ChanServ changed the topic of #linux-rockchip to: Rockchip development discussion | Wiki at http://linux-rockchip.info | Logs at http://irclog.whitequark.org/linux-rockchip | ML at http://groups.google.com/group/linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
Soopaman has joined #linux-rockchip
_whitelogger has joined #linux-rockchip
phh_ has joined #linux-rockchip
dianders has joined #linux-rockchip
dianders has quit [Changing host]
phh has quit [Read error: Connection reset by peer]
_whitelogger_ has joined #linux-rockchip
dianders_ has joined #linux-rockchip
lioka_ has joined #linux-rockchip
phh has joined #linux-rockchip
Soopaman has joined #linux-rockchip
honx_ has joined #linux-rockchip
hipboi has joined #linux-rockchip
honx has quit [Ping timeout: 272 seconds]
phh_ has quit [Ping timeout: 272 seconds]
dianders_ is now known as dianders
RayFlower has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
hipboi has quit [Ping timeout: 244 seconds]
dlan has quit [Ping timeout: 260 seconds]
dlan has joined #linux-rockchip
Soopaman has quit [Read error: Connection reset by peer]
Soopaman1 has joined #linux-rockchip
hipboi has joined #linux-rockchip
Soopaman1 has quit [Quit: Leaving.]
Bludot has joined #linux-rockchip
Soopaman has joined #linux-rockchip
GriefNorth has joined #linux-rockchip
naobsd has joined #linux-rockchip
Astralix1 has joined #linux-rockchip
Astralix has quit [Ping timeout: 250 seconds]
FreezingCold has quit [Read error: Connection reset by peer]
FreezingCold has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
FreezingCold has quit [Read error: Connection reset by peer]
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
FreezingCold has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
FreezingCold has quit [Ping timeout: 260 seconds]
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
FreezingCold has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
FreezingAlt has joined #linux-rockchip
RayFlower has joined #linux-rockchip
FreezingCold has quit [Ping timeout: 260 seconds]
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
Bludot has quit [Quit: Connection closed for inactivity]
FreezingAlt is now known as FreezingCold
FreezingAlt has joined #linux-rockchip
FreezingCold has quit [Ping timeout: 260 seconds]
<naobsd> oops, OTG cannot be used for power input on Firefly...
<naobsd> usually I only use 2cables, OTG and UART, for flashing & testing "boot or not" quickly...
AstralixNB has joined #linux-rockchip
hipboi has quit [Read error: Connection reset by peer]
lioka_ is now known as lioka
lioka has joined #linux-rockchip
lioka has quit [Changing host]
Soopaman has quit [Read error: Connection reset by peer]
Soopaman1 has joined #linux-rockchip
rperier has quit [Remote host closed the connection]
rperier has joined #linux-rockchip
rperier has joined #linux-rockchip
rperier has quit [Changing host]
<rperier> hi all
<naobsd> mmind00: can you try my u-boot binary for rk3288? I got a report that it doesn't work on Tronsmart Orion R28 :(
<naobsd> doesn't work -> doesn't boot, no detail info due to lack of uart
hipboi has joined #linux-rockchip
hipboi has quit [Quit: Leaving]
FreezingCold has joined #linux-rockchip
FreezingAlt has quit [Ping timeout: 265 seconds]
hipboi has joined #linux-rockchip
cristian_c has joined #linux-rockchip
<cristian_c> Hello
<cristian_c> How can I autostart command/scripts in cwm recovery?
<cristian_c> Any ideas?
mrcan has joined #linux-rockchip
mrcan has joined #linux-rockchip
<naobsd> cristian_c: service /path/to/command. see existing lines in /init.rc
<cristian_c> ok
mrcan__ has quit [Ping timeout: 246 seconds]
hipboi has quit [Ping timeout: 260 seconds]
mrcan_ has joined #linux-rockchip
<naobsd> does anyone have Netxeon R89(Beelink R89/Orion R28)?
