has joined #imx6-dev
rz2k has quit []
eebrah has joined #imx6-dev
cnxsoft has quit [Quit: cnxsoft]
werebutt has joined #imx6-dev
werebutt has left #imx6-dev [#imx6-dev]
datagutt has quit [Ping timeout: 245 seconds]
<hste> otavio: how do I build the new beta in yocto? is it in master-next now?
<otavio> hste: the parts that are ready, yes
<otavio> hste: but I keep rebasing it
<otavio> jas-hacks: they ae not
<hste> ok
<otavio> jas-hacks: many bugfixes
<otavio> hste: I notify the mailing list when I rebase it and post the new version of the patches
<otavio> hste: but in general it is fine
<hste> otavio: I saw the new X had support for Randr
<otavio> hste: yes
<otavio> hste: sorry but I gotta run ...
<otavio> ttyl
<hste> ok
jas-hacks has joined #imx6-dev
diego_r has quit [Ping timeout: 272 seconds]
<JBD1986> jas-hacks, you there?
<JBD1986> jas-hacks, I got your jessie rootfs loaded
<jas-hacks> JBD1986: yes
<JBD1986> It seems a bit laggy when draggin around windows
<JBD1986> not quite as bad as my rootfs
<JBD1986> but definitely not workstation smooth
<JBD1986> also, resizing windows is terribly laggy
<jas-hacks> JBD1986: I'm not seeing that much lag
<JBD1986> what about with resizing windows?
<JBD1986> it's not smooth
<JBD1986> for me atleast
<jas-hacks> JBD1986: which kernel version?
<JBD1986> It definitely seems like some form of compositing is enabled
<JBD1986> 3.0.35_4.1.0
<jas-hacks> JBD1986: compositing should be disabled
<JBD1986> is there a way to verify that?
<jas-hacks> under windows tweaks
<JBD1986> jas-hacks, hmmm, even if I enable that, performance is still choppy
<JBD1986> it's not QUITE as bad as my rootfs, but still very unpleasant
<jas-hacks> JBD1986: you don't need to enable it.
<jas-hacks> I don't see the same effect as you
<JBD1986> jas-hacks, the reason I bring it up is I have installed xcompmgr along with openbox. The following items have changed/been observed:
<JBD1986> Dragging windows around is MUCH smoother.
<JBD1986> OpenGL performances drops slightly.
<JBD1986> Resizing windows is VERY SLOW.
<JBD1986> on my rootfs
<JBD1986> so your rootfs is giving somewhat similar behavior to mine with xcompmgr running
<JBD1986> oh glxgears is performing extremely bad
<JBD1986> on your rootfs
<jas-hacks> what cpu are using?
<JBD1986> mx6q dual
<JBD1986> ah jas-hacks , it's using sw gl rendering
<JBD1986> for glxgears
<jas-hacks> looks like it
<JBD1986> i wonder why
<jas-hacks> what does glxinfo
<jas-hacks> report
<JBD1986> sw rendering
<JBD1986> Mesa project
<jas-hacks> on which roofs?
<JBD1986> on yours
<JBD1986> once we added the arm patch when you helped me a week or 2 ago, that solved this issue
<jas-hacks> can you pastebin xorg.log
<JBD1986> jas-hacks, sure just a minute or two
<jas-hacks> also can you do 'find / -name libGL.so'
<JBD1986> jas-hacks, i'm havin an issue mounting my drive, actually moreso that I don't see it coming up in deviecs
<JBD1986> devices
<JBD1986> jas-hacks, libGL.so is there
<JBD1986> usr/lib/
<jas-hacks> but is it anywhere else?
<JBD1986> yes, /root/viv/mesa_backup/
<jas-hacks> can you pastebin glxinfo output
<JBD1986> yes, without flash drive that will be difficult however
<jas-hacks> are you using a imx6dl or a imx6d cpu?
<JBD1986> i.mx6q dual core
<JBD1986> jas-hacks, I remember libdrm being an issue before
<jas-hacks> glxinfo output should be similar to this http://pastebin.com/xDqah0ea
<JBD1986> jas-hacks,
<JBD1986> OpenGL vendor string: Vivante Corporation
<JBD1986> OpenGL renderer string: Vivante GC2000
<JBD1986> that part is wrong for me
<JBD1986> it's Mesa Project
<JBD1986> Software Rasterizing
<jas-hacks> that means it failed to load the exa driver, need output from xorg.log
<JBD1986> xorg errors
<JBD1986> {EE} [drm] drmOpen failed
<JBD1986> (EE) VIVANTE(0): [dri] DRIScreenInit failed. Disabling DRI
<jas-hacks> Is it failing to find "Vivante GCCore:00"
<hste> JBD1986: what kernel? from git?
<JBD1986> jas-hacks, no mention of that
<JBD1986> hste, freescale's 3.0.35_4.1.0
<jas-hacks> any chance you can post the output?
<jas-hacks> difficult without it
<JBD1986> not sure how, as my flash drive is not showing up
<jas-hacks> you just need to mount the flash drive, it should appear as /dev/sd...
<hste> JBD1986: I dont think that patch is in freescale's 4.1.0 but must be added
<hste> jas-hacks: Did u see this <diego_r> for those who may be interested: https://lists.yoctoproject.org/pipermail/meta-freescale/2014-January/006736.html
<JBD1986> jas-hacks, the drive does not appear. I suspect that my kernel is not compatible with your lib modules
<jas-hacks> JBD1986: what does lsusb show?
