#htc-linux

<2012>  <January>  Show everything  Back to index      
1234567891011121314151617
1819202122232425262728293031

00:02 detule emailed, i wouldn't look too hard at it
00:03 detule should i start a clean 3.2 branch on linux-msm-rhod
00:03 jonpry what is it precisely? just the first commit to 3.1?
00:04 detule it is plus adding the missing headers, mach-msm and drivers/usb/gadget from 3.1, also this might be an issue one of the MACHINE parameters "boot_param" seems to be renamed to "atag_offset" in 3.2 (this at the end of board-htcrhodium.c)
00:04 detule perhaps i got that wrong
00:06 Alex[sp3dev] iirc, boot_param was not renamed. boot_param was an actual physical address of params, adn atag_offset is an offset from PHYS_OFFSET
00:07 jonpry the stuff in mach-msm and usb is a little sketchy
00:08 Alex[sp3dev] jonpry: do you have usb working with updated 3.1 usb stack? or you copied all usb stuff from 3.0?
00:08 detule Alex[sp3dev], struct machine_desc doesn't have boot_params member any more in 3.2
00:08 jonpry i used 3.0 usb because at the time there was no android 3.1 stuff out
00:09 Alex[sp3dev] detule: i know that, i just wanted to notice that atag_offset is not a direct replacement
00:10 detule from what i saw in whatever .S file was relevant it looked like an in-situ replacement
00:10 jonpry > - .boot_params = S5P_PA_SDRAM + 0x100,
00:10 jonpry > + .atag_offset = 0x100,
00:10 jonpry so something is different
00:10 jonpry i am not sure our phys_offset is non zero
00:10 Alex[sp3dev] it is 0x10000000, check memory.h
00:11 jonpry fare enough
00:11 Alex[sp3dev] also, a bro tip: disable all machines except rhodium. when there is only one machine in kernel, mtype is not used. so it should boot even with without atags
00:12 jonpry i want to get rid of machine_fixup
00:13 Alex[sp3dev] then fix haret
00:13 jonpry do mem=78M@0x1000000 kind of stuff
00:13 Cotulla noone wants to presave compatible
00:13 Cotulla :P
00:39 detule i think it needs this in kerne/setup.c
00:39 detule - tags = (void *)(PAGE_OFFSET + mdesc->atag_offset);
00:39 detule + tags = (void *)(mdesc->atag_offset);
00:40 detule arm/kernel/setup.c
00:45 Cotulla lol that's useless change
00:45 Cotulla they want to broke bootloaders too
00:46 jonpry detule, why not just change + .atag_offset = 0x10000100, to 0x100 ?
00:47 detule that's probably a better idea, it doesn't boot either way
00:52 jonpry i don't remember needing this stuff in debug.S or head.S
00:52 detule it doesnt compile with errors
00:52 detule i mean it doesn't compile without it
00:53 detule they changed something about the arguments
00:54 jonpry that might need to be understood better
00:54 jonpry also this defconfig is unprocessed?
00:54 detule it's the defconfig from the subsequent commit in 3.1
00:56 jonpry these changes to head.S could blow it up imho
00:59 jonpry best to use stock 3.2 arch/arm/mach-msm/include/mach/debug-macro.S
01:11 detule jonpry, cool that should fix that compilation error
01:18 detule you'll probably have an easier time starting from scratch than wading through that junk
01:18 detule dinnertime
02:13 jonpry d3tul3, you push that 3.2?
02:14 d3tul3 wait didn't know if you wanted me to - just the clean vanilla stuff right?
02:14 jonpry i don't see anything terribly wrong with this patch you did
02:14 jonpry if it shows up as more than 1 commit the whole thing is fine
02:14 jonpry can always revert it. but i like it
02:14 d3tul3 hard to debug it when it doesn't boot
02:15 jonpry yeah but that stuff happens
02:16 jonpry i'll put debug code in decompress and arch init
02:23 d3tul3 jonpry, it's up there
02:23 jonpry cool thanks
02:24 d3tul3 i guess it needs the vanilla head.S debug-macro.S and the atag_offset fix
02:24 jonpry no problem
02:24 jonpry found something on the mailing list about failures with old toolchain
02:25 jonpry 2011q1 wouldn't work. had to use q3
02:25 d3tul3 for 3.2?
02:25 jonpry yup
02:26 d3tul3 arm-none-eabi 2010q3 doesn't work on the mac, but 2010q3 arm-none-linux-gnueabi compiled it fine
02:27 jonpry i think the old toolchain compiled it but the resultant kernel was useless
02:28 d3tul3 ugh if that ends up being the problem...
