**** BEGIN LOGGING AT Fri Sep 02 02:59:57 2011 Sep 02 03:02:10 if the version number doesnt change it shouldn't re-display the changelog Sep 02 05:10:49 oil: it seems to be, here. I'm using "make DEVICE=1 test" to run it. Sep 02 05:11:12 (i also pushed up larger screenshots on the TouchPad, which since there is plenty of room, are much nicer, IMO.) Sep 02 05:57:19 ka6sox, were you able to get the WiFi chip powered on? Sep 02 08:15:53 cryptk, I think I have the right bits Sep 02 08:16:06 but I can't verify yet. Sep 02 08:16:41 what are you using to turn on bt? Sep 02 08:17:49 booto, the BT power is well defined Sep 02 08:18:05 but its more hidden to get wlan working Sep 02 08:18:17 and I haven't found "status" back on that. Sep 02 08:19:10 there are defined GPIO bits for it and they show up (where I can't remember @ this moment as I"m chasing wlan) Sep 02 08:19:26 I think it maybe in /sys/devices/ Sep 02 08:20:25 I remember seeing something called bt_power.0 (or something similar) in sysfs Sep 02 08:21:44 I stopped the userland bt stack and tried hciattach'ing csr and bcsp to /dev/bt_uart, but just got timeouts making me think I wasn't turning it on properly Sep 02 08:25:03 ya, I will work on BT stack *after* I get wlan working with cryptk Sep 02 08:27:02 are you guys trying to get it going from within the ubuntu chroot, or just decouple to base webos env from proprietary stuff? Sep 02 08:27:59 booto, we aren't chrooting... we are booting Angstrom natively Sep 02 08:28:11 oh, ok Sep 02 08:47:17 cryptk, its in 1 of these 4 dirs Sep 02 08:48:07 /sys/bus/platform/drivers/wlan_ar6000_pm_dev Sep 02 08:48:12 /sys/bus/platform/drivers/wlan_ar6000_pm_dev Sep 02 08:48:24 /sys/bus/sdio/drivers/ar6k_wlan Sep 02 08:48:30 /sys/module/ar6000/parameters/wlanNodeCaching Sep 02 08:48:35 /sys/module/ar6000/parameters/wlaninitmode Sep 02 08:48:55 I suspect #2 or #4 Sep 02 08:49:08 they are all "writable" but no feedback. Sep 02 09:04:27 cryptk, unfortunately I haven't found one labeled "power" Sep 02 09:06:18 cryptk, you can try toggling: /sys/devices/platform/msm_sdcc.4/mmc_host/mmc1/mmc1:0001/mmc1:0001:1/power/control and see what happens. Sep 02 09:07:40 control is set for "auto" and wakeup is some control char. Sep 02 09:11:15 bimbo.... Sep 02 09:11:43 /sys/module/ar6000/parameters/wlaninitmode Sep 02 09:12:23 goes from "no such file or directory" (off) to "3" (on) Sep 02 09:13:22 NICE! Sep 02 09:13:42 so is it as easy as echoing 3 into there to bring it on? or is it you set something else that that changes? Sep 02 09:14:01 unknown right now Sep 02 09:14:35 unfortunately the boardfile exports the API to turn it on and off out to a generic thing in kernel land. Sep 02 09:15:29 there are a number of write only nodes...but none marked as "power" Sep 02 09:18:09 cryptk, based on what I saw earlier in the source: Sep 02 09:18:30 I would try: echo '1' > /sys/devices/platform/msm_sdcc.4/mmc_host/mmc1/mmc1:0001/mmc1:0001:1/power/wakeup Sep 02 09:18:49 as all references in the board file are to "wakeup" Sep 02 09:38:07 ok, ticket queue is under control... lemme see if that works... Sep 02 09:44:46 /sys/devices/platform/msm_sdcc.4/mmc_host/mmc1/mmc1:0001/mmc1:0001:1/power/wakeup doesn't exist in angstrom... Sep 02 09:46:37 wait.. it does after the modprobe, duh Sep 02 09:47:28 he he Sep 02 09:51:21 echo 1 > /sys/devices/platform/msm_sdcc.4/mmc_host/mmc1/mmc1\:0001/mmc1\:0001\:1/power/wakeup doesn't do it Sep 02 09:52:47 does /sys/module/ar6000/parameters/wlaninitmode exist? Sep 02 09:53:54 what is the value of /sys/devices/platform/msm_sdcc.4/mmc_host/mmc1/mmc1\:0001/mmc1\:0001\:1/power/control? Sep 02 09:56:12 