<naobsd> ^with serial console
<cristian_c> naobsd, now, I've found an interesting thing
<naobsd> I really have no idea why any newer bootloader doesn't boot on R28/R89
<cristian_c> in init.recovery.rk30board,.rc
<cristian_c> on early-init
<cristian_c> instead in init.rc, I see:
<cristian_c> on early-init
<cristian_c> on init
<cristian_c> on boot
mrcan has quit [Ping timeout: 258 seconds]
<cristian_c> naobsd, i could create 'on init' or 'on boot' section in the first file and put the commands I need
<cristian_c> so these are excuted late
<cristian_c> *executed
<naobsd> cristian_c: yes, you can add "on boot" to rk30board.rc
<cristian_c> ok
mrcan__ has joined #linux-rockchip
hipboi has joined #linux-rockchip
mrcan has joined #linux-rockchip
mrcan has quit [Changing host]
mrcan has joined #linux-rockchip
mrcan_ has quit [Ping timeout: 260 seconds]
mrcan__ has quit [Ping timeout: 260 seconds]
FreezingAlt has joined #linux-rockchip
FreezingCold has quit [Ping timeout: 260 seconds]
<mmind00> naobsd: I'm at the LinuxCon Europe currently, so don't have the box with me
<naobsd> mmind00: I see. I just got a info from netxeon
<naobsd> probably it can be fixed
<mmind00> naobsd: was the reporter able to go back to the original uboot?
<naobsd> mmind00: yes. I think there is holes to force mask rom mode
<mmind00> naobsd: oh, so you're in contact with netxeon the board manufacturer? ... cool
<naobsd> mmind00: very thin connection, but kind person
<mmind00> I really like how this slowly gains momentum
Soopaman1 has quit [Quit: Leaving.]
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
<naobsd> ok, u-boot patch for R89 fixed/worked/confirmed :)
<naobsd> I'm asking it can be opened ;)
FreezingAlt has quit [Read error: Connection reset by peer]
<hramrach_> hello
<hramrach_> I just tested the Android 4.4 firmware. Besides the totally annoying lock screen that cannot be replaced the media player works even worse than on 4.2
<hramrach_> but you can switch to software decoding which solves all problems
<hramrach_> so the vpu driver code from rk is soo immensely useful that it's better to not use it at all :/
<hramrach_> hope the v4l driver is better than tha
<hramrach_> t
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
<hramrach_> naobsd: so is reading the bootloader and dram blob from a board possible ?
<hramrach_> you may say that only boards for which firmware is provided should be supported but this is not the way to go. Usable support should work with any board. If something cannot be recovered from original firmware for technical reason so be it but saying users to go and get a firmware just because is not such a great idea
FreezingAlt has joined #linux-rockchip
<naobsd> hramrach_: reading data on NAND is possible since some monthes(years?) ago, and there is no info how to reconstruct something from read data
<naobsd> hramrach_: please ask vendor for "usable support". I/we are not vendor, we don't give such a support
<hramrach_> yes, I can presumably attach a programmer to the NAND leads and read all the physical blocks which is utterly useless
<naobsd> it's totally wrong idea to get such a support for commercial product from community
<hramrach_> but since the l and L commands write the bootloader and dram area it must be known what format and where the data is
<hramrach_> so you say that I should only run linux-rockchip on non-commercial hardware? What hardware would it be?
<naobsd> l/L doesn't anything with NAND, don't presume.
<hramrach_> it writes bootloader which is stored in nand. how does it do that without doing anything with nand?
<hramrach_> or does it do this by issuing some 'write bootloader' command to the mask rom laoder without any knowledge where that loader is actually written?
<naobsd> hramrach_: oh, for example, if you got defective product, you ask refund to community? wanderful!
<naobsd> hramrach_: oh, you know more than me about l/L, it seems you are the author! great!
<hramrach_> why do you say such nonsense?
<naobsd> hramrach_: why do you say such nonsense?
<hramrach_> I am asking about the l L commands because I did not write them so I do not know how they work
<naobsd> hramrach_: any persona in community doesn't offer commercial support for any product someone bought
<naobsd> <hramrach_> it writes bootloader which is stored in nand. how does it do that without doing anything with nand?
<naobsd> hramrach_: you everytime objection about what I explained=the fact
<hramrach_> and instead of giving an answer you say go fuck yourself or ask the vendor. Ok, I thought the point of this # was to share knowledge about the rk hardware and building and running fitrmware for it. Maybe I was wrone assuming that
mrcan has quit [Ping timeout: 260 seconds]
<naobsd> hramrach_: you always talk wrong assuming as like as the fact
<naobsd> hramrach_: you always object my explanation
<naobsd> l/L doesn't anything with NAND
<naobsd> I explained
<naobsd> you can confirm but you never confirm for everything
<hramrach_> so how does it write the bootloader to nand when it does nothing with nand?
<naobsd> it doesn't write bootloader to nand
<naobsd> I never said that
<naobsd> it's your assuming
<hramrach_> rkflashtool L" for uploading (encrypted) loader.
<hramrach_> so where is that loader stores if not nand?
