<Kwiboo>
how should breaking changes be coordinated between dt and driver unless they are part of a single series? also hard to keep track when review/merge practises vary between subsystems
<marex>
Kwiboo: DT compatible addition is not a breaking change
<marex>
Kwiboo: the practice is very simple -- subsystem patches go through subsystem trees -- if you try to circumvent that established process, you only create chaos and more work for others
<marex>
well, and if you create mixed series which should go through multiple trees, then the patches get mostly ignored
qschulz has quit [Remote host closed the connection]
<Kwiboo>
in this case the compatible addition cause a breaking change, DT node currently contains two compatible, the new one added and an exsisting one, problem is that updated nodes use a single glue+ctrl node, and old compatible expect parent/child nodes
deathmist has joined #u-boot
qschulz has joined #u-boot
<marex>
Kwiboo: so handle both cases in the usb driver, this kind of thing is nothing new really
naoki has joined #u-boot
<marex>
Kwiboo: frankly, this just shows the problem right there -- DT is hardware description, it should be possible to update DT and system software separately
<marex>
Kwiboo: clearly this is not the case here and should be fixed
naoki has quit [Client Quit]
naoki has joined #u-boot
<Kwiboo>
agree, issue here is that DT node got merged in u-boot before it had a proper binding in linux, so breakage is to be expected, and I tried to minimize breakage with a mixed dt+driver update series
<Kwiboo>
guess I could separate DT + driver change if I piggy back on another compatible that follow same structure in the DT update and delay usb driver change until after DT have been merged, will send a v2 without the usb driver part :)
<GrantM11235>
I'm trying to port mainline u-boot on a new board, but I'm having trouble getting u-boot to recognize the onboard spi nand flash. When I run sf probe it says "invalid chip select"
monstr has quit [Remote host closed the connection]
ex-parrot has quit [Read error: Connection reset by peer]
ex-parrot has joined #u-boot
wyre has joined #u-boot
tnovotny has quit [Quit: Leaving]
sakman has joined #u-boot
vagrantc has quit [Quit: leaving]
matthias_bgg has quit [Read error: Connection reset by peer]
frieder has quit [Remote host closed the connection]
<Forty-Bot>
sjg1: I am working on some SPL tests which need the test devicetree; currently I have a patch to enable OF_REAL in sandbox_spl_defconfig; but would it make more sense to just use sandbox_noinst_defconfig?
Stat_headcrabed has joined #u-boot
matthias_bgg has joined #u-boot
flom84 has joined #u-boot
flom84 has quit [Remote host closed the connection]
flom84 has joined #u-boot
f_ has joined #u-boot
vagrantc has joined #u-boot
mckoan is now known as mckoan|away
Stat_headcrabed has quit [Quit: Stat_headcrabed]
<Forty-Bot>
huh, I guess neither of these have OF_REAL