dragonn has quit [Remote host closed the connection]
dragonn has joined #linux-sunxi
<dragonn>
Hi! How can I download (read) the complet nand to my PC by using the fel mode?
awafaa has quit [Ping timeout: 246 seconds]
<mripard>
Turl: hi
Yaku321 has joined #linux-sunxi
hramrach_ has quit [Remote host closed the connection]
hramrach_ has joined #linux-sunxi
mripard_ has joined #linux-sunxi
Liberato1 has joined #linux-sunxi
<Turl>
mripard_: I was reading the mails on the list and talking with oliv3r yesterday
<Turl>
mripard_: what's the main difference between of_iomap and the other proposed functions you guys were discussing?
<mripard_>
of_iomap parses the dt to get the resource, and then call ioremap on it
<mripard_>
while you should actually call request_mem_region between the two
<mripard_>
to make sure that no other mapping exist for the area in memory you are requesting
<mripard_>
ie. no other driver has already mapped it, and potentially use it.
<Turl>
mripard_: so that doesn't apply to my driver I'm guessing, as I actually need to share addresses at times
<mripard_>
indeed
<wingrime>
oliv3r:pinh
<wingrime>
ssvb: ping
dragonn has quit [*.net *.split]
BJfreeman has quit [*.net *.split]
mripard has quit [*.net *.split]
Liberator has quit [*.net *.split]
<Turl>
mripard_: so I implemented pll6 yesterday (still untested) and that got me wondering about compatible strings
Undertasker has joined #linux-sunxi
_BJfreeman has joined #linux-sunxi
<Turl>
pll1 and 4 share the same factors, offsets, etc
<Turl>
5 and 6 too, but different from 1 and 4
<Turl>
mripard_: would it be ok to use "...-pll1-clk" compatible for both 1 and 4, and "...pll5-clk" for 5 and 6?
_BJfreeman is now known as BJfreeman
Undertasker has quit [Ping timeout: 264 seconds]
<mripard_>
well, if they are compatible... :)
dragonn has joined #linux-sunxi
Yaku321 has quit []
ganbold___ has quit [Remote host closed the connection]
Undertasker has joined #linux-sunxi
<lkcl>
has anyone started on a mtd nand driver at all for A10?
<Turl>
lkcl: there's one on the ML
<Turl>
mripard_: is there any magic incantation to test a branch on all commits for build breakage?
<mripard_>
git bisect + a script that invoke make ?
<Turl>
no, not that kind of breakage
<Turl>
mripard_: I rebased *a lot* today moving stuff around and I want to test make after every commit to make sure I didn't move stuff that was needed to build
<mripard_>
ah
<lkcl>
Turl: ML? do you have a link at all?
<mripard_>
hmmm
<mripard_>
I'm not aware of anything that could do that
<mripard_>
Turl: if you just do make, it will build a zImage and all the dt
aseigo_home has joined #linux-sunxi
aseigo has quit [Read error: Connection reset by peer]
<Turl>
mripard_: my script arg parsing is broken and it's actually executing bare make heh :)
drachensun has quit [Quit: Leaving]
drachensun has joined #linux-sunxi
<mnemoc>
lkcl: beware it's very raw, so you will wear out the nand fast
drachensun has quit [Client Quit]
<lkcl>
mnemoc: that's the case for all mtd drivers, right?
<mnemoc>
lkcl: the rely on the emmc or implement the randomization. the mtd driver for sunxi doesn't do that or anything fancy yet]
<lkcl>
mnemoc: i thought randomisation was done by e.g. jffs
<mnemoc>
i thought the same, but no.
<mnemoc>
they assume the chip does it on it's own
<mnemoc>
and if not (nand), the mtd driver has to
<lkcl>
jooy!
notmart has quit [Ping timeout: 248 seconds]
wingrime_ has quit [Quit: Lost terminal]
n01_ has quit [Read error: Connection reset by peer]
n01 has joined #linux-sunxi
mab has joined #linux-sunxi
<Turl>
oliv3r: remember the weirdness you had on uboot with registers?
<hno>
lkcl & mnemoc situation is not that bad. Randomisation is done by the NAND controller.