<naobsd> ram, of course
<naobsd> if you can't understand functionality/description, ask it
<hramrach_> I cannot
<naobsd> not say wrong assuming as if the fact w/o confirmation
<naobsd> why you cannot say "I cannot understand your description"
<naobsd> you never say I cannot
<naobsd> you always say like "it write loader to nand" as if the fact
<hramrach_> I read the description and it says upload bootloader. it is normally uploaded to nand so when the description does not say otherwise that is where I expect it to be stored
<naobsd> "normally"? huh
<hramrach_> yes, every rk device I have seen had bootloader in nand
<naobsd> so what?
<hramrach_> so that's where I expect a bootloader on a rk device
<naobsd> then?
<hramrach_> then your description of the command is incomplete and can be misunderstood
<naobsd> I said we (our rkflashtool) cannot write to NAND several times
<naobsd> huh
<naobsd> you cannot admit your fault
<naobsd> good
<hramrach_> which leads to like 2 days of pointless discussion during which you do not point out where does it really write the bootloader
<naobsd> huh
<hramrach_> yes, I misunderstand your description
<naobsd> yes, I was wrong. now you satisfied?
<hramrach_> no, you assumed that everyone knows the rkflashtool as you do so it's obvoius to everyone that the bootloader is stored in ram
<hramrach_> but did not write it in the description and it is not obvious
cristian_c has quit [Ping timeout: 265 seconds]
<naobsd> no I don't assume everyone can understand rkflashtool
cristian_c has joined #linux-rockchip
<cristian_c> naobsd, can I use 'service' with sh script?
<cristian_c> service /sbin/script.sh
<cristian_c> ?
<naobsd> cristian_c: yes if script can run by enter "/sbin/script.sh"
<cristian_c> naobsd, I launch the script with sh script.sh
<naobsd> sh script.sh and /path/to/script.sh is different
<cristian_c> service sh /sbin/script.sh
<cristian_c> this?
<naobsd> you have to write correct path to sh at 1st line in script.sh, and need to set exec bit to script.sh
<naobsd> probaby full path is needed for "sh"
<naobsd> you can try and confirm yourself
<naobsd> ah, you need "oneshot" as a option to that service
<cristian_c> #!/sbin/sh
<cristian_c> naobsd, oneshot?
FreezingCold has joined #linux-rockchip
<naobsd> otherwise that script.sh is called repeatedly (it's default if command exit)
<naobsd> service xxx
<naobsd> oneshot
<naobsd> ^like this
<naobsd> please google it for detail
<cristian_c> ok
FreezingAlt has quit [Ping timeout: 240 seconds]
nighty^ has joined #linux-rockchip
<cristian_c> naobsd, /sbin/sh: /sbin/script.sh: not found
<cristian_c> lol, I'made a mistake, lol
<cristian_c> ok, it's executed
<naobsd> cristian_c: you could solve your problem finally?
<cristian_c> I'm trying
<naobsd> I'm praying
<cristian_c> lol
<cristian_c> service sh /sbin/adjust_screen.sh
<cristian_c> oneshot
<cristian_c> into init.rc, ok I pack it and I flash it
<naobsd> ah
<naobsd> I found mistake...
<naobsd> service <service name> <path name> [<argment> ...]
<naobsd> so "sh" is just a name(identifier)
<naobsd> it's ok if there is no other "sh" service and /sbin/adjust_screen.sh can be run
<naobsd> anyway you should use full path e.g.
<naobsd> service adjust_screen /sbin/sh /sbin/adjust_screen.sh
<naobsd> or
<naobsd> service adjust_screen /sbin/adjust_screen.sh
<cristian_c> ok
bengal has joined #linux-rockchip
<cristian_c> naobsd, ok, worked
<cristian_c> :)
<cristian_c> naobsd, sh script in /sbin works
<cristian_c> in the custom cwm recovery
<cristian_c> naobsd, thanks
<naobsd> service _sh_ <- this "sh" is not command, just service name, but it's ok
<naobsd> cristian_c: good :)
FreezingAlt has joined #linux-rockchip
FreezingCold has quit [Ping timeout: 260 seconds]
cristian_c has quit [Quit: Sto andando via]
<rperier> mmind00: what is the bqcurie2 exactly ? it has the same hardware topology than the marsboard ? this is the dts that I use to work
<rperier> (most of things work fine for now)
<karlp> it's a tablet from mmind's company
<karlp> rperier: you've got ethernet on the marsboard working right? I'd love to see you rmods to the curie dts,
<karlp> I've not managed to find the phy irq pin yet on my board, but I might have still been missing something else there
<rperier> karlp: oh okay. Yeah I am working on it . I add support to the dts to get emac_rockchip works on the board
bengal has quit [Ping timeout: 258 seconds]
<rperier> I found the right regulator to power on the phy, I am investigating to find the right pins for rmii (everything is in the kernel 3.0.8). I did not find the phy irq yet
<karlp> phyirq I thought I found int he marsboard schematics,
<karlp> unfortunately I tried that on my own board and it was't the right now.
