<DidiBsAs>
hglm, I'm back to Ubuntu now, but will remember for next time I try Fedora.
<hglm>
Does Ubuntu run well? I am using a Debian wheezy.
<DidiBsAs>
I think the Ubuntu image I'm using is not as smooth as it can be. Did you do a fresh Debian install, hglm?
<hglm>
Yes, I downloaded a Debian image linked on the wiki, and compiled the latest kernel.
<hglm>
Ubuntu on PCs is not known to be the fastest, they sacrifice performance for compatibility/stability (still, I use it on my PC).
<hglm>
Does anyone know how to enable 16bpp pixel mode? It should be significantly faster.
techn_ has quit [Read error: Connection reset by peer]
<DidiBsAs>
Have you compiledthe kernel on your device or cross-compiled, hglm ?
<hglm>
On my device, takes about 2 hours.
<hglm>
To keep speed/responsiveness up, the key is to reduce file write access to flash memory...so that means few log files and all temporary files in tmpfs (ramdisk) directories
<hglm>
Also the filesystem should use write-back mode (I think the other mode is very slow).
hglm has quit [Quit: leaving]
Yaku has quit []
hglm has joined #linux-sunxi
tinti has joined #linux-sunxi
tinti has quit [Ping timeout: 246 seconds]
hglm has quit [Quit: leaving]
hglm has joined #linux-sunxi
tinti has joined #linux-sunxi
egbert has quit [Disconnected by services]
egbert_ has joined #linux-sunxi
hglm has left #linux-sunxi [#linux-sunxi]
hglm has joined #linux-sunxi
luoyi has joined #linux-sunxi
drachensun_ has joined #linux-sunxi
fredy_ has joined #linux-sunxi
el_ has joined #linux-sunxi
grevaill1t has joined #linux-sunxi
xnxs has joined #linux-sunxi
fredy has quit [Ping timeout: 245 seconds]
el has quit [Ping timeout: 245 seconds]
drachensun has quit [Ping timeout: 245 seconds]
grevaillot has quit [Ping timeout: 245 seconds]
fredy_ is now known as fredy
xenoxaos has quit [Ping timeout: 245 seconds]
xnxs is now known as xenoxaos
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
el_ has quit [Read error: Connection reset by peer]
<hglm>
Anyone know how to enable 16bpp color mode?
tinti has quit [Ping timeout: 252 seconds]
BJfreeman has quit [Quit: had a good time]
hglm has quit [Quit: leaving]
theOzzieRat has quit [Ping timeout: 252 seconds]
theOzzieRat has joined #linux-sunxi
n01 has quit [Remote host closed the connection]
DidiBsAs has quit [Read error: Connection reset by peer]
<oliv3r>
mornin'
n01|away is now known as n01
<n01>
'morning
rellla has joined #linux-sunxi
<n01>
oliv3r: this morning I submitted a couple of patches to buildroot, maybe can be of interest for you
tkoskine has quit [Quit: hop]
hansg has joined #linux-sunxi
Yaku has joined #linux-sunxi
eebrah|away is now known as eebrah
<oliv3r>
ssvb: rellla should verify this buggy video #7 and inform AW since he's talking to them currently anyway
<oliv3r>
hno: nice! i'll pull it immediatly
<oliv3r>
hno: so fel-boot.bin replaces spl-boot.bin in a sense?
<oliv3r>
n01: never used builroot; i really should try it out
focus has quit [Remote host closed the connection]
focus has joined #linux-sunxi
hglm has joined #linux-sunxi
eebrah is now known as eebrah|away
shineworld is now known as shine|meeting
<luoyi>
what's the difference between gcc-arm-linux-gnueabihf-bin and arm-none-linux-gnueabi- ?
<luoyi>
the BSP require the hf version . and u-boot require gnueabi version
<oliv3r>
there is no floats in u-boot
<oliv3r>
so for u-boot, you can use either
<oliv3r>
u-boot should compile just fine with HF
<oliv3r>
i compile u-boot and the kernel with the same compiler
<hglm>
Anyone know whether 16bpp color mode works? It could speed up the system a lot if you don't need photorealistic graphics.
<hglm>
When porting some software I noticed a negative value of a char data type in C doesn't seem to work...is this common to all Cortex CPUs?
<n01>
hglm: weird o_o
jukivili has quit [Quit: "-leaving-"]
<hglm>
Could it be a compiler bug then?
<luoyi>
oliv3r: I've tried to use the guneabi version of gcc to compile the u-boot. and got some linker error
<hno>
oliv3r, it's the same only different packaging.
<oliv3r>
hno: thoguht so, really really awesome though
<hno>
note, just fixed a silly typo in the usb-boot script.. failed to even attempt to load script.bin.
<oliv3r>
i found some other textual typo in the readme, nothing major though
<oliv3r>
luoyi: if you use the sunxi-bsp, it's as easy as make CROSS_COMPILE=arm-none-linux-gnueabi-
<oliv3r>
or whatever your cross compiler is called :)
<oliv3r>
having 2 installed can be quite confusing for both you and the PC :p
<luoyi>
oliv3r: yes. I've try that. and it give me some linker error
<oliv3r>
luoyi: what cross-compilers are you using?
