<jernej>
so I wonder if Android has some efi binary somewhere
<jernej>
but still, if it bails out, PXE should still successfully boot
<apritzel>
I don't see any EFI tries being successful, though
<apritzel>
No EFI system partition; BootOrder not defined; EFI boot manager: Cannot load any image
<apritzel>
and the load addresses and even the final copy targets are the same
tuxillo has quit [Ping timeout: 272 seconds]
tuxillo has joined #linux-sunxi
<apritzel>
aha! If I run "bootefi bootmgr" before trying to manually TFTP, it hangs as well!
<apritzel>
(bootmgr coming up with the same negative result as in your case)
<apritzel>
jernej: so I guess this always happened with the EFI bootmgr, just wasn't used (widely) before
<jernej>
yeah, I think so too
<apritzel>
so at least I can reproduce it now
<apritzel>
and initrd_high and fdt_high didn't help
<apritzel>
but bootefi instead of bootz worked
dev1990 has quit [Excess Flood]
dev1990 has joined #linux-sunxi
Mangy_Dog has quit [Ping timeout: 265 seconds]
asdf28 has quit [Ping timeout: 272 seconds]
<apritzel>
jernej: so I traced it down to switching to nonsecure state during the bootmgr initialisation
<apritzel>
jernej: if I had to take a guess, I'd say this collides with U-Boot trying to install the PSCI handlers in secure world, just before it launches the kernel
<apritzel>
which sounds like a nasty problem
<apritzel>
wens: do you know if it's feasible to install our PSCI handlers very *early* in U-Boot, or at least prepare everything in the secure world already?
<apritzel>
so that bootm/bootz can run in non-secure state?
gediz539 has quit [Remote host closed the connection]
gediz0x539 has joined #linux-sunxi
Mangy_Dog has joined #linux-sunxi
\\Mr_C\\ has quit [Quit: (Read error: Connection reset by beer)]
vagrantc has quit [Quit: leaving]
camus1 has joined #linux-sunxi
kaspter has quit [Ping timeout: 264 seconds]
camus1 is now known as kaspter
tmlind has joined #linux-sunxi
<hlauer>
jernej: the change on banana pro rgmii->rgmii-id activates the RX/TX_DELAY flags in rtl8211e_config_init, right?
<apritzel>
jernej: dammit, I was tempted to just disable EFI_LOADER for sunxi-v7, as a stop-gap measure, but bootefi itself works, with SMP, and there is no extra option for just this bootmgr function
gaston1980 has quit [Quit: Konversation terminated!]
luke-jr has quit [Remote host closed the connection]
luke-jr has joined #linux-sunxi
ganbold_ has joined #linux-sunxi
ganbold has quit [Ping timeout: 240 seconds]
<jernej>
hlauer: yes, but that is not good commit message, if you're asking for that
<jernej>
apritzel: that would still be stop gap measure, although slightly more intrusive
<hlauer>
jernej: yes - what would be a good one? I only know it works with the change and not without...
<apritzel>
jernej: yeah, thinking about that ...
<jernej>
hlauer: proper commit message explain why are you doing that change. What is done should be visible from diff itself or from comments in code
<jernej>
hlauer: for example, you can explain that BPi Pro needs TX and RX delay in order to be able to communicate and that that can be seen from pull ups from board schematic
<jernej>
so I guess it's possible to disable bootmgr only
<apritzel>
jernej: ah, perfect, this is what I was looking for
<hlauer>
jernej: first ok - also that RX+TX is needed for Gbit only, not 100Mbit makes sense for me. But about pull ups I'm not deep enough into high frequency electronics to make that conclusion...
<jernej>
hlauer: those are configuration pull ups, not something in communication path
<jernej>
and yes, it's confusing - phy is properly configured already by pull ups, but driver still overrides decision
<jernej>
so override must match configuration done by pull ups
<jernej>
don't ask why...
<hlauer>
jernej: Ah, ok, thats helps! Will send a new patch. Thanks!
<jernej>
technically, if you use generic PHY driver (realtek phy driver disabled), it will work without fix
toaster has quit [Quit: Ping timeout (120 seconds)]
<hlauer>
v2 is out, thanks. Also for the explanation - stumbled across the rx/tx delay when getting m2u working with the generic phy.
<hlauer>
Whats the ada
<hlauer>
...advantage of the realtek phy driver, if any?
ganbold_ has quit [Ping timeout: 240 seconds]
<mru>
advantage of realtek, is that a joke?
<KotCzarny>
it's cheap
<KotCzarny>
:)
<apritzel>
the question was about the specific Realtek driver of the generic PHY driver
<apritzel>
s/of/over/
<mru>
in general, the specific drivers can access non-standard features of the phy
<apritzel>
jernej: so explicitly calling "bootefi bootmgr" still kills it (with CMD_BOOTEFI_BOOTMGR disabled), but it looks like it should be gone from the distro boot
<apritzel>
jernej: so can you confirm that *your* use case now works when you disable it in menuconfig?
cmeerw has joined #linux-sunxi
lucascastro has quit [Remote host closed the connection]