tegra-target-flash-scripts: fix UDC discovery and QSPI path for t264

[mihai-chiorean]

Replace hardcoded known_udc_dev checks in nv_enable_remote.sh with
dynamic /sys/class/udc/* enumeration, matching how init-flash.sh
already discovers the UDC.

Add QSPI device symlink (/dev/810c5b0000.spi -> /dev/mtd0) for t264
initrd flash, gated on tegra264 compatible.

UDC issue

During initrd flash on AGX Thor, nv_enable_remote.sh fails to find
the USB Device Controller because it only checks hardcoded device
names (known_udc_dev1="3550000.usb", known_udc_dev2="..."). Thor's
UDC device name doesn't match either, so the script falls through
with "no UDC found" and the ADB connection never comes up.

init-flash.sh already handles this correctly with
ls -1 /sys/class/udc | head -n 1 — this patch applies the same
approach to nv_enable_remote.sh.

QSPI issue

During initrd flash on AGX Thor, nvdd fails to write QSPI NOR because
it looks for /dev/block/810c5b0000.spi (the t264 SPI5 MMIO address).
The Yocto kernel exposes QSPI NOR via Linux MTD as /dev/mtd0 instead.
Since /dev/block -> /dev in the initramfs, a symlink fixes the path
resolution. Gated on tegra264 to avoid affecting other platforms.

[madisongh]

@mihai-chiorean The master-l4t-r38.4.x is the current active branch for R38.4, can you retarget your PR there?

It would also help if you could provide a bit more background on the failure you're seeing that requires these changes.
Thanks!

[madisongh]

@mihai-chiorean Thanks for the updated description... I've never had an issue with flashing my AGX Thor development kit with the way the script is currently. Are you using a development kit, or some kind of custom hardware setup? And is this with the stock NVIDIA kernel (linux-noble-nvidia-tegra recipe), or are you using something else?

[mihai-chiorean]

@mihai-chiorean Thanks for the updated description... I've never had an issue with flashing my AGX Thor development kit with the way the script is currently. Are you using a development kit, or some kind of custom hardware setup? And is this with the stock NVIDIA kernel (linux-noble-nvidia-tegra recipe), or are you using something else?

To be honest, I need to test it again. I started out with our os using 38.2, then @ichergui suggested going the 38.4 route and this might be leftover. will get back once I test again.

[ichergui]

Hi @mihai-chiorean
Thanks for the PR. I agree with @madisongh
I didn't have an issue with the flashing process. Please test it again without updates then let us know.

[mihai-chiorean]

ok, jsut got a new thor and confirmed we don't need this.