<hno>
luoyi, u-boot builds just fine with an gnuabihf toolchain.
<luoyi>
hno: yes. with ghueabihf, it's OK. and without hf, the link failed.
<oliv3r>
luoyi: i'd recommend to uninstall all unrelevant cross-comppilers
<oliv3r>
so only keep the 'hf'
<hno>
oliv3r, see sunxi-tools/felboot/Makefile if you are interested in how much the same u-boot SPL and fel-boot is.
<n01>
hglm: can you produce the error and paste the problem somewhere?
<luoyi>
not on this computer. I'll try to post it 4 hours later maybe
<hglm>
OK, I'll investigate that. I fixed the problem by using unsigned char instead of signed char.
tinti has joined #linux-sunxi
<Yaku>
was war nochmal der name des bootmanagers fuer a10 und den rPI ?
<hramrach_>
rellla: the [7] video on the rendering chart is possibly incorrectly encoded
<hglm>
It's very easy to reproduce: char c = -1; printf("%d", (int)c); prints out 255.
<hramrach_>
mplayer can play it but recoding to avc1 causes artifacts
<Yaku>
what´s been the name of the bootmanager for a10 as well as the rPI to have a multiboot sd card ?
<hramrach_>
and cedar can then play the avc1 encode as much as mplayer on x86
<hramrach_>
if you play [7] with mplayer using softcoded it complains about missing frames
<hglm>
I just read man gcc, it seems char can be unsigned or signed, you hav to use "signed char" to get guaranteed signed char.
<hglm>
I am guessing this feature is a major reason why some apps break in armhf.
tinti has quit [Remote host closed the connection]
<hglm>
Compiling with -fsigned-char may solve some problems.
<oliv3r>
hno: i did look at felboot/main.c
<oliv3r>
Yaku: i know rPi has it, not sure if the A10 has it too
<oliv3r>
but it's kinda fake multiboot I suppose, does work though
<oliv3r>
Yaku: but I hink what you are looking for is 'berryboot'
<oliv3r>
hglm: the kernel (and its drivers) should use u8 and s8 for that reason
<hglm>
I guess so, yes. I think there may be apps too that are not in Debian etc. armhf because they need to be compiled with -fsigned-char.
jukivili has joined #linux-sunxi
<hglm>
It's a bug in gcc that it doesn't give a warning when a negative value is assigned to a char on machine where char is unsigned by default.
<hglm>
Has anyone tried installing on internal NAND of an A10 tablet instead of SD-card? Is it safe or faster?
<hno>
oliv3r, what?
<hno>
I wrote it.
<hno>
ah, n license blurb. fixing that.
<hglm>
I just did a test and it seems A10 sdcard speed is at least as good as internal NAND. I did get one "too much ecc err" warning when reading internal NAND.
<rellla>
hramrach_: does it play correctly in native android?
egbert_ is now known as egbert
Undertasker has joined #linux-sunxi
<hglm>
I noticed there's a lot of reserved memory after you boot...(VE, G2D, LCD) -- is there any way to make this available to the OS when you don't need video acceleration for example?
<oliv3r>
hglm: yes, read hansg's README for his fedora 18 image; i THINK it might also be documented on the wiki
<oliv3r>
but the readme is probably the fastest/easist way
<hglm>
Thanks, 80MB for VE (video acceleration?) is a bit much.
<hansg>
hglm, Add something like this to the kernel boot cmdline: sunxi_g2d_mem_reserve=0 sunxi_ve_mem_reserve=0 sunxi_fb_mem_reserve=16 sunxi_no_mali_mem_reserve
<hansg>
hglm, assuming that like me you're only using the framebuffer and not any of the other gfx / video blocks
<hglm>
OK, thanks, I think the X server uses only fb by default (although I use the console mostly).
<hansg>
hglm, right, unless you've installed the mali blob
<hglm>
You mean compiled and installed sunxi-mali for X?
<hansg>
yes
<hansg>
If you've done that you will want to drop the "sunxi_fb_mem_reserve=16 sunxi_no_mali_mem_reserve" parts
<hansg>
(and loose 80 MB)
<hglm>
And w
<hglm>
what about mali for framebuffer?
<hansg>
same
<hglm>
OK
eebrah|away is now known as eebrah
<hansg>
Although there was a patch set in the works to get rid of the static reservation for mali, once that is in you would loose only 16MB (fb needs to be twice as large for double buffering)
<hglm>
I noticed double buffering isn't really working (or hard to get working) in OpenGL ES2 anyway.
<hansg>
I've never bothered to look into the details, but the mali stuff is doing buffer swapping, so it needs twice the mem
<hramrach_>
rellla: I don't have nativa android anymore. Reportedly the original bb2.mp4 plays as slideshow there as it does on Linux/CedarX. The recoded one is slightly broken by the recoding but plays smoothly. I guess you can write off this one as incorect until more samples of stuff encoded with this codec are collected.
<hglm>
The extra fb memory is handy if you want to change modes to a higher resolution after boot.
<rellla>
hramrach_, so i'll remove it from the list as the file encoding is broken.
