00:20.57 | detule | jonpry, config STRICT_MEMORY_RWX |
00:20.57 | detule | bool "restrict kernel memory permissions as much as possible" <-- wtf!!!! |
01:00.49 | jonpry | lol |
01:02.52 | jonpry | detule, was that the problem? |
01:04.32 | *** join/#htc-linux FlawlesStyle (~LOL@unaffiliated/flawlesstyle) |
01:05.36 | detule | yeah that was that particular problem |
01:06.02 | detule | now it just shuts down .... and i lose the ram console when i boot back up |
01:06.49 | detule | though it could be because my machine_kexec looks nothing like stock right now |
01:08.49 | *** join/#htc-linux FlawlesStyle (~LOL@unaffiliated/flawlesstyle) |
01:10.05 | jonpry | thats possible |
01:10.28 | jonpry | your sure ramconsole is lost, or just nothing ever got put in it? |
01:11.22 | detule | like the ramconsole header gets corrupted somehow |
01:11.27 | detule | and it doesn't even export last_kmsg |
01:11.45 | detule | need to figure out how to bypass that header check and always export last_kmsg -> this is a problem |
01:12.11 | jonpry | just remove the ram console driver from your host kernel |
01:12.15 | jonpry | just have the memory hole |
01:12.23 | jonpry | then you can dump the region with /dev/mem |
01:13.38 | detule | i think i am just not clearing some flag |
01:15.09 | detule | hm got a ramconsole that time |
01:26.48 | detule | this is just strange |
01:26.49 | detule | http://pastebin.com/rmRrWTVn |
01:32.02 | jonpry | looks like no initrd |
01:32.52 | jonpry | your ramconsole must not be very big or somehing |
01:32.59 | jonpry | missing all the good stuff |
01:43.11 | *** join/#htc-linux ccube (ccube@nx.ccube.de) |
01:50.02 | detule | there's def stuff missing |
01:50.08 | detule | here's one after a normal reboot http://pastebin.com/7xBqkL7g |
01:52.07 | detule | i think initrd is there --> [ 1.707462] init: Cannot open real block device |
01:53.06 | jonpry | yeah lloks like it mounted root and stuff |
01:53.10 | jonpry | but no display? |
01:54.11 | detule | no, from my perspective it blanked, heard a vibe and it went to battery charging |
01:54.34 | jonpry | adb works? |
01:57.15 | detule | why didn't it probe msm_ts |
01:57.32 | jonpry | shouldn't use it. thats resistive only |
01:58.38 | detule | my normal kmsg has something like [ 1.472394] mms_ts 3-0048: [TSP] ISC Ver [0xbb] [0x20] [0x20] |
01:58.38 | detule | [ 1.472455] mms_ts 3-0048: [TSP] fw is latest. Do not update. |
01:59.46 | detule | i think i am just mucking up the cmdline probably since it didn't parse init.qcom.rc |
02:00.21 | detule | i think it gets that init.*.rc value from something like androidboot.hardware=qcom |
02:00.35 | jonpry | how are you setting command line? |
02:00.56 | jonpry | is it devtree enabled and what about atags? |
02:01.21 | detule | cmd=`cat /proc/cmdline` |
02:01.21 | detule | cmd="$cmd kexec" |
02:01.40 | detule | <PROTECTED> |
02:01.59 | detule | i think the kexec binary auotmatically reads atags from /proc/atags |
02:02.31 | jonpry | only if your host kernel has /proc/atags |
02:02.36 | detule | it does |
02:02.39 | detule | when i cat it though |
02:02.40 | detule | oot.nvdata_backup=1 androidboot.boot_recovery=0 level=0x574f4c44 androidboot.emmc=true androidboot.serialno=2c1c90c6 a |
02:02.50 | detule | i get some garbled shit |
02:02.56 | jonpry | thats normal |
02:02.58 | jonpry | its binary |
02:03.19 | jonpry | i'm not sure if kexec command line override the atag one or not |
02:03.35 | jonpry | i used fixed command line in the target kernel because i'm too lazy to figure it out |
02:10.03 | detule | alright devtree....no idea...do i even need that? |
02:18.23 | *** join/#htc-linux rob_w (~bob@unaffiliated/rob-w/x-1112029) |
02:27.42 | jonpry | you need it if the kernel uses it |