<karlp> I used the marsboard schematics to help find the right power supply for the phy on my board too :)
<rperier> and you found "vaux33" ?
<karlp> hrm, not in front of it now, but no, I don't think so
<karlp> on my board the phy was direclty powered of vccio
<AstralixNB> rperier, what board do you have?
<rperier> AstralixNB: marsboard-rk3066
<karlp> rperier: check with me again in about 6 hours or so and I can put up my .dts, but it's not working for ethernet yet, it's just a lightly tweaked bqcurie file
<AstralixNB> rperier: CM or SOM?
<rperier> okay, thanks
<karlp> what's CM stand for?
<AstralixNB> guess one is a single board, th other a system on module that requires a carrier board
<AstralixNB> SOM
<AstralixNB> Ah, I see... CM ist the carrier module providing the interfaces, SOM ist the plugin board carrying SOC, Memory and such
<AstralixNB> RMII_INT is at GPIO1_D2/CIF1_CLKIN
<AstralixNB> Schematics are
<rperier> yeah this is the schematics I used
<AstralixNB> so you need emac irq? that is the one I told above
<AstralixNB> ok, it is the IRQ... of the LAN8720A
<rperier> this is what we were looking for, the phy irq
<AstralixNB> SO I found your IRQ... Can I have a working driver in an hour?
<AstralixNB> :)
<karlp> AstralixNB: yeah, easy to get the irq for the marsboard :) schematics help :)
FreezingAlt has quit [Ping timeout: 260 seconds]
<rperier> AstralixNB: in an hour ? with popcorn and an ice scream too ?
<rperier> I do this in my free time
<rperier> this is not my job
<rperier> ;)
<AstralixNB> I am just making fun
<rperier> (it was ironic) yeah I know, in the week I think it's possible ;)
<rperier> well, in fact, I need the phy irq, its base register, all the bank gpio and the pins for mdio, the right regulator to power on the phy... with schematics and the kernel 3.0.8 (https://github.com/powermcu/rockchip-marsboard-rk3066-linux-3.0.8) I should find all informations I need
<rperier> :)
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
<naobsd> btw, rk 3.0 kernel doesn't use phy irq, it does polling
<karlp> yeah, that was my fear, that the phy irq wasn't even connected.
<karlp> man, I tried reading that file myself, never noticed the poll line,
<rperier> naobsd: thanks for these informations, I did not know. good to know
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
<hramrach_> naobsd: if l/L is for uploading bootloader to memory how do you run it?
<naobsd> hramrach_: why do you still believe my description?
<naobsd> hramrach_: my description is wrong. you can see what happen if you try it
<hramrach_> I am trying to get to some intelligible description of the new command
<hramrach_> so that I or anyone else can try it and see if it works as described or not
<naobsd> hramrach_: no need to read description. try and see it if you can believe your eyes.
<hramrach_> I want to know what it is *supposed* to do so that I can try it out and if it really works write it down in the readme
<hramrach_> so that everybody can see if that command is useful to them and worth even trying just looking at the description
<hramrach_> and BTW I possibly do not have the correct loader for it to do anything useful on *my* device so in case I try and it just crashes I would still have to rely on your description only
<naobsd> hramrach_: when I found error on wiki you wrote, you said "you can fix wiki"
<naobsd> hramrach_: you found my error, you can fix it
<hramrach_> and you never did
<hramrach_> for that I have to know what is the correct, full description of the feature
<naobsd> I never write anything only from assuming
<hramrach_> seriously, I can never test every feature with single piece of hardware. So either I will have to assume that description you or somebody else provides here or elsewhere is correct and intelligible or the wiki will never have any information at all
<hramrach_> if the wiki had some information already that would allow me to try something I could try it and correct whatever needs correcting
<naobsd> hramrach_: what you did is not reading somebody's description, you don't read anything, you don't believe anyone, you just believe your assuming is always correct
<naobsd> hramrach_: no excuse please, just do try it yourself, even if you cannot get "usable support" from community
<hramrach_> but as it is there is pretty much nothing that gives practical, step by step guide for anybody who knows nothing about particulars of rk devices to do something useful
<naobsd> I just talking about l/L thing now
<hramrach_> yes, the description of that feature is missing a lot of details
<naobsd> I never say everyone must try every feature himself
<naobsd> hramrach_: no need to use time to excuse, try it
<FergusL> still no hacked 3288 devices with a decent linux running on it ?
