deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
Akagi201 has joined #linux-sunxi
Akagi201 has quit [Ping timeout: 255 seconds]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
Akagi201 has joined #linux-sunxi
deasy has joined #linux-sunxi
deasy has quit [Remote host closed the connection]
lkcl has quit [Ping timeout: 272 seconds]
FreezingCold has joined #linux-sunxi
<libv>
pffff...
<libv>
many libavcodec bits, and other stuff scrounged from the net, some other bits are not kosher indeed
lkcl has joined #linux-sunxi
<naobsd>
ah, I forgot to mention
<naobsd>
that patch is for u-boot
<libv>
naobsd: that's clear from the patch itself :)
<naobsd>
:)
FreezingCold has quit [Ping timeout: 260 seconds]
FreezingCold has joined #linux-sunxi
TheSeven has quit [Ping timeout: 260 seconds]
TheSeven has joined #linux-sunxi
Faisal has quit [Ping timeout: 250 seconds]
lioka has quit [Ping timeout: 272 seconds]
akaizen_ has quit [Ping timeout: 240 seconds]
naobsd has quit [Quit: Page closed]
Black_Horseman has joined #linux-sunxi
keebler has quit [Ping timeout: 245 seconds]
TheSeven has quit [Ping timeout: 272 seconds]
TheSeven has joined #linux-sunxi
rz2k has joined #linux-sunxi
keebler has joined #linux-sunxi
keebler is now known as Guest34161
Akagi201 has quit [Read error: Connection reset by peer]
zeRez has joined #linux-sunxi
<bbrezillon>
petrosagg: can you try to add nand-rnd-mode = "hw"; to your boot0 partition ?
<bbrezillon>
petrosagg: and what exactly is failling when you try to use nandwrite/nanddump ?
nabblet has joined #linux-sunxi
<bbrezillon>
petrosagg: did you erase your partition before writing on it ?
<bbrezillon>
petrosagg: and yes I did some test on boot0 boot0-rescue a long time ago (on the v3 of my driver), but more quitte recently managed to make it work too (read/write on this partition and boot from it after writing it from Linux IIRC)
Skaag is now known as Skaag
rainbyte has quit [Quit: Leaving]
lkcl has quit [Ping timeout: 272 seconds]
kuldeepdhaka has joined #linux-sunxi
zeRez has quit [Remote host closed the connection]
zeRez has joined #linux-sunxi
skoperst has joined #linux-sunxi
lkcl has joined #linux-sunxi
zeRez has quit []
PulkoMandy has joined #linux-sunxi
lkcl has quit [Ping timeout: 245 seconds]
libcg has joined #linux-sunxi
Quarx has joined #linux-sunxi
lkcl has joined #linux-sunxi
Quarx has quit [Ping timeout: 250 seconds]
Quarx has joined #linux-sunxi
jemk has joined #linux-sunxi
23LABCQW2 has joined #linux-sunxi
23LABCQW2 has quit [Client Quit]
libcg has quit [Remote host closed the connection]
<petrosagg>
bbrezillon: yes, this is what I ended up doing. I had to set "nand-rnd-mode = "hw";" to my boot0 partition
<petrosagg>
bbrezillon: otherwise I would write data with nandwrite and nanddump would output different data
<petrosagg>
bbrezillon: even running nanddump two times without actually writing would produce different results
<petrosagg>
bbrezillon: and also nandtest would always fail until I added the nand-rnd-mode
<petrosagg>
bbrezillon: I did erase my partition before writing it. My current problem is that it seems like I deleted my boot0 bootloader that was in there
<petrosagg>
bbrezillon: and I can't find any resources on the internet on how to bring it back
<petrosagg>
bbrezillon: I found out that with the old sunxi-3.4 this NAND region wasn't even accessible
<petrosagg>
bbrezillon: I also tried using the new mtd driver to write a boot0 partition with the eGON.BT0 header both at offset 0 and at 0x2000 (which is the offset on the SD) but none worked
<ssvb>
paulk-aldrin: based on reading the irc log, do I understand it right that you are having problems with dram config?
<petrosagg>
bbrezillon: so I'm wondering if you have any info on how to restore this bootloader. LiveSuit on Linux did work for me so yesterday I started a windows vbox image download and I'll test that now
<paulk-aldrin>
ssvb, seems like it
<ssvb>
paulk-aldrin: which version of u-boot are you using?
<paulk-aldrin>
ssvb, first, script.bin and meminfo don't match
<paulk-aldrin>
ssvb, u-boot-sunxi master
<paulk-aldrin>
so not mainline
<paulk-aldrin>
s/master/sunxi/
<ssvb>
paulk-aldrin: one of the risks is that boot0 and u-boot very likely have slightly different dram init code
<ssvb>
and don't set "dram_odt_en" to anything other than 0
<paulk-aldrin>
ok
<paulk-aldrin>
in script.bin?
<paulk-aldrin>
(does script.bin matter at all here?)
<ssvb>
in u-boot dram_para
<ssvb>
script.bin does not matter
<paulk-aldrin>
right
<ssvb>
"dram_zq = 0x7b" is just a default reset value (123 in decimal, which is kind of their way to say 0xbadf00d), and with dram_odt_en=1 it would be taken into use to kill your dram reliability
kuldeepdhaka has quit [Read error: Connection reset by peer]
<paulk-aldrin>
oh ok
<paulk-aldrin>
emr1 = 0 didn't help
<ssvb>
does reducing the dram clock speed help?
<paulk-aldrin>
lower than 336?
<ssvb>
oh, you already have it hat low
<ssvb>
*that
<paulk-aldrin>
yep
<ssvb>
how do you know that it is not a legitimate oops because of some other problem?
<paulk-aldrin>
well, I don't
<paulk-aldrin>
but last time, Unable to handle kernel paging request was because of a dram misconfiguration
<ssvb>
does it fail after initializing the nand driver?
<paulk-aldrin>
yes
<paulk-aldrin>
could it be the nand driver?
<ssvb>
you can try to disable nand and boot from the sd card
lkcl has quit [Ping timeout: 250 seconds]
<paulk-aldrin>
it does say: "wrong Rb connect Mode, chip = 255 ,RbConnectMode = 1 "
<paulk-aldrin>
sure
Quarx has joined #linux-sunxi
lkcl has joined #linux-sunxi
<paulk-aldrin>
ssvb, good call, it was the nand driver
<bbrezillon>
quitte: I'd say you've correctly applied this patch
<bbrezillon>
;-)
<quitte>
bbrezillon: there is no drivers/mtd/nand.c in my kernel source tree. nor drivers/mtd/nand/nand.c
<quitte>
i did not get a patch from you. that is the actual thing you gave me. which will be a pain because if i wget it there is all kinds of html mess-up in there
<bbrezillon>
quitte: well, actually it's nand_base.c, don't why it's name differently in your head comment...
