uvos | i know nothing of sutch | 00:00 |
---|---|---|
arno11 | Wizzup: you should try overclocking as well | 00:01 |
arno11 | no difference with pm | 00:01 |
Wizzup | arno11: if you can make an issue on gh and share kernel patches that will help | 00:01 |
arno11 | why not | 00:02 |
Wizzup | I will find a new sd card and flash chimaera to it meanwhile | 00:02 |
arno11 | overclock stuff is already on the n900 wiki | 00:04 |
arno11 | fyi | 00:04 |
Wizzup | aight :) | 00:04 |
Wizzup | issue would be useful, just because it's easier for me/us to track that way | 00:04 |
arno11 | ok | 00:04 |
Wizzup | it can point to the wiki for all I care :D | 00:04 |
arno11 | :) | 00:05 |
Wizzup | hm I get no wifi on chimaera image on n900 | 01:02 |
Wizzup | oh, something happens now | 01:04 |
Wizzup | I think I need to give it more time :) | 01:04 |
Wizzup | Removing xserver-xorg-legacy (2:1.20.11.2-1+m7) ... | 01:33 |
Wizzup | Killed | 01:33 |
Wizzup | no swap- oof :) | 01:33 |
freemangordon | uvos: please have a look https://github.com/maemo-leste/libhildonmime/commit/aa4ab14b370268d7e0fe83c434466bf83797470a | 08:53 |
freemangordon | sorry, that one https://github.com/maemo-leste/libhildonmime/commit/6b28559fca1659904fa369aec4bd7550a18730c1 | 08:55 |
uvos__ | freemangordon: looks good, one small concern i have is that you removed the check for the return value of g_desktop_app_info_launch_uris_as_manager | 09:33 |
uvos__ | this will return false if a .desktop file is availble to handle the sheme, but its exec line points to an application that dose not exist. | 09:33 |
uvos__ | err g_app_info_launch_uris_async | 09:34 |
freemangordon | uvos: yes, that was on purpose, because the original code does the same | 10:42 |
arno11 | Wizzup: i forgot to mention the missing swap on first boot | 11:44 |
arno11 | apologies | 11:44 |
Wizzup | arno11: what did you find causes the slowness? | 11:44 |
Wizzup | or did you just solve that with overclocking | 11:44 |
arno11 | after two reboots no more slowness | 11:45 |
arno11 | with swap enable and extand root | 11:46 |
arno11 | with no overclock it is relatively smooth | 11:46 |
arno11 | overclock helps espacially for video decoding | 11:49 |
arno11 | sdcard under class 10 causes slowness as well especially with aptitude | 11:51 |
arno11 | iirc the slowness after booting is due to automatic apt update every 5 min | 11:55 |
arno11 | disabling it helps a lot | 11:56 |
arno11 | sorry to forgot all of that | 11:56 |
freemangordon | uvos: also, I changed to use g_app_info_launch_uris_async() which returns void | 11:57 |
Wizzup | arno11: I see | 12:26 |
Wizzup | arno11: interesting that it's so slow the first few times | 12:26 |
arno11 | yes weird | 12:27 |
Wizzup | like, after I dd'd it, it was unbearably slow | 12:27 |
Wizzup | like I tried to connect to wifi 5 times but the dialog wouldn't even show | 12:27 |
Wizzup | and I did wait a while | 12:27 |
Wizzup | this could be swap related, potentially | 12:27 |
Wizzup | or rather, the lack of it | 12:27 |
arno11 | yes probably | 12:27 |
uvos__ | what, are we oom, or close to it on boot on n900 now? | 12:35 |
Wizzup | probably | 12:35 |
uvos__ | uff | 12:35 |
Wizzup | I thought we had zswap enabled by default | 12:35 |
arno11 | yes close to oom | 12:36 |
uvos__ | dosent bode well | 12:36 |
uvos__ | we still have a lot of features to implement and any move away from gtk2 will use more ram | 12:36 |
Wizzup | sure | 12:37 |
Wizzup | fremantle also has 750MB of swap | 12:37 |
Wizzup | I doubt it would work without | 12:37 |
arno11 | honestly after few tweaks it runs well | 12:37 |
arno11 | i have created a simple 1G swap with no zram and it works perfectly | 12:41 |
arno11 | even a big ram consuming app like kodi 17 works great | 12:42 |
arno11 | be optimistic guys your work on leste is fantastic even on n900 | 12:43 |