yes, wlaninitmode exists... it is set to 3 after the modprobe Sep 02 09:56:21 and the value of control is "auto" Sep 02 09:56:29 then it should be "UP" Sep 02 09:56:40 but dmesg reports timeout while waiting for card to init Sep 02 09:57:31 is the sdio driver complaining? Sep 02 09:58:16 not sure Sep 02 09:58:23 I am back in webos now investigating Sep 02 09:58:34 can you paste the dmesg line next time? Sep 02 09:58:43 yeah Sep 02 09:59:01 then I can grep for that message. Sep 02 10:00:47 hrm... I forget where I found the insmod line for the module at... Sep 02 10:01:29 ugh Sep 02 10:03:21 oh wait.. i know where I found it... one sec Sep 02 10:04:26 insmod: insmod /lib/modules/`uname -r`/kernel/net/wifi/ar6000.ko regcode=0x4067 palm_mac_address=00:1D:FE:DA:B0:39 Sep 02 10:05:18 same line as the other night Sep 02 10:05:34 yeah, but I wanted that line... and I didn't have it handy Sep 02 10:05:44 wanted to grab it while I was in webOS Sep 02 10:06:35 one thing I've seen occasionally with these kind of things is you need to poke them with the cal values or download a blob. Sep 02 10:06:48 it is getting the cal values Sep 02 10:06:57 from the token data Sep 02 10:07:56 going into novaterm in webOS and running `stop PmWiFiService` and then `PmWiFiService -d` will reveal lots of great logging output Sep 02 10:10:08 cat that all up and I'll poke thru that for some tidbits. Sep 02 10:10:16 it is a TON of data Sep 02 10:10:38 let me copy over the PmWiFiService binary and see if I can get WiFi up that way... Sep 02 10:10:53 if so, then we are on the right track... if not, then we are missing something else... some kind of pre-req Sep 02 10:11:41 ya, might want to strace it....I assume its not stripped like the rest arent' Sep 02 10:31:15 ok, I have PmWiFiService and it's required libs copied over to angstrom... time to rebooterate Sep 02 10:38:16 ok, getting PmWiFiService to work in Angstrom is becoming a library hell Sep 02 10:38:38 I swear... that thing depends on almost every library on the device Sep 02 10:48:48 before I get any further into this... I am setting up my ext3fs Angstrom as a "Reference" and my ext2fs Angstrom as a "Development" Sep 02 10:49:03 I hack in the ext2fs one... apply what works to the ext3 one Sep 02 10:59:58 ok, I have my ext3fs angstrom booting now Sep 02 11:00:18 this will become my "Reference" install... only working things go here... Sep 02 11:00:32 then if I break my ext2fs install I can easily roll back to a working version Sep 02 11:01:03 and then when we are done, we can do a recursive diff across the original angstrom tarball and the ext3fs install to make sure we have all of the changes Sep 02 11:09:12 ka6sox, if you like I can tar up my "reference" copy of angstrom for you Sep 02 11:09:23 but it can't be widely distributed as it has novacomd in it Sep 02 15:11:17 cryptk|offline, yes, please Sep 02 16:20:44 * ka6sox didn't hear what cryptk was doing with wlan power... Sep 02 16:21:11 and would like the latest angstrom rootfs. Sep 02 16:26:28 cryptk|offline, did you get a manifest of your build of Angstrom? Sep 02 16:26:43 if you have one I'll ask narc to build me that image. Sep 02 16:28:18 scoutcamper, now that we have the shim done lets start working on "Apps" Sep 02 16:28:29 ok Sep 02 16:28:45 an app really consists of: Sep 02 16:28:52 1) an upstart script Sep 02 16:28:59 btw my touchpad is at home on my desk Sep 02 16:29:06 2) a chroot_wrapper Sep 02 16:29:25 3)an x-window? Sep 02 16:29:36 3) a post inst that calls the shim with 2 arguments (app and action) Sep 02 16:30:01 4) a prerem that calls the shim to remove the package. (leaving any config info alone) Sep 