02:27.48 | jonpry | CONFIG_OF i think |
03:25.18 | *** join/#htc-linux rob_w_ (~bob@ppp-46-244-128-99.dynamic.mnet-online.de) |
03:55.31 | *** join/#htc-linux AstainHellbring (AstainHell@unaffiliated/astainhellbring) |
04:00.35 | *** join/#htc-linux G4y (~G4y@160.8.202.84.customer.cdi.no) |
04:17.09 | *** join/#htc-linux stybla (stybla@pylon.zeratul.org) |
04:24.36 | *** join/#htc-linux Bry8Star (~Bry8Star@gateway/tor-sasl/bry8star) |
04:56.56 | *** join/#htc-linux detule (~detule@unaffiliated/d3tul3) |
05:08.40 | *** join/#htc-linux DuperMan (~Duper@46-116-127-163.bb.netvision.net.il) |
05:25.23 | *** join/#htc-linux DuperMa (~Duper@93-173-48-18.bb.netvision.net.il) |
05:32.44 | *** join/#htc-linux DuperMan (~Duper@93-173-181-110.bb.netvision.net.il) |
05:54.27 | *** join/#htc-linux kiozen (~kiozen@p578a42db.dip0.t-ipconnect.de) |
05:59.58 | *** join/#htc-linux mes (~mes@sentry.lazo.ca) |
06:06.51 | *** join/#htc-linux jianC (4c141154@gateway/web/freenode/ip.76.20.17.84) |
06:32.06 | *** join/#htc-linux balans2 (~user@82-170-217-205.ip.telfort.nl) |
06:56.02 | *** join/#htc-linux Alex[sp3dev] (~alexander@nat.rnd.stcnet.ru) |
07:10.10 | *** join/#htc-linux ALoGeNo (~alogeno@unaffiliated/alogeno) |
07:11.47 | *** join/#htc-linux eR^zeRa` (~zzeratul@88.103.98.168) |
07:13.23 | *** join/#htc-linux DuperMa (~Duper@93-173-181-110.bb.netvision.net.il) |
07:21.01 | *** join/#htc-linux Ondalf (~ondalf@cable-roi-fff8dd00-39.dhcp.inet.fi) |
07:36.22 | *** join/#htc-linux conantroutman (~chatzilla@cpc5-pert4-2-0-cust205.sgyl.cable.virginmedia.com) |
07:40.19 | *** join/#htc-linux helicopter88 (~helicopte@host146-117-dynamic.55-79-r.retail.telecomitalia.it) |
08:10.48 | *** join/#htc-linux toer_ (~tore@179.81-166-86.customer.lyse.net) |
08:10.48 | *** join/#htc-linux NetRipper (~netripper@tikkie.net) |
08:10.48 | *** join/#htc-linux arrrghhh (~arrrghhh@unaffiliated/arrrghhh) |
08:10.48 | *** join/#htc-linux nrirclog (~nrirclog@netripper.com) |
08:15.02 | *** join/#htc-linux toer_ (~tore@179.81-166-86.customer.lyse.net) |
08:15.02 | *** join/#htc-linux NetRipper (~netripper@tikkie.net) |
08:15.02 | *** join/#htc-linux arrrghhh (~arrrghhh@unaffiliated/arrrghhh) |
08:15.02 | *** join/#htc-linux nrirclog (~nrirclog@netripper.com) |
08:15.51 | *** join/#htc-linux lamikr (lamikr@nat/nokia/x-eblyehmwdvrzuvdj) |
08:28.17 | *** join/#htc-linux arif-ali (~arif-ali@81.23.53.226) |
09:06.08 | *** join/#htc-linux marc1706 (~Marc@phpbb/modifications/marc1706) |
09:22.53 | *** join/#htc-linux ychavan (ychavan@nat/redhat/x-usscadfslatfarik) |
09:27.08 | *** join/#htc-linux lamikr (lamikr@nat/nokia/x-eokrqpeufzogdqni) |
09:30.46 | *** join/#htc-linux walter79 (~walter79@31-17-62-189-dynip.superkabel.de) |
10:03.55 | *** join/#htc-linux TheCrowe (~Thomas@2001:470:1f05:c19::89) |
10:36.30 | *** join/#htc-linux balans2 (~user@82-170-217-205.ip.telfort.nl) |
10:37.37 | *** join/#htc-linux leviathan (~quassel@2001:470:26:484:6ef0:49ff:fee6:8dca) |
10:50.51 | *** join/#htc-linux walter79 (~walter79@31-17-62-189-dynip.superkabel.de) |
11:08.48 | *** join/#htc-linux helicopter88 (~helicopte@host112-113-dynamic.26-79-r.retail.telecomitalia.it) |
12:11.39 | *** join/#htc-linux paulk-desktop (~paulk@lib33-1-82-233-88-171.fbx.proxad.net) |
12:15.40 | *** join/#htc-linux detule (~detule@unaffiliated/d3tul3) |
12:51.20 | *** join/#htc-linux kiozen (~kiozen@p578a42db.dip0.t-ipconnect.de) |
13:37.36 | *** join/#htc-linux rajkosto (~rajkosto@cable-94-189-239-212.dynamic.sbb.rs) |
14:14.53 | *** join/#htc-linux rajkosto (~rajkosto@cable-94-189-239-212.dynamic.sbb.rs) |
14:22.52 | *** join/#htc-linux Rajko (~rajkosto@cable-94-189-239-212.dynamic.sbb.rs) |