<rellla>
i think, there are more files uploaded to this server which are broken itself
<hansg>
hglm, with the 16M I use you can go upto 1920x1200, and more then that the hdmi out cannot handle. Actually I've had to add code to kick the hdmi encoder real hard to make it do 1920x1200 in the first place
<hglm>
OK, by default the fb is too small if your disp init mode is small. That option is useful.
<hglm>
I wrote a little utility to switch modes using the new disp driver (switch LCD/HDMI and between HDMI modes).
<hglm>
16M should be just enough for double buffering on 1920x1080.
vinifm has joined #linux-sunxi
<hglm>
I have a 1920x1080 monitor but using a lower resolution like 1280x720 (scaled by the monitor) speeds up all aspects of the A10 by increasing memory bandwidth.
hglm has quit [Quit: leaving]
<shine|meeting>
hansg, same for my 800x480... :)
<shine|meeting>
sorry hglm I mean
<oliv3r>
mnemoc: timezone setting might be wrong on dl.linux-sunxi.org; i just uploaded soemthing and it's off by 2 hours
<oliv3r>
also it looks pretty now :)
hglm has joined #linux-sunxi
<oliv3r>
mripard: Turl yay! i've booted. One small step for linux-sunxi, one giant leap, for oliver :D
<oliv3r>
Linux cubieboard 3.10.0-rc1 #1 Thu May 16 10:25:21 CEST 2013 armv7l GNU/Linux
<hglm>
I have 950MB free now, that's a nice boots (was 835MB). It would really make a differnence on a 512MB device.
<slapin_nb>
oliv3r: vanilla?
<slapin_nb>
hglm: disabled mali?
<hglm>
No, Mali is still enabled, VE is disabled and G2D is disabled.
<slapin_nb>
hglm: I don't see a point disabling G2D, disabling mali is good, or disable all this things on headless device
<slapin_nb>
hglm: G2D is good for watching videos it seems...
<hglm>
I thought G2D is 2D acceleration (which is not enabled by default in X and doesn't work too well), VE is video acceleration (CedarX), I think.
<slapin_nb>
was anybody able to record and use cedarx encoder for video from camera?
<slapin_nb>
hglm: as soon as G2D is well supported, it will be mandatory, I think
<hglm>
I wouldn't mind fast accelerated 2D graphics in X.
<Turl>
oliv3r: what was it?
<oliv3r>
Turl: many things :p first, cygwin + dd == nogo
<oliv3r>
earlyprintk isn't enabled in the kernel by default :p
<oliv3r>
slapin_nb: yep, vanilla 3.10; well mripards sunxi-next branch
<Turl>
:P you build linux on cygwin?
<oliv3r>
oh hell no
<oliv3r>
but i'm at work
<oliv3r>
so my resources are limited :p
<slapin_nb>
oliv3r, mripard: thanks for good work! will check that
<oliv3r>
and the cubieboard's stock android didn't have console login nor terminal.apk
<oliv3r>
i'll write some inital stuff up on the wiki so people can test
<oliv3r>
Turl: so I couldn't do dd from within android on the cubieboard so had to use a work PC to dd
<oliv3r>
anyway, my driver isn't loading :p
<Turl>
oliv3r: do you have it posted/commited somewhere?
<oliv3r>
does dev_info always output on the console? or do I need to enable some debugging something
<oliv3r>
i'll push it to my github, i posted some older versions to the ML
<Turl>
oliv3r: I don't see any entry on the Makefile to compile your driver when the config option is enabled
<Turl>
oliv3r: something like this should do obj-$(CONFIG_EEPROM_SUNXI_SID) += sunxi_sid.o
<oliv3r>
did I forget to add that
<oliv3r>
that would be a big 'doh'
<oliv3r>
i messed up a rebase/merge so i did a format-patch, re-checkout and output
<mripard>
oliv3r: about your emac patches, you should use format-patch -M when sending renaming patches like this
<mripard>
it's much easier to process :)
<oliv3r>
what does -M do?
<oliv3r>
that makes it easier to process
<mripard>
git help format-patch ? :)
<oliv3r>
yeah i'm googling as we speak :p
<Turl>
oliv3r: detects renames or something like that, and doesn't print all the +++---
<mripard>
it just says that one file as been renamed
<mripard>
and only shows the difference inside the file
<mripard>
so that you don't have all the noise about deleting/adding this new file
<oliv3r>
ahhh
<Turl>
the only drawback is that patch doesn't like them
<oliv3r>
yeah i find that horribly annoying
<mripard>
just that it has been renamed, and maybe the differences if there's some
<oliv3r>
I even made the 'rename' a seperate patch, as I thought it tried to avoid that
<oliv3r>
guess not
<oliv3r>
but the actual file rename is a seperate patch, so nothing is modified in the file
<oliv3r>
btw, format-patch says for the -M: -M[<n>]
hno has quit [Ping timeout: 256 seconds]
<mripard>
yeah, it has an optional argument to set the "fuzzyness" of the renaming detection
<oliv3r>
ah ok
<oliv3r>
well so -M breaks patch, but makes the patch easier to read
<oliv3r>
I think my solution isn't so bad
<oliv3r>
the actual file rename in a seperate patch :D
<mripard>
well, it breaks the "patch" binary
<mripard>
but if you use git apply/am, it works fine
<oliv3r>
but then people who prefer to use 'patch' or want to test a patch without applying it to their git complain I shouldn't use -M :(
<Turl>
mripard: btw does -M work on send-email?