02 16:30:26 5) the package Icon. Sep 02 16:31:05 okay Sep 02 16:31:51 we can use the ubuntu-natty-chroot app as a template Sep 02 16:32:05 okay Sep 02 16:32:07 then we can mod the bits. Sep 02 16:32:53 ka6sox, in screen ... Sep 02 16:54:43 so is the idea of the Xapp to launch X apps from inside of a chroot but give it its own card with all the manual steps and launching it from an app icon in Webos? Sep 02 16:55:05 without* Sep 02 16:57:26 specific xapps, yes Sep 02 16:57:35 like currently we are working on xchat Sep 02 16:58:32 oh alright got it, nice. will it launch through xecutah or just regular webos launcher? Sep 02 16:58:43 launcher is the plan Sep 02 16:59:14 nice Sep 02 17:01:44 scoutcamper, this uses xecutah Sep 02 17:01:57 right Sep 02 17:02:12 I am working up to modding xecutah to bring up multiple x-servers and assign apps to them. Sep 02 17:02:13 * scoutcamper was thinking of a different thing Sep 02 17:02:29 so each app *will* have its own card. Sep 02 17:02:59 if we get multiple display version of x-server then I just will use :0, :1, etc using the same server. Sep 02 17:04:41 definatly looking forward to seeing the progress on this project Sep 02 17:05:22 which project? Sep 02 17:05:47 Bascially anything having to do with X Sep 02 17:06:12 ah, I am not a fan of having WM inside of a WM. Sep 02 17:07:29 I agree I would think it is better just to have individual Xapps seem like regular webos apps Sep 02 17:08:21 yup Sep 02 17:08:43 has anything for then been posted in the repo? or still planning phase? Sep 02 17:08:58 eventually xecutah may become a daemon, keeping track of which apps are assigned to which display. Sep 02 17:09:53 elpollodiablo1, it needs to be documented in the wiki, I already have "hacks" that I've done to make it run xchat, libreoffice, and other apps. Sep 02 17:10:20 from Xecutah? or just through the chroot? Sep 02 17:10:34 now I am formalizing it by making the shim, and creating xapps "packages" Sep 02 17:10:40 using Xecutah Sep 02 17:10:53 I hacked up chroot_wrapper.sh Sep 02 17:11:15 for the ubuntu-natty-chroot Sep 02 17:11:26 but thats not the preferable way to handle this Sep 02 17:11:45 also it doesn't let me "install, control, and remove" them from preware. Sep 02 17:12:54 ah so right now it is just launch something that has already been installed. well when you get to a beta point I would be willing to help out with testing and documenting if need be. Sep 02 17:13:01 the shim allows me to use preware to install apps in a chroot, based upon what the chroot needs Sep 02 17:14:59 never thought about installing through preware but that seems like the logical choice Sep 02 17:15:16 I hope to have all this ready for the testing feed tonight Sep 02 17:15:42 so each kind of chroot can use its own package management system. Sep 02 17:15:49 didnt realize you were that far along Sep 02 17:16:55 when we first started seeing folks use chroots to get apps that they couldn't otherwise I realized that a wm inside a wm was not good. Sep 02 17:17:18 also they were trying crazy apps that would never work on a TP Sep 02 17:17:35 (like no hardware support for x86 things) Sep 02 17:17:58 made sense to "guide" folks to apps that are known to work with the TP. Sep 02 17:20:08 also xecutah currently uses a hardcoded "whitelist" Sep 02 17:22:02 so every app recipe also has to modify xecutah recipe too... Sep 02 17:25:04 Xecutah needs a rewrite too...to more gracefully handle app additions and (currently) multiple x-server sessions. Sep 02 17:25:23 I am working up to that. Sep 02 17:34:26 does anyone here mount /any/ remote fileshare to their tp for video watching? friend got