maemish_ | I am following this everyday and I so much appreciate your work. Did not get any answer for my question about getting with new release instructions for getting the swap and for overclocking. The 250/900 setup for overclock worked perfectly on Maemo 5. | 12:47 |
uvos__ | there are instructions on overclock on the wiki, and activateing swap is no different than any other linux distro | 12:54 |
uvos__ | but it seams we will have a swap partition on sdcard soon anyhow... | 12:54 |
arno11 | hi Maemish. yes you could use wiki to overclock 250-800 | 12:55 |
arno11 | and for swap i will update the wiki asap | 12:57 |
freemangordon | uvos__: I still think we shall use emmc dedicated swap | 13:05 |
Wizzup | I think ideally we would have these patches in our kernel | 13:05 |
Wizzup | 12:44 < arno11> be optimistic guys your work on leste is fantastic even on n900 | 13:05 |
Wizzup | :) | 13:05 |
freemangordon | :) | 13:05 |
Wizzup | freemangordon: for emmc dedicated swap, if you can share fstab line we can just put it in our images | 13:06 |
freemangordon | arno11: seems you played a bit, do you have any thoughts about whether it is better to use swap on uSD or emmc | 13:06 |
Wizzup | and hope that people didn't reformat their emmc | 13:06 |
freemangordon | it is another partition anyways | 13:06 |
Wizzup | or have one on sdcard by default, and make it easier for people to switch | 13:06 |
uvos__ | what patches exactly? | 13:07 |
Wizzup | uvos__: https://leste.maemo.org/Nokia_N900#Overclocking_.2F_Undervolting | 13:08 |
Wizzup | looks like 'just' some dts changes | 13:08 |
freemangordon | arno11: given that rootfs is on uSD too | 13:08 |
freemangordon | Wizzup: I am reluctant to patch the kernel to OC by default | 13:09 |
uvos__ | Wizzup: wat? you want to overclock our stock kernel? | 13:09 |
uvos__ | no way | 13:09 |
arno11 | freemangordon: swap better on usd | 13:09 |
freemangordon | why? | 13:09 |
uvos__ | we can provide another dtb to boot sure | 13:09 |
Wizzup | we can provide a package for it in any case | 13:09 |
freemangordon | ok | 13:09 |
Wizzup | uvos__: right, I didn't mean by default per se | 13:09 |
freemangordon | that's better | 13:09 |
Wizzup | but it's better than requiring users to manually modify images | 13:10 |
Wizzup | that will cause trouble later | 13:10 |
freemangordon | uvos__: what about providong dtb that enables OC frequencies, but limit through debugfs by default? | 13:10 |
freemangordon | *providing | 13:10 |
freemangordon | hmm, was it on debugfs | 13:11 |
freemangordon | wherever it is | 13:11 |
freemangordon | arno11: why is better on uSD? | 13:11 |
arno11 | freemangordon: better on usd to avoid mistakes on emmc and break everythig from user point of view | 13:13 |
freemangordon | my question was rather from performance POV | 13:13 |
arno11 | oh ok | 13:13 |
Wizzup | freemangordon: I think we can just add two lines to /etc/fstab, one with swap on usd, and one with swap on emmc, the emmc one commented | 13:14 |
Wizzup | freemangordon: do you have such a fstab line for me? I will try it now | 13:14 |
freemangordon | we can make smart enough script that enables emmc swap if there is one and create swap loop file or somesuch otherwise | 13:14 |
freemangordon | Wizzup: no, but I can provide one | 13:14 |
freemangordon | I gues | 13:14 |
freemangordon | s | 13:14 |
freemangordon | not now though, sorry | 13:14 |
Wizzup | ok, i'll figure it out | 13:14 |
arno11 | sounds good | 13:14 |
Wizzup | swapon /dev/mmcblk1p3 | 13:15 |
Wizzup | works in any case | 13:15 |
arno11 | Wizzup: freemangordon: maybe a bash script is enough to provide overclock and avoid issue with users | 13:17 |
Wizzup | right, but we would want it to run on kernel upgrades | 13:18 |
freemangordon | umm, why? | 13:19 |
Wizzup | uvos__: is patch for idlest1 included in our kernel? | 13:19 |
freemangordon | it should run on boot | 13:19 |
Wizzup | uvos__: I don't see it | 13:19 |