<quitte>
this is your head comment. thanks. I know where to put it and that it's the right thing now.
<RaYmAn>
libv: tbh, if allwinner was to implement fastboot finally, I wouldn't be surprised if they messed it up so much it required a seperate fastboot exe ;)
<libv>
MY123: what?
<libv>
RaYmAn: :)
<petrosagg>
quitte_: I see you have a uboot partition as well. If I have no uboot written on the nand would I see any output to my serial port?
<libv>
RaYmAn: in such a case, we need more than "here is a dropbox link to a windows binary with no history or nothing"
<petrosagg>
quitte_: because if not, then it might be loading fine but not be able to continue and no way of giving me logs back
<quitte_>
petrosagg: you would get output from boot0 alone
<MY123>
libv: Is Lima in good shape?
<quitte_>
err sorry sunxi-spl
<skoperst>
libv: i'm that guy who uploaded the fastboot.exe
<libv>
skoperst: ok, good, as the wiki email didn't work
<MY123>
(I'm currently REing the VideoCore)
<libv>
skoperst: what is this, where does it come from, what is its history, why is there only a windows binary, why is there just this .exe?
<libv>
MY123: then why are you here, in linux-sunxi?
<MY123>
libv: Because I asked about how to port GNU/linux to an A20 tablet ( I have the script.fex)?
<petrosagg>
quitte_: when you say boot0 you mean the u-boot-spl.bin I just compiled, right?
<libv>
MY123: new device howto
<libv>
MY123: it is referenced from the front page
<quitte_>
petrosagg: no. the spl/sunxi-spl.bin
<quitte_>
petrosagg: i guess that's your problem
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
<petrosagg>
quitte_: sorry, that's what I meant
<libv>
paulk-aldrin: so it was a nand problem?
<quitte_>
oh
<petrosagg>
quitte_: That's the file I wrote. spl/sunxi-spl.bin
<libv>
skoperst: if you do not like explaining this in public, feel free to msg me
skoperst has quit [Ping timeout: 246 seconds]
<libv>
*sigh*
<petrosagg>
quitte_: I'm just verifying that by boot0 you didn't mean the stock firmware from allwinner
<quitte_>
petrosagg: i'm getting errors trying to nandwrite right now. i forgot to update that, i guess. I'll send my sunxi-spl.bin as a reply in the meantime
<MY123>
libv: Seen that. It just boots with with no screen and USB-OTG doesn't have the microHDMI. ( it's the Kurio 7S)
<paulk-aldrin>
libv, indeed. I'll need to fix that however
<libv>
paulk-aldrin: so the meminfo stuff was ok?
<petrosagg>
quitte_: ok. I'm also trying again using your dts
<libv>
paulk-aldrin: seems like we both need to work with libnand today
<MY123>
* microHDMI cable
<libv>
MY123: that's not on our wiki
<libv>
MY123: start working through the new device howto
<paulk-aldrin>
libv, well I'm still running with a very low dram clock, but it seems alright
<MY123>
Started, I even generated a BSP.
<libv>
MY123: start with documentation
<paulk-aldrin>
libv, I probably won't work on it in the next few hours, I'm writing docs…
<libv>
like the ndh prescribes.
<libv>
MY123: bsp is pretty late in the process
<libv>
pictures need to be taken soon
<MY123>
libv: I can't void my warranty for UART/
<libv>
MY123: why can you not void warranty?
<MY123>
( there is still a year and a half)
<libv>
it's a 50EUR tablet
<libv>
especially the images that google turns up...
<MY123>
libv: It's 200EUR. The Kurio 7S.
<libv>
that looks like one of the 2012 mid forfactors
<petrosagg>
quitte_: also, why do I need to use mtdparts when I have the partitions defined in my dtb? plus in mtdparts you can't define the randomiser values
<libv>
MY123: start filling in the ndh, and take all round pictures :)
<quitte_>
petrosagg: u-boot stores it's partittion information in its environment
<quitte_>
it does not tell linux about the partitions. so you have to make those match yourself. mtdparts is just a fancy way of setting environment variables
<petrosagg>
quitte_: oh, ok. got that
<petrosagg>
quitte_: So basically I have to give sunxi-spl the right configuration to find and load u-boot with the correct environment
<petrosagg>
quitte_: and then u-boot will load linux with a dtb
<quitte_>
as it is now it will try to load u-boot from offset 2MiB off the nand
<quitte_>
the README of u-boot has good information about the variables in sunxi common config
<petrosagg>
ok, one last question. What read the uEnv? the spl part of the uboot part?
<quitte_>
I actually don't recall. I'd have to check
<quitte_>
and since there will be lots of read errors it's a lot nicer to have it in mmc - since that works reliably
lkcl has quit [Ping timeout: 250 seconds]
<quitte_>
having u-boot in the spl partition would be nice, however. I haven't tried but it should work. just write spl with randomization and u-boot without
<mnemoc>
libv: feel free to unpack the SDKs in /srv/http
<petrosagg>
I don't understand why u-boot doesn't need randomisation. Doesn't it use the special 0x4a80 value as well?
<quitte_>
petrosagg: the reason that spl exists is that in the early boot stages sdram is not accessible. so everything needs to work with 24K of sram
<libv>
MY123: cool!
<quitte_>
so spl basically just enables ram and loads the next program into it
<libv>
mnemoc: turl has unpacked the important bits already
<libv>
mnemoc: which suits all our needs :)
<quitte_>
this sunxi-spl reads the next part from a partitin with randomization disabled
<MY123>
libv: Another edit.
<petrosagg>
quitte_: is this what disables randomisation in your dts? nand-randomizer-seeds = /bits/ 16 <0x00>;
<MY123>
How to embeed images?
<libv>
"Upload file"
<quitte_>
petrosagg: supposedly it is. however i never tried that. I wrote u-boot to nand from within u-boot
<libv>
on the left menu
<quitte_>
petrosagg: which should make writing a non-randomized u-boot to the spl partition a lot easier
<mnemoc>
libv: cool :)
<libv>
MY123: there is this thing called "Recent Changes" on our wiki, i spent many a bored moment watching that, so i see the changes you make
<quitte_>
petrosagg: having a tiny kernel image is important because of the missing read-retry support. when you try to nand boot it use one that is xz compressed. doesn't fix it, but makes it bearable
<libv>
quitte_, petrosagg, bbrezillon: can you guys write up a separate wiki page about the mtd driver, its status today, how to use it and what are the pitfalls?
<petrosagg>
quitte_: I remember reading about read-retry issues in past logs and also remember reading about in my NAND's datasheet
<libv>
you seem to be spreading a lot of info around here on irc, but that will be lost pretty soon
<petrosagg>
libv: I've acquired a huge amount of knowledge the past week that I'm trying to make that work
<libv>
petrosagg: document as you go
<petrosagg>
libv: I'll make a draft with what I know
<libv>
petrosagg: otherwise bits will get lost
<petrosagg>
libv: should I just register in the wiki?