a tp and trying to help, but a)somewhat linux noob and b)i don't know what works/doesn't/recommended Sep 02 17:36:24 another project I've had no time to finish. Sep 02 17:37:14 dtzWill, people have reported that they are using CIFS shares from windows computers Sep 02 17:37:31 now whether the apps can see them is another thing Sep 02 17:38:03 I guess if you put the mountpoint *inside* of /media/internal that would work. Sep 02 17:41:29 dtzWill, is now a good time to ask questions about how to handle multiple cards using x-server? Sep 02 17:42:35 ka6sox: i was wondering same thing re:mounting inside /media/internal, but tools that use fileindexer might not work, shrug Sep 02 17:43:02 ya, but they both would be "windows" ish fS Sep 02 17:44:12 right but if it uses some mechanism that relies on the fs to ping it when there are changes (inotify, etc) then mounting over portions of it might not do what you'd like? i obviously don't know Sep 02 17:44:43 me neither Sep 02 17:46:10 and yeah go ahead, although still waking up so might be a tad slow :) Sep 02 17:47:47 currently, as I see the way that x-server is spawned, the call to Xsdl includes a display=:0 argument. Sep 02 17:48:17 so theoretically I could modify that call to instead be display=:X Sep 02 17:49:14 then xecutah would either have to have a pool sooo big of numbers that it never could use the same one again or... Sep 02 17:49:32 keep track of which ones are up, which ones are in use, and reassign numbers. Sep 02 17:50:57 this is where I got confused Sep 02 17:51:20 how to know when a x-server has died or had its card thrown away. Sep 02 17:52:41 yes, that is a challenge. only good way afaik is to try the display itself :/ Sep 02 17:53:08 you mean attempt to spawn one on a display and watch for errors? Sep 02 17:54:21 so I guess I need to manually attempt to spawn an x-server on top of an existing one and see what feedback I get. Sep 02 17:54:36 well, no, we'd write something key'd against xlib that tries to open a connection to the specified display Sep 02 17:54:41 but its' not pretty no Sep 02 17:54:56 thrte of course are other signs, locks in /tmp, etc, but they aren't alwas cleaned up nicely Sep 02 17:56:49 19566 ? S the :0.0 is the display number and window right? Sep 02 17:57:25 if by window you mean screen, then yes Sep 02 17:58:12 okay I can easily parse down to the :X Sep 02 17:59:11 then I need to make xecutah pass a display argument to the app being started. Sep 02 17:59:42 I really didn't want to have to keep track of things. Sep 02 17:59:57 just figure out what available and assign it. Sep 02 18:00:09 might be cool if we could have the X server open additional screens in cards Sep 02 18:00:21 so cards would be :0.0 :0.1 etc. Sep 02 18:00:42 yeah, ideally for all this we'd have like 1 root x server Sep 02 18:00:44 errr rootless Sep 02 18:00:49 and spawn cards for windows Sep 02 18:00:51 thats waaay over my head... Sep 02 18:00:57 ya, that would be ideal. Sep 02 18:01:15 my only problem with the cards for windows idea is multi-window apps, and with stacks that might work well Sep 02 18:01:25 but for now I'm going to work on having xecutah be able to spawn multiple X-servers and assign apps to them. Sep 02 18:01:34 ka6sox: agreed Sep 02 18:02:15 and I really didn't want to have to keep track of state. Sep 02 18:02:26 ka6sox: agreed. i have no idea how to do rootless hehe :) Sep 02 18:02:54 dtzWill, I have already *manually* hacked this to work. Sep 02 18:03:20 :D great! Sep 02 18:03:46 ka6sox: i can probably script up a "ensure this display is really dead" thing, if you want that for sanity-checking the state :/ Sep 02 18:03:47 