| 00:25.16 | *** join/#hawkboard CHli_ (~CHli@212-98-42-40.static.adslpremium.ch) |
| 02:31.15 | *** join/#hawkboard khasim (~a0393720@192.163.20.231) |
| 03:41.06 | *** join/#hawkboard jackman (~jackman@c-24-21-216-140.hsd1.or.comcast.net) |
| 05:10.12 | *** join/#hawkboard khasim (~a0393720@192.163.20.231) |
| 05:21.04 | *** join/#hawkboard virals (~Viral_Sac@59.97.0.145) |
| 05:46.16 | *** join/#hawkboard khasim (~a0393720@192.163.20.231) |
| 06:06.07 | *** join/#hawkboard awarth (~4c5b180c@gateway/web/freenode/x-kracdterwnuxzlht) |
| 06:08.10 | awarth | hey guys... is anyone else having trouble getting the right bootargs for making the console work out of tty0 (out of the vga port)? i've tried "console=tty0" but when I boot it's still only working on the serial port. |
| 06:08.50 | awarth | could it be that the kernel i'm using does not have support for what I want to do? |
| 06:21.56 | eFfeM_work | CHli_: saw your msg from yesterday afternoon, didn't have time yesterday, but can try to boot my card tonight, feel free to remind me after 7 pm cet (nick from home is eFfeM) |
| 06:22.20 | eFfeM_work | awarth: never tried that |
| 06:22.39 | awarth | thanks, eFfeM |
| 06:23.42 | awarth | btw, do you know if there is a step-by-step tutorial somewhere that shows how I might get one of the angstrom distros running off of a usb pen drive? |
| 06:31.42 | eFfeM_work | awarth: don't think there is one, but running the rootfs from usb pen is not difficult, booting the kernel from usb might be |
| 06:32.03 | eFfeM_work | not sure if hawk u-boot has usb support |
| 06:40.04 | *** join/#hawkboard franktango (~a0746747@nat/ti/x-twjpajaonraetbdx) |
| 07:05.26 | CHli | Hello everybody |
| 07:05.58 | CHli | Today : Try to load kernel through serial (like Gaston|Home) |
| 07:06.06 | Gaston|Trvl | back friday night the latest |
| 07:06.32 | CHli | ok, thanks for the help ! |
| 07:06.54 | Gaston|Trvl | I dont think yur issues are solved yet |
| 07:07.52 | CHli | No as I said I suspect bad hardware / or kernel deadlock, race condition :-| |
| 07:09.21 | Gaston|Trvl | as you can boot on the original kernel I think your board is ok, but the new kernel might need some tweeking to work with your particular chip revisions. |
| 07:09.38 | Gaston|Trvl | my 2 cents |
| 07:10.05 | Gaston|Trvl | Good luck and I will follow on hawkboard group |
| 07:10.19 | Gaston|Trvl | leaves |
| 07:12.31 | *** join/#hawkboard khasim (~a0393720@192.163.20.231) |
| 07:19.20 | CHli | Gaston|Trvl, Arghhh too late :( |
| 07:19.37 | CHli | I wanted you to give me your OMAP revision :) |
| 07:19.46 | Gaston|Trvl | I'm still here packing my stuff |
| 07:21.56 | *** join/#hawkboard khasim (~a0393720@192.163.20.231) |
| 07:22.17 | Gaston|Trvl | first line says XOMAPL138ZCE |
| 07:23.01 | CHli | same here :-| |
| 07:23.19 | Gaston|Trvl | second line $7-97A2GAJW |
| 07:23.33 | CHli | and then something like $7 - 99AIR8W |
| 07:23.35 | eFfeM_work | where is that info ? |
| 07:23.47 | CHli | on the OMAP itself |
| 07:23.54 | eFfeM_work | ah ok |
| 07:24.21 | Gaston|Trvl | third line GI or G1 all underscored |
| 07:24.27 | CHli | yeap same here |
| 07:25.05 | CHli | thanks |
| 07:28.49 | CHli | X = Experimental Device :-| |
| 07:44.08 | *** join/#hawkboard pepere_ (~539eaa89@gateway/web/freenode/x-qstynauzjigbygrj) |
| 07:44.38 | pepere_ | Hi all |
| 07:53.43 | CHli | Hi pepere_ |
| 07:54.04 | pepere_ | I've seen email from stuart :( |
| 07:54.16 | pepere_ | I'm a guy with a problematic baord :) |
| 07:54.56 | pepere_ | I've tried yesterday kernel from Gaston and still i was freezing... |
| 07:55.35 | pepere_ | I'm wondering If I can build and OE recent tree and still use old kernel (2.6.32)? |
| 07:55.40 | pepere_ | I thnik so... |
| 07:55.52 | pepere_ | for now it could be a sufficient workaround.... |
| 07:56.29 | pepere_ | CHli have you manage to get uboot compiled by yourself? |
| 07:56.41 | CHli | yes |
| 07:56.45 | CHli | ! |
| 07:56.48 | CHli | AT LEAST ! :D |
| 07:56.51 | pepere_ | good :) |
| 07:57.08 | pepere_ | Next step will be the kernel ;) |
| 07:57.54 | CHli | But the U-Boot was not the problem it seems. Wether I was doing something wrong with the AISGen GUI or they've got a bug in it :) (because I only managed to have a working image with the command line tool) |
| 07:58.06 | pepere_ | ha ok |
| 07:58.30 | CHli | Yes the Kernel...it seems the problem is known but no one wants to debug it. :( |
| 07:58.39 | pepere_ | or can... :) |
| 07:59.01 | pepere_ | perhaps with a lot of printk :) |
| 07:59.26 | CHli | yeah :( |
| 08:46.42 | *** join/#hawkboard wolfspraul (~wolfsprau@c-24-5-80-179.hsd1.ca.comcast.net) |
| 09:01.29 | eFfeM_work | what version kernel is this problem with ? |