<libv>
petrosagg: definitely!
<petrosagg>
libv: cool, I'll do that
<petrosagg>
quitte_: where should this read-retry logic be? in sunxi_nand.c?
<libv>
anyway, you will find yourself referring to what you documented just a few days earlier, pretty quickly, instead of having to wonder or having to dig through irc logs
<petrosagg>
libv: that's true, working memory is limited
<quitte_>
petrosagg: you'll have to ask bbrezillon about that. however I don't think it makes sense to patch that in. the proper course of action would be to port or merge brezillon's driver.
<MY123>
libv: Adding photo.
<petrosagg>
quitte_: wait, you mean the linux drivers handles that properly but the read-retry logic is only missing from u-boot?
<quitte_>
I'm not even sure if the dma headers are free. those should be taken from the dma mainlining thing for that reason alone
<quitte_>
petrosagg: yes
<petrosagg>
libv: quitte_: ok, I'm making a skeleton wiki page now
<quitte_>
petrosagg: thanks. please don't hesitate to ask about anything, especially if you're going to put it on the wiki
<MY123>
libv: Did you see the back picture? The quality is quite bad.
<libv>
MY123: those pictures are very bad quality indeed
<libv>
MY123: provide decent all round lighting, stick some white paper underneath the device
<libv>
and get in close enough but far enough to allow for focus
<libv>
and try to get the hw from straight above
<libv>
in case of the front, but without leaving a reflection
<petrosagg>
libv: I can't seem to be able to create a User:petrosagg/nand_draft page in the wiki
<libv>
cut down the image in a image editor to just show the device
<libv>
MY123: and try not to get smudged pictures (this is my biggest problem, i smudge due to pressing the button)
<libv>
petrosagg: just a sec, i need to make you a into a real boy
* petrosagg
waits to become a man
<libv>
petrosagg: and don't bother with a user page
<libv>
petrosagg: just don't name it nand
<MY123>
libv: They are taken from a 3,2MP tablet sensor. For the image editor: Okay.Smudged pictures, will apply the retardator to 5sec.
<libv>
petrosagg: MTD_Driver perhaps
<MY123>
(I have not a camera)
<petrosagg>
libv: alright
<libv>
MY123: i actually get better pictures out of my cameraphone that with a dedicated but cheap camera
<quitte_>
libv: do you think you can do the same for me, jsut in case I actually can force myself to add content some day?
<MY123>
libv: The first Galaxy Tab. Bought it in December 2011.
<libv>
petrosagg: "a real boy" is a reference to pinocchio :)
<quitte_>
"I want to be a real doll!"
<petrosagg>
libv: ah, it's been a long time since I saw it :)
<libv>
quitte_: done
<quitte_>
thanks
<libv>
quitte_: add to that page as soon as petrosagg has set up the basics
<libv>
so that you do not generate too many conflicts
<petrosagg>
libv: I'm getting a permission error
<petrosagg>
libv: do I have to logout/login ?
<libv>
petrosagg: on your user page?
<libv>
petrosagg: perhaps, but be sure to copy the content in some text editor first as to not lose it
<petrosagg>
libv: no, on the MTD_Driver page
<petrosagg>
libv: I thought I shouldn't use a user page
<libv>
ok, so non people cannot create pages, at all
<libv>
that i didn't know
<petrosagg>
libv: ok, I'll create a user page
<libv>
petrosagg: try logging out and in
<petrosagg>
I just need somewhere to do a brain dump
<petrosagg>
ok
<libv>
but keep the current tab open
<petrosagg>
libv: same issue
* libv
digs around
<petrosagg>
libv: I guess you can create an empty page and then I'll edit it
<libv>
petrosagg: i would like to get to the bottom of this
<petrosagg>
libv: no prob :)
<libv>
petrosagg: could you keep your first edit open in a tab
lkcl has joined #linux-sunxi
<libv>
then open a new tab to create the page
<libv>
and then copy paste the content over?
<petrosagg>
libv: I haven't written any content yet
<bbrezillon>
quitte_, petrosagg: back for a few minutes :-)
<mnemoc>
Turl: shows me $6.79 with free cnp shipping to .ar
<Turl>
they now make you go to the post office for those and pay a flat fee for the handling though :p
<Turl>
yep I see it as well :)
<mnemoc>
Turl: how much is that flat fee?
<mnemoc>
< mnemoc> .ar is hell for online shopping :(
<bbrezillon>
quitte_, petrosagg: when you tried to write the spl from linux, which image did you try, u-boot-spl.bin ?
<quitte_>
sunxi-spl.bin
<bbrezillon>
quitte_, petrosagg : can one of you try to format the spl from AW tool (or from u-boot if it works) then dump it from Linux, check the eGON pattern and format the dumped image (after a flash erase of course)
<bbrezillon>
quitte_, petrosagg: one other interesting thing would be to compare the dumped image with the one generated by u-boot
<bbrezillon>
IIRC there's only 1K of data per page (in CT case 8K page)
<deepe>
Hi guys, I need, little help, or advice
<quitte_>
bbrezillon: what are you trying to accomplish?
<deepe>
I am setting display on A10-olinuxino-Lime board, Display initialize well, but I am having trouble to show some real data on in (console text)
<bbrezillon>
quitte_: I haven't all read the backlog, but did petrosagg you manage to get a working SPL ?
<deepe>
after trying to send just random data /dev/urandom to framebuffer device
<Turl>
mnemoc: I think it was 40 ARS last I went to pick sth up
<quitte_>
bbrezillon: petrosagg ended up flashing mine. but the problem probably was that he used a nand less configuration to build spl
<quitte_>
otoh I don't see why nand less spl wouldn't run...
<bbrezillon>
quitte_: indeed, if you have a valid MMC inserted it should work
<quitte_>
bbrezillon: I'm fairly confident that apart from badblock problems u-boot spl and yours are completely interoperable
<deepe>
display change it appearance, also checking hsync, vsync, etc with osciloscop all is good (according to display datasheet)
<quitte_>
bbrezillon: it should output something at the very least
<bbrezillon>
quitte_: yes, but I'm not sure you can directly flash an SPL from u-boot
<deepe>
so What I want to as you, to ged display work, I need to setup fex file to use LCD, setup LCD parameters
<bbrezillon>
quitte_: sorry from Linux
<quitte_>
bbrezillon: you can. nand write.1k works fine
<deepe>
and display should works?
<quitte_>
and yes that works, too
<quitte_>
(i meant flashing from linux, deepe)
<bbrezillon>
quitte_: you mean flashing mtd0 with sunxi-spl.bin works ?