I created modded scripts that allowed me to create 2 x-servers and then assign an app(manually) to :0 and the other app to :1 Sep 02 18:04:01 but those were hardcoded into scripts. Sep 02 18:04:16 saw this the otehr day "newx() { DISPLAY=:1.0; xinit $* -- :1; }; newx [app]" Sep 02 18:04:24 probably like what you're doing Sep 02 18:04:29 yep Sep 02 18:04:34 do wehave xinit? Sep 02 18:04:45 not that I know of Sep 02 18:04:52 I am just doing the equivelent of htat Sep 02 18:05:00 nah, but that's same as DISPLAY=:1.0; Xsdl :1; app Sep 02 18:05:12 so I had xchat in 1 card and xterm (chrooted) in a second card. Sep 02 18:05:17 (assuming the DISPLAY is exported so it's the value when 'app' is run) Sep 02 18:05:50 it was all script hackery Sep 02 18:06:07 and I only did that because I wanted to prove to myself this would work. Sep 02 18:06:24 eventually I could see xecutah being a daemon. Sep 02 18:06:46 service? Sep 02 18:06:48 ka6sox: what do you mean? Sep 02 18:06:57 app says" gimme a display" to the xecutah daemon Sep 02 18:06:58 it already is a service :) Sep 02 18:07:02 ah Sep 02 18:07:11 xecutah says: okay, lemme start :1 and give you that. Sep 02 18:07:35 well the xecutah gui is also useful, i like the idea of being able to start multiple apps on different display's if you'd like Sep 02 18:07:45 so the xecutah gui would have a list of the active displays Sep 02 18:07:49 you click one, then click what you wanna run Sep 02 18:07:58 dtzWill, this is 1 click Sep 02 18:08:00 but you can also swipe away displays and it kills them Sep 02 18:08:01 click on app Sep 02 18:08:06 or spawn a new one Sep 02 18:08:23 hmmm yeah Sep 02 18:08:26 app asks for a display, one is spawned and app runs against that. Sep 02 18:09:11 render all apps to same off-screen display, then have cards be windows on that display Sep 02 18:09:22 would allow moving apps between cards Sep 02 18:10:00 and cut/paste? Sep 02 18:10:04 linuxjacques: what's a 'window'? Sep 02 18:10:11 linuxjacques: you mean like workspaces/etc? Sep 02 18:10:21 i forget what the terminology is now Sep 02 18:10:35 window == view Sep 02 18:11:27 I need to figure out how to make apps aware of BT vs VKB and resize themselves upon start. Sep 02 18:11:32 let's say we had an off-screen X display which was 2048x1536 Sep 02 18:11:39 split it in 4 Sep 02 18:11:54 have one card showing each quadrant Sep 02 18:12:31 geometry would be enforced to keep apps within quadrants Sep 02 18:13:11 but apps would be movable Sep 02 18:17:51 app geometry is another issue with vkb being variable size (or missing) Sep 02 18:18:17 but thats not one I'm going to fight today. Sep 02 18:18:24 yeah, not sure of a general way of handling that Sep 02 18:18:31 agreed. Sep 02 18:18:42 apps like Gimp are impossible. Sep 02 18:18:42 even with xrandr resizing the display, I'm not sure all apps will heed it Sep 02 18:19:09 I need a few months to play with all this stuff :-( Sep 02 18:19:17 haha yeah... Sep 02 18:19:19 no kidding Sep 02 18:19:29 i see, interesting idea. mayb export via xinerama or something that we have 4 screens Sep 02 18:19:37 so that apps can 'maximize' within one Sep 02 18:20:01 i also don't know/thing palm supports multiple windows for a single pdk app? Sep 02 18:20:25 i'm not saying we can't, but if they don't support it it might not work very well, for example how do you konw *which* window is active at a moment? etc Sep 02 18:20:26 hehe Sep 02 18:20:31 but otherwise a really nifty idea Sep 02 18:21:54 okay so for now we are all agreed that each xapp gets its *own* x-server :X Sep 02 18:24:29 ? Sep 02 18:26:31 i mean taht's a great model and i saw go for it and i'll