<hramrach_> ok, I give up on this. I don't need loading bootloader into memory
<hramrach_> whoever wants it and has big enough collection of rk bootloaders to have one that likely works for this purpose let them try
<hramrach_> naobsd: seriously, I don't ask these questions to just disturb you. You succeeded in doing something and wrote a feature for that in rkflashtool. But from the description I cannot tell what exactly it is and how I would use it.
<hramrach_> so I ask questions to clarify the description so that when somebody else who does not understand rk hardware can read it and understand what the damn command is for
<hramrach_> maybe I make stupid assumptions about the nature of the feature
<hramrach_> but there is nothing in its description to contradict what I assume so the description is incomplete at the very least
<rperier> the code from kernel 3.0 is really ugly :/
hipboi has quit [Ping timeout: 245 seconds]
naobsd_ has joined #linux-rockchip
naobsd has quit [Ping timeout: 246 seconds]
GriefNorth has quit [Ping timeout: 240 seconds]
<rperier> karlp: look at http://git.jp.linux-rockchip.org/?p=rk3188_rk3066_r-box_android4.4.2_sdk.git;a=blob;f=kernel/arch/arm/mach-rk30/board-rk30-sdk-vmac.c;h=5d254b4ff93bbdd804ac964296b1bda32d31d2a9;hb=refs/heads/master , regulator vaux33 is used
<rperier> (ethernet support seems to be splitted between arch/arm/mach-rk30/board-rk30-sdk-vmac.c and drivers/net/rk29_vmac.c)
hipboi has joined #linux-rockchip
<rperier> I/O handling are in mach-rk30 and the rest is in rk29_vmac.c
<rperier> at least, this is what I understood
<rperier> must of the iomux stuffs are shared between pinctrl and the devicetree now
<rperier> s/must/most/
<naobsd_> try to always-on for all regulator output
<naobsd_> if ethernet doesn't work, try to change every free gpio out/high and out/lo
<naobsd_> ;)
FreezingCold has joined #linux-rockchip
<naobsd_> power may be supplied from regulator directly, or via gpio
<karlp> rperier: yeah, the marsboard phy is powered differently than on the minix x5 mini I have, I traced the power pin on phy to the regulator myself
<rperier> naobsd_: I agree, this is what I was doing yesterday at home
<rperier> naobsd_: assuming the mac does not connect to the phy correctly (because I don't know all its pins yet), if I force the regulator to be always-on, the phy should be powered anyway, right ? (the led next to the ethernet plug should be powered)
<karlp> depends what the led is for :)
<karlp> might be a link light, might be nothing if you don't configure it in the phy?
<rperier> most of the time you have a powered on led and a led for tx/rx :)
<rperier> mhhh...
<karlp> "most" ?
<karlp> on most switch gear I've worked with you get one for _link_ and one for traffic
<karlp> not for power on the phy,
<karlp> because only a developer would care if it was on or off, and it's normally always on for any useful purpose for n indicator light
<naobsd_> mac-phy connection should be good if phy can be probed
<naobsd_> phy irq is input(no need to configure) and can be ignored when polling mode
<rperier> yeah if you pass the right gpio banks and pins to the mac in the devicetree :)
<naobsd_> power may be supplied via gpio
<naobsd_> well
<naobsd_> pins for mii is fixed
<naobsd_> ^if my memory is collect
<naobsd_> power/irq line can be vary
<naobsd_> default pinctrl for mii should work
<rperier> pins are fixed ? seriously ?
<karlp> naobsd_: yeah, it can probe, but I guess the question for rperier is whether the mainline emac code supports polling mode?
<naobsd_> (if my memory is correct)there is a set of pins for mii, if ethernet is not needed, it can be used for other purpose such as gpio
<naobsd_> but any other pins can be used for mii
<naobsd_> ^wrong
<naobsd_> but any other pins cannot be used for mii
<rperier> the arc_emac driver does both
<rperier> irq and polling
<rperier> (polling throught napi)
<rperier> (arc_emac is the core driver, emac_rockchip is on top of it)
<rperier> I don't want to modify the driver for that, the phy is the same than on the rock, the mac works almost on the same way. So irq should be supported
<karlp> I'm more curious about if it works in polling mode if my board doesn't have the irq,
<karlp> and how to configure the dts entry to make it use polling
<naobsd_> GPIO3_C0-D1 can be configured for mii
<rperier> karlp: the irq just tells to the driver that packets have been received, then the irq handler schedule an napi call to call the "poll" function ;)
<naobsd_> no other pins cannot be configured(used) for mii
<naobsd_> phy irq and power is not included in GPIO3_C0-D1
<naobsd_> same for other pins
<rperier> which means, that it is possible to receive/send packets and access all information you need in polling mode outside a irq context
<naobsd_> generally, pinX, it can be used for function Y, or gpio. other pins cannot be used for funtion Y
<naobsd_> packet tx/rx interrupt is different
<naobsd_> mac is not in polling mode
<naobsd_> (^probably it can be, but another story)
<karlp> I'm only talking about phy, not mac, whether it's _possible_ that minix didn't connect the phy irq line, because they're using it in polling mode.