14:23.33 | *** join/#htc-linux detule (~detule@unaffiliated/d3tul3) |
14:31.01 | *** join/#htc-linux bitrot (~rajkosto@wan.rajkonet.info) |
14:42.30 | *** join/#htc-linux rajkosto (~rajkosto@wan.rajkonet.info) |
15:45.03 | *** join/#htc-linux Alex[sp3dev] (d5551202@gateway/web/freenode/ip.213.85.18.2) |
15:53.07 | *** join/#htc-linux detule (~detule@unaffiliated/d3tul3) |
16:08.01 | *** join/#htc-linux jonpry (~jon@c-24-17-200-206.hsd1.wa.comcast.net) |
16:24.52 | detule | jonpry, is it normal that when the device is off and you plug in a charger, it actually boots from the /boot zImage |
16:25.01 | detule | automatically |
16:27.26 | *** join/#htc-linux helicopter88 (~helicopte@95.234.115.212) |
16:33.11 | *** join/#htc-linux rob_w (~bob@ppp-46-244-128-99.dynamic.mnet-online.de) |
16:33.11 | *** join/#htc-linux rob_w (~bob@unaffiliated/rob-w/x-1112029) |
16:36.53 | Alex[sp3dev] | detule: which device? |
16:38.07 | detule | Alex[sp3dev], sgs iii |
16:38.13 | detule | perhaps this is standard on samsung devices |
16:38.41 | Alex[sp3dev] | either way that's a normal behavior. it is reasonable to make the bootloader as small as possible and move complex hw init and charging algos to kernel. on s2 and nexus the bootloader appends a cmdline param and launches kernel and init launches the charger splash screen |
16:39.14 | Alex[sp3dev] | detule: are you overall happy with the new exynos? doing any hacks? |
16:39.31 | detule | yeah that's what i get.....makes me slightly more worried about flashing /boot |
16:39.45 | detule | this is the U.S. msm 8960 edition |
16:40.01 | Alex[sp3dev] | with 2G ram? |
16:40.09 | detule | yes |
16:40.13 | Alex[sp3dev] | dang |
16:40.29 | Alex[sp3dev] | you MUST get it running ubuntu or meego. with hdmi and usb host support :) |
16:40.34 | detule | trying to get kexec going to have an easier way of testing 3.4 |
16:40.58 | Alex[sp3dev] | doesn't it have fastboot? |
16:42.57 | detule | there's CWM and what not out there for it....however the one time i tried actually flashing 3.4 onto /boot (and of course it failed), i lost the ramconsole afterwards |
16:43.52 | jonpry | mine turns on when charger is inserted |
16:44.03 | jonpry | seems to depend on what kind of off its in |
16:44.51 | detule | yeah those logs from last nigh weren't from booting with the kexec-ed kernel....it was from after the fact it going into charging mode |
16:45.23 | jonpry | i think you need to remove ramconsole support from the /boot |
16:45.55 | detule | but i think kexec might be failing before it hits cpu reset again in which case i need it there still |
16:46.04 | jonpry | not really |
16:46.24 | jonpry | you can write into the identity map |
16:46.34 | jonpry | even from relocate kernel |
16:46.48 | jonpry | your ramconsole is below 0xc*? |
16:46.58 | detule | no way up |
16:47.03 | jonpry | bleh |
16:47.14 | jonpry | gotta move it down |
16:47.32 | jonpry | like 0xb |
16:48.13 | detule | i don't really get it, i register ramconsole at a physical address...the identity maping it takes out before kexec is on the userspace memory from 0 - 0xb* right? there's no guarantee what's in there |
16:49.04 | jonpry | first off there will be no more ramconsole. just a hole |
16:49.23 | jonpry | second one thing will be gauranteed. the hole |
16:49.51 | jonpry | so you can do stuff like *(u32*)RAM_CONSOLE='a' |
16:51.48 | detule | i need to read up on linux MM, i thought virtual addresses from 0 to page offset are process specific and refer to dynamically managed |
16:51.56 | detule | memory |
16:52.07 | jonpry | not after setup_mm_for_reboot |
16:52.20 | jonpry | it zonks the whole process memory |
16:52.28 | jonpry | no process can ever run again |
