<abrasive>
the ATAG that the VPU driver uses is, as far as I know, the board revision
<abrasive>
which we report as one of the MX6Q types
<ajayr>
we hard code get_board_rev to 0x63011 which means that the last 1 is cpu rev 1, from what I can see the rev 1 was in the initial lot of GK802 sold in Jan but the later lots in mar have v 2
<abrasive>
it is a dirty, dirty hack on the driver's part.
<abrasive>
pretty sure the kernel reads the version independently
<ajayr>
i can see y cpu is a type 2 so ist that one of the issues for not seeing hdmi edid?
<abrasive>
if you look at the start of the boot messages it tells you the correct chip revision
<abrasive>
it certainly looks to be revision related, but is there any code in the tree that acts differently on different cpu revisions?
<abrasive>
you could always try changing the get_board_rev value to the -2 revision and then see if that fixes anything.
<ajayr>
I dont know ... I am still looking at why my LG TV LM6200 pust out EDID of 00
<abrasive>
it's not just you :)
<abrasive>
a lot of newer revision GK802 have that problem
<ajayr>
that is a small consolation
<abrasive>
the board itself has a different revision number
<abrasive>
well, yes, but it means that only one person has to find the cause and a lot of people will be very happy.
<ajayr>
my inital project for fedora fell flat since I cannot find a x 1.11 in F15 and F17 has 1.13 so I am waiting for yocto to work
<abrasive>
me too.
<ajayr>
till then the sticks are in a draw waiting for some love!
<abrasive>
i have a lovely arch instgall but with no VPU
<abrasive>
well, if you have a bus analyser or oscilloscope, and a bit of time... it'd be great if you could figure out the i2c issue ;)
<ajayr>
unfortunately no 'scope and stuff for me at home ... havent done hardware for 10 years now, I doubt if my eyes and hands are steady enoght to to smt rework
<abrasive>
i wish i had a new revision gk802 around to fix that problem
<abrasive>
it's been bugging me.
<ajayr>
i am planning that once i get the new u-boot working I may see what upstream linux wit DT looks like on iMX6 after all gpu vpu does not work for me and I dont even need it - that may be a fun way to tear some hair out!
<abrasive>
hahaha
<abrasive>
i like your definition of fun.
<abrasive>
i need to get that uboot patchset merged upstream
<abrasive>
so many things to do.
<ajayr>
well best of luck - its best to wear your best flame retardant suit before you post on the u-boot list, reviews a quite harsh there
<abrasive>
yeah, i don't doubt it
<abrasive>
most of the gk802 stuff is similar to what is elsewhere in the tree
<abrasive>
but i expect to get a vicious beating over the recovery mode hook
<abrasive>
i'm considering tossing it just for that reason.
<ajayr>
hopefully I will have a test and log for you in the next 6 - 8 hrs, I should get some time to work with the stick in a bit and will try some options and report back
<ajayr>
bye for now
<abrasive>
thanks. take care!
<projectgus>
abrasive: i missed this before. so the problem with i2c/edid not reading on some TVs is probably hardware you think?
<projectgus>
and some older GK802s can read TVs that newer ones cannot?
<abrasive>
no, I don't know why it is
<abrasive>
but it's certainly board revision related
<abrasive>
suspect that i2c is just not being enabled right on some chip revisions
<abrasive>
needs more information.
rz2k has joined #imx6-dongle
<dgp>
abrasive: old version == the one with the can top package?
<abrasive>
can top?
<dgp>
there is a can version of the imx6
<dgp>
i thought that was the version in the old boards
Noskcaj has quit [Read error: Connection reset by peer]
<abrasive>
can as opposed to bare-backed flip die?
<abrasive>
mine is old and bare die
<dgp>
mmm I thought that as how you could tell which revision it was.. oh well.. anyhow I still have a new one with broken BT that I haven't depopulated yet
<dgp>
I could send you that if it would help
<abrasive>
that would help... it has broken edid too, right?
<dgp>
it works with my stuff but I've only tested it with one display
<dgp>
and I think I have the display mode hardcoded
fossxplorer has joined #imx6-dongle
fossxplorer has quit [Changing host]
fossxplorer has joined #imx6-dongle
<abrasive>
oh right
<abrasive>
if it reads the EDID as all zeroes then it's a candidate for debugging
<dgp>
I'll check that when I next boot it up
datagutt has joined #imx6-dongle
cnxsoft has quit [Remote host closed the connection]
MindBeat has joined #imx6-dongle
ajayr has quit [Remote host closed the connection]