00:14 <notu> I dumped the mtd of the old flash, desoldered it and switched to a W25Q128FV following this guide. Uboot cannot write to the new chip and I am trying to boot initramfs to see if its some issue with partitioning

00:15 <notu> https://openwrt.org/toh/tp-link/td-w8970_v1#mb_flash_mod

00:17 <notu> U-Boot 2010.06-LANTIQ-v-2.0.40-dirty (Dec 22 2024 - 14:31:45)

00:17 <notu> DRAM: 64 MiB

00:17 <notu> CLOCK CPU 500M RAM 250M

00:17 <notu> Using default environment

00:17 <notu> In: serial

00:17 <notu> Out: serial

00:17 <notu> Err: serial

00:17 <notu> Net: Internal phy(GE) firmware version: 0x841d

00:17 <notu> vr9 Switch0 KiB W25Q128 at 0:3 is now current device

00:17 <notu> I see the size from there

03:22 <russell--> notu: does u-boot support that flash chip?

03:28 <russell--> "note that the modified U-Boot is still needed"

03:33 <russell--> notu: what i do is create a new 16MB image, with blobs extracted from the original image, written to the new locations, and write that to the new flash with the same programmer you used to read the original flash part

04:08 <mrnuke> russell--: it's under development. See https://github.com/Hurricos/realtek-poe/pull/35

04:25 <russell--> mrnuke: without a patch to adjust to 115200 baud on top of what's in packages feed, i noticed that my gs1900-10hp doesn't provide power to the ports, but does with just that change. i don't have any control over the ports, but ... so far that's good enough.

04:48 <russell--> mrnuke: will i be sad if i try out your mrnuke-rtl8238b-devel branch?

05:25 <mrnuke> russell--: I just saw your github comment. if the mrnuke-rtl8238b-devel branch reports a useful status, then you have a realtek dialect PoE controller -- and we haven't cracked those yet

05:47 <russell--> somehow the master branch + baudrate 115200 patch at least enables poe on the ports

05:47 <russell--> what is needed to make progress?

14:53 <mrnuke> russell--: "baudrate 115200 enables poe" is misleading. There are many command IDs that are recylced beetween the dialects. You might say "enable warp factor nine", which is interpreted as "fire lasers" byu the Poe controller. I suspect that is what is happening

14:54 <mrnuke> to make progress, we'd need

14:54 <mrnuke> (1) to implement a way to automatically detect the dialect and switch baudrate

14:55 <mrnuke> (2) figure out what we;re doing different on the PoE init sequence. As far as I can tell from packet dumps, nothing different

14:55 <mrnuke> ---end of list--

18:19 <Habbie> SetupHostCommand is turning staging_dir/host/bin/grep into a symlink to itself

18:19 <Habbie> using an sdk snapshot from this weekend

19:39 <Habbie> ok, not putting the host/toolchain paths at the front of $PATH helps

19:39 <Habbie> looks to be a bug introduced in include/prereq*.mk mid 2023

19:54 <Habbie> the sdk does -not- like <pkgdir>/src to be a symlink to ../..

21:01 <swalker> updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html

21:39 <russell--> mrnuke: does it need to by dynamic? can't we just say "oh, this is v2 hardware, use 115200, use dialect $foo"? i can capture serial traces running vendor firmware, if that's helpful.