08:16 <qschulz> robmur01: could we have the sram 0-4K SRAM be reserved in the DT maybe? Or at least add a comment in the device tree so we don't have to wonder in the future :) ?

08:19 <qschulz> catcream: I've had an empty atf-N for a few years already on puma rk3399, no clue if that's expected, some hack or something unnecessary /me shrugs

08:19 <qschulz> catcream: also, if you're compiling TF-A from source, you need to make sure you don't have BL31 env variable set, otherwise make will happily tell you it's already compiled

08:19 <qschulz> (been bitten SO. MANY. TIMES. by this already)

09:59 <diederik> qschulz: The first paragraph of your latest ML response seems to stop abruptly "Considering that for 125MHz TXC (for 1GbE), the clock period is 8ns and that two" ...

10:00 <diederik> (and +1 on the SRAM comment in the DT)

10:02 <qschulz> diederik: thought I had removed that sentence :)

10:02 <qschulz> Didn't manage to find the maths to figure out max delay based on the clock period :)

10:03 <diederik> Yep, your request looks fair/justified :)

10:21 <robmur01> qschulz: From a quick scan, I think RK3308 already answers that question ;)

10:26 <qschulz> robmur01: I for sure won't read rk3308's device tree when working on rk3588 :)

10:31 <robmur01> heh, my editor still had the search results for "sram" in dts/rockchip open from Friday :D

14:39 <catcream> qschulz: yes the $UBOOT thing happened to me many times yesterday lol

15:50 <TahomaSoft> Hi all, is there a general rule of thumb or standard procedure when dealing with pcie devices? As in, if the board ships with wifi, ethernet, etc devices attached to a pcie host, should such devices be explicitly listed in the DTS?

15:57 <CounterPillow> no

15:57 <CounterPillow> PCIe is enumerable, DT is only for things that can't be enumerated at runtime

16:18 <TahomaSoft> Thank you!

16:43 <robmur01> although of course that *can* sometimes include out-of-band properties for PCI endpoints like power supplies, MAC addresses, etc.

16:44 <robmur01> but in general, unless you know you need to describe the PCI hierarchy beyond the controller, then you don't

17:18 <TahomaSoft> Thank you; in my case with an rk3568-based board I am working to support, I needed to put the two RTL8125 devices into my draft DTS to get them to show up. But given what I understood about PCIe, that seemed like it should not have been needed.

19:14 <Dex2x> Has anyone tried to get a 16k Page kernel running on the rk3588 ?