<mripard>
Turl: I *think* the options send-email doesn't know about are given to format-patch
hno has joined #linux-sunxi
<n01>
I always use format-patch -M + send-email
<mripard>
so to me, it works, but I definitely not sure about it
<n01>
oliv3r: what do you mean the it breaks patches?
<mripard>
n01: patch can't apply them
<n01>
well you are supposed to use git to apply the patches I think
<n01>
"The output of this command is convenient for e-mail submission or for use with git am"
<Turl>
I don't use format-patch because it doesn't do the cool diffstats for me and the cover letter is not precompleted either
<mripard>
n01: which was exactly what I was saying :)
<oliv3r>
n01: yeah, but what If I post a patch to a mailing list, and someone wants to 'test' it, but hasn't cloned the repo, but just used the 'download zip file' (think windows users god forbid or otherwise)
<mripard>
Turl: it is
<mripard>
git format-patch --cover
<n01>
yep
<mripard>
but anyway, send-email uses format-patch as a backend to generate the patch
<mripard>
so any option send-email doesn't know about is "forwarded" to its internal format-patch call
<Turl>
I always use send-email --amend --cover-letter
<oliv3r>
what does --ammend do?
<Turl>
opens each patch on the editor
<oliv3r>
Ah, probably a good idea to review the message
<Turl>
so you can add comments over the diffstat or otherwise review it
<oliv3r>
I get use the 'confirmation' thing to check the message, but can't really change it :p and if you abort half way through, half of the mails have gone out
<oliv3r>
well the coverletter + diffstats you can edit anyway
<Turl>
I also have a email alias file so I can do stuff like --cc linux-arm
vicenteH has quit [Ping timeout: 252 seconds]
<hglm>
Has anyone tried any edgy kernel optimizations like 2G/2G split/no highmem, or Thumb2?
<Turl>
I run 2G/2G split at times
<oliv3r>
mripard: our a10's do thumb 2 do they not? why are the thumbEE and thumb-2 options disabled in the default multi-config?
<oliv3r>
jinx! sorta
<Turl>
with no highmem that is
<oliv3r>
CONFIG_THUMB2_KERNEL for example I mean
<Turl>
last I tried one of those the kernel would not build
<Turl>
so I dunno :)
<hglm>
Turl: Is it stable? I ran 2G2G with nohighmem and it was about 5% faster in some memory-intensive benchmarks, but I had some kernel oops that might be related.
<oliv3r>
Turl: once i can test this somehow, i'll try to bulid it
<Turl>
hglm: I didn't see anything weird happening
<hglm>
I might try it again sometime.
<hglm>
Lowering framebuffer resolution has a bigger effect on performance though.
vicenteH has joined #linux-sunxi
<Turl>
well, I bet using half of the memory bw helps :)
<oliv3r>
repushed the fixed repo :p
<oliv3r>
now it's time to go fix bugs
<Turl>
hglm: you could use a lower color depth too
<oliv3r>
cause, sunxi-sid 1c23800.eeprom: Unable to create sysfs bin entry
<oliv3r>
is not good :(
<oliv3r>
not sure if I know how to fix it either
<Turl>
oliv3r: at least it's loading now ;)
<hglm>
Turl: I have tried, but I don't know how, kernel command line options for 16bpp for format and seq didn't seem to work.
<Turl>
hglm: you're using X right?
<Turl>
there's an option on xorg.conf to set it up
<oliv3r>
yep, absolutly no clue why it's not working :(
<hglm>
Turl: So 16bpp is supported in the X but not console fb?
<Turl>
hglm: ask ssvb to be sure
<hglm>
OK, if X/framebuffer run in different depth that could cause problems when switching VTs.
<Turl>
oliv3r: maybe 'key' is too generic? really long shot guessing here
<oliv3r>
Turl: i can try
<oliv3r>
i'll test with something very specific
<Turl>
err, facepalm
<Turl>
oliv3r: forget that :p
<Turl>
oliv3r: you have the error checking inverted
<Turl>
so it's actually working right
<oliv3r>
Turl: tumb2 don't work
<oliv3r>
linux-sunxi/arch/arm/boot/compressed/head.S:485: Error: selected processor does not support requested special purpose register -- `mrs r0,spsr'
<oliv3r>
Turl: i do?
<oliv3r>
it returns 0 on success?
<oliv3r>
i guess so :p
<oliv3r>
stupid me :(
<oliv3r>
anyway, with gcc 4.6.3 tumb2 doesn't build
<oliv3r>
i really thought a10 had thumb2; maybe a20
<mripard>
it does.