<naobsd_> you can access phy w/o interrupt
<naobsd_> well, I have to return to home now
<naobsd_> later
naobsd_ has quit [Quit: Page closed]
<karlp> I know I can access it, and probe it, but will it actually be _useable_ with the mainline arc_emac driver?
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
<rperier> rk29_vmac.c seems to use both... irq and polling...
<karlp> so, how do I tell the dts it doesn't need an irq?
<rperier> not sure that the poll function will be called by the kernel if it is not explicitly "scheduled" by the irq handler (I don't find informations about that...)
<rperier> (or scheduled by something explicitly)
<karlp> well, the irq line is probably hooked up, I just have to find it :)
<karlp> I just wasns't having much luck finding where in the 3.0 code it was declared which pin it was
Soopaman has joined #linux-rockchip
Soopaman1 has joined #linux-rockchip
Soopaman has quit [Read error: Connection reset by peer]
<rperier> the idea behing this polling system with netif_napi is : the interrupt simply informs the kernel that something happened on the phy , then the irq handler asks the kernel to call the polling function when it has time, the polling function is called once to collect all rx/tx packets, it avoids to interrupt the kernel all the time when a lot of packets are coming. the polling function itself should process all rx/tx packets and clean the BD
<rperier> ring buffer, nothing more. The irq is not only useful for packets but also for stats and errors for example. So that's important. (I am still a beginner in the linux kernel, so I might have forgot something...)
FreezingCold has quit [Ping timeout: 245 seconds]
naobsd has joined #linux-rockchip
<naobsd> generally, there is "interrupt driven" and "polling"
<rperier> [15:53:56] <rperier> http://www.linuxfoundation.org/collaborate/workgroups/networking/napi is very interesting about this part
<rperier> read this article
<naobsd> "interrupt driven": when something happen, interrupt handler access device to see and handle event
FreezingCold has joined #linux-rockchip
<naobsd> "polling": function access device periodically to see something happen and handle event if it happened
<karlp> "arc-emac": irq only sets a flag saying to come back and poll :)
<naobsd> interrupt handler shouldn't do so much work
<naobsd> it should just enqueue task when cpu is idle
<naobsd> task which should be done when cpu is idle
<rperier> this is exactly what happens with netif napi
<rperier> and polling functions
<naobsd> keywords: interrupt context, softirq, workqueue, and something I forgot
<rperier> when an irq is received, irq are disabled, the irq handler calls napi_schedule to tells to the kernel that something happens. The kernel then calls this polling function when it is in idle state and when it has time. Once everything have been ended and the workflow reduced, irq are re-activated
<rperier> s/ended/handled/
<naobsd> irq should be re-activated just after task is queued
<rperier> no
<naobsd> ah, yes and no
<rperier> again
<rperier> read this
<rperier> if the workflow is too important, irq should not be re-activated :)
<rperier> this how it works in arc_emac on mainline and how it seems to work in rk29_vmac.c (napi_schedule is called from vmac_intr)
<rperier> this is *
<naobsd> (I'm talking about generic thing) if there is more space to store next event, irq can be activated. if resource is full, irq cannot be activated handler/task/something else hadle event and make free space
<naobsd> some hardware may assert interrupt when space is full
<naobsd> some hardware may assert interrupt for every event
<naobsd> some hardware may have only space for 1 event
<naobsd> some hardware has "interrupt coalescing" function which reduce # of interrupt but it may make response speed worse
<naobsd> software can do polling regardless of interrupt line availability but polling interval should be enough short to avoid overflow
<naobsd> receive interrupt, set flag, but handler runs by polling == no need to see device every interval
<naobsd> arc_emac_intr() is for MAC
<rperier> oh yeah, about generic things, sure
<naobsd> well, sorry, if there is no space irq cannot be activated, is not correct.