<jas-hacks> hste: I saw diego_r comments
<JBD1986> jas-hacks, lsusb shows the drive, but dmesg doesn't show the storage device
<JBD1986> jas-hacks, i'm guessing that my kernel rejects your version of usb_storage.ko
<jas-hacks> JBD1986: you should copying your kernel modules to the rootfs
<jas-hacks> /s/should/should be
<JBD1986> jas-hacks, sorry for the confusion
<JBD1986> jas-hacks, i did not do that
<JBD1986> although that will be somewhat difficult without usb storage lol
<JBD1986> I should be able to do that with mfgtools however
<jas-hacks> hste: the glacore kernel changes are back ported into 4.1.0 for the new viv libs
<hste> jas-hacks: yes I saw that. But without a rootfs.. :)
<hste> waiting for it in yocto
<hste> jas-hacks: stephan has some nice patches too https://github.com/wolfgar/xbmc/commits/imx-wip
<hste> jas-hacks: still problems with Xorg?
<jas-hacks> hste: it was working but I seem to have broken it again
<JBD1986> jas-hacks, got it
<JBD1986> my modules are loaded
<JBD1986> glxgears now runs properly
<jas-hacks> can you post xorg.log
<JBD1986> yes, i can mount my card now
<JBD1986> just a min
<JBD1986> jas-hacks, http://pastebin.com/ajdsqfwe
<jas-hacks> JBD1986: you don't need to set 'fbmem=32M'
<JBD1986> jas-hacks, I've started with AND without that
<JBD1986> jas-hacks, no difference
eebrah has quit [Ping timeout: 272 seconds]
<jas-hacks> what does glxinfo report?
<JBD1986> jas-hacks, the correct info, vivante
<JBD1986> jas-hacks, can I turn off the window snapping?
<jas-hacks> yes
<JBD1986> I feel like that could be making the issue appear worse than it is
<JBD1986> jas-hacks, it appears that the window also floats behind the mouse
<JBD1986> is there a way to make it move exactly with the mouse?
<jas-hacks> JBD1986: you still seeing slow window dragging?
<JBD1986> jas-hacks, yes, even with a tiny window, moving the window around the screen is not smooth
<jas-hacks> JBD1986: not sure you will get completely smooth
<JBD1986> jas-hacks, does your window follow your mouse?
<JBD1986> or does it lag behind the mouse?
<jas-hacks> JBD1986: can't test at the moment ... have a broken build
<JBD1986> jas-hacks, even if I make the window SUPER small, it's minimum size, it's still not perfect.
<JBD1986> now, if i enable compositing on your rootfs, and reboot
<JBD1986> it's slightly better
<JBD1986> almost tolerable
<jas-hacks> JBD1986: another options would be to try compton for compositing
<JBD1986> but this is working perfectly for you on your board right?
<JBD1986> without compositing?
<hste> JBD1986: if u try using bpp=16 instead of bpp=32 do u notice any difference?
<JBD1986> hste, i will give it a shot
<JBD1986> hste, same issue with 16bpp
<JBD1986> high cpu utilization of xorg while dragging windows around
<JBD1986> even dragging around the tiniest window gets cpu utilization over 80%
<hste> if i had the box here I should test, I'll see if I can test that later
<JBD1986> hste, thanks, it's appreciated
<JBD1986> jas-hacks, thanks for your help through this issue
<JBD1986> it's very frustrating lol
<JBD1986> I can't find any similar problems on the freescale forum
<JBD1986> or if I do, they have not been answered
<JBD1986> It would be interesting to see if my 512MB of ram is causing an issue
<JBD1986> it seems unlikely since I can run glmark2-es2 fullscreen to completion
<JBD1986> but I'm running out of ideas
<jas-hacks> what score do you get with glxgears?
<JBD1986> glmark2-es2 --fullscreen gets 73
<JBD1986> i think windowed mode was 149, but I don't remember
<jas-hacks> also 512MB is on the low side
<JBD1986> 1080p
<hste> JBD1986: can u paste the dmesg?
<JBD1986> yes just a minute
<jas-hacks> JDB1986: is your board using 32 or 64bit ddr setup
<JBD1986> 32bit I believe, But i'd have to verify
<JBD1986> it should be the same as the sabresd
<jas-hacks> if it 32bit then that will limit GPU/memory throughput
<JBD1986> jas-hacks, It is able to output fullscreen 3D rendering and video very well though at high framerates
<JBD1986> and once I enable xcompmgr, it works smoothly
<JBD1986> I believe there is something wrong with my 2D rendering setup
<jas-hacks> it doesn't use 2D rendering, xcompmgr relies on x extensions for composite
jas-hacks has left #imx6-dev [#imx6-dev]
dshankar has joined #imx6-dev
bfederau has quit [Remote host closed the connection]
bfederau has joined #imx6-dev
frankbutt has joined #imx6-dev
frankbutt has left #imx6-dev [#imx6-dev]
dshankar has quit [Quit: dshankar]
<newell> anyone here use the global timer with i.MX6? If so, I am trying to create a linux kernel driver to test the interrupt latency using this timer. Has anyone done something like this before?
<newell> Just wanted to get a sanity check on what I am trying to do.
dshankar has joined #imx6-dev