16:53.27 | jonpry | and all addresses below 0xC are now idmapped. and you can pretend that mmu is off so long as your only accessing the bottom 3g |
16:56.58 | detule | alright that makes sense....first thing though i think i'll ifdef ram_console_probe if the poweroff_charging flag is set in commandline so it doesn't mess with it then |
17:04.42 | *** join/#htc-linux kiozen (~kiozen@ppp-93-104-93-117.dynamic.mnet-online.de) |
17:09.42 | jonpry | i finally managed to bring up aux core on kexec |
17:09.55 | jonpry | and i have a 3.0 kernel in the works |
17:10.05 | detule | nice |
17:10.41 | jonpry | just a bunch of other weird stuff is not working |
17:11.13 | jonpry | i'm convinced that no omap4 has ever kexecd with aux core |
17:11.56 | jonpry | there is zero infrastructure for dealing with what actually happens |
17:39.08 | *** join/#htc-linux paulk-desktop (~paulk@lib33-1-82-233-88-171.fbx.proxad.net) |
17:53.14 | *** join/#htc-linux skodde (~skodde@unaffiliated/skodde) |
17:54.40 | *** join/#htc-linux jianC (4c141154@gateway/web/freenode/ip.76.20.17.84) |
18:23.45 | *** join/#htc-linux balans2 (~user@82-170-217-205.ip.telfort.nl) |
18:24.34 | *** join/#htc-linux Bry8Star (~Bry8Star@gateway/tor-sasl/bry8star) |
18:36.05 | *** join/#htc-linux milaq (~milaq@h2041382.stratoserver.net) |
18:37.51 | detule | hm [ 1.261803] CPU1: failed to come online |
18:48.57 | *** join/#htc-linux BabelO_ (~wdlxtv@AMontpellier-553-1-176-108.w92-133.abo.wanadoo.fr) |
19:07.03 | jonpry | detule you got boot? |
19:07.10 | *** join/#htc-linux MethoS- (~clemens@134.102.106.250) |
19:11.55 | *** part/#htc-linux marc1706 (~Marc@phpbb/modifications/marc1706) |
19:12.38 | *** join/#htc-linux marc1706 (~Marc@phpbb/modifications/marc1706) |
19:19.37 | *** join/#htc-linux detule (~detule@unaffiliated/d3tul3) |
19:21.28 | jonpry | detule you got boot? |
19:29.09 | detule | yeah |
19:29.17 | detule | but watchdog stalled me |
19:30.11 | detule | clock problems too |
19:30.28 | detule | it was too much to hope for that it would be smooth on the other side of cpu reset |
19:41.02 | *** join/#htc-linux raymonddull (~raymonddu@c-69-245-114-102.hsd1.mi.comcast.net) |
19:48.55 | *** join/#htc-linux BHSPitMonkey (~stephen@unaffiliated/bhspitmonkey) |
20:00.26 | *** join/#htc-linux gassed (faxed@c-76-30-186-37.hsd1.tx.comcast.net) |
20:22.14 | *** join/#htc-linux ALoGeNo (~alogeno@243.Red-217-125-20.staticIP.rima-tde.net) |
20:22.14 | *** join/#htc-linux ALoGeNo (~alogeno@unaffiliated/alogeno) |
20:37.16 | *** join/#htc-linux Bry8Star (~Bry8Star@gateway/tor-sasl/bry8star) |
20:37.54 | *** join/#htc-linux zeusk_ (77eb3613@gateway/web/freenode/ip.119.235.54.19) |
20:41.17 | *** join/#htc-linux ALoGeNo (~alogeno@unaffiliated/alogeno) |
20:41.45 | *** join/#htc-linux DuperMan (~Duper@93-173-181-110.bb.netvision.net.il) |
20:53.45 | *** join/#htc-linux walter79 (~walter79@31-17-62-189-dynip.superkabel.de) |
21:07.28 | jonpry | clock issues abound |
21:11.48 | detule | there's some kind of a cpu logical map -> it always designates the booting cpu to be 0 |
21:14.18 | jonpry | yes |
21:14.36 | jonpry | bringing it online could be very difficult |
21:15.16 | jonpry | imho you don't really need to worry because when the kernel is finished. aux core will probably just work when not kexec'd |
21:16.56 | detule | if i cpu_down(1) before kexec, it hangs....otherwise, it boots but cpu_up(1) fails |
21:17.36 | zeusk_ | is it cortex-a9 ? |
21:18.23 | zeusk_ | there was some wierd bus init for A9-SMP in arm arm docs |
21:18.26 | detule | i think 8960 is based on something newer |
21:18.50 | zeusk_ | qualcomm... |
21:18.54 | jonpry | well |
21:19.15 | jonpry | the fundamental problem is that cpu startup is different in two ways in kexec |