02:38 jonpry why is gitorius so slow :(
03:10 jonpry d3tul3, found something
03:10 d3tul3 what did i mess up
03:10 jonpry prototype is now void (*fixup)(struct tag *, char **,
03:10 jonpry struct meminfo *);
03:12 jonpry think just machine_desc* needs to be removed
03:14 d3tul3 uh
03:14 d3tul3 why oh why
03:14 jonpry double vibe
03:14 jonpry video
03:14 jonpry rootfs
03:14 d3tul3 perfect
03:14 jonpry :D
03:14 d3tul3 dammit i almost had it
03:15 arrrghhh 3.2 tiem?
03:15 jonpry yup
03:16 d3tul3 was that 4 hrs?
03:16 jonpry no it just synced like 5 minutes ago
03:17 arrrghhh trololololol
03:17 jonpry just looked at the build warnings
03:17 arrrghhh didn't even need coffee
03:17 jonpry prototype errors are actually quite dangerous
03:17 jonpry especially in machine fixup :p
03:17 d3tul3 :)
03:18 d3tul3 wish it would do more than a warning
03:18 jonpry yeah you have to make sure to fix them
03:18 d3tul3 alright i'll get 3.3 right
03:20 jonpry well this wasn't bad
03:20 jonpry i don't have a working android yet though
03:20 arrrghhh gbx0b
03:21 jonpry where do i get that?
03:21 arrrghhh http://forum.xda-developers.com/showthread.php?t=1420264
03:22 arrrghhh jonpry gets handfed links w/o me bitching. i think he's the only one...
03:22 arrrghhh :P
03:23 jonpry i tried to search for it yesterday
03:23 jonpry oic. the gbx0a is at the top of the forum for no particular reason
03:24 arrrghhh well
03:24 arrrghhh 0b is in testing
03:24 arrrghhh 0a.. i guess i need to desticky
03:24 arrrghhh done.
03:25 jonpry how do you feel about h2o?
03:25 d3tul3 water? i kind of like it
03:27 arrrghhh nourising
03:27 arrrghhh jeebus
03:27 arrrghhh nourishing
03:28 jonpry no wireless
03:31 jonpry this gbx0b is 3.1?
03:31 d3tul3 .27 is bundled
03:32 jonpry can't have that
03:33 d3tul3 obviously
03:33 arrrghhh 3.1 is rhod-only...
03:33 jonpry so is life
03:33 arrrghhh sorry, there are other devices that we attempt to poorly support.
03:33 arrrghhh lol
03:35 arrrghhh blackstone users are having to evidently do build.prop hacking to get the soft kb to show up correctly. so that's awesome.
04:04 jonpry urg my touch screen is reversed
04:14 d3tul3 touch-screen reversed?
04:15 jonpry top is bottom
04:17 d3tul3 0b with 3.1?
04:18 jonpry 27
04:21 d3tul3 where's arrrghhh when you need him
04:25 jonpry d3tul3 3.2 working for you?
04:26 d3tul3 you put the patch up?
04:26 rpierce99 jonpry: isn't the touchscreen direction determined by the calibration file? where did you get your ts-calib?
04:27 jonpry the thread
04:27 jonpry d3tul3, pushed
04:28 jonpry should i just delete the calib and it will generate one?
04:29 rpierce99 i still don't think that works because of the kernel spam
04:29 jonpry kspam eh
04:31 jonpry k got it
04:32 jonpry just no scbs,gposit or 3.2
04:32 d3tul3 lol
04:34 jonpry phone calls don't seem to work right either
04:35 d3tul3 oh?
04:35 d3tul3 with is drivers/ssb ?
04:36 jonpry takes a long time before it rings when calling. and nothing seems to happen when people call me
04:37 d3tul3 it takes a bit to settle at the beginning, or is this every call
04:37 d3tul3 ERROR: "__msm_ioremap" [drivers/ssb/ssb.ko] undefined!
04:38 jonpry should i push my config?
04:38 jonpry maybe a make clean
04:44 jonpry wasn't there some known problem with the phone not waking up for calls?
