11:52 AndrewR: Hello. Is there possibility ddx commit f0fdc156039eaa79fb1ad82fcc6437f2432306c6 ("Try using EXA ops harder in drmmode_fbcon_copy fallback") broke X startup with nv43/AGP? I bisected it down to exactly next commit, "Actually do a clear instead of doing a bogus copy." but guess it was just fixup for first one?
11:56 AndrewR: also, mesa-demos related to point sprites quite broken, and even multitexturing demo (multiarb) broken, too! (with mesa-git (git-42298b0) )
12:17 imirkin_: AndrewR: point sprites are quite unrelated, i think :)
12:18 imirkin_: AndrewR: can you describe how X startup was broken?
12:20 imirkin_: mlankhorst: can you elaborate why it should be a GXclear and not a GXcopy btw? i'm sure you're right, but would like to understand
12:20 imirkin_: oh, coz it's not a memcpy, it's a memset!
12:22 imirkin_: AndrewR: are you saying that the GXclear fix is what broke things, or that the first commit is what broke things?
12:22 AndrewR: imirkin, it simply froze X with accel, I can shut down machine with power button, after some time. On screen there was some noise (static). It worked without acceleration. I only use startx from console login (slackware 14.0, mostly - with new kernel, libdrm, xf86-video-nouveau..and few other packages). I froze like this on earlier kernels, not tried with nouveau/4.1 yet (running this kernel now, seems to work, but with ddx set to
12:22 AndrewR: safe version)
12:23 AndrewR: imirkin, moment ....
12:23 imirkin_: i assume it's that first one, it appears to have a glaring issue
12:24 imirkin_: can you throw in a PUSH_KICK(pNv->pushbuf) right before the nouveau_bo_wait?
12:25 AndrewR: http://pastebin.com/WKRNuEWp - bissect log ...yes, will try to add this ...
12:27 imirkin_: oh, so f0fdc156039eaa79fb1ad82fcc6437f2432306c6 is good? that's surprising...
12:29 AndrewR: imirkin_, you mean nouveau_bo_wait at line 404?
12:29 imirkin_: no
12:29 imirkin_: the one that was added in the fallback
12:30 imirkin_: btw, i assume you're getting one of those warnings?
12:30 imirkin_: [in your xorg log]
12:32 AndrewR: imirkin_, sorry not saved X log from freezing version
12:32 imirkin_: even non-freezing would be fine
12:34 imirkin_: anyways... that code is pretty sloppy
12:34 imirkin_: i bet that PrepareSolid is failing
12:35 imirkin_: AndrewR: are you using 16-bit color by any chance?
12:36 AndrewR: no, I hope? 47.086] (==) NOUVEAU(0): Depth 24, (--) framebuffer bpp 32
12:36 imirkin_: ok good
12:36 imirkin_: can you pastebin your xorg log?
12:40 AndrewR: imirkin_, http://pastebin.com/2uAKaQe9
12:40 imirkin_: this code seems bogus btw: http://cgit.freedesktop.org/nouveau/xf86-video-nouveau/tree/src/nv04_exa.c#n45
12:41 imirkin_: planemask |= ~0 << ppix->drawable.bitsPerPixel;
12:41 imirkin_: if bitsPerPixel == 32, then it just sets all the bits of planemask!
12:41 AndrewR: imirkin_, but this one with working nouveau version, I added you suggested mod , compiled new driver, time to restart X ....
12:41 AndrewR: *your
12:42 imirkin_: hold on
12:42 imirkin_: it won't work
12:53 imirkin_: AndrewR: something like this? http://hastebin.com/tidofiyuje.coffee
13:01 AndrewR: imirkin_, testing ...
13:05 AndrewR: imirkin_, seems to work
13:05 imirkin_: AndrewR: yay! do you see a warning in your xorg log btw?