help, but i don't think we should /regress/ on being able to run multiple apps in a card Sep 02 18:27:01 if for no other reason than having a "gnome" "xapp" would be nice--but worthless if each panel had it's own card... etc Sep 02 18:27:07 but i don't think that's what you're suggesting at all :) Sep 02 18:28:25 no, I am saying 1app/card Sep 02 18:28:32 however we can do that Sep 02 18:28:54 as :X or :X.y either is fine by me. Sep 02 18:30:45 for now I suspect it will be :X Sep 02 18:32:38 I think a good goal would be :X.y as that reduces the number of x-servers taking up space. Sep 02 18:43:06 I didn't mean limit to one app per card, only that all apps would have to stay within quadrant boundaries Sep 02 18:43:20 so you wouldn't see parts of windows in a card Sep 02 18:43:59 and xinerama is definitely something we would want to look at, FAIK it might already do most of what we would need Sep 02 18:44:37 also, I'm not sure multiple X servers would be that inefficient; they should share a lot of pages Sep 02 18:45:41 I haven't looked at all. Sep 02 18:46:02 I just saw that both were running and that both apps were in their own x-server cards. Sep 02 18:46:03 I think compositing managers work a lot like what I was describing; rendering off-screen and compositing to views Sep 02 18:46:24 ka6sox: but I definitely agree for now please continue doing what you;re doing Sep 02 18:46:39 it might be never before we get the wild ideas working Sep 02 18:47:55 okay let me get an xapp to be able to be installed, and launched (from a hacked whitelist) and then I'll worry about multiple displays Sep 02 18:50:21 linuxjacques: yeah, we already strongly separate the render buffer and the screen itself Sep 02 18:50:26 since we don't have any direct ahrdware interaction Sep 02 18:50:46 so what you said is very straightforward, but i think the xinerama is imprtoant so any wm running in that 4-card X would know what to do Sep 02 18:51:25 and i am still uncertain how i'd demux the input hehe Sep 02 18:56:48 dtzWill: heh I hadn't thought about input Sep 02 18:57:14 I need to read up on xinerama Sep 02 18:59:42 linuxjacques: well i just mean whatever thing X normally exposes to clients Sep 02 18:59:55 on multi-monitor displays Sep 02 19:00:10 linuxjacques: actually only the mouse would be the issue..... hehe Sep 02 19:02:48 ka6sox, so where are we? Sep 02 19:08:41 scoutcamper, trying to get a git repo, working out high level details of how to attach x-displays to Xapps. Sep 02 19:09:10 i saw the x-display stuff Sep 02 19:09:53 next thing we need a test in the postinst of an Xapp to find out whether the webos user is present or not in the chroot. Sep 02 19:09:55 go do that :D Sep 02 19:10:44 I would check /etc/passwd and /home of the chroot personally. Sep 02 19:11:10 I don't want Xapps running as rooooooooooot Sep 02 19:11:22 (the apps themselves) Sep 02 19:12:06 so if the user isnt there we want to make it? Sep 02 19:12:12 I'm also trying to figure out how to bind mount /media/internal to the chroot as /home/webos (or if that is a good idea) Sep 02 19:12:25 nope I do NOT want to do the chroot's job. Sep 02 19:12:37 the chroot script shoud be idempotent. Sep 02 19:12:45 (postinst) Sep 02 19:13:01 ka6sox, dtzWill and i had talked about doing that before, we were thinking /media/internal to /media/internal lol Sep 02 19:13:45 okay, we can discuss that later..its a minor point. Sep 02 19:14:09 or /media/internal/home as /home/ Sep 02 19:14:24 iirc i was picky about not cluttering root of /media/internal with the normal dotfiles :) Sep 02 19:14:26 what I"m concerned about is linking to it and what happens when it switches