mmind00 changed the topic of #linux-rockchip to: Rockchip development discussion | public log at https://libera.irclog.whitequark.org/linux-rockchip
<dsimic> it's related to splitting the DTs for different board versions
<dsimic> let's see what will Krzysztof respond with
Perflosopher has quit [Ping timeout: 264 seconds]
Perflosopher has joined #linux-rockchip
stikonas has quit [Quit: Konversation terminated!]
<naoki> dsimic: I see
<naoki> (close my eyes)
Daanct12 has joined #linux-rockchip
<dsimic> hmm, what's the problem there?
<dsimic> I mean, why is it sad news? I see that the patches have been accepted
<naoki> ZERO 2 and ZERO 2 Pro is 100% compatible both software and hardware
<naoki> just product was renamed
<naoki> we are trying to find out better way
<naoki> so I think zero 2 pro patch was bad precedent...
<naoki> well
<naoki> it might be okay for linux-amlogic
<naoki> then I don't care
<naoki> btw... could you help me to write correct commit message, anyone? https://lore.kernel.org/linux-leds/A85312FB70235D56+bd5fad03-36bf-4df9-ad44-7f7eaa7b2aa9@radxa.com/T/#t
<naoki> (actually I couldn't understand what Lee Jones want...)
<dsimic> hmm, but why did you submit those patches for ZERO 2 Pro?
<dsimic> you could've just renamed the model name from "Radxa ZERO 2" to "Radxa ZERO 2 / ZERO 2 Pro" in the original board dts file, for example
<dsimic> if the boards are 100% compatible, there's no point in introducing a new board dts file
<dsimic> I'd suggest that you submit new patches that basically revert those changes
<dsimic> actually, I'd _highly_ suggest that you submit those patches
<dsimic> regarding the commit message, I can help, please ping me in about a couple of hours
<naoki> I wanted to change file name like zero-2-pro, but now I'm feeling just change model name is better...
<dsimic> basically, DTs describe hardware
<dsimic> so if two boards are identical, having two DTs is simply wrong
<dsimic> especially because the board was just renamed for marketing purposes
System_Error has quit [Ping timeout: 260 seconds]
<dsimic> maybe you can still ask the maintainers to drop those patches
<dsimic> anyway, it's up to you and I don't care much about Amlogic boards, but having the additional board dts file is simply wrong
<naoki> I sent email to the maintainer
<naoki> dsimic: please help me about commit message...
<dsimic> sure, in about an hour or so
<dsimic> I'm a bit busy at the moment :)
<naoki> I see, thanks
kevery has quit [Ping timeout: 248 seconds]
<dsimic> anytime :)
kevery has joined #linux-rockchip
shailangsa has joined #linux-rockchip
sjoerd5 has joined #linux-rockchip
sjoerd has quit [Ping timeout: 252 seconds]
sjoerd5 is now known as sjoerd
Daanct12 has quit [Ping timeout: 252 seconds]
Daanct12 has joined #linux-rockchip
System_Error has joined #linux-rockchip
<dsimic> naoki: regarding the commit message, you were asked not to "recycle" a comment from another file, but to describe the reasoning in prose, using your own words
<dsimic> basically, you need to describe what you do in your patch, and why you do that
<dsimic> but using your own words, providing as much detail as possible, including the specific platform on which the issue was observed and fixed, i.e. E25
warpme has joined #linux-rockchip
<naoki> oh... I don't know "why", I just found difference between leds-pwm an leds-pwm-multicolor, and that comment sounds reasonable...