| 09:01.54 | eFfeM_work | mine is probably .32 that would explain why I had no problem |
| 09:08.12 | *** join/#hawkboard _roger_ (~a0740758@nat/ti/x-mcxjovjouizkxsfu) |
| 09:13.10 | CHli | eFfeM_work, it's the one in OE based on arago. It's at 2.6.33-rc4 |
| 09:26.03 | *** join/#hawkboard virals (~Viral_Sac@59.97.1.135) |
| 09:26.12 | eFfeM_work | CHli: will check what i have |
| 09:31.26 | eFfeM_work | CHli: checked in the oe git, there are several changes for hawk in openembedded from april 12, my kernel is definitely older, guess the person to bother is Koen Kooi, he committed the patches. Also some of the changes are from Roger Monk |
| 09:31.59 | eFfeM_work | there is also an increase of the kernel partition |
| 09:32.10 | CHli | Yes to 4MB |
| 09:32.46 | eFfeM_work | see http://cgit.openembedded.org/cgit.cgi/openembedded/log/recipes/linux/linux-davinci/hawkboard/patch-2.6.33rc4-psp-to-hawkboard.patch and http://cgit.openembedded.org/cgit.cgi/openembedded/log/recipes/linux/linux-davinci_git.bb |
| 09:33.48 | CHli | Yeah I saw these already |
| 09:34.09 | CHli | I was talking to Koen yesteray on the openembedded devel mainling list |
| 09:34.30 | CHli | he HAD an hawkboard (prototype) but it's broken now so he can't help I think :( |
| 09:34.44 | eFfeM_work | you probably also can find him on #beagle |
| 09:35.10 | eFfeM_work | thinks it is not good to commit untested changes |
| 09:35.59 | *** join/#hawkboard CHli_ (~CHli@212-98-42-40.static.adslpremium.ch) |
| 09:36.33 | CHli_ | yeah I see him, thanks |
| 09:36.34 | eFfeM_work | CHli_: got this? |
| 09:36.35 | eFfeM_work | (11:34:40 AM) eFfeM_work: you probably also can find him on #beagle |
| 09:36.36 | eFfeM_work | (11:35:06 AM) ***eFfeM_work thinks it is not good to commit untested changes |
| 09:37.02 | CHli_ | he probably had it working on his prototype before it died :) |
| 10:12.42 | *** join/#hawkboard jackman (~jackman@c-24-21-216-140.hsd1.or.comcast.net) |
| 10:13.13 | *** join/#hawkboard wolfspraul (~wolfsprau@c-24-5-80-179.hsd1.ca.comcast.net) |
| 11:11.55 | pepere_ | Chli do you know a simple way to compile a specific uImage with oe , I mean where can i find recipes for compiling an older kernel for example? |
| 11:12.43 | _roger_ | pepere_ - what problems are you seeing with the latest kernel? |
| 11:12.44 | pepere_ | Perhaps a good thing to do is to find from which patch our baord started to no more work :) |
| 11:12.49 | pepere_ | ha |
| 11:13.07 | pepere_ | let me show you on paste bin |
| 11:13.11 | _roger_ | k |
| 11:13.44 | pepere_ | http://pastebin.com/72YXif6G |
| 11:13.52 | pepere_ | I've enabled early printk |
| 11:14.19 | _roger_ | ok, I've seen that one reported befoew |
| 11:14.35 | _roger_ | glad it's not a different one |
| 11:14.35 | pepere_ | and we (Chli, another guy and me) always freeze @0.35s when Switching to clock source timer0_1 |
| 11:14.43 | _roger_ | yep, ok |
| 11:14.44 | pepere_ | :) |
| 11:14.57 | _roger_ | I don't see that on any of the boards I have |
| 11:14.57 | pepere_ | Can I help to debug it? |
| 11:15.20 | pepere_ | or must I wait until someone skilled wil debug it? |
| 11:15.29 | pepere_ | arf... |
| 11:15.39 | _roger_ | yes, help is greatly appreciated |
| 11:15.49 | _roger_ | (especially since I can't reproduce it) |
| 11:15.59 | _roger_ | is waiting for a failing board to arrive |
| 11:16.13 | _roger_ | The things I'd recommend... |
| 11:16.37 | _roger_ | edit the board file and start removing pieces from the board startup |
| 11:16.53 | pepere_ | ok |
| 11:17.01 | _roger_ | take the latest oe kernel |
| 11:17.06 | pepere_ | sorry i'm early beginner with oe |
| 11:17.09 | _roger_ | unpack it with :- |
| 11:17.16 | pepere_ | where is baord fiel? |
| 11:17.39 | _roger_ | bitbake -b linux-davinci_git.bb -c compile -f -D |
| 11:17.46 | pepere_ | ok |
| 11:18.01 | _roger_ | that will compile you the latest kernel and put it under work/hawkboard/kernel-* |
| 11:18.18 | pepere_ | perfect |
| 11:19.01 | _roger_ | the board file is under /git/arch/arm/mach-davinci/board-da850-omapl138-hawkboard.c (or similar) |
| 11:19.29 | _roger_ | edit and recompile with bitbake -b linux-davinci_git.bb -c compile -f -D |
| 11:19.46 | pepere_ | ok I get the idea |
| 11:19.49 | _roger_ | then copy the uImage from /git/arch/arm/boot/uImage to the target |
| 11:19.53 | _roger_ | and let me know :) |
| 11:20.04 | pepere_ | ok thx :) |
| 11:21.56 | _roger_ | I would remove everything after davinci_serial_init() and see if it still hangs |
| 11:22.55 | _roger_ | it's possible it was a change in the base da850 support though, but we need to rule out boardfile first |