<quitte_>
yes
<bbrezillon>
quitte_: okay, thought this might be the problem
<quitte_>
for both me and petrosagg
<bbrezillon>
quitte_: just a clarifaction about randomization => if you really want to disable it you can just set nand-rnd-mode to "none" instead of setting the seed to 0x0
<deepe>
quitte_: sorry I do not get it
<quitte_>
bbrezillon: since I applied the patch ubiformat worked fine on both first tries
BorgCuba has quit [Quit: leaving]
<bbrezillon>
quitte_: though disabling randomization on such NAND is not a good idea :-(
<quitte_>
bbrezillon: is it possible to have 2 partitions mapped in the same nand area, btw?
<quitte_>
it would be nice if spl could be accessed both randomized and non-randomized
<quitte_>
to put both spl and u-boot there
<bbrezillon>
quitte_: AFAIR I don't check for that, but I REALLY should
<quitte_>
okay so it would be better to have spl read randomized
<bbrezillon>
quitte_: having part of the NAND mapped on 2 MTD region is very dangerous
<quitte_>
there'sno reason why spl couldn't read from randomized spl, after all eGON can do it. except it doesn't work right now.
<bbrezillon>
quitte_: it would be better, yes, but I'm more concerned by your big partitions storing the kernel and UBI
<quitte_>
how so?
<quitte_>
ubi is randomized
<bbrezillon>
this is not related to the thing stored on your partition, but the size of your content
<quitte_>
it's pretty much read only. rootfs_data is just huge because rootfs is still tiny
<bbrezillon>
the more pages you write in a block the more chance you have to get bitflips, and this is even worst when you write the same pattern on several pages
<quitte_>
my vocabulary isn't goos enough to understand that
<quitte_>
however the ubi image I currently write is just 6MiB
<deepe>
so somebody with good knowledge with sunxi framebuffer driver or LCD to help?
<quitte_>
bbrezillon: rootfs-data is set to expand itself when it is attached the first time to the free partition space
<quitte_>
the comments aren't mine, I should mention. since as far as I can tell they don't fully match what the options actually do
<bbrezillon>
quitte_: your bootloader will at most take a few (say < 40) pages, but kernel and UBI will fill entire blocks and the data stored there might contain similat and thus generate more bitflips
<libv>
deepe: did you set console=tty0 as a kernel command argument?
<libv>
deepe: what linux distribution are you using?
<bbrezillon>
quitte_: randomization is here have a good distribution of data over a NAND block in order to limit bitflips
<MY123>
libv: What should I do for photos, retrying?
<quitte_>
okay. so what's the problem with ubi in my case then?
<quitte_>
it has randomization enabled the same as your data partition
<libv>
MY123: try both front, back and connectors and buttons
<libv>
MY123: get in close that the tablet can focus
<libv>
try to make the pictures sharp and not wobbly
<bbrezillon>
quitte_: that's not related to UBI, but I remember seeing some of your partitions with randomization disabled or with seeds set to 0x0 (which is exactly the same)
<MY123>
libv: My custom rom does not have autofocus support.
<skoperst>
Its driving me crazy, CT(A20),Marsboard(A10) and now Pcduino8(A80) all have some 'blank' boot up, like an old car they need a few warmup boots until they can boot up. sometime it takes 1, sometime 2,3,4..
<bbrezillon>
quitte_: your ubinize.cfg seems correct
<libv>
MY123: these pictures are really poor the way they are
<libv>
MY123: move your tablet until your focus is good :)
<rz2k>
skoperst: check your PSU
<libv>
skoperst: psu?
<skoperst>
rz2k: good idea, will check that
<skoperst>
power supply?
<libv>
skoperst: 2A at least
<libv>
MY123: these pictures are meant for identification, they need to be halfway decent
<quitte_>
yes all of the non-randomized partitions are for compatibility with yuq u-boot, which as far as I know does not support randomization except for spl. I'm not going to figure out if that is true, either. I don't know of anyone that wants to fix yuq. so u-boot needs your mtd implementation anyways
<libv>
not beautiful, not perfect, but good enough that other people can use it to identify the hw
<rz2k>
skoperst: my CT cant boot from anything except ATX PSU that I use as a bench one. every crappy USB charger-like ones I tried make CT fail to boot or hang randomly
lkcl has joined #linux-sunxi
<quitte_>
...or the nand XXXX.1k commands, rather
<libv>
i have a 1.2A nokia usb charger which suits most of my tablets
<libv>
but those load from the pc as well
<skoperst>
rz2k: that makes perfect sense for why my boards dont work, i have 1.5A chinese charger..
<bbrezillon>
quitte_: so the only partition without randomization (or with seeds set to 0) are the u-boot ones ?
<skoperst>
rz2k: Most of the time I also connect my PC to the board together with the power, still have some boot fails
<bbrezillon>
quitte_: well, actually it should work most of the time, but this is definitely not recommended (and the bitflips you're seeing when you try to load a big kernel, could possibly be fixed with randomization, even if read_retry will most likely fix those)
<quitte_>
however I'm not at all happy with that scheme. spl and u-boot should be in the first erase block. a env partition is fine, i guess. but i want dtb and the kernel integrated with ubinize somehow
<libv>
i have spent a lot of time turning those devices pages around, but they are about as far as i can get them without the hw in hand
<MY123>
Currently, will be unavailable for 30min (Freeblob for VideoCore)
<bbrezillon>
quitte_: that's definitelty a good approach (having the spl and u-boot in the same patition, and the kernel and dtb in the same UBI volume)
kuldeepdhaka has quit [Read error: Connection reset by peer]
<quitte_>
bbrezillon: now that openwrt has a rw overlay I'm getting close to providing a demo where all the pieces are in one place. hopefully this attracts people to working on u-boot
<bbrezillon>
quitte_: by working on u-boot, you mean porting my driver ?
<quitte_>
yes
<bbrezillon>
quitte_: I though rgwan and you planned to do so ?
<quitte_>
bbrezillon: for me this escalated and I'm waiting for a devkit for different hardware that i was actually planning to work on. rgwan is as far as i can tell a highschool kid. and the language barrier makes it near impossible to judge him properly. so i won't rely on his work.
<bbrezillon>
quitte_: then I hope ddc is stil interested in this project :-)
<quitte_>
I hope so, too.
<deepe>
libv: Using linux-sinxi 3.4
<deepe>
with 3.4.90+ kernel
<bbrezillon>
quitte_: I have to go (again :-)), thanks for helping petrosagg
<deepe>
libv: And yes I set console=tty0 as kernel command argument
<mnemoc>
that reminds me I need to merge stage-3.4
<quitte>
wigyori: If you could have a look and give some tipps that help increasing the chances of that at some point being used upstream - that would be great.
akaizen has joined #linux-sunxi
<rz2k>
libv: problem is, I dont have that hw in hand :c
<rz2k>
a710 died long time ago and q88 was given away half a year ago because I barely used it
<rz2k>
libv: I saw your comment on the benchmarks page and it needs to be rewritten indeed. I smashed random pieces of it just only be the order I got the results. I might get there soon and form some sort of table structure.