04:44 rpierce99 yes, it's slow for some, but not others, it doesn't start vibrating for a good 3 or 4 rings, i can't pick it up for about 10
04:45 rpierce99 and there are audio issues that were just squashed this morning, so if vibe is off you might not hear a ring at all
04:45 rpierce99 but the phone should wake up and vibe without issue
04:45 d3tul3 yeah make sure you have the vibe on
04:47 d3tul3 but it varies across devices, over the last couple of days while mucking with acoustic i must've called my phone over a hundred times
04:47 d3tul3 the screen comes on with in a vibe or two probably > 90% of the time for me
04:48 jonpry gotta beet this into shape
04:49 d3tul3 i wonder if all the acoustic spam as the screen comes on is causing an issue
04:50 rpierce99 didn't we at one point find that the dmesg spam overall was causing some other significant issue
04:52 rpierce99 but specifically for the phone call issue, wouldn't that manfiest the same on devices with the same kernel, at least approximately
04:53 jonpry some people just have crappy phones
04:53 rpierce99 like you and I?
04:53 d3tul3 i am on a partitioned build, don't think that would matter
04:53 d3tul3 VZ radio
04:54 rpierce99 you also run ext4
04:54 jonpry no like other people
04:54 jonpry 3.2 with big block ext4 is going to rule
04:54 rpierce99 so all 3 of those things are different for me, but it's not the vz radio because arrrghhh and i are both in the slow calls boat
04:55 jonpry and i haz gsm
04:56 d3tul3 rpierce99, have you tried that hacked "for_broken_rhods" libaudio
04:56 rpierce99 no
04:56 d3tul3 there is a function i am not calling in there
04:56 d3tul3 isStreamActive
04:57 d3tul3 which cycles through the active audio streams and tries to figure out which one is active could be that one is blocking
04:57 jonpry well this is useless. it goes to voicemail before it ever rings and i have the phone awake
04:58 d3tul3 do you see the call in logcat
04:58 d3tul3 or it doesn't even register
04:58 rpierce99 yeah that's not normal
04:59 jonpry no call in logcat
05:00 rpierce99 wow my phone just woke up super duper fast
05:00 d3tul3 t-mo is screwing you
05:00 rpierce99 hm, whatever was broken before is working fine now
05:01 jonpry i'm not on tmo anymore. got h2o
05:01 jonpry works fine in winmo
05:01 d3tul3 rpierce99, which kernel
05:01 rpierce99 hm
05:02 rpierce99 .39.3-gee04dfa
05:02 d3tul3 i am going to charge my Skype expenditures to xdandroid
05:02 jonpry i bet 27 is just too lame to work
05:03 rpierce99 i was certainly having the issue on .39 before though
05:08 jonpry bleh, i'm switching to the hot stuff
05:08 jonpry looks like no hw3d and no touch
05:09 jonpry sound works
05:10 d3tul3 3.2?
05:10 jonpry yup
05:10 d3tul3 you building the modules?
05:10 jonpry worked on first ring though
05:10 jonpry no
05:11 d3tul3 ok coz i am still getting that error when i try to build the modules, but i'll ignore it for now probably a ton of dependencies missing from
05:12 d3tul3 jonpry, probably the first few patches form 3.1 (touchscreen and hw3d) will go in clean
05:12 jonpry so is this assorted android patches current or same vintage as the first 3.1?
05:12 d3tul3 mach-msm is current with mach-msm on 3.1
05:12 d3tul3 as is usb/gadget
05:12 jonpry ic
05:12 d3tul3 everything else is vintage 3.1
05:33 jonpry got 3d
05:34 d3tul3 the touchscreen commit from 3.1 applies cleanly
05:34 jonpry yeah i pushed it
05:34 jonpry and 3d
05:35 d3tul3 cool
05:36 d3tul3 it would probably make life easier if we try to apply those patches from 3.1 in order when possible
05:36 jonpry i think this is in order
05:37 d3tul3 other than kexec that patch might have to go in early to make life easier :)
05:38 d3tul3 yeah i am sure they are i was looking forward
05:38 jonpry things were going ok until it slept. not i have no phone calls in or out
05:39 d3tul3 not sure if missing wake locks would be causing issues
05:39 jonpry doubt it
05:39 jonpry i think the ril actually lost control
05:41 d3tul3 i think i read that 3.0 is the next LTS (after 2.6.32)
05:41 d3tul3 3.1 apparently has only one or two more updates in store
05:42 jonpry good thing we have 3.2 already
05:42 jonpry it thinks i have an ongoing phone call
05:44 d3tul3 that defconfig is kind-of wacko the cpufreq stuff i know made 3.1 kind of unusable early on
05:46 jonpry just make a new one after all the code is ported?
05:51 d3tul3 yeah, k i am out thanks for getting 3.x off the ground…again...
05:51 jonpry np
16:42 detule i guess staging/android is making its way back to 3.3
17:33 GNUtoo pong Alex[sp3dev]