| 11:28.36 | pepere_ | ok first compilation done |
| 11:28.52 | pepere_ | now I will check baord file |
| 11:30.12 | CHli | Hi _roger_ |
| 11:31.02 | pepere_ | file is angstrom-dev/work/hawkboard-angstrom-linux-gnueabi/linux-davinci-2.6.32+2.6.33-rc4-r4 4+gitr76fcecb83d562608bcebba44774f92b6c62d3593-r44/git/arch/arm/mach-davinci/board-da850-hawk.c |
| 11:36.22 | _roger_ | right |
| 11:36.37 | _roger_ | look for davinci_serial_init |
| 11:36.56 | _roger_ | and #if 0 everything else after it in that function |
| 11:36.59 | pepere_ | yep got it |
| 11:37.04 | pepere_ | ok |
| 11:39.59 | pepere_ | yep it boots |
| 11:40.00 | pepere_ | :) |
| 11:40.02 | eFfeM_work | you might also use git bisect or so to see what commit is the bad guy |
| 11:40.31 | pepere_ | I'm a gitnewbie ;) |
| 11:40.46 | pepere_ | ok now I will find the bad line :) |
| 11:40.50 | pepere_ | let's go |
| 11:43.28 | CHli | go pepere_ go go go ! :) |
| 11:43.28 | _roger_ | i suspect it cpu_freq or _idle... |
| 11:43.28 | eFfeM_work | pepere_: i'm not too good at it either, what you're doing is also a good way (actually I tend to scatter additional printk's_ |
| 11:43.43 | _roger_ | yeah, there's no git history in there anyway |
| 11:43.58 | _roger_ | (since it's a patch) |
| 11:44.13 | _roger_ | there were only a couple of changes though |
| 11:44.21 | pepere_ | ok I refreezed |
| 11:44.23 | _roger_ | RTC/CpuFreq/CpuIdle |
| 11:44.29 | _roger_ | what did you add? |
| 11:44.34 | pepere_ | and i just reenabled cpufreq.. |
| 11:44.38 | _roger_ | bingo |
| 11:44.42 | pepere_ | i will remove it now to check |
| 11:45.08 | _roger_ | Ok, let's comment that baby out... - it's possible something changed in base da850 support that has now broken hawk on some boards |
| 11:45.34 | _roger_ | (there's no pmic on hawk anyway) |
| 11:46.09 | pepere_ | nop it is not cpu freq |
| 11:46.12 | pepere_ | I continue... |
| 11:47.43 | _roger_ | back later - going for lunch - check _PM_ |
| 11:47.50 | pepere_ | ok |
| 11:48.12 | CHli | no PM but we can still vary the freq right ? |
| 11:48.48 | pepere_ | I still need to check four line and we will knwo wich line is the bugy one |
| 11:49.27 | pepere_ | 2 lines |
| 11:50.46 | eFfeM_work | go pepere_, go |
| 11:51.01 | pepere_ | ok seems it is : |
| 11:51.02 | pepere_ | ret = da8xx_register_lcdc(&vga_monitor_pdata); |
| 11:51.34 | eFfeM_work | if you comment that one out, the boot continues ? |
| 11:51.38 | eFfeM_work | file/line ? |
| 11:51.46 | pepere_ | one moement please |
| 11:51.49 | pepere_ | I'm jsut doing that |
| 11:52.11 | CHli | great I don't need VGA :D |
| 11:52.20 | CHli | ;) |
| 11:52.26 | pepere_ | ok that is this one |
| 11:52.29 | pepere_ | file: |
| 11:52.44 | pepere_ | angstrom-dev/work/hawkboard-angstrom-linux-gnueabi/linux-davinci-2.6.32+2.6.33-rc4-r4 4+gitr76fcecb83d562608bcebba44774f92b6c62d3593-r44/git/arch/arm/mach-davinci/board-da850-hawk.c |
| 11:53.31 | CHli | is rebuilding without VGA support :D |
| 11:54.11 | CHli | thinks _roger_ and pepere_ are the heroes of the day :) |
| 11:54.22 | pepere_ | wait oit does not work for now :) |
| 11:54.49 | pepere_ | I'm double checing currently |
| 12:00.26 | pepere_ | ha no it is not vga sorry |
| 12:00.36 | pepere_ | it is related to this line: |
| 12:00.46 | pepere_ | ret = da8xx_register_rtc(); |
| 12:01.15 | pepere_ | so perhaps CONFIG_RTC_DRV_OMAP is not disabled by default now? |
| 12:02.35 | pepere_ | yep i confirm this is related to this line |
| 12:02.46 | pepere_ | it says it should eb disabled by defaukt |
| 12:02.54 | pepere_ | and when i supress it i boot |
| 12:03.06 | pepere_ | so I suspect an issue with config file guys |
| 12:03.57 | pepere_ | comented: // Default hawkboard doesn't have RTC battery - so CONFIG_RTC_DRV_OMAP |
| 12:04.03 | pepere_ | commented: // should be disabled by default |
| 12:08.16 | pepere_ | in angstrom-dev/work/hawkboard-angstrom-linux-gnueabi/linux-davinci-2.6.32+2.6.33-rc4-r44+gitr76fcecb83d562608bcebba44774f92b6c62d3593-r44/git/.config |
| 12:08.35 | pepere_ | I got # CONFIG_RTC_DRV_OMAP is not set |
| 12:08.47 | pepere_ | so perhaps a problem in the .c file? |
| 12:09.05 | pepere_ | the fact is I don't where to find da8xx_register_rtc(); :) |
| 12:09.23 | pepere_ | I will wait until _roger_ come back |
| 12:09.25 | *** join/#hawkboard _roger_ (~a0740758@nat/ti/x-nppgexfrgdbddctu) |
| 12:09.26 | CHli | arch/arm/mach-davinci/devices-da8xx.c, line 584 |
| 12:09.33 | pepere_ | ok |
| 12:09.39 | _roger_ | back |
| 12:09.40 | pepere_ | ho _roger_ |
| 12:09.43 | _roger_ | pc crash |
| 12:09.45 | pepere_ | we've got it |
| 12:09.48 | _roger_ | what's the latest? |
| 12:09.56 | pepere_ | ret = da8xx_register_rtc(); |
