n6pfk has quit [Remote host closed the connection]
tzafrir_laptop has joined #arm-netbook
popolon has quit [Quit: Quitte]
leszcz has quit [Ping timeout: 245 seconds]
termleech has left #arm-netbook [#arm-netbook]
mSquare has joined #arm-netbook
<Turl>
ZaEarl: ping
cheng has joined #arm-netbook
revident has quit [Quit: Combustible lemons? Bah, I bring you weaponized asparagus!]
Gumboot has quit [Read error: Operation timed out]
tekzilla has quit [Ping timeout: 260 seconds]
tekzilla has joined #arm-netbook
<lundman>
was there any CPU speed fixes since the original 3.0.8+ branch?
ibrah has joined #arm-netbook
<Turl>
lundman: there was a fix for the hdmi detection thread
<Turl>
it was running too fast I think, and they slowed it doen a bit
<Turl>
I don't recall any other
<Turl>
down*
<Turl>
lundman: why?
<lundman>
just felt a little slower after I put my kernel on
<lundman>
but it is moving around so i guess it is working
<lundman>
but video playback dies after 2-3 seconds, something about audio
<lundman>
you know what, I think it is time for me to give up
<lundman>
so, i wonder if ouya is something to look at too
ZaEarl has quit [Ping timeout: 255 seconds]
<Turl>
lundman: ouya is tegra :P if you want to hack on it I'd forget it
<Turl>
lundman: it'd be better if you used the mele config btw, the generic ones on the kernel tree are.. generic
<Turl>
usually people bundle it and you can grab it at /proc/config.gz
arete74 has quit [Ping timeout: 265 seconds]
<discopig>
hi
arete74 has joined #arm-netbook
Gumboot has joined #arm-netbook
ibrah has quit [Ping timeout: 260 seconds]
Quarx has joined #arm-netbook
<lundman>
turl: afaik, I am using sun4_crane_defconfig, its not particularly generic
<lundman>
but audio is plain breaks
<lundman>
tegra is worse than a10? they've already promised opensource, so that would be better than allwinner
<Turl>
remember that tree isn't mele's by itself
<Turl>
lundman: tegra? open source? PFFFffff ahahaha good joke
<Turl>
tell me another one
<lundman>
heh nah, I mean "ouya" has promised fully open
<lundman>
promise is one thing of course
<Turl>
"ouya" is nobody
<Turl>
nvidia barely releases kernel sources for their development platforms only
<Turl>
which doesn't get you much if you don't have one of those exact platforms, or a design based on one of them w/ little changes
<lundman>
well, that is what they claim, so we'll see if that happens
<Turl>
and you still have the issue that the kernel by itself doesn't do anything without the crapton of matching userspace blobs
<Turl>
which often have specific hardware stuff tied in them
<lundman>
not sure what your argument there is. its what we have now, except we also dont have a kernel
<Turl>
lundman: you do have a kernel
<Turl>
lundman: kernel does stuff in the kernel
<lundman>
i would argue I dont
<lundman>
if all I am trying to do is replace the kernel, with one I built
<lundman>
and everything breaks
<lundman>
then, we do not have the kernel
<Turl>
because your config isn't right
<Turl>
in any case
<Turl>
on sunxi you also have userspace components using standard interfaces
<Turl>
cameras over V4L, audio over alsa
<Turl>
nvidia has improved a bit on that side lately, but it's still pretty ugly
<Turl>
you basically have a crapton of .so which talk with the hardware over their own weird undocumented interfaces
<Turl>
there's no userspace source whatsoever
<lundman>
we dont even have a kernel from allwinner, it was a leak. so so far, I fail to see the difference
<Turl>
AW then GPL'd it
cat1 has quit [Ping timeout: 248 seconds]
<Turl>
nvidia releases the kernel because if they didn't they'd get sued
<lundman>
heh sounds the same
<lundman>
not they gpled what was leaked, not what they have
<lundman>
since they clearly fixed audio since then
<Turl>
lundman: in any case, if you want a device to hack on, get an OMAP
<Turl>
lundman: audio works on my tablet fwiw
<lundman>
whats an OMAP?
ibrah has joined #arm-netbook
<Turl>
lundman: TI's processor series
<Turl>
beagleboard, pandaboard
<Turl>
those have OMAP processors
<lundman>
one thing that COULD work with ouya, is so many people will have the hardware, so that size alone you are almost guaranteed to get xbmc and good support
<Turl>
buy a pandaboard
<Turl>
it has XBMC already
<lundman>
nobody here has one :)
<lundman>
similarly odroid-x also has JB and xbmc
<Turl>
pandaboard has linaro support too
<lundman>
what is linaro anyway
<Turl>
and it's supported-ish directly on AOSP too
<lundman>
at the moment I guess I just wait and see what evryone does
<rz2k>
their code in kernel was ok, they got to 3.6 with DT support, they have "official" u-boot fork, playing with EV2 SoC was fun.