<oliv3r>
then i'll enable only 1 of the 2 thumb options :p
<oliv3r>
horrible location, but! i get data, not what I had expected however :(
<oliv3r>
but data never the less
<oliv3r>
reboot segfaults though
<oliv3r>
gotta check if poweroff oopses too
mdp_ has joined #linux-sunxi
mdp has quit [Read error: Connection reset by peer]
mdp_ is now known as mdp
<hglm>
ssvb: It didn't quite work, display was messed up (half screen), I think it's because SCALER mode wasn't disabled (it should be disabled in 16bpp).
shine|meeting has quit [Quit: Leaving]
<ssvb>
hglm: yes, with scaler mode enabled, 16bp is not supported, so fbcon thinks that it uses 16bpp but the scanout is still done as 32bpp
ganbold_ has joined #linux-sunxi
<ssvb>
hglm: do you really need scaler mode? it wastes one hardware scaler which could have a better use
<ssvb>
hglm: a10 has two scalers, but a13 has only one
<hglm>
ssvb: I had the impression that scaler mode helps with the stability of higher resolutions like 1920x1080, not sure though.
<ssvb>
hglm: and if you enable scaler mode on a13, you can say goodbye to the hardware scaled video playback
<oliv3r>
how likly is it, that if I use md 0x12345 10, i get complete 32 bit words, but when I try to access that same data usin md.b 0x12345 40, I get only partial bytes, i byteordering is different (not important) but i only get the last 8 bits, then 4x 0x00, then the next 8 last bits of the 2nd word
<oliv3r>
hno: is this 'expected' behavior? or why do you think that is?
<oliv3r>
or am I simply only allowed to read those registers in 'word' mode
<ssvb>
hglm: yes, there is some issue apparently related to Mali GPU starving the framebuffer scanout for high resolutions, so that you can observe rolling waves on screen
<ssvb>
hglm: but if you are interested in 16bpp, then it should not be affected
<hglm>
ok
<hglm>
It seems to work now! I disabled scaler mode.
<ssvb>
hglm: btw, scaler mode does not fully solve the problem, I guess it just introduces a larger intermediate buffer between reading the data from the framebuffer and sending it over hdmi, so underruns are less likely (but still can happen)
<ssvb>
hglm: good :)
<ssvb>
oliv3r: are you really in such a high need for saving every last bit of RAM that you want thumb2 in the kernel? :)
<oliv3r>
ssvb: no, but if it works, and it makes certain things faster, why not enable it :D
<oliv3r>
to be fair, my tablet has only 512 mb ram, so yeah, it could be usefull
<oliv3r>
mripard: btw, Jazelle has been replaced by ThumbEE, which is actually supposed to be 'active'
<ssvb>
oliv3r: thumb2 is mostly a code size reduction thing, it generally makes performance a bit worse (at least with the older versions of gcc)
<oliv3r>
ah, well it's broken currently :(
<ssvb>
oliv3r: the heavyweight bloatware monsters such as firefox actually benefit from it though, because of less I-cache misses
<oliv3r>
well, i do have firefox on my tablet :) i stoped using it 4 versions ago cause it kept crashing/was slow
<oliv3r>
most probably due to oom; 512 is really low
<oliv3r>
so generally speaking thumb2 could be a good thing :p
<ssvb>
oliv3r: yes, if the compilers do a good job
<oliv3r>
4.6.3 to old?
<ssvb>
oliv3r: thumb2 is a variable length (16-bit or 32-bit) encoding alternative to the traditional 32-bit arm mode with almost complete instructions coverage (a few deprecated instructions got dropped)
<ssvb>
oliv3r: in theory just assembling all the same instructions as thumb2 instead of arm should reduce the size and keep performance the same
<oliv3r>
does armv8 still support it? or is it more like amd64, where you have 'x32' mode now
<ssvb>
oliv3r: but the "problem" is that the compilers tend to favor size and prefer smaller instructions, trading off the performance
<oliv3r>
memory usage vs execution speed
Skaag has joined #linux-sunxi
<oliv3r>
the eternal dilema
<oliv3r>
only a problem when you lack the memory :)
<Skaag>
how do I switch from stage/sunxi-3.0 to 3.4?
<oliv3r>
git checkout stage/sunxi-3.4
<ssvb>
oliv3r: also it seems fun, but thumb2 has higher power consumption on cortex-a8 at least for cpuburn application :)
<oliv3r>
and welcome Skaag
<Skaag>
I have a problem with the X files no longer being generated during my kernel build
<Skaag>
thanks
<oliv3r>
ssvb: hmm, that of course is a downside again
<Skaag>
they used to be produced under rootfs/lib/
<oliv3r>
Skaag: make clean; ./configure; make
<Skaag>
now when I start Xorg I get: Fatal server error: no screens found
<oliv3r>
should reset/restore/rebuild pretty much everything
<Skaag>
that's what I'm trying to figure out... how did I break it? :)
<Turl>
ugh, journalctl is a cpu hog
<Skaag>
I cleaned the whole thing, reconfigured, and suddenly those libraries were no longer generated
<oliv3r>
ssvb: what gcc version did you use to compile? mine barfs atm
<oliv3r>
Skaag: did you git pull?
<oliv3r>
ssvb: is that a 'arm-version-rewrite' of the old cpuburnK6?