| 12:10.03 | pepere_ | is th one taht make me crash |
| 12:10.03 | _roger_ | hmmm, ok |
| 12:10.16 | _roger_ | that should work as long as the OMAP_RTC is not enabled in CONFIG... |
| 12:10.18 | pepere_ | should be disabled for hawkboard it says |
| 12:10.23 | pepere_ | yes... |
| 12:10.23 | _roger_ | yep |
| 12:10.25 | pepere_ | exactly |
| 12:10.37 | pepere_ | and I've just checked .config just in case |
| 12:10.44 | pepere_ | I got # CONFIG_RTC_DRV_OMAP is not set |
| 12:10.47 | _roger_ | hawkboard doesn't have an RTC battery so that module is not powered |
| 12:10.58 | _roger_ | strange |
| 12:11.05 | pepere_ | if I uncomment I got no problems.. |
| 12:11.14 | pepere_ | let's have a look inside the .c function? |
| 12:11.47 | eFfeM_work | no ifdef for config in that one, guess you found it |
| 12:11.59 | _roger_ | in config? |
| 12:12.13 | pepere_ | ?_? |
| 12:12.17 | pepere_ | i've done: |
| 12:12.31 | CHli | yes the CONFIG is taken care of probably |
| 12:12.38 | CHli | *not |
| 12:12.40 | _roger_ | I enabled it in the board file so that users could add a battery and set the CONFIG and have it work |
| 12:12.48 | eFfeM_work | do'tn have the real sources handy but google gave me this: |
| 12:12.49 | eFfeM_work | + +int da8xx_register_rtc(void) +{ + /* Unlock the rtc's registers */ + __raw_writel(0x83e70b13, IO_ADDRESS(DA8XX_RTC_BASE + 0x6c)); + __raw_writel(0x95a4f1e0, IO_ADDRESS(DA8XX_RTC_BASE + 0x70)); + + return platform_device_register(&da8xx_rtc_device); +} |
| 12:13.24 | eFfeM_work | no test for the config var in it, and direct access to regs, so guess that or the caller should have an ifdef for the config |
| 12:13.32 | pepere_ | yes |
| 12:13.34 | _roger_ | interesting |
| 12:13.36 | _roger_ | checks |
| 12:13.56 | eFfeM_work | onlyu 90% sure the code I pasted is actually in hawk, please verify |
| 12:14.11 | pepere_ | i've just verifed |
| 12:14.17 | pepere_ | no check for vars |
| 12:14.46 | eFfeM_work | _roger_: can you make a fix? |
| 12:15.02 | pepere_ | --> fix = // everything :) |
| 12:15.40 | pepere_ | I'm happy to have found something for this bug :) |
| 12:16.53 | _roger_ | yes, will update - need to review a bit first - will do later today |
| 12:16.55 | _roger_ | and push to OE |
| 12:16.57 | eFfeM_work | pepere_: decent fix is ifndef CONFIG_RTC_DRV_OMAP around the clock code |
| 12:17.04 | _roger_ | right |
| 12:17.07 | pepere_ | ha thx :) |
| 12:17.12 | pepere_ | will check right now |
| 12:17.16 | eFfeM_work | _roger_: can test tonight if you want to |
| 12:17.22 | eFfeM_work | _roger_: you have write access ? |
| 12:17.28 | _roger_ | thanks |
| 12:18.26 | pepere_ | what should th efunction return ? |
| 12:18.29 | pepere_ | 0 if ok? |
| 12:18.39 | eFfeM_work | good work guys (and nice to see some activity here, it has been very quiet for a long time) |
| 12:18.39 | pepere_ | because ret is not initialized |
| 12:22.34 | CHli | well I will try also to mount a battery :) |
| 12:24.03 | pepere_ | _roger_ , eFfeM_work : a working patch is: |
| 12:24.06 | pepere_ | #ifdef CONFIG_RTC_DRV_OMAP |
| 12:24.20 | pepere_ | at teh begining of register_rtc function |
| 12:24.33 | pepere_ | #endif at the end |
| 12:24.38 | pepere_ | and ti works for me now |
| 12:25.06 | pepere_ | thanks _roger_ , eFfeM_work for your help |
| 12:25.22 | pepere_ | I go to lunch |
| 12:25.25 | eFfeM_work | np, yw |
| 12:25.29 | eFfeM_work | and enjoy lunch |
| 12:25.40 | pepere_ | (my first kernel hack :D) |
| 12:25.50 | eFfeM_work | gratz ;-) |
| 12:35.22 | *** join/#hawkboard virals (~Viral_Sac@122.166.2.35) |
| 14:03.49 | *** join/#hawkboard cmurillo1 (~cmurillo@186.32.57.22) |
| 14:43.02 | pepere_ | Hey guys anyone used to play with DSP modules? |
| 14:43.16 | pepere_ | I've seen there is DSP_LINK and CMEM modules |
| 14:43.27 | pepere_ | which one will I need to use? |
| 14:43.53 | pepere_ | I mean in which case someone want to use DSP_LINK in place of CMEM? |
| 14:45.03 | pepere_ | Are you used to put linux after the first 16MB of ram? Is 16MB for DSP (code?) sufficient? |
| 14:45.16 | pepere_ | yeah newbies questions in fact, first time i use dsp from TI |
| 14:47.22 | _roger_ | prepere_ - yep, that all works |
| 14:47.29 | _roger_ | best place to start is ti-codec-engine-examples |
| 14:48.17 | _roger_ | on your target, just do opkg install ti-codec-engine-examples, and you'll get the lot in binary form to play with |
| 14:48.21 | CHli | It's the first time I use DSPLINK also but my understanding is that CMEM is a kernel module that will manage a reserved memory space (that's why you have to use the boot arg mem=SOMETHING SMALLER THAN THE MEMORY YOU HAVE to leave some space for CMEM. |