<dsimic> well, you _need_ to know why
<naoki> sure
<dsimic> in other words, all patches should be based on knowing exactly why, how and where
<naoki> otherwise I cannot say my fix is correct ;)
<dsimic> and allo those details should be provided in the commit summary
<dsimic> s/allo/all/
<dsimic> exactly, that's how you prove that your fix is reasonable and correct
<dsimic> and how you provide as much detail as possible for possible regression debugging later
<dsimic> regarding this specific patch, you may even want to connect a scope to the PWM outputs that drive the LED on E25, to see exactly what's happening
<dsimic> (if needed to go that far, of course)
<naoki> problem is that I don't know much about pwm, electrical circuits, etc
<dsimic> in that case, start learning today :)
<naoki> anyway, thank you very much for you help
<dsimic> I'm happy to help
<dsimic> would you like me to recommend you a book?
<naoki> written in Japanese? ;)
<naoki> 2nd problem is it's difficult to explain detail in English ;)
<dsimic> then start learning English today :)
<naoki> haha
<naoki> sure
<dsimic> I'm not a native speaker either, and it took me many years to reach the current level
<dsimic> e.g. I spent 2.5 years on English Wikipedia, writing articles and learning written English
<dsimic> English *
<naoki> oh
<dsimic> about 40,000 edits :)
<dsimic> and about 10,000 people reading daily what I wrote there
<dsimic> wow, did I really edit 5,500 pages and create almost 1,850 new pages (which includes redirectcs)? :D
<dsimic> s/redirectcs/redirects/
<naoki> great work
<dsimic> thanks
<mps> also for me (non native speaker and self taught in english) sometimes is hardest part is writing commit message
<dsimic> well, I love writing, so that's rather enjoyable to me
<dsimic> a nice patch is like a beautiful painting, and the commit summary is your signature on the painting :)
* mps prefer to speak instead of write :)
<dsimic> spoken English is even harder
<mps> agree fully
<mps> or better say "English is hard" (I would like if the some of old langs is used for international communications - Latin, old Greek or Sanskrit)
<naoki> nothing is explained lol
<naoki> (it's a kind of routine like savedefconfig)
<dsimic> mps: I'd say that English is actually deceptively simple
<dsimic> naoki: such commit descriptions leave a lot to be desired :)
<dsimic> mps: I tried learning Latin, and it's waaay harder than English, IMHO
<mps> dsimic: yes, so simple that you cannot express deep thinking in it ;-)
<mps> also, nearly half of words and terms in english are from latin
<mps> knowing latin helps a lot to understand other european langs
kevery has quit [Ping timeout: 252 seconds]
ldevulder has joined #linux-rockchip
<dsimic> no, no, you can express whatever you want in English :)
<dsimic> I wanted to say that many people think they know English, while they actually don't
<dsimic> BTDT... you need to know it very well, before realising why you didn't know it very well before :)
<dsimic> regarding Latin, perhaps knowing it helps
<diederik> FTR: your English doesn't have to be 'perfect' to write a good commit message
<dsimic> true
<diederik> Just explain to the best of your (English) abilities what you did and most importantly *why* you did it. The more justification for why, the better
<mps> yes, I often excuse for "my bad english" but readers usually says it quite fine (maybe somewht 'speial' :) )
<mps> s/speial/special/
<warpme> guys: can anybody hint me how to say to kernel to NOT use given memory region? I'm trying with Kernel command line: root=/dev/mmcblk0p2 rw rootwait earlycon=uart8250,mmio32,0xfeb50000 console=ttyS2,1500000n8 debug memmap=0x80000000$0xBFFFFFFF but i'm getting: Unknown kernel command line parameters "memmap=0x80000000$0xBFFFFFFF", will be passed to user space. ....
<dsimic> warpme: are you going to upstream your eMMC LED patch?
<mps> warpme: is memmap supported on arm arch
<warpme> ah - i'm not sure it is ready to upstream: it not correctly handles capability to drive separate read vs. write leds :-(
<warpme> mps : i'm not sure and....it look like not :-(
<mps> also I'm not sure but looks like it is not
<warpme> is there any other way to disable mem region from use @ arm?
<warpme> (beside declaring in DT)
<warpme> asking as my rock5a started to report https://gist.github.com/warpme/8b6a0a1b9970d2c636aa4a36318a514b before throwing it to trash - i want to make it semi-usable by excluding 3rd bank....