<rz2k>
not sure how that would look like, but it needs a table instead of unending list
astr has quit [Ping timeout: 240 seconds]
skoperst has quit [Quit: Page closed]
akaizen has quit [Remote host closed the connection]
lkcl has quit [Ping timeout: 240 seconds]
akaizen has joined #linux-sunxi
rz2k has quit []
akaizen has quit [Ping timeout: 240 seconds]
astr has joined #linux-sunxi
<MY123>
libv: Viewed the new photo?
lkcl has joined #linux-sunxi
lkcl has quit [Ping timeout: 245 seconds]
<petrosagg>
mnemoc: libv: back. I confirmed my email and I can create the MTD_Driver page, thanks!
<ptitnuage>
i immagine something is going worng at the early stage of boot process, but i don't know witch one. I would like to know at witch stage of the boot process does the screen might normally ligth on: SPL? Uboot?
<ptitnuage>
or can it be normal that the screen remains black?
<ptitnuage>
my other question is: "can my problem have something to deal with the quality of the Sd card i'm using" (do i have to use a class 10 sd crad for exemple, or can any kind of sd card (even a crappy chinese one) might boot fine
<ptitnuage>
if someone could help me with this issue, i will be very kind (sorry for my poor english)
rainbyte has joined #linux-sunxi
<quitte>
ptitnuage: i had effectively the same symptoms with one sd card. it's unlikely but if you have others readily available it's worth a try.
<quitte>
ptitnuage: however the most helpful thing you can do is attach a serial console
<ptitnuage>
quitte: thanks for answer
<ptitnuage>
quitte: i think i'll try to get an uart usb dongle (i don't own one now)
<Turl>
ptitnuage: the screen is light pretty late, the kernel does it
<Turl>
there's a lot of things that can go wrong until that happens
<ptitnuage>
quitte: another question: if the spl and u-boot are loading corectly, should the screen light on) in others words: at witch stage of the boot, may the screen light on?
FreezingCold has quit [Ping timeout: 245 seconds]
<libv>
ptitnuage: which q8 board?
<ptitnuage>
libv: it's a point of view mobii 700 tablet
<libv>
703?
<libv>
ok, i have the exact tablet
<libv>
so it won't boot properly?
<ptitnuage>
libv: sorry, 703. I built u-boot using "POV_TAB_P703" target.
<libv>
the sd card is not a likely cause
<libv>
ptitnuage: does the backlight come on
<libv>
did you add console=tty0 to the kernel commandline?
<ptitnuage>
libv: the backlight doesn't come on...
oliv3r has joined #linux-sunxi
jemk has quit [Quit: leaving]
<libv>
ptitnuage: hrm.
<libv>
ptitnuage: does android still boot?
<ptitnuage>
libv: no problem with android (booting from the nand)
<libv>
ok
<libv>
so the hw is ok
<libv>
good, let me throw u-boot and script.bin onto my sun5i sdcard and go find out for myself
<MY123>
paulk-aldrin: In which one?
<paulk-aldrin>
MY123, front picture
<libv>
MY123: i will edit these pictures myself later on, as they currently really are not too useful
<libv>
MY123: the low quality of the pictures apparently cannot be helped with your tablet camera
<libv>
deepe: sorry, supper and things...
<libv>
deepe: which distribution are you using?
<ptitnuage>
liv: my boot.cmd file states the following: setenv bootargs console=ttyS0,115200 root=/dev/mmcblk0p2 rootwait panic=10 ${extra}
<libv>
ptitnuage: add another console=tty0 there
<libv>
ttyS0 will not be useful as you have no uart.
lauri has quit [Remote host closed the connection]
<ptitnuage>
libv: i have to add it after ${extra} ?
<libv>
ptitnuage: just change console=ttyS0,115200 to console=tty0
<ptitnuage>
ok, and next i need to run again the mkimage tool on boot.cmd, right?
<libv>
yup
<ptitnuage>
libv: i give it a try now and tell you
<ptitnuage>
libv: thanks for helping
<MY123>
paulk-aldrin: The photo was took near the door.
<MY123>
( my child was playing a game)
<paulk-aldrin>
maybe try taking another one in natural light
<paulk-aldrin>
maybe you don't have a good camera available though
<libv>
heh, i still have to dig out libnand
<libv>
MY123: yeah, the camera is at fault
<libv>
paulk-aldrin: even
<libv>
paulk-aldrin: its a camera from another tablet
<paulk-aldrin>
libv, I may just copy the nand settings from tzx-q8-713b as it seems to boot fine when I tried with its script.bin on the ampe a76
<paulk-aldrin>
libv, ouch, too bad
<libv>
with poor camera software to boot
FreezingCold has joined #linux-sunxi
<libv>
paulk-aldrin: so the script.bin ram settings from the other tablet are fine on the ampe?
<paulk-aldrin>
that too
<libv>
err
<libv>
nand!
<libv>
nandsettings
<libv>
pfff
<paulk-aldrin>
yes
<paulk-aldrin>
it didn't crash
<paulk-aldrin>
with the other tablet's script.bin
<libv>
and the nand is accessible?
<paulk-aldrin>
not sure it actually worked
<libv>
complete script.bin?
<libv>
hrm
<libv>
that's living dangerously
<paulk-aldrin>
yes complete script.bin :)
<paulk-aldrin>
I know, I'm a moron
<libv>
ah, you made an error
<paulk-aldrin>
more or less
<libv>
well, at least you learned something new :)
<paulk-aldrin>
yep
<MY123>
libv: Should I give you the script.fex ?
<libv>
MY123: follow the howto
<MY123>
libv: Followed it but the SD does not boot, (adapted u-boot myself). I can't bother flashing the NAND.
<ptitnuage>
libv: ouch, still not working
<libv>
both of you, you must've done something wrong
<libv>
try running through it again
<libv>
be very exact about everything
<libv>
ptitnuage: yours i can test
<ptitnuage>
libv: if you have the same tablet than mine, and if you have any working linux sd card image , if you could send it so i can test and compare, i would be very kind
<ptitnuage>
libv: do you mean you wand me to send and image of my sd card?
<MY123>
libv: About the NAND, I lost the factory image. Before, I tested the A20-Olinuxino-Micro images, they worked without touchscreen and WiFi(attached a mouse with USB-OTG).
<ptitnuage>
libv: *sorry, mistyping: do you mean you want me to send to you an image of my sd card?
<MY123>
( Android )
<libv>
ptitnuage: no
<libv>
MY123: and you cannot ask the manufacturer for a livesuit image?
<libv>
MY123: why are you worried about that warranty again?
<MY123>
libv: It was the 120a update beta-test . It stalled so they gave me a Livesuit image to recover.
<MY123>
Wait. Have stored it.