Wizzup | there was a patch for extra pm info | 13:20 |
freemangordon | Wizzup: so, do we have -testing repo now? | 13:20 |
uvos__ | freemangordon: btw fast sdcard was this onehttps://geizhals.de/sandisk-extreme-r100-microsdxc-64gb-kit-sdsqxaf-064g-gn6ma-sdsqxaf-064g-gn6aa-a1620820.html | 13:20 |
Wizzup | uvos__: this is 'Allows seeing the deeper idle state blockers in...' | 13:20 |
uvos__ | it saturates d4 bus in all modes as tested by "iozone -s 10M -r 512 -i 0 -I -i 1 -i 2 " | 13:20 |
uvos__ | maybe run that on n900s emmc | 13:21 |
Wizzup | uvos__: 8ee02090a5bd936bf0feb806fbfb6de5366ac6b8 on maemo-5.15 | 13:21 |
uvos__ | Wizzup: ok | 13:21 |
freemangordon | I don;t have n900 around to play with, maybe arno11 can do it | 13:21 |
Wizzup | let me see if it applies cleanly | 13:21 |
freemangordon | arno11: can you test emmc on n900 with that uvos__ provided? | 13:22 |
uvos__ | what dose this do exactly? | 13:22 |
Wizzup | uvos__: it exposes the blocker bits from omap to userspace | 13:22 |
Wizzup | so we can see what blocks RET, OFF, etc | 13:22 |
freemangordon | gives more info on who blocks idles states | 13:22 |
uvos__ | Wizzup: ok, yeah i just read it from the registers | 13:22 |
Wizzup | let me clean up the commit, just a moment | 13:23 |
* freemangordon is back to work work | 13:23 | |
freemangordon | ttyl | 13:23 |
Wizzup | uvos__: I pushed the commit to branch maemo-idle-report-6.1.y - when you have a chance please look at including it | 13:24 |
uvos__ | Wizzup: o | 13:24 |
uvos__ | k | 13:24 |
Wizzup | with this we can use n900-pm scripts to debug blockers | 13:25 |
Wizzup | arno11: I see the power jump constantly from 0.04A to 0.18A with 3.7V with screen off (no way to read it out programatically with this PSU I am afraid), so I guess that gets us to about what you were seeing | 13:28 |
Wizzup | wait, let me turn off htop over ssh :) | 13:28 |
uvos__ | lol | 13:28 |
Wizzup | :D | 13:28 |
Wizzup | seems to stay at 0.04A or so, and my serial module always draws a bit of power there | 13:29 |
arno11 | cool :) | 13:29 |
arno11 | sorry guys i have to go | 13:29 |
Wizzup | ttyl | 13:29 |
arno11 | i can help with emmc stuff this evening | 13:29 |
Wizzup | I have swap on emmc already, it's ... fine, I guess | 13:30 |
arno11 | if needed | 13:30 |
sicelo | Wizzup: that 0.04A is with the blacklist in place? | 13:32 |
Wizzup | yes | 13:34 |
Wizzup | shall I push renewed blacklists (for pm and remove modem) to chimaera-testing ? | 13:38 |
Wizzup | or just to chimaera? | 13:38 |
Wizzup | (for n900) | 13:38 |
uvos__ | are we blacklisting something important? | 13:38 |
Wizzup | pm blacklist is omap3_isp / ehci_omap / omap_hdq / hci_nokia / hci_uart | 13:39 |
Wizzup | I am not sure (anymore) what these are used for | 13:39 |
Wizzup | maybe ehci_omap could be relevant for charger detection or something | 13:39 |
uvos__ | ehci_omap sounds bad | 13:39 |
uvos__ | do we have 1w devices? | 13:40 |
Wizzup | one-wire ? | 13:40 |
uvos__ | otherwise load it with the same trick as on d4 | 13:40 |
uvos__ | that disables bus scanning | 13:40 |
uvos__ | omap_hdq is the 1w bus driver | 13:40 |
uvos__ | used on d4 to comunicate with the battery | 13:40 |
uvos__ | we add a module param there | 13:40 |
uvos__ | to make it behave pm wise | 13:40 |
uvos__ | the hci stuff probubly breaks bluetooth | 13:41 |
uvos__ | so not to great all in all | 13:41 |
Wizzup | right, I don't mind so much breaking bt atm | 13:41 |
Wizzup | we also blacklist it on the d4 after all | 13:41 |
Wizzup | for similar reasons (idle) | 13:41 |
uvos__ | no | 13:41 |
uvos__ | we backlist it because its broken | 13:41 |
Wizzup | well, that and the loading race | 13:41 |
uvos__ | right | 13:42 |
sicelo | the blacklist is safe. none of those modules are critical for anything on n900 | 13:42 |
Wizzup | I am not sure if omap hdq is actually used for battery btw | 13:42 |
