<apritzel>
MasterR3C0RD: trying to merge and polish your A133 DRAM code (from the allwinner-a133 branch in your github), and scratching my head about some small details
<apritzel>
don't know how to comment on just a file in github, so trying this here:
<apritzel>
in mctl_auto_detect_dram_size(), the range of rows and cols seems to be off by 1 when compared to jernej's latest code (now in mainline's dram_dw_helpers.c): is this expected?
<apritzel>
also in the same function, the alleged loop over the banks is not really a loop, isn't it? for (banks = 2; banks < 3; banks++)
<apritzel>
and in auto_detect_ranks(), found_config doesn't seem to be used?
<MasterR3C0RD>
apritzel: Don't remember all the details off the top of my head, but I probably based rows/cols ranges off whatever I saw in boot0, and the ranks loop was me trying to be "consistent" for my own understanding; likely would've been cleaned up later
<apritzel>
ok, thanks, doing those clean ups now, but was wondering if I was missing something
<MasterR3C0RD>
Not sure if bank groups are handled by the mainline helpers yet (I'll take a look later), so probably won't be able to shift it over completely
<MasterR3C0RD>
And as for auto_detect_ranks, might've been a leftover from refactoring
<apritzel>
also trying to adapt mctl_auto_detect_dram_size() to the latest mainline changes, to at least come close to the generic helper, as indeed that's missing some code
<apritzel>
so using this larger pattern instead of the simple check, as I now suddenly see our infamous double detection again
warpme has joined #linux-sunxi
<MasterR3C0RD>
I'm sure you noticed it already, but will need to pull out the simple test, as I had pulled that and modified it for debugging from somewhere else
<apritzel>
MasterR3C0RD: and is this correct: it configures say config->cols to 12, but then loops only over 8 to 11?
<MasterR3C0RD>
Apologies for not being able to work on upstreaming the last few months; life likes to keep happening :(
<MasterR3C0RD>
Ah now that I'm looking at the code, I likely messed up on both of those and they should be <=
<apritzel>
don't worry, I am always happy if people have a real life and don't need to spend their time on pathetic SoCs ;-)
<MasterR3C0RD>
Oh, actually I think I was a little more clever than I should have been
<apritzel>
though jernej's code does the same, wondering if this is a bug or a feature, as in: if the but-last doesn't match, it must be the last, no need for testing?
<MasterR3C0RD>
Essentially if no other configuration is found, loop ends at 12
<apritzel>
right
<MasterR3C0RD>
Yeah and that would be the same for the banks for loop, if it doesn't succeed with banks = 2, it will exit with banks = 3
<apritzel>
ah
<MasterR3C0RD>
I wonder what the compiler turns that into
<apritzel>
btw: I made some adjustments to mainline (struct sunxi_ccm_reg and sunxi_prcm_reg are gone now), will put my changes somewhere ASAP, so you don't have to redo this
warpme has quit [Ping timeout: 480 seconds]
<MasterR3C0RD>
apritzel: Oh, and if you want me to sign off anything from the later commits, just let me know. Won't be able to do it tonight or tomorrow, but should have a chance Sunday
<apritzel>
MasterR3C0RD: yes, please, that's crucial for mainline. It would be as easy as to just reply to the patch of yours I posted, and add your signature
<spawacz>
The thermal sensor works! Bluetooth however does not - I loaded hci_uart module, the configs CONFIG_BT_HCIUART=m and CONFIG_BT_HCIUART_BCM=y are set
chewitt has quit [Quit: Zzz..]
<electricworry>
exit
<electricworry>
I've been away from here for a while. Of the top of your heads, do you know the status of graphics on H616/H618? A few months back there were some HDMI patches pending in tokyovigilante's fork, IIRC.
<spawacz>
Oh bluetooth works as well - I forgot to set the HCIUART_RTL config