01:00 <MasterR3C0RD> Hmm, it appears the offset for the DRAM parameters in boot0 is different than expected. It gets a different set of DRAM parameters for DDR3 when the tool is run on it

01:13 <MasterR3C0RD> Yeah, there's sets of DRAM parameters at 0x038 (incorrect), 0x474 (correct), and 0x674 (incorrect). Interestingly, there are 3 different dram_para sections in the sys_config.fex as well: "dram_para", "dram_para3", and "dram_para7"

01:22 <MasterR3C0RD> Ahhhh looking at the SPL headers these are being stored in an additional "spare header"

02:41 <MasterR3C0RD> Looks like there's a method (sunxi_dram_handle) for selecting which boot mode is in use, and that changes which DRAM parameters are utilized. In the case of my device, the boot0 "extended header" states that selection is done by checking GPADC channel 1

05:27 <MasterR3C0RD> apritzel: Submitted a PR to sunxi-fw to add A133 and DRAM.ext support. They're combined in one PR, as I'm unsure whether or not this is a common theme on A133 devices, but it very well could be. https://github.com/apritzel/sunxi-fw/pull/6

05:29 <MasterR3C0RD> parthiban: If you could give this a try on the boot0 for your board, I'd be interested to see what results you get, especially if you get additional parameter sets in your results.

05:58 <Parthiban> MasterR3C0RD: Morning, am new to this tool. Should I just run extract?

05:59 <MasterR3C0RD> parthiban: Nope, you'd run `sunxi-fw info -v [boot0.bin | phoenixsuite-image.img]`

05:59 <MasterR3C0RD> Either the boot0 or a PhoenixSuite flash image will work

06:01 <Parthiban> not much info, I think am doing it wrong

06:01 <Parthiban> ./sunxi-fw info -v /tmp/pn/boot0_sdcard.bin

06:01 <Parthiban> @ 0: boot0: Allwinner boot0

06:01 <Parthiban> size: 4096 bytes

06:01 <Parthiban> eGON checksum: 0xa98d87f0, programmed: 0x5f0a6c39

06:01 <Parthiban> DRAM parameters: A64 H616 A133 H6

06:01 <Parthiban> running against the elf gives the same out

06:02 <Parthiban> got it

06:02 <Parthiban> ./sunxi-fw info -v /tmp/pn/boot0_sdcard.fex

06:02 <Parthiban> @ 0: boot0: Allwinner boot0

06:02 <Parthiban> eGON checksum matches: 0x7ae47132

06:02 <Parthiban> size: 61440 bytes

06:02 <Parthiban> DRAM parameters: A64 H616 A133 H6

06:02 <Parthiban> DRAM clock : 0x318 0x318 0x318 -

06:02 <Parthiban> DRAM type : 0x8 0x8 0x8 -

06:02 <Parthiban> ZQ value : 0x7070707 - - -

06:02 <Parthiban> ODT enabled : 0xd0d0d0d 0xd0a050c - -

06:02 <Parthiban> DX ODT : - 0x7070707 0x7070707 -

06:02 <Parthiban> DX DRI : - 0xd0d0d0d 0xd0d0d0d -

06:02 <Parthiban> CA DRI : - 0xe0e 0xe0e -

06:02 <Parthiban> PARA0 : - - 0xd0a050c -

06:02 <Parthiban> PARA1 : 0xe0e 0x30ea 0x30ea -

06:03 <Parthiban> PARA2 : 0xd0a050c 0x1000 0x1000 -

06:03 <Parthiban> MR0 : 0x30ea 0 0 -

06:03 <Parthiban> MR1 : 0x1000 0x34 0x34 -

06:03 <Parthiban> MR2 : 0 0x1b 0x1b -

06:03 <Parthiban> MR3 : 0x34 0x33 0x33 -

06:03 <Parthiban> MR4 : - 0x3 0x3 -

06:03 <Parthiban> MR11 : - 0x4 0x4 -

06:03 <Parthiban> MR12 : - 0x72 0x72 -

06:03 <Parthiban> MR14 : - 0x7 0x7 -

06:03 <Parthiban> MR22 : - 0x26 - -

06:03 <Parthiban> TPR0 : 0x1b 0x6060606 0 -

06:03 <Parthiban> TPR1 : 0x33 0x4040404 0x26 -

06:03 <Parthiban> TPR2 : 0x3 0x1f080403 0x6060606 -

06:03 <Parthiban> TPR3 : 0 0 0x4040404 -

06:03 <Parthiban> TPR4 : 0 - 0x1f080403 -

06:03 <Parthiban> TPR5 : 0x4 - 0 -

06:03 <Parthiban> TPR6 : 0x72 0x48010101 0x48010101 -

06:03 <Parthiban> TPR8 : 0x7 - - -

06:03 <Parthiban> TPR10 : 0 0x273333 0x273333 -

06:03 <Parthiban> TPR11 : 0x26 0x241f1923 0x241f1923 -

06:03 <Parthiban> TPR12 : 0x6060606 0x14151313 0x14151313 -

06:03 <Parthiban> TPR13 : 0x4040404 0x81d24 0x81d24 -

06:03 <Parthiban> TPR14 : - - 0x2023211f -

06:04 <MasterR3C0RD> Might've been a better idea to post that to a Pastebin HAHA

06:04 <Parthiban> Sorry, posting it

06:05 <Parthiban> https://pastebin.com/raw/vrJjALbp

06:07 <MasterR3C0RD> I can see yours doesn't have the additional DRAM headers; looks like it's isolated to my device. Good to know

06:07 <Parthiban> meaning, it's not lpddr4?

