03:40 <lucenera> I chose to spend the night flashing OpenWrt on Linksys MX5300. How do I switch between boot partitions if something goes wrong? What is the way to connect via serial cable?

09:24 <mrkiko> robimarko: hi!

09:24 <robimarko> Howdy

09:26 <mrkiko> robimarko: I wanted to ask you for come recommendations. I plan to improve little bit support for the gl-mv1000 to make it easier to boot from a micro SD card. so I would need to make somechanges in the boot script to try mmc dev 1 before dev 0 if some circumnstances arise. Do you think I'm better adding a different bootscript or modify the generic arm64 mvebu one somehow?

09:27 <robimarko> I think you are going to have to add another one

09:29 <mrkiko> robimarko: thanks; well, I would have liked so much for gl.iNet to add setexpr :D

09:31 <mrkiko> robimarko: thanks a lot. However, I hope it's being a good Sunday for you :)

09:35 <robimarko> It will be one day that I figure out how to get RM520N to work with LPAC

09:45 <mrkiko> robimarko: LPAC?

09:46 <robimarko> Open source eSIM LPA

09:46 <robimarko> https://github.com/estkme-group/lpac

09:47 <robimarko> I need to integrate eSIM support at work and the reference Quectel LPAd is horrible

09:47 <mrkiko> robimarko: uuuh! Cool

09:47 <robimarko> But Quectel somehow messed up the FW (Or my modem is broken) and it just refuses to access the ISD-R applet on eUICC

09:47 <robimarko> It worked for a short time and now nope

09:51 <mrkiko> robimarko: eh ... sorry I can't help

09:51 <robimarko> No worries, just need to vent :)

09:52 <mrkiko> robimarko: furthermore, I will need to get openwrt to use upstreamDTS for gl-mv1000 instead of our one

09:54 <robimarko> That shouldnt be hard

09:54 <robimarko> But look for the diff between them first

10:06 <mrkiko> robimarko: sure

10:38 <mrkiko> robimarko: I am on a strange situation - I would like to restore u-boot-env to writeable, but due to it not ending at an rase boundary the kernel marks it read-only; it's a good reason to do so, but what's the right thing to do here? Or should I just recommend in the commit messaged to use mtd-rw to force read-write and change the env?

10:38 <mrkiko> robimarko: the user needs to set an mmc_dev variable to decide from where to boot from, SD or MMC

10:40 <robimarko> Well, U-boot env is always multiple of erase blocks

10:41 <robimarko> So it should end on a erase boundary

10:42 <mrkiko> robimarko: ok, Iìll investigate the thing again

10:50 <mrkiko> robimarko: but, may I ask you if you could have a look at the layout? Even looking at it now, I don't see holes or something allowing me to round it up

10:54 <robimarko> What is the NOR model being used?

10:55 <mrkiko> robimarko: mhm, it's the same device. Nor is used for booting and to hold a rescue firmware, OpenWrt (and the stock firmware) are in the internal MMC

10:55 <robimarko> Not what I asked, what is the exact model used?

10:56 <robimarko> Cause 32k for U-boot env is rather weird

10:59 <mrkiko> robimarko: oh sorry

11:00 <robimarko> It could be that 4K erases are not enabled in the kernel

11:00 <robimarko> And that NOR doesnt support 32K natively

11:01 <mrkiko> robimarko: w25q128fw - 16384

11:01 <mrkiko> kb

11:06 <mrkiko> robimarko: but I cann tell you u-boot reads and updates env just fine, and factory doesn't contain things that seem related to u-boot-env in any way. The layout seems determined correctly.

11:06 <mrkiko> robimarko: oh sorry - I lost your messages. I will have a look

11:40 <stefanct> btw, how realistic is it to replace the vendor's uboot (to improve flash support)?

11:59 <robimarko> On which device?

12:00 <robimarko> mrkiko: Hm, that flash supports 32K erases, so its weird that it complains

12:00 <robimarko> Try enabling CONFIG_MTD_SPI_NOR_USE_4K_SECTORS but it should not be neeed

12:05 <stefanct> rt3888-based with spi flash

12:05 <stefanct> belkin f9k1109v1

12:06 <stefanct> but i would like to know more generally. is that something that is done sometimes?

12:31 <robimarko> Its not really often, maybe on those platform that have upstream support

14:11 <stefanct> the custom config files from files/ should be found in the staging_dir after build, right?

14:38 <mrkiko> robimarko: regarding the DTS - my plan is to sync upstream with openwrt one after the nvmem enhancements, fix validation of the dts and submit upstream. And then drop the openwrt version and use upstream in OpenWrt. Does this make sense?

14:52 <stefanct> hm, no, it should only be in build_dir and it is there. but it doesnt end up on the device

14:52 <stefanct> and i wonder if that's related to the kernel warning i get on sysupgrade before it reboots

14:53 <stefanct> that's the uart output i get after launching the upgrade on an ssh session: https://pastebin.com/NTSYYGv9

15:15 <stefanct> the files are actually included in the squashfs image in the sysupgrade.bin... and i am using sysupgrade -n. shouldnt that be enough?

15:50 <Mangix> robimarko: krb5 fails everywhere now

15:50 <Mangix> 7c32295b0036be425ba0cd527eb06316a87d0ec0 needs reverting looks like.

15:50 <Mangix> I have no idea why.

15:54 <Mangix> hmmm I think I figured out why. Not sure how to fix though

15:56 <Mangix> the SDK's compile_et has DIR=/builder/shared-workdir/build/staging_dir/host/share/et , which is a nonsensical value. Needs replacing with $(STAGING_DIR_HOST)/share/et

16:17 <Mangix> HOST_INSTALL_DIR is used nowhere. Interesting.

16:17 <robimarko> Hm, this is a weird one

16:18 <Mangix> HOST_INSTALL_DIR is also used wrongly by several host packages...

16:20 <mrkiko> robimarko: following your hint, it works fine.

16:20 <robimarko> Mangix: But why does only SDK have the issue?

16:21 <Mangix> because DIR gets populated based on the build system

16:21 <Mangix> the hostpkg version worked because it gets built by the SDK

16:21 <Mangix> but now it's pre-built

16:21 <robimarko> Ahh

16:22 <robimarko> Its using datadir from configure.ac to populate it

17:25 <swalker> updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html