<Skaag>
yes I git pulled to the latest 3.0
<Skaag>
maybe that's what broke it
<oliv3r>
well you can change the kernel to 3.4, but that won't generate all the libs i think you miss
<ssvb>
oliv3r: not quite, it tries to achieve the same goals, but relies on cortex-a8 microarchitecture features
<ssvb>
oliv3r: what does (arm) gcc say?
<oliv3r>
does it also test memory bw?
<Skaag>
is there maybe something I need to enable in make linux-config that would get it to build the X libraries?
<oliv3r>
linux-sunxi/arch/arm/boot/compressed/head.S:485: Error: selected processor does not support requested special purpose register -- `mrsr0,spsr'
<oliv3r>
Skaag: no idea, but make linux-config configures the kernel :)
<ssvb>
Skaag: do you have /var/log/Xorg.0.log ?
<Skaag>
yes, I see now that fbdev is not found
<Skaag>
will try to figure out where to configure it to build it
<Skaag>
(I mean that's what Xorg.0.log says)
<ssvb>
Skaag: I'm sure it says a lot, do you mean it can't find /dev/fb0 ?
<ssvb>
Skaag: have you copied the compiled modules to /lib/modules ?
hglm has quit [Ping timeout: 252 seconds]
<ssvb>
Skaag: it might be better to just use a pre-compiled hwpack if you are not very confident compiling the kernel yourself
<ssvb>
Skaag: and the recent kernels actually should have CONFIG_FB_SUNXI=y by default
<Skaag>
I see
<Skaag>
where do I find this in the menu?
<ssvb>
press '/' to start a search in the menuconfig, then enter FB_SUNXI and see what it shows
<ssvb>
but as I said, if you "git pulled to the latest 3.0", then you should CONFIG_FB_SUNXI=y set by default
<ssvb>
*should have
<Skaag>
I think I found it
<vinifm>
cat .config | grep FB_SUNXI
vinifm has quit [Quit: Saindo]
shineworld has joined #linux-sunxi
hglm has joined #linux-sunxi
<Skaag>
it was on =m
<Skaag>
I found where to change it to =y and am recompiling
<Skaag>
see, it used to be that it generated files in specific locations, now I can no longer find them there:
<Skaag>
cp: cannot stat ‘build/a13_olinuxino_hwpack/rootfs/lib/x11/*’: No such file or directory
<Skaag>
cp: cannot stat ‘build/a13_olinuxino_hwpack/rootfs/lib/framebuffer/*’: No such file or directory
<Skaag>
but let's see if it works now
paulk-desktop has joined #linux-sunxi
<hno>
oliv3r, The I/O registers only support full word access (32 bit). If you try accessing them using another size then zeroes is returned on any unaligned address.
<hno>
that's why we have fel-pio in sunxi-tools for accessing the PIO registers over FEL. (FEL only does byte access)
<Skaag>
now it no longer boots at all
<Skaag>
:)
<hno>
Skaag, progress?
<Skaag>
no :(
shineworld has quit [Quit: Leaving]
<Skaag>
I'm going to clean everything and restart
<Skaag>
yah, linux-config seems to have better defaults now
<Skaag>
building
luoyi_ has joined #linux-sunxi
hansg has quit [Remote host closed the connection]
<techn_>
fel-boot-$board.bin generation must be automated :/
<Skaag>
weird, it works now after a reboot
<Skaag>
the difference is that I disconnected this USB hub that I had hooked up to the board
<Skaag>
which has a printer on it, a usb camera, and a serial->usb converter
<Turl>
techn_: you can boot without initramfs
<Turl>
assuming you pass a suitable root= on the cmdline
<Skaag>
will try to reboot again with the hub, now that I have console, and see what's bothering the boot process when the hub is hooked up
<Skaag>
waddayaknow, it all just works now
* Skaag
hates when things just work without explanation
<mnemoc>
it's called "magic" :)
eebrah|away is now known as eebrah
<Skaag>
I would much rather know why it suddenly started working
<Skaag>
I mean I did clean and rebuild, but with the same image, it didn't boot once, then it suddenly booted on the second attempt.
<Skaag>
I'm just wondering why it would not boot on the first attempt
<Skaag>
maybe it was fsck'ing the FS in that first boot, I just wasn't on the console on that boot...
<Skaag>
I guess I may never know.
<hglm>
Just wondering, is it safe to mount internal NAND partitions on a tablet, or even write to them? (Linux is on the sdcard)
<rz2k>
techn_: mega idea is integrate fel-boot and flash mtd partitions
<rz2k>
skipping whole livesuit generation and etc
<mnemoc>
hopefully in a way compatible with existing flash-over-usb tools :p
<mnemoc>
.oO(what's the name of android's thing?)o
<oliv3r>
hno: so how do I know which registers only do 32bit access? And why does fel mode does byte access only? I tested with u-boot, thought that came reasonably close to fel mode
<Turl>
mnemoc: fastboot?