<naobsd> input event from external cannot be stopped ;) overflow can happen regardless of active/not active irq
<naobsd> I'm not sure about napi, but irq re-reactivation is not needed until hander is called
<naobsd> because flag is set
<naobsd> no need to set more than once
<naobsd> ah, until handler is called -> until handler finishes his work
<naobsd> re-activate irq == "please call me when next event come"
<rperier> yeah this is how it works with napi, and how should work all recent network drivers
Soopaman1 has quit [Read error: Connection reset by peer]
<rperier> (according to the above article)
Soopaman has joined #linux-rockchip
<naobsd> (general talk) "interrupt or polling" is not only for network
<naobsd> and it's possible "only polling, no irq"
<naobsd> periodically access device and handle event, activate/deactivate irq can be omitted totally
<naobsd> ^device dependent, not for all
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
<naobsd> mmind00: ping
<naobsd> mmind00: R28/R89 dies here http://git.jp.linux-rockchip.org/?p=rk3288_r-box_android4.4.2_sdk.git;a=blob;f=u-boot/drivers/power/pmic/pmic_act8846.c;hb=HEAD#l271
<naobsd> mmind00: it should do change direction(in->out, level: low) and change level(high) http://git.jp.linux-rockchip.org/?p=rk3288_r-box_android4.4.2_sdk.git;a=blob;f=u-boot/drivers/gpio/rk_gpio.c;hb=HEAD#l131
<naobsd> mmind00: but kernel should do same on boot http://git.jp.linux-rockchip.org/?p=rk3288_r-box_android4.4.2_sdk.git;a=blob;f=kernel/drivers/regulator/act8846.c;hb=HEAD#l926
<naobsd> mmind00: gpio for power hold is defined in dts for R28/R89
<naobsd> mmind00: u-boot died, need to disable that line. but kernel works with that line
FreezingCold has quit [Read error: Connection reset by peer]
FreezingCold has joined #linux-rockchip
<naobsd> mmind00: probably power hold gpio is not used for shutdown (there is a change in Orion R28 SDK, it touches act8846 0xc3 register for shutdown)
<naobsd> mmind00: I think u-boot issue can be fixed by removing gpio definition for power hold line from R28/R89 dts
<naobsd> mmind00: but I'm not sure that gpio line is for what
<naobsd> mmind00: or is it better to disable gpio_direction_output() in u-boot by like "#ifdef NETXEON_R89" ?
<naobsd> or that line can be removed for all. default setting should fine for _boot_ because all board can run til gpio_direction_output(power hold)
<naobsd> I'm not sure it's R28/R89 specific issue or not...
<naobsd> my hotack board has definition for that line too, but it seens no effect, I can toggle that line and board still alive
<karlp> are there any tools/plugins to help with keeping the patch V6,7,8,9 log in mail clients?
<karlp> git-send-email doesn't seem to have it that I can see
naobsd has quit [Ping timeout: 246 seconds]
naobsd has joined #linux-rockchip
<naobsd> oops
naobsd has left #linux-rockchip [#linux-rockchip]
naobsd has joined #linux-rockchip
<naobsd> oops
naobsd has left #linux-rockchip [#linux-rockchip]
naobsd has joined #linux-rockchip
<naobsd> hm
<naobsd> let read act8846 manual...
<naobsd> ah, I see
<naobsd> pwrhld need to be asserted after power-on, and if pwrhld is deasserted, regulator output will be off
<naobsd> netxeon person said it's pull-ed up
<naobsd> it should be mistake
<naobsd> gpio_direction_output() deassert it = cannot boot
RayFlower has joined #linux-rockchip
<naobsd> :(
<naobsd> well, I cannot understand why kernel is ok
RayFlower has quit [Read error: Connection reset by peer]
<hramrach_> karlp: git-send-email has subject prefix in which you can specify V6 or foobar or whatever
<hramrach_> it does not automagically track which version you have, unfortunately
<naobsd> mmind00: it seems there is no definition about "gpio 6(A6)" in your mainline r89 dts
<naobsd> mmind00: can you try to change direction via sysfs and see what happen?
<naobsd> mmind00: when you return home, of course :)
arokux has quit [Remote host closed the connection]
arokux has joined #linux-rockchip
<naobsd> ok, I'll not disable gpio_direction_output() in u-boot, but change dts not to toggle gpio0-a6
<naobsd> ah
<naobsd> mmind00: "gpio 6" -> "gpio0 6"
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
arokux has joined #linux-rockchip
arokux has quit [Remote host closed the connection]
<karlp> hramrach_: yeah, was hoping something could track my past mailings, but I guess you have to just copy it in from old ones...