to being a USB drive. Sep 02 19:14:45 probably bad things Sep 02 19:14:55 ka6sox: can we just disable usb drive mode? Sep 02 19:15:02 nope Sep 02 19:15:09 lots of headache, and the expected use case for usb drive mode is disabing functionality Sep 02 19:15:11 but who goes into usb while a chroot is running? Sep 02 19:15:13 does usb drive even keep cards open? Sep 02 19:15:23 suuure we can, just remove the usb mode buttons >_> Sep 02 19:15:24 lol Sep 02 19:15:25 yep Sep 02 19:15:30 it does Sep 02 19:16:00 just make a note, "NO USB DRIVE IF YOU LEAVE THIS OPEN, COMPLAIN AND BE SLAPPED" in the changelog Sep 02 19:16:07 LSOF might keep things from being dismounted. Sep 02 19:16:23 so I am hesitant on this. Sep 02 19:16:28 i remember the beta optware bootstrap that broke usb mode Sep 02 19:16:35 right Sep 02 19:16:44 then we put it into the cryptofs Sep 02 19:16:52 yep Sep 02 19:17:26 so the real answer is "probably not" for /media/internal access from a chroot. Sep 02 19:17:38 i know, have 2 scripts, one that has the bind mount and one doesnt, you have to go into xecutah prefs to enable the bind media/internal Sep 02 19:18:06 scoutcamper, I am trying to *eliminate* xecutah as something anybody sees. Sep 02 19:18:17 :) Sep 02 19:18:17 reduce it to a daemon. Sep 02 19:19:07 lets just make a note on the wiki page that users can edit the chroot-wrapper.sh to add a bind mount of media/internal Sep 02 19:19:29 s/users/darring devs not afraid of the webos doctor when things borke/ Sep 02 19:19:40 no Sep 02 19:19:50 I do NOT want to promote breaking stuff. Sep 02 19:20:07 let XDA-Developers do that :P Sep 02 19:20:27 lol Sep 02 19:20:32 that works too Sep 02 19:20:38 heh Sep 02 19:21:17 okay bbiaf..heading into the office to see if I can read these DVD's someone sent. Sep 02 19:22:00 I need to see if there is a partition table on it...probalby not Sep 02 20:19:44 scoutcamper|away, where on fiona is the entire xchat tree? Sep 02 20:20:17 nm....I'll just clone xterm and go from there. Sep 02 23:46:13 ka6sox, under /home/tking Sep 02 23:46:34 scoutcamper, thanks Sep 02 23:46:52 okay I am going to create different "apps" under xapps Sep 02 23:47:27 I need to find out which assistant in xecutah has the whitelist Sep 02 23:47:38 scoutcamper, you wanna look @ cgit for that? Sep 02 23:47:43 xecutah's cgit Sep 02 23:47:57 probably in sources Sep 02 23:48:03 (but I could be wrong) Sep 02 23:48:09 its gonna be a js file Sep 02 23:48:16 *-assistant.js Sep 02 23:48:44 k Sep 02 23:50:49 ka6sox, http://git.webos-internals.org/applications/xecutah/tree/app/assistants/main-assistant.js line 31 Sep 02 23:51:04 line 31-37 Sep 02 23:51:40 scoutcamper, okay now is the time to document this on the wiki. Sep 02 23:52:03 if you will start a page (similar to wIRC or other apps) I'll add stuff there. Sep 02 23:52:13 ka6sox, i am off to buy a truck, no wiki time for me Sep 02 23:52:30 fun fun Sep 03 00:31:29 ka6sox: have an app inside the chroot (like VMWare Tools) which sends an ls2 message to xecutah service to mount and unmount /media/internal Sep 03 00:33:03 good idea. Sep 03 00:34:11 what I was hoping for is something *inside* the chroot that would be able to see the a call to USB_Storage app was initiated and close open FDs and umount the partition. Sep 03 00:34:54 or is that not necessary... Sep 03 00:36:32 let me open a file with Tide and see what happens when I initiate a USB_storage event. Sep 03 00:36:36 inclined to think it'd be easier (better?) to instead prevent usb mode while a chroot is mounted, and just tell the user too bad Sep 03 00:36:36 lol Sep 03 00:36:56 like on the usb mode button have a hijack that