<mnemoc>
Turl: thanks :)
<Turl>
bbl :)
<oliv3r>
can I make a git 'worktree' in a .git/config (or .git/modules/repo/config) made relative? now i have the full absolute path, and over nfs that kinda fails
<mnemoc>
i think you can make .git a symlink
<mnemoc>
but over nfs... meh
<oliv3r>
well, right now my repo on my server is at /silo/build/sunxi-bsp; and that's set in the 'worktree' var
<oliv3r>
but on my desktop, it's in /home/build/sunxi-bsp
<oliv3r>
so some commands fail as it tries to go to /home
<mnemoc>
sudo ln -s /home /silo
<oliv3r>
lol
<oliv3r>
yes, that would work :)
<oliv3r>
but not really flexible
<mnemoc>
flexibility and freedom are overrated
<techn_>
hmm
<techn_>
it requires u-boot from sunxi-current branch?
<techn_>
mnemoc: hno: why we have sunxi as master in u-boot?
<mnemoc>
sunxi is the boring "stable" branch
<mnemoc>
good default imo
<oliv3r>
hno was saying he wanted to push sunxi-current to sunxi
<mnemoc>
ah, sorry. missed that part
<techn_>
but it lacks a lot of features.. and is really old
<oliv3r>
techn_: it will get pushed soon
<oliv3r>
or atleast that's what hno said :D
shineworld has joined #linux-sunxi
<ssvb>
hramrach_: a guy with blue xbmc videos in #cubieboard, it's time to interrogate him :)
<oliv3r>
blue (corrupted) vids?
<oliv3r>
hno: is it because fel-pio copies all registers to sram as cache and reads from there?
n01 has joined #linux-sunxi
<hno>
oliv3r, yes that is why fel-pio copies to/from sram. sram can be read at byte level.
<oliv3r>
are all registers limited to 32bit access?
<hno>
Not sure. There is very many different I/O modules.
<hno>
on different busses.
<oliv3r>
but generally speaking?
<oliv3r>
if the user manual defines it more or less as a 32bit variable
<hno>
All the normal ones we have been dealing with is 32-bit access only.
<oliv3r>
ok
<oliv3r>
any example of 8 bit register?
<oliv3r>
(both read and write I assume)
<hno>
I don't know.
<oliv3r>
ok no prob
<oliv3r>
atleast that explains why my registers where wrong :)
<oliv3r>
i'll fix my code
<hno>
There several 8-bit I/O modules, but those are 32-bit padded so each regiser is on a 32-bit register boundary.
<hno>
And I do not exclude that there is I/O modules that can be addressed at byte boundaries. But don't know,
<oliv3r>
yeah, but I think it's safe to assume nearly all would be 32-bit, with exceptions (that we don't know yet)
LoCoZeNoz_ZUE has joined #linux-sunxi
<LoCoZeNoz_ZUE>
Hi. Whats the diff betwen inet97f-ii and inet97f-ii-android?
sanka has joined #linux-sunxi
<oliv3r>
defualt config
<oliv3r>
default*
<LoCoZeNoz_ZUE>
?
<LoCoZeNoz_ZUE>
didnt understand
<hno>
oliv3r, yes, 32-bit is always safe.
<oliv3r>
LoCoZeNoz_ZUE: there's 2 kernel config files, 1 for android, 1 for non-android usage
<oliv3r>
LoCoZeNoz_ZUE: think video drivers etc
<LoCoZeNoz_ZUE>
I am trying to out gentoo there. Should i use non-android one?
<LoCoZeNoz_ZUE>
s/out/put
<LoCoZeNoz_ZUE>
yes?
* hno
instead thinks of free mind vs googles nih mind.. but that's me.
<hno>
yes
<hno>
both are andoid kernels btw. But differ in drivers because of binary blobs syndrome.
<hglm>
Loco, gentoo is not ideal for smaller/slower systems -- compiling everything takes a long time. It's a nice concept though, I've used it on a PC.
<LoCoZeNoz_ZUE>
Do you think i cant crosscompile?
<hno>
yes
<hno>
at least some of it.
<oliv3r>
I love gentoo on my desktop :)
<oliv3r>
crosscompiling the kernel will be easy
<oliv3r>
installing stage3 armv7 shouldn't be a problem either
<oliv3r>
nativly compiling everything you use, is just a matter of patience (and fixing bugs if encountered)
<LoCoZeNoz_ZUE>
If i will succed at cross compiling, it will be good. If not, i can wait a day or 2 for it to finish compiling :3
<oliv3r>
crosscompiling everything is possible, but much much harder
<oliv3r>
LoCoZeNoz_ZUE: anyway, use the non-android config
<LoCoZeNoz_ZUE>
already sir
<oliv3r>
or, use make linux-config CROSS_COMPILE=yourcompiler
<oliv3r>
he can follow that, but will need the bsp (or something) to compile kernel + bootloader
<LoCoZeNoz_ZUE>
currently i am semi-blindly follow the "FirstSteps" at you wiki
<LoCoZeNoz_ZUE>
and compiling another gentoo system for my laptop. But thats another story
<hno>
LoCoZeNoz_ZUE, you'll soon be a gentoo master.
<LoCoZeNoz_ZUE>
sure :3
<LoCoZeNoz_ZUE>
btw, i lost my micro sd adapter. So i think il need to stop my proccess when it will come to SD cards and search under tables for this small piece of plastic and... what ever metal they use there
<hglm>
lol, I had the same problem a few days ago, those microSD slots have a lot of recoil, I found it fortunately.
