07:20 <krzk> GNUtoo: if you cannot change the pin also on other kernels, maybe the modem pulls them to specific state? Maybe it should be like that after initialization?

07:20 <krzk> GNUtoo: Anyway, I would not compare that match with v3.18 kernel but rather with the first, vendor kernel for this device.

07:22 <krzk> GNUtoo: That would be probably v3.0 or v3.4. In my archives I have for one S3 flavor - SCH-S968C: https://github.com/krzk/linux-vendor-backup/tree/samsung/galaxy-s3-sch-s968c-exynos4412-dump

07:24 <krzk> GNUtoo: although some sources are saying that SCH-S968C is with Qualcom Snapdragon, so maybe I was mistaken and it's not Exynos4412

07:25 <krzk> GNUtoo: For S3, I would also recommend to get the Tizen kernel for TRATS2 or RD-PQ. These should be equal to S3 and they had working modem, AFAIR.

07:32 <krzk> GNUtoo: damn, SCH-S968C is really qualcomm, I need to rename the branch

07:49 <krzk> GNUtoo: you can take a look at Mokee branch - I noted it consist code for Exynos 4412 and it is still a vendor kernel: mokee/android-3.0-samsung-galaxy-i777-i9100-i9300-n7000-exynos4210-exynos4412

13:39 <GNUtoo> Thanks, I used the smdk Replicant kernel which is based on LineageOS because it has been tested a lot it with libsamsung-ipc

13:39 <GNUtoo> Replicant users use it daily

13:40 <GNUtoo> I'll look at the other kernel as well

13:41 <GNUtoo> That LineageOS kernel was made from code dumps from opensource.samsung.com

13:41 <GNUtoo> Unfortunately theses are not archived anywhere

13:41 <GNUtoo> I've also opened a bugereport on that in Heritage but as usual it needs someone to work on it

13:42 <GNUtoo> (and there is only 24h in a day)

13:42 <krzk> GNUtoo: I'll put Galaxy S3 kernel from opensource.samsung.com soon on my vendor-backup

13:42 <GNUtoo> wow I didn't manage to get it I think, let me see what I have

13:43 <GNUtoo> here's the heritage bugreport: https://forge.softwareheritage.org/T2523

13:46 <GNUtoo> I've only GT-N7105 (N7105XXALID N7105XXDMB6 N7105XXUEMK9 N7105XXUFND3 N7105XXUFQH1)

13:49 <GNUtoo> For some reason my website is down so I've to fix that but I archived them in archive.org/web: https://web.archive.org/web/2020*/https://gnutoo.cyberdimension.org/mirrors/opensource.samsung.com

13:54 <krzk> GNUtoo: here you have https://github.com/krzk/linux-vendor-backup/tree/samsung/galaxy-s3-gt-i9300-exynos4412-dump

13:54 <krzk> Thats a v3.0 kernel

14:17 <GNUtoo> Thanks

14:21 <GNUtoo> For the pin maybe it's indeed normal that it doesn't change but between the two kernel I compared they end up in a different state (according to register reads with devmem2): in one case it's high and in the other it's low

14:22 <GNUtoo> What I wonder is what could produce this behavior

14:23 <GNUtoo> If you set it low but that at the electrical level it stays high, is what should happen when reading the pin known?

14:23 <GNUtoo> Is there a way to safely test that (I've also an odroid U3)

14:24 <GNUtoo> Note that I badly need to not destroy that odroid U3, I'd prefer frying a GT-I9300 instead as we need it for potential u-boot work and/or testing

14:37 <GNUtoo> Ahh sorry I made a confusion

14:37 <GNUtoo> The smdk kernel I was using was 3.0.101

14:38 <GNUtoo> There has been a 3.18 port too I needed to try but I've not validated it yet so I cannot compare with that

14:42 <GNUtoo> The issue is that I'm not confortable tracing GPIO configuration from within the SMDK kernel

14:44 <GNUtoo> In this kenrel there is for instance /sys/class/gpio/gpiochip0, cat ngpios returns 8, and then the next gpiochip is gpiochip9

14:44 <GNUtoo> and base is 0

14:45 <GNUtoo> next, it should start at gpiochip8 AFAIK

15:00 <krzk> GNUtoo: I think only soldering and attaching a LED to the pin :)

15:18 <GNUtoo> I've done something like that, I wrote a tool to dump the registers through /dev/mem and validated that aginst the camera flash

15:19 <GNUtoo> but I can't see what happens when you pull the pin high from one side and low from the other for instance as here I've no reference to compare with

15:20 <GNUtoo> In any case I'll do more testing and fixes before trying that risky route

15:20 <GNUtoo> like simply having the same config before starting up the modem, as it's not the case yet

15:57 <GNUtoo> Changing the pullup / pulldown configuration to match the one in Android works though

15:57 <GNUtoo> Maybe I should try with the stock bootloader too to see how it behaves

17:50 <belgin> ls

17:50 <belgin> oops sorry

17:51 <belgin> is otg working in mainline for exynos4412? i know someone was working on it but i can't find anything anymore