<Turl>
gnight :)
<Turl>
I sent the cpufreq cleanup to the ML btw
<lundman>
thanks for your comments
cat_n9 has joined #arm-netbook
<lundman>
hmm the panda is dual for 182. odroid is quad for 130
<lundman>
just that waiting isnt as much fun
ZaEarl has joined #arm-netbook
<discopig>
lundman, odroid is about to be awesome because mali400 3d accel has been working recently
<discopig>
or at least semi-working
<lundman>
explain? it wasnt working before?
<discopig>
it only worked on android
<discopig>
it had no hw accelerated xorg or anything on linux
<lundman>
ok so same mali hassle as a10
<discopig>
but from what i've heard these past few days there's been a lot of work going into it and there's working implementations
<lundman>
you have a board?
<discopig>
i only have cubieboard and hackberry as far as A10/mali go
<discopig>
they,re both headless so i didnt really try any graphic driver
<lundman>
yeah
<lundman>
I would look at JB with xbmc though, with the future hope of being able to rent/buy tv/movies from google play
<discopig>
supposedly xbmc works flawlessly with the mali drivers
hipboi has joined #arm-netbook
pawel5870 has joined #arm-netbook
pawel5870 has quit [Ping timeout: 245 seconds]
<mnemoc>
moin
<Maqs>
moin
<andoma>
morning
leszcz has joined #arm-netbook
<rm>
hey all
cat_n9 has quit [Ping timeout: 246 seconds]
<rm>
what do you think will last more under a regular logfile-style write scenario
pawel5870 has joined #arm-netbook
<rm>
an USB flash stick or an SD card?
cat_n9 has joined #arm-netbook
<hno>
rm, none of them unless you accept disabling syncronous writes for the logs.
<ZaEarl>
I would think usb stick and sd card to have similar wearing performance.
<ZaEarl>
ie very little
<Triffid_Hunter>
rm: as ZaEarl says, I think they'd be similar since they both use the same type of storage, the same wear levelling algorithms, just different host interfaces
<andoma>
i ran my HTPC (using ubuntu installed in a USB drive) lasted 1-2 years maybe
<rm>
so the question is really "which one typically has better wear-levelling and smaller erase block sizes"
<rm>
e.g. I wouldn't expect much smarts from a microSD card, given its size
<rm>
USB sticks can be pretty sizeable with a separate controller and flash chips etc
<rm>
(otoh)
<Triffid_Hunter>
rm: sure, physical size alone doesn't mean that they're better though.. if you're looking for longevity, I suppose you'd best be getting stuff from reputable manufacturers
popolon has joined #arm-netbook
<mnemoc>
rellla: where did you send your fix for the sata fix? i can't find it :|
<rm>
also what a useless benchmark, comparing it only to itself
<rm>
not to at least e.g. Atom
<mnemoc>
rellla: found
<Maqs>
ok.. zeroing the first mb of the uSD card (except the partition table) did not work... U-Boot SPL still seems to be stuck after the first line of output
sspiff has joined #arm-netbook
sspiff has joined #arm-netbook
<mnemoc>
Maqs: i have that problem after a reboot, never of the first pass...
<Maqs>
i tried a reboot and a cold start.. same result
<Maqs>
i think i'll just use the uboot that came with the image and modify the kernel with a hardcoded dram size.. i guess that should be the easiest way :-/
<mnemoc>
hno: ----^
cheng has quit [Quit: Leaving]
<rm>
afaik One Does Not Simply hardcode RAM into the kernel
<rm>
the portions of RAM you plan to use have to be pre-initialized (by u-boot? by boot0/1?...)
newbie has joined #arm-netbook
newbie is now known as Guest75986
<mnemoc>
Maqs: that probably means you are using bad dram initialization parameters
<mnemoc>
Maqs: what board?
hp__ has quit [Ping timeout: 276 seconds]
<Maqs>
mk802, a10
<Maqs>
there is a script.bin in on the first partition, called evb.bin.. i generated dram.c for uboot from that one
<mnemoc>
problem with the dram_para info in script.bin is that in reality boot1 doesn't use it :|
hipboi has quit [Ping timeout: 256 seconds]
pawel58701 has joined #arm-netbook
pawel5870 has quit [Quit: Leaving.]
<mnemoc>
livesuit probes the ram and write whatever it finds proper
<mnemoc>
hno has have some success grabbing rests of boot1 from the nand's uboot
<mnemoc>
cat_n9: are you still happy with the usb changes? or have some doubts?
<cat_n9>
mnemoc, i am always happy with my changes :) but seriously, did anybody try it on sun5i?
newbie1 has joined #arm-netbook
<mnemoc>
cat_n9: i'm preparing an stage branch for 3.0 and 3.4 with a bunch of things to test (including buf2) and was wondering if it's a good idea to get the usb cleanup in yet or not
<cat_n9>
mnemoc should be relatively safe as usb basic stuff does not seem to be broken.