21:19.28 | jonpry | first what turns on aux core is different |
21:19.37 | jonpry | second the reset vector is different |
21:19.57 | jonpry | so you end up in a situation where it kind of boots up at the wrong time and then does a hyperspace jump |
21:20.59 | jonpry | good news is that if you hanging when booting up you've already figure out how to turn it on |
21:21.11 | jonpry | whatever happens at the hang is your secret sauce |
21:21.49 | jonpry | then you just need to move that code into the wakeup function. and figure out how to program the reset vector |
21:22.42 | detule | i'll be processing that for a day or two and I am sure i'll as you questions about it later |
21:23.17 | jonpry | https://gitorious.org/droid3-kexec/droid3-kexec-kernel/commit/6140054821ece4d9bd69d43f5d7a9f4d3cb08b7f |
21:23.39 | zeusk_ | droid 3 is 8960 ? o.o |
21:23.49 | jonpry | omap4 |
21:24.11 | detule | msm has some secret dual_boost() that gets called on boot |
21:24.18 | detule | that tries to bring up non-boot cpus |
21:24.35 | detule | but that's no good |
21:24.43 | zeusk_ | secret ? maybe it's just a simple spinlock ? |
21:28.43 | zeusk_ | why the hell is lk modifying T bit in cpsr, isn't that illegal ? |
21:30.57 | jonpry | so dual_boost() causes it to hang? |
21:33.17 | zeusk_ | no, during boot, it checks cpu id, if it is not cpu0, the core goes into a spin loop till a value is changed in memory, dual_boost must be setting that value and releasing the core from loop and adding it to linux's scheduler |
21:33.39 | zeusk_ | or it can be even implemented using wfi and ipi |
22:06.40 | *** join/#htc-linux rajkosto (~rajkosto@wan.rajkonet.info) |
22:16.01 | detule | yeah i think it's dual_boost...calling cpu_up(nonboot) which then executes mach-msm/platsmp.c: boot_secondary and it fails there....need to get better debugging output to figure out what exactly fails |
22:25.52 | *** join/#htc-linux rajkosto (~rajkosto@wan.rajkonet.info) |
22:27.57 | jonpry | detule, sounds like it might be turning on at the right time. just doing the hyperspace jump |
22:31.13 | *** join/#htc-linux Ondalf (~ondalf@cable-roi-fff8dd00-39.dhcp.inet.fi) |
22:49.44 | jonpry | detule, you try the platsmp.c from stock kernel? that looks like some powerful stuff |
22:53.27 | detule | from CA? |
22:54.24 | jonpry | no i was looking at your sgsiii repo |
22:55.05 | detule | that's the platsmp i am using |
22:55.15 | detule | that repo is stock samsung source |
22:55.21 | jonpry | hrm |
22:57.02 | jonpry | it looks like all this stuff should be able to reset it |
22:58.08 | jonpry | https://gitorious.org/~detule/linux-msm-rhod/detules-linux-msm-rhod/blobs/local-3.4-sgs3/arch/arm/mach-msm/platsmp.c looks not the same as sgsiii |
22:59.08 | detule | i am trying to kexec the sgsiii kernel |
22:59.24 | jonpry | oh |
23:03.07 | jonpry | this scm_set_boot_addr is the key |
23:03.33 | jonpry | need to make sure that is happening before the best decides to come alive |
23:04.41 | detule | thanks i'll look into that -> this is what i have right now http://pastebin.com/ikjyvN9g |
23:05.22 | *** join/#htc-linux DuperMa (~Duper@93-173-181-110.bb.netvision.net.il) |
23:06.34 | jonpry | sounds like you have some kind of irq storm |
23:08.36 | detule | cpu doesn't come up and modem goes into reset |
23:12.50 | zeusk_ | jonpry, looks like they're using spin locks instead of wfi, why don't you try booting without cpu1 ? (only cpu0 enabled, don't release cpu1 from holding pen in platform_smp) |
23:49.06 | *** join/#htc-linux constintine (~constinti@216-111-105-244.dia.static.qwest.net) |
23:54.34 | *** join/#htc-linux ALoGeNo (~alogeno@243.Red-217-125-20.staticIP.rima-tde.net) |
23:54.34 | *** join/#htc-linux ALoGeNo (~alogeno@unaffiliated/alogeno) |