| 14:48.35 | _roger_ | the examples will get installed to /usr/share/ti/ti-codec-engine-examples |
| 14:48.37 | CHli | DSPLINK then uses CMEM to store buffer that will be exchanged with the DSP. |
| 14:48.40 | _roger_ | then run ./loadmodules |
| 14:48.47 | _roger_ | and ./scale as first test |
| 14:49.02 | _roger_ | CHli right |
| 14:49.23 | _roger_ | there's also ti-dsplink-examples as well |
| 14:49.27 | _roger_ | go wild :) |
| 14:49.45 | cmurillo1 | you basically need to reserve memory for the kernel, cmem and dsp |
| 14:50.10 | CHli | :) I think there is a DSP LINK loopback example that's worth looking at also. |
| 14:52.19 | cmurillo1 | pepere_: this is a good start to understand how linux, cmem and dsp memory are placed. http://processors.wiki.ti.com/index.php/Changing_the_DVEVM_memory_map |
| 14:52.37 | cmurillo1 | figure 1 will show you |
| 14:53.27 | CHli | _roger_ do you have an explanation why the fact that the battery is not mounted lead to a crash when configuring the RTC ? Is it when it's trying to access the registers that the CPU stall ? |
| 14:53.40 | pepere_ | Thx all |
| 14:53.48 | pepere_ | will check |
| 14:54.14 | CHli | I'm still confused why it would fail on some boards and not on another. |
| 14:54.22 | _roger_ | right, DSPLINK needs some reserved memory for basic comms between dsp and arm |
| 14:54.42 | _roger_ | for larger buffers we use CMEM to allocate big message buffers |
| 14:55.05 | _roger_ | We also need some space to store the DSP code and data memory, |
| 14:55.05 | pepere_ | ha ok |
| 14:55.06 | _roger_ | so... |
| 14:55.19 | pepere_ | so DSPLINK basic comunication |
| 14:55.24 | pepere_ | CMEM big buffers? |
| 14:55.25 | _roger_ | CMEM + DSPLINKMEM + DSP DATA + DSP CODE needs to be reserved |
| 14:55.30 | pepere_ | ok |
| 14:55.43 | _roger_ | and we reserve using mem = on the cmd line to stop the kernel using that memory |
| 14:56.00 | _roger_ | in our examples, we make the memory map pretty big, so you can get most things working |
| 14:56.13 | pepere_ | I've seen that we can say to linux to sart using @16MB |
| 14:56.15 | _roger_ | in a 'real' system - you'll probably want to reduce the size |
| 14:56.34 | pepere_ | for that i need to change kernel |
| 14:56.35 | _roger_ | yeah, but we don't use that |
| 14:56.39 | pepere_ | ok |
| 14:56.40 | _roger_ | exactly |
| 14:56.43 | pepere_ | so for now |
| 14:56.43 | _roger_ | it's easier to reserve |
| 14:56.47 | pepere_ | i just go simple |
| 14:56.54 | pepere_ | 0-100MB for linux |
| 14:56.59 | pepere_ | other for DSP |
| 14:57.11 | pepere_ | and modify dsp to boot @120MB for exmaple |
| 14:57.20 | _roger_ | if you want to |
| 14:57.27 | _roger_ | but I'd start with what we already have configured |
| 14:57.33 | _roger_ | it works pretty well |
| 14:57.42 | pepere_ | ha ok |
| 14:57.43 | _roger_ | DSPLINK gives you basic comms |
| 14:58.07 | _roger_ | Codec Engine on top, give you ability to run algos easily and hide all the DSP interaction and message passing |
| 14:58.26 | _roger_ | so... depending on your app, you could use base DSPLINK or CE |
| 14:58.55 | cmurillo1 | I think the DSP has 14MB assigned by default, you can change it though |
| 14:59.01 | _roger_ | I'd recommend playing with the ti-dsplink-example and ti-codec-engine-examples packages, to a) make sure you're system is working ok, and b) evaluate which one is best |
| 14:59.15 | pepere_ | ok will do that |
| 14:59.25 | pepere_ | i needed a basic overview of the whoel mess :) |
| 14:59.30 | _roger_ | ;) |
| 14:59.33 | _roger_ | it's not that bad |
| 15:00.58 | pepere_ | my basic need will be use upp from DSP to grab data from FPGA, process data, save in a buffer, linux get teh buffer and send data over the network |
| 15:01.09 | cmurillo1 | _roger_: have you tried changing the memory map? assigning more space for the dsp |
| 15:03.33 | _roger_ | yes, many times |
| 15:04.13 | _roger_ | pepere_, so DSPLINK probably best for you, but take a look at the CE examples as well, since they do simplify things a good deal |
| 15:04.27 | _roger_ | cmurillo1 - you seeing problems with it? |
| 15:06.33 | cmurillo1 | _roger_: yes, I need to integrate new AAC codecs, and assign 2 extra MB to DDR2. So first I decided to use the whole 128MB of ram since I was only using 64MB |
| 15:07.24 | cmurillo1 | I tried just moving the memory map, without increasing the DDR2 size. This caused my aac decoder to stopped working |
| 15:08.15 | cmurillo1 | I tried leaving the memory map where it was and still increasing to 128MB, by passing two mem parameters to the kernel |
| 15:10.02 | cmurillo1 | that allowed me to increase DDR2 size, but my codecs still fail during the codec engine creation |