<MY123>
( in a backup)
<libv>
MY123: so what is your goal here in #linux-sunxi?
<libv>
iirc, you said earlier that your display wouldn't come up anymore?
<MY123>
There is (or was) a PhoenixSuit IMG inside.
<MY123>
libv: Still here ?
<libv>
hrm, time for me to attach a soldering iron
<libv>
MY123: your android works, you probably did something wrong with the setup of the sd-card
<libv>
i too am not getting anywhere with this a13 tablet with the kernel that is installed on it, so i will now go find out what that is about
<MY123>
libv: It contained the result of the new device guide.
<paulk-aldrin>
libv, btw, when using CONS_INDEX=2, should I get getty to spawn on ttyS1 ?
<libv>
paulk-aldrin: no, CONS_INDEX is only for uboot
<libv>
MY123: of the full manual build howto?
<MY123>
libv: Yeah.
<paulk-aldrin>
libv, so somehow, u-boot routes linux's ttyS0 to UART1?
ptitnuage has joined #linux-sunxi
<ptitnuage>
(sorry,, got dsiconnected)
<libv>
paulk-aldrin: linux has its own settings for that
<libv>
which for early printk is in the kernel config, and for the proper uart driver is in script.bin
<paulk-aldrin>
libv, well, the thing is that I don't get any prompt, nor does my Android recovery kernel provides any, while it does on A20 where it uses UART0
<ptitnuage>
the point that keeps not clear to me is to know at witch stage of the boot shoud the backlight light on and the screen start to show anything: spl? u-boot?, or later stage?
<MY123>
libv: Did add the android identification into the page.
<libv>
cool, thanks
<libv>
ptitnuage: the backlight should come on as part of the kernel loading
<libv>
paulk-aldrin: you'll have to try around for yourself
<paulk-aldrin>
yep
<paulk-aldrin>
I have got to say, having a tty on the lcd is freaking cool
<MY123>
libv: What should I do ?
<ptitnuage>
libv: thanks. my boot partition have the following files in it: boot.cmdboot.scr script.bin uImage
<ptitnuage>
libv: does it seem correct?
<libv>
again, verify that all the steps that you took from the manual build howto are correct
<libv>
both of you
<MY123>
libv: I verified and did it 3 times.
<MY123>
(as the A20-Olinuxino-Micro 10'' LCD NAND images boots (without touchscreen), it is an A20)
<libv>
MY123: how did you get display?
<ptitnuage>
if someone could send mr the firts megabytes (includign boot partition with kernel) of an sd card working for Point Of View P703 tablet (Q8 board), i may be very nice...
<MY123>
libv: The A20-olinuxino-micro 10'' LCD NAND images boots without modification before PhoenixSuit.
<MY123>
(I reflashed the Kurio ROM after try)
<MY123>
libv: A recommendation?
<wigyori>
quitte: i'm waiting for a 3.16 patchset into openwrt, once done i'll update the sunxi target
<ptitnuage>
libv: if i understand, the kernel is not booting on my tablet.. i a buy an uart, can it help, or is it too early stage to see anything on the uart?
<wigyori>
quitte: i'll take a look this week at your tree - thanks for noting
<quitte>
wigyori: I'd like to automate the ubinizing of the image. but I'm at a total loss how that is supposed to be done. do you think you can give me some hints?
<MY123>
ptitnuage: It MAY be useful.
<wigyori>
quitte: i'm not at all familiar with ubi - please ask on the -devel mailing list
<quitte>
okay thanks
libcg has joined #linux-sunxi
lkcl has joined #linux-sunxi
PulkoMandy has quit [Quit: Vision[0.9.7-H-20140108]: i've been blurred!]
<paulk-aldrin>
libv, bingo! UART1 was neither enabled nor mapped as ttyS0
<paulk-aldrin>
libv, thanks a lot for the hints
<MY123>
libv: Here ?
<paulk-aldrin>
libv, should I send the script.fex patch to the mailing list for inclusion in sunxi-boards?
<libv>
MY123: that is sooo dangerous what you did there.
<MY123>
Where I mean without touchscreen, I see on the screen but touch doesn't work.
<MY123>
I reflashed the original ROM so no problem.
<Turl>
libv: what's life without the danger? :p
<libv>
paulk-aldrin: yeah, i will take it from there
<MY123>
libv:No danger.
<libv>
MY123: driving a 7" lcd with the data from a 10" one?
<libv>
MY123: script.bin contains loads of important hw settings
<MY123>
libv: I tried the 10'' because it has 1024x600.
<MY123>
The 7'' only have 800x480.
<libv>
MY123: you were still very very lucky
<libv>
MY123: so what about the backlight and the actual panel power?
<libv>
MY123: how are those driven?
<libv>
MY123: and yet you are afraid to open your tablet to get uart?
<MY123>
libv: It worked with the help of God. ( the settings were the same).
<libv>
heh, i am not religious. i check.
lkcl has quit [Ping timeout: 250 seconds]
<MY123>
libv: I'm still afraid and all my converters are 5V level
<libv>
oh, now hno butts in
<libv>
if that "let's ignore the GPL" bullshit goes on, i will pull the emergency break on all this.
<libv>
and seriously scare allwinner away
<Turl>
libv: for whose benefit?
<libv>
Turl: i dunno, i am just sick of this.
<MY123>
libv: When BCM release docs and code in BSD , Allwinner try to make things propretary.
<Turl>
while I do not agree with the 'conveniently ignore gpl violations', I do think directionless ranting will get us anywhere
<Turl>
will not get us*
<libv>
Turl: yet everyone seems to want to do so
<libv>
on a whim
<libv>
for 1 email and 2 email addresses!
<libv>
which mean _nothing_
yann_s is now known as yann_s|AFK
<Turl>
I'd rather have allwinner enabling floss support if it means leaving some violations unresolved
<libv>
Turl: who says that they will?
<Turl>
(ie, something like start releasing full sdks with no blobs by themselves from now on)
<mnemoc>
the libnand violation is very harmful
<Turl>
nobody, but it's too early to tell in any case
<libv>
Turl: all we have now is a public interview with the marketing manager, 1 email from who knows, and 2 email addresses
<libv>
_nothing_ solid
<libv>
we had eva before
<libv>
and she vanished the second marketing thought they had ticked the open source box
<Turl>
libv: you had a marketing contact, not much help
* Turl
goes back to writing unrelated emails
<libv>
whatever triggered this, pressure from linaro on making them look bad and/or pressure from the competition as ssvb rightfully stated, that's why they are talking
<libv>
not because someone sent them an email
<libv>
if the pressure from either side falls away, allwinner is out
MY123_ has joined #linux-sunxi
<libv>
and docs are never the full story, neither is code, but code at least worked at some point
MY123 has quit [Ping timeout: 250 seconds]
<libv>
and that says the guy who got docs out of AMD
MY123_ has quit [Changing host]
MY123_ has joined #linux-sunxi
MY123_ is now known as MY123
<paulk-aldrin>
libv, looking at the 8250_sunxi code, it seems that serials cannot be routed one to the other (i.e. the base addresses for each tty id are hardcoded)
<libv>
paulk-aldrin: then hardcode another :p
<paulk-aldrin>
that's probably the right way to do it, but that's a pain in the neck
<paulk-aldrin>
I'll have to provide a different uEnv.txt for each board with the right console=
<libv>
paulk-aldrin: you can do it properly once you have identified the right tty
<MY123>
libv: Now online. I will go to sleep. Will come back tomorrow.