<mnemoc>
ok
<mnemoc>
also, do you include the unification patch or your set goes after it?
<Maqs>
mnemoc: is it just the dram_size i have to adjust manually?
Guest75986 has quit [Ping timeout: 276 seconds]
hipboi has joined #arm-netbook
<Maqs>
i'm not that far into linux.. the oops was after "<2>kernel BUG at mm/bootmem.c:275!"
<mnemoc>
Maqs: i think it's 4 dram_ things
<mnemoc>
they usually come NULL (empty) in modern script.bin files
<mnemoc>
hipboi: hi! any news code the cedarx/armhf arena? I read on j1nx blog they received new stuff from allwinner
<cat_n9>
mnemoc: it is included, i did some minor changes there to be able to apply against our tree.
<hipboi>
mnemoc: not yet
<mnemoc>
Maqs: dram_chip_density, dram_io_width, dram_bus_width and dram_size
<mnemoc>
hipboi: hope?
<Maqs>
mnemoc: do you know how they get to the kernel/where to hardcode that?
<mnemoc>
Maqs: that BUG feels odd
<mnemoc>
Maqs: the kernel doesn't care
<mnemoc>
Maqs: it's all spl or boot0/boot1 thing
<mnemoc>
Maqs: you might want to disable the mali mem reserving option. CONFIG_SUNXI_MALI_.... iirc
<oliv3r>
if you scroll down to the 'Clock generation section, you see it works now
<oliv3r>
but if I add 1 more letter to the top half of the equation, it breaks :S
cat_n9 has quit [Ping timeout: 246 seconds]
<mnemoc>
give me a <math /> block to test
<oliv3r>
<math>\frac{24 \, \mathrm{MHz} \times N \times}{M}</math>
<oliv3r>
this works
<oliv3r>
but
<oliv3r>
<math>\frac{24 \, \mathrm{MHz} \times N \times P}{M}</math>
<oliv3r>
no longer works ( P added)
<oliv3r>
(my latex skills are very rusty)
<Maqs>
looks okay to me
<libv>
how stable is this pll?
<oliv3r>
libv: i have absoluitly no idea
<oliv3r>
if i ever get a cubieboard, i can hook it up to my scope though
<libv>
all seem to be the same right?
<oliv3r>
all 6 PLL's?
<oliv3r>
nope, far from
<libv>
ok?
<oliv3r>
libv: unless of course i missunderstand you, I only know very very very little about PLL's
<mnemoc>
oliv3r: talk with hipboi. he is shipping the second batch this weekend
<oliv3r>
but looking at the doc, PLL1 does 240 MHz - 2 Ghz, but PLL2 only does 22.5 or 24.5 MHz
<libv>
oliv3r: me too, i only spent some months of my life testing them against 19" scopes
<oliv3r>
mnemoc: i'm poor :p i'm thinking of waiting for an A40 based one :)
<libv>
the sort of scopes that come with a vga connector :p
<oliv3r>
libv: LOL, i actually have a rigol something; lots of money spend on ... a toy really :)
<oliv3r>
i know absolutly nothing about PLL's. I had to read up on the wikipedia yesterday to see what it does :)
<libv>
oliv3r: CRTs are surprisingly good at showing pll instabilities
<oliv3r>
so i'm affraid I cannot answer your question on how stable the PLL controller is :(
<libv>
oliv3r: i think i am the only graphics developer out there who has gone and tested the plls on his display engines to death though
<mnemoc>
oliv3r: they work on wikipedia, so the problem is mine :|
<oliv3r>
lol, because of them being buggy?
<oliv3r>
mnemoc: have you seen the problem on linux-sunxi wiki?
<libv>
even the radeon guys, when they copied radeonhd code, wondered what a weird line cutting off some internal frequencies was for, and removed it
<mnemoc>
oliv3r: yes, tested on my sandbox and got the parsing error
<oliv3r>
libv: they should have asked why it was there for
<mnemoc>
while the same works in my wikipedia sandbox
<oliv3r>
mnemoc: ok, i'll just use the broken text and check if it works when i'm done :)
<oliv3r>
when your done*
<libv>
oliv3r: they also removed copyright statements, so no, they would not do that
<oliv3r>
libv: ouch :( that's just rude imo
<libv>
oliv3r: radeonhd vs radeon had way more ruder things, like using fd.o root to vandalize the repo
<oliv3r>
i remember those stories on phoronix
<libv>
in any case, as soon as the display engine is better known, and we are close to setting modes liberally, i will try to kill another 19"er
<oliv3r>
a bit childish imo
<libv>
and then we will see just how stable these plls really are :)
<oliv3r>
gonna force the PLL's into modes that they shouldn't be in? :)
<oliv3r>
Haha
<libv>
there is nothing like pll testing for killing modes
<oliv3r>
libv: well hopefully, I'll finish writing the PLL bit on the wiki soon!