<shineworld>
I've a A13 powered tablet on hands and ANTUTU say 1200Mhz for frequency... so can A13 work at that speed ?
<LoCoZeNoz_ZUE>
Anyway to use "android" for those SD card manipulations?
<oliv3r>
what manipulations
<hglm>
A10 tablets were advertised as 1.2 GHz when they were running at 1.0 Ghz, is Antutu really measuring the speed or just reporting a spec string?
<oliv3r>
shineworld: not reliably, i think max is 1102 MHz
<shineworld>
so antutu detect a wrong value
<LoCoZeNoz_ZUE>
I lost a MicroSD card. But still got android on my tablet. Is it possible to do all the "SD card related" stuff via android?
<oliv3r>
IF your card is IN your tablet, yes
<hno>
shineworld, all marketing uses theoretical max values, rounded up to closest nice figure.
<hno>
LoCoZeNoz_ZUE, what do you mean?
<LoCoZeNoz_ZUE>
oliv3r: Yeah, my "MICROSD" is in tablet. But i lost the thingy to put microsd to PC
<shineworld>
ah ok, just to compare it with cubieboard
<LoCoZeNoz_ZUE>
hno: I got no MicroSD adapter. So i cant do stuff dirrectly from computer
<oliv3r>
LoCoZeNoz_ZUE: no problem really, you can connect USB to pc, 'share' sdcard with pc, copy files to sd from PC; then do the dd-ing etc from android terminal
<LoCoZeNoz_ZUE>
:3
<hno>
or push contend to the card via adb.
<shineworld>
dram_clk = 432
<LoCoZeNoz_ZUE>
Sure, but i thinked there is some "Hardcore Partitioning" to be done
<oliv3r>
LoCoZeNoz_ZUE: you can do all that from android too :p
<LoCoZeNoz_ZUE>
oliv3r: There are not alot of "Partitioning" tools on android, even with busybox
<LoCoZeNoz_ZUE>
but i think it got dd
tinti has quit [Ping timeout: 245 seconds]
<oliv3r>
you could dd the partition table, if you make it somewhere, but busybox doesn't have fdisk?
<oliv3r>
remember, there's bb and busybox usually
<LoCoZeNoz_ZUE>
i am not sure about this point
<oliv3r>
bb is small with bare stuff; busybox is usually more features
<LoCoZeNoz_ZUE>
aaaanddd..... i am at "make hwpack-install" point
<LoCoZeNoz_ZUE>
What should i show as "SD_CARD"?
<LoCoZeNoz_ZUE>
:3
tinti has joined #linux-sunxi
<hno>
LoCoZeNoz_ZUE, you can export the raw SD as a storage device over USB. Just not via the android GUI.
<LoCoZeNoz_ZUE>
?
<hno>
Make the Android tablet act as if it was an sd card reader.
<LoCoZeNoz_ZUE>
erm..... i dont know how. I tried to figure out before, but it didnt allow "low-lvl access" to it
<LoCoZeNoz_ZUE>
nvm, i better find my card reader
rz2k has quit []
<hglm>
What's the best root file system for an sdcard? ext3? ext4 with write-back mode? Disable journalling? Any experimental flash filesystems?
<specing>
hglm: F2FS
<hglm>
Thanks, I'll look that up.
<oliv3r>
LoCoZeNoz_ZUE: you don't need to 'write' the hwpack thing; copy the files from build/$board_hwpack/bootloader to your somewhere on the tablet; then DD them
<LoCoZeNoz_ZUE>
just dd?
<LoCoZeNoz_ZUE>
btw, android says my sdcard is "broken" and needs reformating. So i better deal with this thingy first
tinti has quit [Remote host closed the connection]
<hno>
LoCoZeNoz_ZUE, from android root shell, echo /dev/block/mmcblk0 > /sys/devices/platform/sw_usb_udc/gadget/lun0/file
<hno>
but you may want to pull first.. did some cleanups
<hno>
(other stuff)
shineworld has quit [Quit: Leaving]
<hno>
but still conflicting with your changes.
<hno>
at patch level, not function.
<techn_>
I'll continue this tomorrow.. started to be interesting :)
<hno>
ok.
<hno>
now if I only could remember which Laurent helped doing the original A13 prototype version... util_printf.c is from there and lacks license information.
hglm has quit [Quit: leaving]
<hno>
found.
paulk-desktop has quit [Quit: Ex-Chat]
n01 has quit [Read error: Operation timed out]
drachensun_ has quit [Quit: Leaving]
_BJFreeman has joined #linux-sunxi
BJfreeman has quit [Ping timeout: 256 seconds]
<oliv3r>
mripard: right, tested my driver and i'm 99% happy with it :) so gonna submit it finally; (sorry for taking so long to test :p) anyway, 1 "issue" remains. When I hexdump/cat the dev entry, which should be a 'read-only' operation, on an S_IRUGO node, with a read function, I get permission denied errors (it tries to write). strace shows: write(2, "hexdump: /sys/bus/platform/devic"..., 79hexdump: /sys/bus/platform/devices/1c23800.eeprom/key: Operation no