AstralixNB has quit [Ping timeout: 250 seconds]
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
<naobsd> ah, RK3188 GPIO0_A6 default is pull-up... I see
bengal has joined #linux-rockchip
hipboi has quit [Read error: Connection timed out]
hipboi has joined #linux-rockchip
arokux2 has joined #linux-rockchip
gb_master_ has joined #linux-rockchip
_whitelogger_ has quit [Ping timeout: 260 seconds]
_whitelogger_ has joined #linux-rockchip
_whitelogger_ has joined #linux-rockchip
_whitelogger__ has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
karlp has quit [Ping timeout: 255 seconds]
karlp has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
c0d3z3r0 has quit [Ping timeout: 246 seconds]
c0d3z3r0 has joined #linux-rockchip
anwar-rekik has joined #linux-rockchip
honx_ has quit [Ping timeout: 246 seconds]
anwar-rekik has quit [Remote host closed the connection]
anwar-rekik has joined #linux-rockchip
anwar-rekik has quit [Remote host closed the connection]
anwardz has joined #linux-rockchip
honx has joined #linux-rockchip
ganbold_ has quit [Ping timeout: 255 seconds]
PASAf has joined #linux-rockchip
<PASAf> Someone could share links to rk28xx SDK please?
<Astralix1> wow... RK28... That was far before my time...
<Astralix1> There are even no details about RK29xx
<PASAf> All I found is this screenshot http://i.imgur.com/y8H4A.png but guy that made it told me that he deleted all rockchip stuff.
<Astralix1> Sorry, checked my archives... There is lots about RK3xxx but no 29. I never had any 28 device
<Astralix1> There where rumors about 29xx devices never really had a software manual as engineers took all information from the ASCI design files directly.
<Astralix1> *ASIC
<Astralix1> I have no rumors about 28xx devices
<PASAf> There are still bunch of RK28xx devices in China and Russia. E-ink readers in the majority.
Soopaman has quit [Quit: Leaving.]
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
gb_master_ has quit [Remote host closed the connection]
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
npcomp has quit [Ping timeout: 272 seconds]
npcomp has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
benja` is now known as benja
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
_whitelogger__ has quit [Ping timeout: 260 seconds]
_whitelogger__ has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
_whitelogger__ has quit [Excess Flood]
hramrach_ has quit [Ping timeout: 264 seconds]
_whitelogger__ has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
arokux2 has quit [Remote host closed the connection]
RayFlower has joined #linux-rockchip
arokux2 has joined #linux-rockchip
_whitelogger__ has joined #linux-rockchip
PASAf has quit [Quit: Page closed]
hramrach_ has joined #linux-rockchip
gb_master_ has joined #linux-rockchip
_whitelogger__ has quit [Excess Flood]
_whitelogger__ has joined #linux-rockchip
_whitelogger__ has quit [Excess Flood]
_whitelogger__ has joined #linux-rockchip
_whitelogger__ has quit [Excess Flood]
_whitelogger__ has joined #linux-rockchip
RayFlower has quit [Quit: RayFlower]
RayFlower has joined #linux-rockchip
_whitelogger__ has joined #linux-rockchip
_whitelogger__ has joined #linux-rockchip
nighty^ has quit [Quit: Disappears in a puff of smoke]
_whitelogger__ has joined #linux-rockchip
gb_master_ has quit [Remote host closed the connection]
anwardz has quit [Ping timeout: 260 seconds]
_whitelogger__ has quit [Excess Flood]
_whitelogger__ has joined #linux-rockchip
benja is now known as benja`
Soopaman has joined #linux-rockchip
_whitelogger__ has quit [Excess Flood]
_whitelogger__ has joined #linux-rockchip
_whitelogger__ has joined #linux-rockchip
_whitelogger__ has quit [Excess Flood]
_whitelogger__ has joined #linux-rockchip
RayFlower has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
_whitelogger__ has joined #linux-rockchip
_whitelogger__ has quit [Excess Flood]
_whitelogger__ has joined #linux-rockchip
_whitelogger__ has joined #linux-rockchip
_whitelogger__ has quit [Excess Flood]
_whitelogger__ has joined #linux-rockchip
arokux2 has quit [Ping timeout: 260 seconds]
FreezingCold has quit [Ping timeout: 250 seconds]
Soopaman1 has joined #linux-rockchip
_whitelogger__ has quit [Excess Flood]
Soopaman has quit [Read error: Connection reset by peer]
_whitelogger__ has joined #linux-rockchip
FreezingCold has joined #linux-rockchip
RayFlower has quit [Read error: Connection reset by peer]
RayFlower has joined #linux-rockchip
_whitelogger__ has quit [Excess Flood]
_whitelogger__ has joined #linux-rockchip
Soopaman has joined #linux-rockchip
Soopaman1 has quit [Read error: Connection reset by peer]