| 15:11.06 | CHli | away, see you tomorrow |
| 15:11.16 | CHli | nick CHli^afk |
| 15:11.56 | CHli^afk | thanks pepere_ , _roger_ for the bug catching ! :D |
| 15:11.58 | CHli^afk | bye |
| 15:17.36 | _roger_ | cmurillo - do you need more DATA or CODE for AAC? |
| 15:17.51 | _roger_ | Isn't AAC already integrated into our DSP combo anyway? |
| 15:23.25 | *** part/#hawkboard franktango (~a0746747@nat/ti/x-twjpajaonraetbdx) |
| 15:24.23 | cmurillo1 | just the AAC decoder |
| 15:24.33 | cmurillo1 | I need to integrate the AAC encoder |
| 15:25.43 | cmurillo1 | more CODE |
| 15:36.54 | _roger_ | i see |
| 15:37.17 | _roger_ | and is your total required memory > that the current DDR + DDRALGHEAP size? |
| 15:37.31 | _roger_ | i.e. could you steal some from DDRALGHEAP to give to DDR? |
| 15:37.41 | _roger_ | we usually make DDRALGHEAP huge |
| 15:37.48 | _roger_ | (this is done in the .tcf file) |
| 15:37.57 | _roger_ | that way you can keep most other things the same |
| 15:42.08 | cmurillo1 | my required memory < DDR + DDRALGHEAP, so yes, I can probably steal 2 MB out of the 10MB assigned to DDRALGHEAP. |
| 15:42.40 | cmurillo1 | I will give it a try |
| 15:44.05 | cmurillo1 | _roger_: thanks! |
| 15:52.26 | _roger_ | np |
| 15:52.29 | _roger_ | good luck! |
| 15:58.45 | pepere_ | hey guys, i'm trying to bitbake ti-dsplink, as the compiler must taken from ti, i've downloaded it and putted it download dir |
| 15:59.06 | pepere_ | do I need to anythin g lese so it will be taken into accoutn by bitbake? |
| 16:01.45 | pepere_ | h afound help on the net... |
| 16:07.32 | _roger_ | nope |
| 16:07.35 | _roger_ | that should be it |
| 16:08.34 | _roger_ | khasim ping |
| 16:09.19 | cmurillo1 | _roger_: my audio encoder is working now, turns out that my memory map was correct but the issue was the codec parameters |
| 16:11.07 | pepere_ | in fact ti-cgt6x_6.1.9.bb fail because it try to fetch file from unpossible location... but other recipes continue ok |
| 16:11.32 | pepere_ | the problem is I need to relaunch each time since ti-cgt6x is failing each time |
| 16:11.48 | pepere_ | is there a way to tell him; hey do not download, you've ever got it.... |
| 16:13.11 | *** join/#hawkboard eFfeM (~frans@j200125.upc-j.chello.nl) |
| 16:14.34 | eFfeM | hi all |
| 16:14.45 | eFfeM | _roger_: is there already a patch ? |
| 16:22.03 | *** join/#hawkboard wolfspraul (~wolfsprau@c-24-5-80-179.hsd1.ca.comcast.net) |
| 16:28.40 | eFfeM | nevermind found the answer on the mailing list, I'm building it now |
| 16:29.14 | pepere_ | yep: 49a1f4bdb309969f2c40b5dbfefdb56c83c0b264 |
| 16:29.21 | pepere_ | it is pushed |
| 16:32.35 | *** part/#hawkboard khasim (~a0393720@192.163.20.231) |
| 16:32.46 | eFfeM | pepere_: i'm building it already |
| 16:32.55 | pepere_ | ok |
| 16:33.13 | eFfeM | already reviewed it, looks good |
| 16:33.27 | pepere_ | is there a way to say to bitbake: hey atke a look at new package and recompile what is needed |
| 16:33.39 | pepere_ | ? |
| 16:33.50 | eFfeM | bitbake package? |
| 16:33.51 | pepere_ | or did you do: bitbake virtual/kernel? |
| 16:33.53 | eFfeM | in this case I did |
| 16:33.58 | pepere_ | ok |
| 16:33.59 | eFfeM | bitbake virtual/kernel |
| 16:34.03 | pepere_ | thx |
| 16:34.38 | eFfeM | due to vacation I didn't pull in 6 weeks or so, so some other stuff that is needed is rebuild as well (e.g. some libs) |
| 16:34.49 | pepere_ | ok |
| 16:36.18 | _roger_ | <PROTECTED> |
| 16:36.28 | _roger_ | doh - wrong window :) |
| 16:36.41 | _roger_ | (5:11:32 PM) pepere_: the problem is I need to relaunch each time since ti-cgt6x is failing each time |
| 16:36.44 | _roger_ | don't understand this |
| 16:37.00 | _roger_ | what's it failing on? |
| 16:37.05 | _roger_ | do-fetch checksums? |
| 16:37.21 | pepere_ | in fact I asked bitbake to do dspè_link |
| 16:37.36 | pepere_ | but it failed each tiem fetching the ti compiler |
| 16:37.45 | pepere_ | so I asked him to bitbake ti cgt |
| 16:37.50 | pepere_ | and it worked... |
| 16:37.58 | pepere_ | and now I don't have any more issue :) |
| 16:38.03 | _roger_ | wierd - don't understand that |
| 16:38.10 | pepere_ | (I put ti_cgt into donloads dir) |
| 16:38.26 | _roger_ | I'd love to see the log for that fail case... |
| 16:38.27 | pepere_ | and in the meantime i also chmod'ed it |
| 16:38.35 | _roger_ | right |
| 16:38.44 | pepere_ | and also... intall in /opt/ ... |
| 16:38.50 | _roger_ | ah-ha... |
| 16:38.53 | pepere_ | so i don't know what resolved the issue :) |
| 16:38.58 | pepere_ | :/ |
| 16:39.09 | pepere_ | but now it works... |
| 16:39.15 | _roger_ | well - sort of... |
| 16:39.20 | _roger_ | you'll see a problem later I think |