calls xecutah "is usb mode safe?" and xecutah goes "no" Sep 03 00:36:56 lol Sep 03 00:37:01 xecutah can rebroadcast such events to xecutah-tools inside the chroot Sep 03 00:37:22 good...thats quite helpful. Sep 03 00:37:22 only because handling this seems like a PITA and who needs usb support *while* X is running :/ Sep 03 00:37:31 well alright if you guys think you can do it nicely i'll be quiet Sep 03 00:37:54 nice to not have to unnecessarily disable things... Sep 03 00:38:04 the USB_storage event I'm talking about is the one where your phone becomes an external drive Sep 03 00:38:51 okay let me look into monitoring ls2 events inside a chroot. Sep 03 00:39:46 dtzWill, the other kind...where you want to use a USB resource inside a chroot is much more "interesting" to me also. Sep 03 00:40:20 ie have libusb catch a gadget for the camera being connected via USB OTG. Sep 03 00:40:42 ka6sox: agreed--but i think i was on board, i was just suggesting instead of this tools/relay/automounting/autofdclosing setup we just dno't let the device go into usb mode while a chroot is using /media/internal Sep 03 00:41:09 maybe have xecutah ahve buttons to mount/unmount it and (and on unmount success let the device go into usb mode) Sep 03 00:41:30 if it can't unmount it don't try to autoclose anything just tell user they need to fix that Sep 03 00:41:48 if its the same as what happens when Tide is editing a file on that partition then it should be okay Sep 03 00:42:05 because LSOF is the same for both things. Sep 03 00:42:15 its "busy" Sep 03 00:42:27 and won't umount Sep 03 00:42:30 guess this seems like an awful lot of error-pronie automagic to not only save a user from themselves but rescue them as well Sep 03 00:42:41 right, it won't umount which is what i was suggesting was the oracle Sep 03 00:42:48 kk Sep 03 00:42:49 ka6sox: are you saying that webos already does that? Sep 03 00:43:05 dunno...about to find out how it handles it Sep 03 00:43:06 ka6sox: like if i edit a file with TIDE on the usb, will webos just refuse usb mode? Sep 03 00:43:09 hehe k Sep 03 00:43:26 that is what I want to find out Sep 03 00:43:54 Tide, editing a file on /media/internal, go into USB mode...see what happens. Sep 03 00:50:57 rwhitby-pre, I think in order to make this work like the jails the chroot would have to live in a cryptofs. Sep 03 00:51:05 and be SLOW. Sep 03 00:51:46 so the SIGUSR1 event would be cause and "drop" all fd's in that partition. Sep 03 00:52:26 s/cause/caught Sep 03 00:53:28 dtzWill, I went in and out of USB mode with Tide, and since its in a jail it didn't care Sep 03 00:53:40 because thats *inside* the crpytofs. Sep 03 01:03:29 so let me open a file in vim from the chroot to the /media/internal Sep 03 01:03:37 and attempt to USB Sep 03 01:05:48 doesn't have to be even from the chroot. Sep 03 01:09:31 dtzWill, that was a wierd experience.... Sep 03 01:09:42 1) it let me keep on editing the open file Sep 03 01:10:19 2) I did a "w /media/internal/foo44.txt" and it did NOT complain Sep 03 01:10:58 3) when I got done with the USB mode I umounted it from the laptop Sep 03 01:11:11 4) no /media/internal/foo* exists Sep 03 01:11:18 so it silently fails. Sep 03 01:12:10 and I was NOT in a jail. Sep 03 01:13:38 ka6sox: if you go back into usb mode is the file still there? Sep 03 01:13:56 they probably unmount /media/internal, you wrote foo44, and then they mounted over /media/internal hiding foo44 Sep 03 01:14:25 oh, thats UGLY...but probably true Sep 03 01:14:52 I can find the upstart script for that event and that is exactly what happend. Sep 03 01:15:00 becuase the mountpoint didn't disappear Sep 03 01:15:46