<paulk-aldrin>
(thankfully, Android puts its uart shell where the console cmdline arguments tells it to)
ninolein has quit [Ping timeout: 260 seconds]
ninolein has joined #linux-sunxi
<ptitnuage>
go to bed
<ptitnuage>
*i go to bed
<ptitnuage>
thanks everyone for answers
ptitnuage has quit [Quit: Leaving.]
lkcl has joined #linux-sunxi
libcg has quit [Ping timeout: 255 seconds]
ricardocrudo has quit [Remote host closed the connection]
bengal has joined #linux-sunxi
<petrosagg>
bbrezillon: quitte: if a page is the minimum IO unit of a NAND then why do we have erase blocks? When we say "writing a page" do we only mean changing 1s to 0s?
<petrosagg>
if I want to flip a bit from 0 to 1 I have to erase the whole erase block, whereas for the inverse I can only write the specific page?
gaby has quit [Read error: Connection reset by peer]
gaby has joined #linux-sunxi
<bbrezillon>
petrosagg: yep, pages are the minimum unit for write/read operations (this is not exactly true as some NAND support read/write on subpages)
bertrik has quit [Read error: Connection reset by peer]
<bbrezillon>
petrosagg: blocks are the min unit for erase operations
<Turl>
bbrezillon: what's the difference between write and erase?
<bbrezillon>
petrosagg: and yes, erasing restore all bits in a given block to 1, while writing is here to set either 0 or 1 in each bit of a given page
<Turl>
(ie, what's erase's reason to exist?)
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Changing host]
Black_Horseman has joined #linux-sunxi
rafaelMOD has quit [Quit: Leaving]
<petrosagg>
Turl: write can only turn bits from 1 to 0
<petrosagg>
Turl: to flip a bit from 0 to 1 you have to erase the whole erase block and rewrite the pages with the bit you want flipped
<Turl>
that sounds slow
<petrosagg>
Turl: also write operations operate on pages, which are smaller than an erase block
<Turl>
so if you want to write almost anything you need to do an erase
<petrosagg>
Turl: You can do some clever things to avoid erasing and doing it in the background
<petrosagg>
Turl: I think this is what UBI achieves
<petrosagg>
Turl: this is a good presentation about how UBIFS handles writing to the NAND
schopen has joined #linux-sunxi
<petrosagg>
bbrezillon: the oob part of a page is not anything special, right? I mean in theory we could draw the data/oob line anywhere we want or not have an oob area at all and use the whole page for data, right?
<bbrezillon>
Turl: I'm definitely not a expert in electronic, but AFAICT the page and block concepts are tightly linked to the internal design of NAND flashes
<bbrezillon>
petrosagg: yes
<bbrezillon>
petrosagg: OOB bytes are just extra bytes in a page (often used to stored ECC bytes, but not necessarly)
<petrosagg>
bbrezillon: cool, making sure I don't write anything stupid in the wiki
<Turl>
petrosagg: nice slides
<paulk-aldrin>
grr, my message must have gotten stuck in moderation again
<libv>
the aliexpress seller marked the device i ordered as no longer available
<libv>
but still is processing my order like nothing ever happened.
<mnemoc>
libv: it happens from time to time to me as well
<mnemoc>
libv: seller loses it's supplier or access to the "right price"
<mnemoc>
libv: they still solve the problem for me, but they won't risk to lose money selling more units
<libv>
do you get to give negative feedback if your order gets cancelled?
<libv>
oh, ok, so this gets resolved usually?
<mnemoc>
in my experience, yes
<mnemoc>
to not lose the face
<libv>
ok
<bbrezillon>
Turl: this does not exactly answers your question, but at least it explains why there is 2 different levels (one for erasing and the other one for writing)
<libv>
mnemoc: does it take 7d to resolve the matter in such a case?
<mnemoc>
libv: they could ship tomorrow if they got enough pieces... depends on how much ali gives them to ship
<mnemoc>
if you get a tracking number before it expires, they are still trying to send it
<mnemoc>
if not you can be contacted to avoid a dispute
<libv>
it's been 7d of "awaiting shipment"
<petrosagg>
bbrezillon: When an erase block gets bad this information is stored by setting some specific bits in the OOB area to 1 (or 0)?
<mnemoc>
libv: ali should tell you how long do they have ship (a count down)
<libv>
yeah, 2d left
<libv>
mnemoc: is that more often used to the max?
<mnemoc>
when sellers actually have stock they say they'll ship in 2-3 days max. if yours announced a week, he didn't have any stock
<libv>
this guy had 7d "processing time" listed
<mnemoc>
but collecting preorders for a bulk buy
<petrosagg>
bbrezillon: and if yes, are these bits defined by the manufacturer? are they specified by the MTD subsystem? and how do we differentiate them from valid ECC data?
<libv>
ah, ok
<libv>
but the item is in the meantime listed as no longer available, so bulk buy seems out
<libv>
there was only my order...
<libv>
mnemoc: have you had orders cancelled on you before?
<mnemoc>
yes
<mnemoc>
but only 2 out of over 100 :p
<libv>
did you get to give feedback in such cases?
<bbrezillon>
petrosagg: what you're refering to is called a bad block marker
<libv>
ok
<mnemoc>
usually 3
<mnemoc>
which is already punishment
<mnemoc>
we westerns put the "good enough" in the middle, they do at the max
<mnemoc>
like with grades at school... if they don't get the max they did something wrong
<bbrezillon>
petrosagg: and a bad block is often marked bad by writing 0x00 on the first 2 bytes of the OOB of the first, last or the first two pages of a block
<bbrezillon>
petrosagg: and the pages chosen to mark BB are manufacturer specific
<libv>
mnemoc: but aliexpress still allows you to give feedback if an order is cancelled?
<mnemoc>
yes
<libv>
ok, this was not clear to me at all :)
<libv>
also, this thing was in portugese all the time for me
<libv>
i hope that removing cookies fixed that now
<petrosagg>
bbrezillon: how are the bad blocks handled after the initial read of the bad block marker of the manufacturer? for example if a block goes bad afterwards?
<petrosagg>
bbrezillon: is this something the MTD driver has to implement?