13:05 imirkin_: perhaps it's the silent fallback if it can't find the fbcon_id
13:08 AndrewR: imirkin_ not one marked with WW ...
13:09 AndrewR: imirkin_, http://pastebin.com/icypxzk9
13:09 imirkin_: AndrewR: ok. are you like disabling fbcon or fbdev or something?
13:10 imirkin_: mlankhorst: would you expect fbcon_id to not get found?
13:10 imirkin_: mlankhorst: or is 0 perhaps a valid value? :)
13:10 AndrewR: imirkin_, no, it all tandard nouveau kms frambuffer ...sorry, dmesg a bit spammy due to high debug loglevel I set while hunting for another hang
13:12 imirkin_: if (fb->depth != pScrn->depth || fb->width != w || fb->height != h)
13:12 imirkin_: could also be that one
13:13 AndrewR: imirkin_, well, if KMS start at 1280x1024 (max for this display), and X start with 1024x768@85 .....
13:14 imirkin_: aha, that's why it's falling back then
13:35 mlankhorst: imirkin_: if you use CONFIG_EXPERT and have drm fbcon as a module..
13:36 imirkin_: mlankhorst: yeah, his case was much simpler -- diff fb sizes :)
13:38 imirkin_: mlankhorst: btw... why is GXcopy not the right thing?
13:39 imirkin_: Solid(GXcopy, fg value) seems like it'd do the clear, no?
13:43 mlankhorst: cant remember
13:43 mlankhorst: think it might have been
13:44 imirkin_: you did it initially and then flipped to GXclear
13:44 mlankhorst: was looking for bugs in tegra
13:44 imirkin_: http://cgit.freedesktop.org/nouveau/xf86-video-nouveau/commit/?id=5439a9905eab8c8d0f86e9963df6c893633bdbfd
13:44 tobijk: imirkin_: r-b on the solid fill two-patch-series from me as well
13:44 mlankhorst: but ended up with something else probably being the issue
13:44 mlankhorst: feel free to revert if it helps you
13:44 imirkin_: well, i think the destroypixmap move is right
13:45 imirkin_: but yeah, i'll flip it back to copy, coz pre-nv50 likes that better
13:45 imirkin_: and i assume nv50+ couldn't care less
13:59 mlankhorst: probably
13:59 mlankhorst: the error i had with tegra appeared to not relate to the ddx so reverting's fine
13:59 imirkin_: ok cool
14:06 manjaro_user2: Hello all! I have problems with changing hz on Manjaro linux with XFCE, have tried that xrandr, but it does not seem to have any effect, what should I try?
14:07 imirkin_: manjaro_user2: xrandr --output foo --set 'scaling mode' None
14:40 manjaro_user: imirkin_: Thanks, it actually works now!
14:40 imirkin_: cool. newer kernel will make that the default.
23:53 RSpliet: skeggsb: mind giving as much feedback on my pm tree as possible? I'd like to get it merged and partially-enabled as much as possible for 4.2
23:54 RSpliet: imirkin, imirkin_: perhaps you can do a code(style) review?
23:55 imirkin_: RSpliet: send patches, i'll review in email
23:55 imirkin_: [i.e. email]
23:55 RSpliet: to you personally? I'm not sure if I want to pollute the mailinglist with it just yet
23:56 imirkin_: then it's not ready for me to review ;)
23:56 imirkin_: the standard for mailing things out isn't that high
23:56 imirkin_: esp if you stick a 'RFC' in the patch name
23:56 imirkin_: i.e. [PATCH RFC 1/5] or whatever
23:57 imirkin_: e.g. look at the junk i send out ;)
23:57 RSpliet: right
23:58 imirkin_: or are you saying i should be more selective with what i send out? heh
23:58 RSpliet: tbh, I can't keep up with the ML myself
23:59 imirkin_: there's like 10 emails a week
23:59 imirkin_: [to nouveau@]
23:59 imirkin_: esp if you don't count the bugzilla updates