<dsimic> warpme: oh wow, look at that :/
<mps> maybe DTB overlay
<diederik> build newer u-boot?
<warpme> option is to reflow or reball but this is costly and not 100% guarantee (e.g. when issue is via on pcb; but iirc mem bus should not use via due impedance imparity between bus lines)
<dsimic> reflowing should be inexpensive in any cellphone repair shop
<dsimic> it's quite suspicious how regular the failed reads are
<warpme> diederik : why new u-boot? Current one works perfect for other 6 rk3588 bords i have in lab. also iirc my dram training blob is most current one....
kevery has joined #linux-rockchip
<dsimic> ch:3 dq0 fail,write:0x1,read:0x2
<dsimic> ch:3 dq1 fail,write:0x2,read:0x4
<dsimic> ch:3 dq2 fail,write:0x4,read:0x8
<dsimic> ch:3 dq3 fail,write:0x8,read:0x10
<dsimic> ch:3 dq4 fail,write:0x10,read:0x20
<dsimic> ch:3 dq5 fail,write:0x20,read:0x40
<dsimic> ch:3 dq6 fail,write:0x40,read:0x80
<dsimic> ch:3 dq7 fail,write:0x80,read:0x0
<dsimic> see the pattern? that's really strange
<warpme> dsimic isn't this exactly points to 1 failing line in dram bus?
<dsimic> the trouble is that read values are twice as large
<dsimic> if they were smaller, then a falied line would make sense
<dsimic> failed *
raster has joined #linux-rockchip
<dsimic> it looks like the reads behave if there was a shift register somewhere
stikonas has joined #linux-rockchip
<dsimic> ch:3 dq0 fail,write:0xfffffffe,read:0xfffffffd
<dsimic> ch:3 dq1 fail,write:0xfffffffd,read:0xfffffffb
<dsimic> ch:3 dq2 fail,write:0xfffffffb,read:0xfffffff7
<dsimic> input, base 16: fffffffe
<dsimic> converted, base 2: 11111111111111111111111111111110
<dlg> does rk356x have hdmi input?
<dsimic> input, base 16: fffffffd
<dsimic> converted, base 2: 11111111111111111111111111111101
<dsimic> input, base 16: fffffffb
<dsimic> converted, base 2: 11111111111111111111111111111011
<dsimic> etc.
<diederik> warpme: it indeed probably won't make a difference for this issue. The build date was (just) the first thing I noticed
<dsimic> dlg: I'd suggest that have a look at the datasheet
<dsimic> s/that/that you/
<dsimic> warpme: based on that, perhaps the row buffer in the DRAM chip failed
<dlg> datsheet doesnt have hdmi rx, but i was hoping i was missing something
<dsimic> RK3588 has it, FWIW
<dlg> yeah
<dlg> has anyone turned an rk3588 into a pikvm kind of thing?
stikonas has quit [Quit: Konversation terminated!]
System_Error has quit [Ping timeout: 260 seconds]
<warpme> dsimic : indeed this look more like dram chip failure than connectivity on pcb (e.g. cold solder). Interesting is that issue is definitely temperature triggered: starts happen after 5..10min of operation. power off for 5min+ brings board back. Using spray freezer on dram chip results with immediate back to operational. thats why my initial hypothesis was cold solder....
raster has quit [Quit: Gettin' stinky!]
<warpme> so - for excluding dram bank usage from kernel - currently i see only DT option (and not by overlay as mainline don't have ov support). Any other ideas?
<dsimic> I don't see another option
<warpme> dsimic : thx!