<mnemoc>
i wonder if I should be ashamed or proud that my orders page in aliexpress is 26 pages long :\
<libv>
not sure how it got there though, probably a link you or Turl posted years ago
<libv>
mnemoc: so you get all of that shipped to berlin?
<libv>
what does customs stop these days, and what does it let through?
<mnemoc>
most of it was to spain
<libv>
ah, right
<mnemoc>
spain doesn't care at all
<libv>
we talked about "ze german vay" before :)
<mnemoc>
they seem to stop packages randomly... but specially by size
<petrosagg>
bbrezillon: I also read that the bad block table is stored in a specific part of the NAND that is supposed to be reliable
<mnemoc>
once in the zoll thing I show the aliexpress order page, if it's under 22€ (converted by them) they let me go
akaizen has joined #linux-sunxi
<petrosagg>
bbrezillon: is this again something that the MTD driver has to implement? I guess not all NANDs have this
<libv>
mnemoc: so how is this now that you are in berlin?
<mnemoc>
sometimes they check if the brands are fake or if the CE thing is there
<libv>
and if it is fake?
<libv>
have you had stuff confiscated so far?
<mnemoc>
i haven't been in that case, I always buy chinese brands
<libv>
ok :)
<bbrezillon>
petrosagg: regarding factory bad block detection this is not related to the NAND controller driver (which actually what NAND drivers are refering to)
<mnemoc>
once the guy was picky about a "swissgear" backpack
<mnemoc>
because it had the swiss flag
<mnemoc>
but it said it was chinese everywhere, no attempt to claim .ch origin
<bbrezillon>
petrosagg: the BB marker position is chosen according to the NAND id retrieved during NAND discovery
<mnemoc>
I believe that they would just confiscate it if they decide it's counterfeit
naobsd has quit [Quit: Page closed]
paz has quit [Quit: No Ping reply in 180 seconds.]
paz has joined #linux-sunxi
<quitte>
what is the significance of badblocks that the manufacturer marked bad? why is everyone so worried about deleting those markers?
<bbrezillon>
petrosagg: regarding on flash BBT, this depends on the NAND chip driver, but if the controller does not support any specific format for the BBT, the nand core code provides a default BBT format, and store the BBT in the last two blocks of the NAND (one primary BBT and one mirror)
<mnemoc>
people in the berlin-schoeneberg zollamt thingy is pretty nice
<quitte>
...the last two non-bad blocks according to the above wiki
<petrosagg>
bbrezillon: hm, it seems like there are two kinds of drivers. NAND controller drivers and NAND chip drivers
<petrosagg>
bbrezillon: Your MTD driver is about the NAND controller that is in the A20 chip?
<bbrezillon>
quitte: factory BB are those marked bad by the NAND chip manufacturer before the NAND leaves its factory (and this is based on some specific tests)
<petrosagg>
bbrezillon,quitte: but I guess it's not the end of the world if you erase them as you can to a simple "erase block & check" pass to detect them
<bbrezillon>
petrosagg: there no NAND chip drivers for now
<quitte>
why would i care if it is factory bad or worn out?
<quitte>
petrosagg: the world has not ended
<Turl>
libv: no portuguese here :)
<petrosagg>
quitte: I confirm that, the world exists here too
<bbrezillon>
quitte: because on factory BB you're not even sure that you can mark a block as bad after it has been erased (this is what we've experienced with ddc)
<quitte>
bbrezillon: on worn bad you can ?
<petrosagg>
bbrezillon: but can't you just store that it is bad in the BBT instead of marking the block itself?
<libv>
Turl: or spanish, i dunno what it was
<Turl>
if you see ç it's portuguese
<bbrezillon>
quitte: I've always been able to do so :-)
<mnemoc>
Turl: or catalan
<libv>
it's gone now :)
<libv>
no matter how often i changed languages, it usually would revert to $language
<libv>
so clearing cookies is the secret
<bbrezillon>
petrosagg: yes, that's exactly what we've done, but still, erasing bad blocks is dangerous, because the BBT is created from these BB markers
<libv>
s/secret/key/
<Turl>
mnemoc: the chances a website is in catalan is low
<Turl>
s/is/are
<mnemoc>
they have their own tld :) .cat
<quitte>
does erasing multiple times in a row count as erasing once in terms of life-span or does any erase wear the chip out, no matter if it changes the state or doesn't?
* Turl
tries lol.cat
<Turl>
heh, it's actually a thing
<bbrezillon>
quitte: I'd say the latter
<petrosagg>
bbrezillon: I'm a bit confused. Let's say that I try to erase a bad block. Won't I detect that some of the bits in the block are still 0? Why is this dangerous?
paulk-aldrin has quit [Remote host closed the connection]
<Turl>
in any case, you're better off using the sites in english
<Turl>
on dx at least, the product translations are so bad they're funny
<libv>
:)
<petrosagg>
bbrezillon: I've certainly ran `flash_eraseall` on my mtd partitions. Does this mean that I erased the OOB area and thus the factory BBMs as well?
<petrosagg>
bbrezillon: or is this area protected by the MTD subsystem?
petr has quit [Ping timeout: 260 seconds]
petr has joined #linux-sunxi
<libv>
petrosagg: i have fixed up the header stuff in that page of yours
<petrosagg>
libv: I thought the '= =' header was only meant for the wiki title
<petrosagg>
libv: that's how wikipedia doesit
<petrosagg>
libv: is there a reason not to follow their recommendation?
<libv>
petrosagg: check wikipedia again
<libv>
they use no single = though
<libv>
but we do
<libv>
the title is the name of the page anyway
<petrosagg>
libv: yes, I was referring to the use of single =
<petrosagg>
libv: should I start my headers with single =?
<petrosagg>
libv: refresh
<petrosagg>
libv: this is my first draft, I need to add more but it's getting late here
<petrosagg>
quitte: Was the reason you said it's good to have u-boot and spl in the first 1k of the NAND that the first 1k is supposed to be reliable?
bengal has quit [Ping timeout: 240 seconds]
<quitte>
petrosagg: I said within the first erase block. so within the first 2MiB. The reason is mostly that I don't like the clutter of all those partitions and 2MiB is plenty of space. I don't see a reason not to update spl and u-boot as a pair,too.
<petrosagg>
quitte: oh, ok. Because I read somewhere (I don't remember where) that some erase blocks are more reliable than others
FreezingCold has quit [Ping timeout: 245 seconds]
<quitte>
on the openmoko wiki?
<quitte>
they are referring to a different kind of flash, though
<petrosagg>
quitte: But I remember that you also said that having an xz'd kernel is a good idea to have a smaller chance of bit flips
<petrosagg>
I really don't remember :/
<quitte>
that was by experience. a kernel bigger than 2,4MiB didn't work once with yuq u-boot
<quitte>
so that is not something to document but to fix
<quitte>
from how nand read felt in u-boot i'm fairly certain that nand outperforms kernel decompression a lot. So what I want is an uncompressed kernel