<libv>
as basically you disable h/v sync temporarily
<libv>
making the monitor switch off briefly
<oliv3r>
we have a few 'broken' monitors here at work that can use some 'testing'
<oliv3r>
19" tft's :D
<libv>
and then reenable h/v sync, which makes the monitor switch on again
<libv>
do that many thousands of times, and you will kill some electronics
<libv>
oliv3r: i have some CRTs reserved just for that sort of work
<oliv3r>
i bet :( poor electrons, you are violating their homes!
<oliv3r>
Should I use | or / to designate an OR, e.g. P: 1/2/4/8 or P: 1 | 2 | 4 | 8
<oliv3r>
libv: why do you still have CRT's! :p
<oliv3r>
hmm, P: {1, 2, 4, 8}
<libv>
oliv3r: exactly for this reason, tfts do not show anything
<libv>
a crt really is a big high frequency scope
<oliv3r>
libv: the only CRT i have left is the one in our bedroom
<libv>
display work needs crts
<oliv3r>
and I better not dare touching that one :)
<libv>
tfts are quite forgiving, but the electronics tend to limit what is allowed even more than with crts
<libv>
spread spectrum works for tfts
<libv>
it's absolute hell for a crt
<libv>
it will display it, but it is completely unreadable
<oliv3r>
I do know that ye'ol osciliscope isn't much more then a television, or rather, vise versa
<oliv3r>
vice*
<libv>
the thing is, the combination crt / eyes is pretty sensitive
Almamuetya has joined #arm-netbook
<libv>
but it is limiting though, especially in the case of the A10, where the plls are probably stuck in a separate block, as you can only test the clocks that can be wired to the vga
<libv>
the good thing is though, that the separate block might mean that the plls internally might be very much the same
<libv>
and that they have the same internal stability limits
<Maqs>
seems to come from ./drivers/staging/rtl8192u/r8192U_dm.c here
Entropy512 has quit [Remote host closed the connection]
<hno>
PBC is from Wifi Protected Setup, Push Button Configure.
<RaYmAn>
it's almost depressingly funny how much that chromebook looks like a macbook, lol
<RaYmAn>
not that that's a bad thing, but.
<mnemoc>
i wish it had the display of a macbook
<RaYmAn>
yeah
<RaYmAn>
though, it doesn't look bad
<mnemoc>
it's the same of my current thinkpad...
<RaYmAn>
as long as it's IPS that's not necessarily bad
<RaYmAn>
also same = ? Same resolution?
<RaYmAn>
or same screen?
<mnemoc>
same resolution
<RaYmAn>
Do you really need more on an 11" screen though?
<mnemoc>
certainly not less. but indeed, don't need more
tmerle has joined #arm-netbook
<RaYmAn>
it kind of feels like it's getting to same issue as phone cameras...Bigger and bigger resolutions, but completely the same quality pictures, because sensor or lcd is too small :P
<Maqs>
kernel debugging was enabled
<Maqs>
for all kinds of things.. so rtl8192 took the liberty to spam console + klog with everything :-)
<mnemoc>
:)
<cat_x301>
Maqs: i think we can switch to rtlwifi driver once we resolve two issues with it: 1) it is crashed while loading fw -- workaround is available; 2) after wlan initialization one need to ping host from outside in order "wake" it up -- i do not know however it is only me who sees this problem.
<cat_x301>
mnemoc: what do you think of comments on cubieboard about allwinner and linaro?wonder if it ever happens..
<mnemoc>
cat_x301: have conflict feelings
<cat_x301>
mnemoc: same here, glad that im not alone :)
<hno>
I would love if Allwinner got involved with Linaro, but fear it's too late for the A10 in that game and have doubts if AW can understand how it brings benefit to them.
<hno>
Even doubt it really brings noticeable benefit to AW in their current market segment.
<Turl>
maybe for sun6i
<hno>
Problem is the product lifespan. What we saw with A10 was a ~8-10 months peek period where it sold in massive volume, and then getting almost completely replaced by a competitor. Slightly extended with the A13, but at much lower profit.
<hno>
Serious Linaro involvement requires a longer expected lifespan to make much sense imho.
<xxiao>
i missed that, AW is committed to Linaro now?
<hno>
no
rsalveti_ has joined #arm-netbook
* xxiao
's ubuntu 12.10 failed to boot in vbox
rsalveti has quit [Ping timeout: 260 seconds]
rsalveti_ is now known as rsalveti
<xxiao>
works now, somehow vbox did not prompt me for crypt-disk-password, not it works
<xxiao>
Linaro wants more chip vendors for sure...