| 16:39.32 | _roger_ | did you update your OE, before this build? |
| 16:39.38 | pepere_ | nop |
| 16:39.51 | pepere_ | I will do it after... |
| 16:39.55 | _roger_ | I pushed a fix a couple of days ago for an installer bug |
| 16:40.00 | pepere_ | yeah I saw |
| 16:40.04 | _roger_ | that fixed the /opt/ issue |
| 16:40.08 | pepere_ | ok :) |
| 16:40.18 | pepere_ | good news ;) |
| 16:40.30 | pepere_ | i will try to pull after he has finished |
| 16:40.58 | _roger_ | cool |
| 16:41.09 | _roger_ | gotta run for today - have fun all! |
| 16:41.18 | pepere_ | bye |
| 16:41.26 | _roger_ | (5:09:19 PM) cmurillo1: _roger_: my audio encoder is working now, turns out that my memory map was correct but the issue was the codec parameters |
| 16:41.28 | _roger_ | cool! |
| 16:41.46 | *** part/#hawkboard _roger_ (~a0740758@nat/ti/x-nppgexfrgdbddctu) |
| 16:54.08 | *** join/#hawkboard ddompe (~ddompe@186.32.57.22) |
| 18:03.26 | *** join/#hawkboard khasim (~a0393720@192.163.20.231) |
| 18:23.44 | pepere_ | EFfeM have you tried last kernel? |
| 18:23.52 | pepere_ | eFfeM |
| 18:24.05 | pepere_ | It is working but I got a lot of: |
| 18:24.28 | pepere_ | end_request: I/O error, dev mtdblock2, sector 0 [ 10.410000] Buffer I/O error on device mtdblock2, logical block 0 |
| 18:24.34 | pepere_ | at boot |
| 18:24.41 | pepere_ | do you have also this? |
| 18:36.12 | eFfeM | pepere_: was afk, haven't checked yet if hte build is completed |
| 18:36.30 | eFfeM | it is, let me see if I can get things started, will keep you posted |
| 18:38.39 | eFfeM | pepere_: what version do you have, I've just build minimal/2.6.32+2.6.33-rc4-r45+gi |
| 18:38.47 | eFfeM | (i've build minimal, not angstrom |
| 18:39.09 | eFfeM | also got mtd errors |
| 18:39.33 | eFfeM | but somewhat different from yours: |
| 18:39.40 | eFfeM | [ 11.800000] end_request: I/O error, dev mtdblock2, sector 0 |
| 18:39.41 | eFfeM | [ 11.800000] Buffer I/O error on device mtdblock2, logical block 0 |
| 18:39.44 | eFfeM | sorry the same |
| 18:39.49 | eFfeM | didn't read well |
| 18:43.09 | pepere_ | yep th e same :) |
| 18:43.24 | pepere_ | will have to check that with mr roger :) |
| 18:43.53 | pepere_ | anyway it work better than before ^_^ |
| 18:50.47 | *** join/#hawkboard ssvb (~ssvb___@a88-112-120-50.elisa-laajakaista.fi) |
| 18:54.59 | eFfeM | pepere_: tested some more, accessing mtd0/1/2/3/4 does not give a problem, mtdblock0/1/2 do mtdblock3/4 don't |
| 18:55.14 | eFfeM | test: dd if=/dev/mtd1 of=/dev/null bs=512 |
| 18:55.47 | pepere_ | I don't undersatnd mtd devixce, it is a moutn of the and flash no? |
| 18:56.09 | pepere_ | so it means a moduel of the kernel taht seek bad sectors is not oaded no? |
| 18:57.12 | pepere_ | loaded |
| 19:00.26 | pepere_ | eFfeM same result there |
| 19:08.32 | eFfeM | pepere_dinner: mtd = memory technology device or soemthing like that |
| 19:09.05 | eFfeM | mtd0 etc are char devices, mtdblock0 etc are block devices |
| 19:09.13 | eFfeM | guess something is wrong with the definition |
| 19:10.23 | eFfeM | here is what is in it (and apolgies for flooding a little bit) |
| 19:10.24 | eFfeM | root@hawkboard:/proc# cat /proc/mtd |
| 19:10.24 | eFfeM | dev: size erasesize name |
| 19:10.24 | eFfeM | mtd0: 00020000 00020000 "u-boot env" |
| 19:10.24 | eFfeM | mtd1: 00020000 00020000 "UBL" |
| 19:10.24 | eFfeM | mtd2: 00080000 00020000 "u-boot" |
| 19:10.25 | eFfeM | mtd3: 00200000 00020000 "kernel" |
| 19:10.25 | eFfeM | mtd4: 07c00000 00020000 "filesystem" |
| 19:18.01 | eFfeM | pepere_dinner: guess I found an answer, not sure if it is *the* answer: see http://markmail.org/message/7b3xay5fbgo63xsr |
| 19:23.04 | eFfeM | reboot, back in a bit |
| 19:28.05 | *** join/#hawkboard eFfeM (~Frans@j200125.upc-j.chello.nl) |
| 19:28.12 | eFfeM | re |
| 19:47.48 | pepere_dinner | eFfeM I'm back |
| 19:49.11 | pepere_ | I've readen somewhere atht we can choose ECC or NONE |
| 19:49.31 | pepere_ | perhaps the answer is around there |
| 19:49.37 | pepere_ | let dig the web :) |
| 19:49.53 | eFfeM | no idea if/how this is, it is an area i did not touch, mtd0 is the u-boot settings, u-boot handles that |
| 19:49.58 | pepere_ | http://www.cnblogs.com/goodloop/archive/2008/11/23/1339621.html |
| 19:50.16 | eFfeM | mtd0 is UBL, this is the secondary boot loader. No idea how that is protected |
| 19:50.25 | pepere_ | seems we can choose or not, I'm wondering if ti is sysctl tunable |
| 19:50.30 | eFfeM | mtd1 = UBL |
| 19:50.33 | eFfeM | mtd2 = uboot |
| 19:50.52 | eFfeM | apparently udev rules can handle it see the link i posted |
| 19:50.53 | pepere_ | perhaps these two do not use ECC for correction check? |
| 19:50.58 | pepere_ | ok |