06:07 <MasterR3C0RD> s/my device/my device for the time being/

06:07 <Parthiban> From the schematics I can see it's lpddr4

06:08 <MasterR3C0RD> No, it's still LPDDR4; this is unrelated to that

06:09 <MasterR3C0RD> Essentially I found that on my device, it has additional DRAM parameters in an extended header stored within boot0

06:09 <MasterR3C0RD> I was thinking it might be more common on A133 devices

06:13 <MasterR3C0RD> Anyways, it's a bit late so I'll have to hop off, but you'll want to keep note of the parameters under the A133 header in that table; you'll need them later when I finish porting over the LPDDR4 timings (dashes mean they aren't present, you can just ignore those).

10:03 <apritzel> MasterR3C0RD: it's actually the same on the recovery image for my A133 tablet: the DRAM parameters at the beginning of boot0 (what sunxi-fw decodes) are for DDR3, and there are more sets of parameters

10:03 <apritzel> and it clearly selects type=7 in the boot0 log

10:07 <apritzel> MasterR3C0RD: thanks for the PR, will have a look later, and try it on one of my images

15:10 <Parthiban> MasterR3C0RD: apritzel: Managed to get the GPU driver functional in A133. here: https://pastebin.com/raw/vPcdQYbK

15:12 <MasterR3C0RD> Awesome! Have you tested running anything on it, i.e, starting up Xorg, glxgears using the latest Mesa, etc?

15:12 <MasterR3C0RD> Or are you still missing TCON?

15:13 <Parthiban> MasterR3C0RD: Yeah, still DE, TCON is not enabled yet. With no details in the manual, trying to figure it still.

15:14 <MasterR3C0RD> You could probably try using the TCON devicetree definitions for the H6/H616 with the A133's offsets

15:14 <Parthiban> ideal plan is to run glmark2. but haven't compiled the mesa 3D. Imagination is using 23.0 version of it, whereas scrathgap yocto of mine using 24.x. Need to patch some yocto stuff there

15:14 <Parthiban> Yeah, a64, H6, H616, this is all am checking.

15:16 <MasterR3C0RD> Note that the open-source PowerVR drivers only provide Vulkan, and not completely comforming Vulkan yet at that; OpenGL support is through Zink.

15:18 <Parthiban> Yeah, am aware. I want the DE + TCON to work first.

15:20 <Parthiban> Are there any on-going effort for G2D? Just curious. I did check this https://github.com/cheofusi/sunxi-g2d

15:29 <MasterR3C0RD> Acording to the status matrix on the wiki, no: https://linux-sunxi.org/Linux_mainlining_effort#Status_Matrix

15:32 <Parthiban> ok thanks

17:15 <apritzel> yeah, no one ever tried upstreaming G2D, I guess it's also somewhat questionable how useful this would be these days in the Linux context

17:16 <apritzel> parthiban: don't know if you have seen, but this is the latest series for H616 DE33: https://lore.kernel.org/linux-sunxi/cfc66e71-4af8-43c8-8b86-0bf5208b5c42@gmail.com/T/#r9627ea397a508b5a7c9856f5ae1dd39b5ab2d78f

17:17 <apritzel> there are separate documents for the display engine, for instance this one here: https://linux-sunxi.org/images/d/dd/Allwinner_DE3.0_Spec_V1.0.pdf

17:26 <MasterR3C0RD> apritzel: Based on the reference to "SmartColor 2.0" in the A133 documentation, it seems likely it might be the DE2.0: https://linux-sunxi.org/images/7/7b/Allwinner_DE2.0_Spec_V1.0.pdf

17:26 <MasterR3C0RD> parthiban: ^

17:27 <MasterR3C0RD> It may even be older, since the maximum supported size is 2048x2048, instead of 4096x4096

17:41 <MasterR3C0RD> Closest match I've found so far is the T133, but it's missing some of the features claimed by the A133 user manual

18:04 <MasterR3C0RD> Considering the reference to mixers, it might actually be DE2.0, more specifically the H5 version (writeback but no rotation)

19:21 <MasterR3C0RD> Yeah, looks like it's DE2.0 from a quick look at the Linux tree. Should mean the correct bindings may do the trick

20:43 <apritzel> MasterR3C0RD: don't read too much into features like screen size, this is sometimes artificially limited by the target market (tablets = no HDMI = no 4K)

20:44 <apritzel> but good work, I was hoping it being close to something already supported

20:47 <MasterR3C0RD> apritzel: It wasn't the screen size I had been looking at, but rather the output size from the DE; though perhaps those are related, I'm still getting a handle on the hardware blocks. In any case, the only other difference I could find is that there is only a single mixer mentioned in the user manual, rather than 2, and that the DE blocks are

20:47 <MasterR3C0RD> "independent", with DE1 being located outside the memory map.

20:48 <MasterR3C0RD> Likely a hack to disable DE1 so that only DE0 is present

20:51 <MasterR3C0RD> s/so that/since/