<lkcl>
hno: whew :)
notmart has joined #linux-sunxi
notmart has quit [Changing host]
notmart has joined #linux-sunxi
Undertasker has quit [Read error: No route to host]
Undertasker has joined #linux-sunxi
<oliv3r>
Turl: yeah; what abou tit
<oliv3r>
hno: it's unmodified, except for the emac renaming patch that i've posted
<oliv3r>
mripard_: the thing is, first you told me and n01 to use one thing, then now, it turns out we have to use something else, but even turl doesn't understand why; also, there's no ldd version that includes DT things, ldd3 is all pre-dt
jemk has joined #linux-sunxi
<mripard_>
DT changes only very minor details. The whole logic stays there, and is described in LDD
mab has quit [Ping timeout: 240 seconds]
<mripard_>
and yes, I overlooked something, I'm just like every one here, I can overlook stuff or be wrong about others.
<mripard_>
If I told you this, it's because you raised some very valid questions, but also some basic ones, that are addressed, and should be known before doing kernel stuff. I didn't mean to be rude or anything.
<Turl>
oliv3r: well, that's why writel, if you use the others I bet weird stuff like that would happen too
<\\Mr_C\\>
i will repeat the process again and see what it does
<mnemoc>
you dumped that image over the full logical nand, not just the boot partition
<mnemoc>
don't know what it's intended to be used, but looks awful
<hno>
oliv3r, have you tried the i2c command I gave?
<\\Mr_C\\>
its in that link you just showed me
<\\Mr_C\\>
First, we need to get the boot partition we downloaded earlier on onto the NAND. You can either copy it over using SCP, then use dd to write it to /dev/nand (dd bs=4096 if=./cubie_nand_uboot_partition_image.bin of=/dev/nand), or just copy it directly there with SCP (on my Mac, scp ~/Downloads/cubie_nand_uboot_partition_image.bin root@<CUBIEBOARD IP ADDRESS>:/dev/nand).
<mnemoc>
ok
* mnemoc
doesn't do the image dumping dance
<\\Mr_C\\>
rebooting it like it said
<\\Mr_C\\>
moment...
<mnemoc>
if you want to repartition, repartition. don't dd unknown stuff...
<mnemoc>
awful
<\\Mr_C\\>
well thats from the sunxi site
<\\Mr_C\\>
and i see it in all the documations
<mnemoc>
it's a wiki, anyone can write whatever they want
<hno>
\\Mr_C\\, sounds like the partition table in that image is not what you want.
* mnemoc
shuts up and tried to fix the math extension
<\\Mr_C\\>
oh
<mnemoc>
s/tried/tries/
<hno>
u-boot supports loading uImage from partitions as well. I find that easier than dealing with Android boot images.
shineworld has joined #linux-sunxi
<\\Mr_C\\>
93 0 3989504 nand
<\\Mr_C\\>
93 2 3988992 nandb
<\\Mr_C\\>
93 1 508 nanda
<\\Mr_C\\>
thats what it created
<hno>
mnemoc, yes that cubie image is awful, trying to shrinkwrap it all in an binary image.
<hno>
\\Mr_C\\, yes, and not what you want I supose. But you can load kernel from /boot in nandb just fine if you want.
<\\Mr_C\\>
hmm
<hno>
or at least in theory. Not sure what the state of the NAND u-boot is really.
<oliv3r>
mnemoc: i think your wrong, i belive mtd simply talks directly to the flash(controller). IF the controller does Wear Leveling, good, you can use mtdblock. otherwise, you use jffs etc
<hno>
maybe doesn't support ext*, and very likely can't support ext4 even with a rebuild.
<hno>
oliv3r & mnemoc MTD is just an interface to the raw flash. It does not do anything fancy, just raw access. It is up to any upper layers to do bad block management, wear leveling etc.
<mnemoc>
but you told me we needed to add randomness and some other fanciness to the mtd driver...
<hno>
and there is not usable block device for MTD in the Linux kernel. To use MTD you MUST use an MTD aware filesystem.
<oliv3r>
Turl: ahh, you mean because write/read allignment
<oliv3r>
Turl: i took that into account
<hno>
mnemoc, I think you misunderstood me then. MLC flash do need randomization, but that is handled by the flash controller. But I have not verified if the posted MTD driver do it properly.
<lxsameer>
after about 4 days finally i finished cloning the linux-sunxi repo :D
<oliv3r>
mripard_: i didn't say you where rude, but I do read _alot_ but there's no 'easy to understand' reading matterial, most of the stuff is 'half' where you allready have to know most on how to use :)
<oliv3r>
mnemoc: it's new, i used CCM just last night