| 19:51.21 | eFfeM | no idea, ask roger or maybe khasim knows about it |
| 19:51.53 | pepere_ | ok I've understood |
| 19:52.05 | pepere_ | so perhaps modifying an udev rule shoudl do the trick |
| 19:52.14 | pepere_ | let have a look a them... |
| 20:35.41 | *** join/#hawkboard _roger_ (~a0740758@nat/ti/x-owomzjllhzcbazps) |
| 20:39.55 | pepere_ | anyone know where is the dsplink module in oe ipk? |
| 20:40.15 | pepere_ | i've doen ipk install ti_dsplink_modulexxxx.ipk |
| 20:40.33 | pepere_ | and ipkg install ti_dsplink_1_64xxx.ipk |
| 20:40.41 | pepere_ | but modprobe still does not work :/ |
| 20:43.41 | pepere_ | ha seems it's an issue with ipkg... |
| 20:43.51 | pepere_ | he don't want to install the module |
| 20:43.52 | _roger_ | what did you install? |
| 20:44.01 | pepere_ | i'm trying to install; |
| 20:44.06 | _roger_ | opkg install ti-dsplink-apps |
| 20:44.14 | _roger_ | (or examples) |
| 20:44.21 | _roger_ | opkg install ti-dsplink-module |
| 20:44.22 | pepere_ | ipkg or opkg? |
| 20:44.25 | _roger_ | opkg |
| 20:44.29 | pepere_ | !! |
| 20:45.21 | pepere_ | Package ti-dsplink (1:1_64-r45f.5) installed in root is up to date. |
| 20:45.38 | pepere_ | modprobe dsplink , modprobe: Module dsplink not found. |
| 20:45.43 | pepere_ | arf :( |
| 20:46.02 | pepere_ | hum |
| 20:47.04 | pepere_ | ok i've doen opkg remove ti-dsplink |
| 20:47.23 | pepere_ | I think i'm using sadly opkg... |
| 20:47.41 | pepere_ | because opkg install ti-dsplink now says: |
| 20:47.53 | _roger_ | is the module there in /lib/modules/... |
| 20:47.55 | _roger_ | ? |
| 20:47.56 | pepere_ | Collected errors: * opkg_install_cmd: Cannot install package ti-dsplink. |
| 20:48.11 | _roger_ | anything else? |
| 20:48.13 | pepere_ | so I think i misuse opkg |
| 20:48.26 | pepere_ | i've put .ipkg fiel in /ipg |
| 20:48.50 | pepere_ | I have to read opkg help I think |
| 20:49.01 | pepere_ | I think this is a dumb newbie error.... |
| 20:49.25 | _roger_ | you should just be able to copy the ipk from your deploy directly onto the target fs, and do opkg install ti-dsplink-module.....ipk? |
| 20:49.37 | _roger_ | sounds like something else is going wrong |
| 20:49.40 | pepere_ | ok i try |
| 20:50.44 | pepere_ | Multiple packages (ti-dsplink-module and ti-dsplink-module) providing same name marked HOLD or PREFER. Using latest. Multiple packages (ti-dsplink-module and ti-dsplink-module) providing same name marked HOLD or PREFER. Using latest. |
| 20:50.57 | pepere_ | I 've got this error when installing |
| 20:51.02 | pepere_ | can it be the error? |
| 20:51.28 | pepere_ | I've buried my fs ? :) |
| 20:51.48 | pepere_ | Ok let roll back everything to a clean state.... |
| 20:54.22 | _roger_ | hmm - strange |
| 20:54.45 | _roger_ | yep, sounds like you should unpack a clean fs |
| 20:54.54 | pepere_ | hey _roger_ by the way |
| 20:55.02 | pepere_ | th e new kernel boots well |
| 20:55.28 | pepere_ | I and eFfeM get only an error on mtdblkxxx |
| 20:55.35 | pepere_ | when kernel boot |
| 20:55.46 | _roger_ | yeah, going to look at the mtd issue tomorrow |
| 20:56.15 | _roger_ | but, thanks for the info - glad the kernel is working, and thanks for the help to track it down! |
| 21:01.36 | pepere_ | yeah it works |
| 21:01.49 | pepere_ | another big thx for you roger... |
| 21:02.05 | pepere_ | I buried my fs using ipkg instaed of opkg.... |
| 21:02.12 | pepere_ | thx _roger_ :) |
| 21:04.06 | pepere_ | arf no I still have an issue ... |
| 21:04.22 | pepere_ | hum |
| 21:04.30 | pepere_ | let's look if module is present |
| 21:05.29 | pepere_ | ha ok |
| 21:05.38 | pepere_ | modprobe does not found th emodule |
| 21:05.47 | pepere_ | but insmod'ing it by hand works |
| 21:05.48 | _roger_ | try depmod -ae |
| 21:05.54 | _roger_ | then modprobe |
| 21:06.10 | pepere_ | good catch |
| 21:06.17 | _roger_ | hmmm |
| 21:06.30 | _roger_ | looks like we've got a bug there - depmod should happen automatically |
| 21:06.35 | pepere_ | is it an ipkg issue or pepere_ issue :) |
| 21:06.58 | _roger_ | is an OE module post_install problem |
| 21:07.15 | pepere_ | is there a bug tracker anywhere? |
| 21:07.15 | _roger_ | use depmod for now |
| 21:07.19 | pepere_ | yeah sure |
| 21:07.31 | pepere_ | soemwhere we we can track the bugs |
| 21:07.53 | _roger_ | will look |
| 21:08.21 | pepere_ | examples really work now |
| 21:08.23 | pepere_ | thx |
| 21:08.25 | _roger_ | cool |
| 21:08.51 | pepere_ | now i'ev got to master ti toolchain for DSP ;) |
| 21:12.15 | *** join/#hawkboard Raph_ (~5d158b8e@gateway/web/freenode/x-szltzdaajtgmlbfj) |
| 21:13.24 | pepere_ | hi Raph_ ;) |
| 21:14.40 | Raph_ | Hello pepere |
| 21:15.40 | Raph_ | Hi Pepere_ |
| 23:06.16 | cmurillo1 | khasim: ping |