sicelo | no. it isn't. it's there because omap2plus has bq27xxx-hdq, which N900 doesn't use. our bq27 is on i2c | 13:43 |
Wizzup | check | 13:43 |
uvos__ | why dont you look at /sys/bus/w1/ | 13:43 |
Wizzup | I guess the isp we know we don't use now, and for ehci_omap, I guess it could impact potentially charger vs pc detection? | 13:43 |
uvos__ | to see if anything is useing this bus | 13:43 |
sicelo | Wizzup: no, not that either :-) | 13:43 |
Wizzup | uvos__: will check once I finish the pm test | 13:44 |
sicelo | check what? 1-wire? | 13:47 |
Wizzup | /sys/bus/w1/ doesn't exist on my system now, but I guess that is because I blacklisted the module :) | 13:50 |
uvos__ | yes | 13:50 |
uvos__ | you have to load it | 13:50 |
uvos__ | just load it by hand | 13:51 |
Wizzup | # find /sys/bus/w1/devices/ | 13:54 |
Wizzup | /sys/bus/w1/devices/ | 13:54 |
Wizzup | /sys/bus/w1/devices/w1_bus_master1 | 13:54 |
Wizzup | /sys/bus/w1/devices/01-000000000000 | 13:54 |
Wizzup | # cat /sys/bus/w1/devices/01-000000000000/name | 13:54 |
Wizzup | 01-000000000000 | 13:54 |
sicelo | i think i already mentioned a number of times - nothing uses it on N900. it's insufficient? | 13:54 |
Wizzup | sicelo: no, that's fine, I'm just following up on uvos' question | 13:54 |
sicelo | i'm referring to it too | 13:55 |
uvos__ | i hardly see the harm looking at the bus to make sure | 13:55 |
sicelo | we're sure already | 13:55 |
uvos__ | you are sure | 13:56 |
sicelo | Wizzup: ehci_omap isn't connected with charger detection in any way on N900 | 13:56 |
uvos__ | anyhow yes apearantly nothing uses it | 13:56 |
Wizzup | I'll make the changes, thanks guys | 13:56 |
uvos__ | so ehci_omap, used for host mode i presume? | 13:56 |
sicelo | the blacklist is global or n900-only? | 14:00 |
uvos__ | n900 only ofc | 14:00 |
uvos__ | otherwise you would be breaking d4 pretty hard | 14:00 |
uvos__ | yes i know n900 dosent do otg (ie autoswiching host mode) | 14:01 |
Wizzup | rafael2k: did the pkg rename | 14:12 |
Wizzup | I get p=230 or so on n900 | 15:40 |
freemangordon | I think it is better to not blacklist 1w, but instead to the same as on d4 | 16:10 |
freemangordon | as we cannot be sure in what state bootloader leaves it | 16:11 |
Wizzup | freemangordon: what do you suggest? | 16:20 |
Wizzup | I don't see the point of not blacklisting a module that isn't even used for anything | 16:20 |
Wizzup | like, not sure why we would complicate it | 16:21 |
freemangordon | because it initializes the HW | 17:26 |
freemangordon | and puts it in a well known state | 17:27 |
freemangordon | in theory ofc | 17:27 |
freemangordon | otherwise we count on the bootloader to init it | 17:27 |
freemangordon | and not, it is not more complex than blacklisting | 17:27 |
freemangordon | *no | 17:27 |
freemangordon | Wizzup: this https://github.com/maemo-leste/leste-config/blob/master/leste-config-mapphone/etc/modprobe.d/wire.conf.leste | 17:28 |
freemangordon | is not more complicated than https://github.com/maemo-leste/leste-config/blob/master/leste-config-mapphone/etc/modprobe.d/blacklist-bluetooth.conf.leste | 17:29 |
freemangordon | for n900 we need https://github.com/maemo-leste/leste-config/blob/master/leste-config-mapphone/etc/modprobe.d/wire.conf.leste as is | 17:29 |
Wizzup | freemangordon: ok, can you make the leste-config changes then? | 17:58 |
Wizzup | (don't forget gen-displace) | 17:58 |
Wizzup | fwiw I won't be around much tonight, back tomorrow | 18:57 |
freemangordon | Wizzup: done | 21:47 |
freemangordon | someone shall test on n900 though | 22:04 |
freemangordon | ugh, no iphb module :( | 22:05 |
freemangordon | I wonder why current draw increases drastically when I enable my sip account | 22:05 |
freemangordon | Wizzup: iphb-dkms is installed, but module is not build, I wonder why | 22:07 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!