<dsimic> it's interesting that such regular DRAM chip failure is triggered by temperature
<dsimic> I'd suggest that you don't throw your Rock 5A in trash, keep it for spare parts :)
<dsimic> though, maybe I'm wrong and there's other option
<warpme> yeah - especially reusing FCBGA1088 chip (SoC) might be handy spare part :-) I already see this long queue of hobbyst asking me for sell it to them as they want to resolder it as hobby :-))))))
<dsimic> one man's trash is another man's treasure :)
<dsimic> or s/man/person/ if we want to be gender neutral
<diederik> ha, I guess I should've asked out loud whether cold vs warm boots makes a difference
<dsimic> cold and warm literally :)
<diederik> doesn't that mean that updating u-boot is back on the table as possible solution? It seems at least worth a try ...
<qschulz> if the TPL propagates the proper info through ATAGS, a new U-Boot should be able to handle that and mark those as memory to not use (or just not add the area as available)
<warpme> diederik : cold vs. warm reboot: no difference. Diff is in cold board vs. warm board: cold boots ok. warm: stops after some time
<diederik> warpme: my "won't make a difference for this issue" was based on the assumption it always fails ... so (most of) u-boot things aren't even run
<diederik> as that appears not the case, then a bug in u-boot *could* cause it, in which case a newer u-boot could fix it
<warpme> yeah - indeed idea of "exclude bad mem reg. by dt" is an half solution: it will work only when dram training succeeded. Evident issue correlation with board temp gives me highest probability for hw defect - so i stopped checking sw. route. (additional args i had myself that this is not sw: my other 6 boars never had this; can't find any other report of this such issue; on this device also kernel gets mem paging ops after 1..2h of
<warpme> operation).
<warpme> i'm really curious how kernel will go with added mem reserv. in dt
naoki has quit [Quit: naoki]
psydroid has quit [Read error: Connection reset by peer]
psydroid has joined #linux-rockchip
ldevulder has quit [Quit: Leaving]
Daanct12 has quit [Quit: WeeChat 4.4.1]
Stat_headcrabed has joined #linux-rockchip
Stat_headcrabed has quit [Quit: Stat_headcrabed]
raster has joined #linux-rockchip
dsimic has quit [Ping timeout: 252 seconds]
dsimic has joined #linux-rockchip
warpme has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
warpme has joined #linux-rockchip
warpme has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
warpme has joined #linux-rockchip
warpme has quit [Remote host closed the connection]
warpme has joined #linux-rockchip
raster has quit [Ping timeout: 246 seconds]
Bahhumbug is now known as Gerdesas
Gerdesas is now known as Bahhumbug
raster has joined #linux-rockchip
warpme has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
warpme has joined #linux-rockchip
mps has quit [Quit: leaving]
mps has joined #linux-rockchip
warpme has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
stikonas has joined #linux-rockchip
stikonas has quit [Client Quit]
detlevc1 has joined #linux-rockchip
sre4 has joined #linux-rockchip
detlevc1 has quit [Quit: The Lounge - https://thelounge.chat]
sre4 has quit [Quit: The Lounge - https://thelounge.chat]
detlevc8 has joined #linux-rockchip
raster has quit [Quit: Gettin' stinky!]
stikonas has joined #linux-rockchip
VladVysko has joined #linux-rockchip
stikonas has quit [Remote host closed the connection]
stikonas has joined #linux-rockchip
naoki has joined #linux-rockchip
crabbedhaloablut has quit []
crabbedhaloablut has joined #linux-rockchip
VladVysko has quit [Quit: Client closed]
crabbedhaloablut has quit []
crabbedhaloablut has joined #linux-rockchip
Sario has quit [Ping timeout: 604 seconds]
crabbedhaloablut has quit []
crabbedhaloablut has joined #linux-rockchip
kevery1 has joined #linux-rockchip
kevery has quit [Read error: Connection reset by peer]
kevery1 is now known as kevery
VladVysko has joined #linux-rockchip
Perflosopher has quit [Ping timeout: 255 seconds]
_hramrach has quit [Ping timeout: 252 seconds]
_hramrach has joined #linux-rockchip
Perflosopher has joined #linux-rockchip
Sario has joined #linux-rockchip