f_ changed the topic of ##raspberrypi-internals to: The inner workings of the Raspberry Pi (Low level VPU/HW) -- for general queries please visit #raspberrypi -- open firmware: https://librerpi.github.io/ -- VC4 VPU Programmers Manual: https://github.com/hermanhermitage/videocoreiv/wiki -- chat logs: https://libera.irclog.whitequark.org/~h~raspberrypi-internals -- bridged to matrix and discord
<bonda_000> somehow cpi.h and cam0.h both are at 7e800000 and register names are different, but this CPI bit may change the way how these registers are used
<bonda_000> tried poking at those AVEIN_ clocks but no success
<bonda_000> I probably did it all wrong and I need to set clocks to ALT3 rather than ALT4
<bonda_000> but, tomorrow. now I'm gonna get some rest
<clever> bonda_000: alt 3/4 on the first 16, is the aveout and avein from https://elinux.org/RPi_BCM2835_GPIOs#
<clever> its possible that cpi is only present on other chips, that dont have unicam
<clever> and its a choice by broadcom about which one to include in the chip
bonda_000 has quit [Ping timeout: 256 seconds]
bonda_000 has joined ##raspberrypi-internals
<bonda_000> clever: I get it
<bonda_000> but if I'm taking ADC values in, and supplying the clock to it, data bits should be AVEIN_ and clocks should be AVEOUT_
<bonda_000> by the logic of things, if that's how it works
bonda_000 has quit [Quit: Leaving]
bonda_000 has joined ##raspberrypi-internals
jcea has quit [Ping timeout: 240 seconds]
bonda_000 has quit [Remote host closed the connection]
bonda_000 has joined ##raspberrypi-internals
<bonda_000> clever: what is the start_db.elf file? It weight an extra 1.1MB compared to my current elf
<clever> bonda_000: a debug build, it has a lot more logging statements
<bonda_000> oh nice
<clever> the idea being, if you have a bug with the firmware, set start_db=1, reproduce the bug, and send it to the engineers
<clever> i'm not sure if db has the extra stuff from start_x
<bonda_000> I just tried with AVEOUT_ clocks and there's nothing on the scope
<bonda_000> I fed CM_CAM0CTL with PLLD as a source
<bonda_000> then did something similar to the unicam_reset_cpi()
<bonda_000> there are descriptions of CAM bitfields in the android drop in mach-rhea but its all acronyms, hard to understand
<bonda_000> here
<bonda_000> maybe you can recognize some of these since you've been doing this for several years now
<bonda_000> What I see so far, is there is a set of registers that starts with an 'I': ICTL, ISTA, IDI and that stands for CTL, STA, DI for some "I"
<clever> bonda_000: nothing immediately stands out from there
<clever> but the DAT0 and DAT1 kind of feel like this is a 2 lane CSI interface
<bonda_000> and a set of registers that starts with a 'D': DCS, DBSA, DBEA that stands for CS, BSA, BEA for some "D"
<bonda_000> BEA is a buffer I believe
<bonda_000> so a D-buffer and an I-buffer
<bonda_000> #define CAM_MISC_CPI_SELECT_SHIFT 17
<bonda_000> that's the one I mentioned earlier
<bonda_000> #define CAM_ANA_PTATADJ_SHIFT 8
<bonda_000> #define CAM_ANA_CTATADJ_SHIFT 4
<bonda_000> PTATADJ is a '7' when nothing's going on
<bonda_000> same as CTATADJ
<bonda_000> also
<bonda_000> #define CAM_ANA_AR_SHIFT 2
<bonda_000> #define CAM_ANA_BPD_SHIFT 1
<bonda_000> #define CAM_ANA_APD_SHIFT 0
<clever> i should get off to bed now, its 27 past midnight
<bonda_000> makes a '7'
<bonda_000> right
<bonda_000> I couldnt go sleep
<bonda_000> nothing works and I can't even take a scope capture
<bonda_000> trying some linux freeware atm
dolphinana has joined ##raspberrypi-internals
dolphinana has quit [Remote host closed the connection]
bonda_000 has quit [Ping timeout: 260 seconds]
f_ has joined ##raspberrypi-internals
f_ has quit [Quit: To contact me, send a memo using MemoServ, PM f_[xmpp], or send an email. See https://vitali64.duckdns.org/.]
f_ has joined ##raspberrypi-internals
f_ has quit [Changing host]
f_ has joined ##raspberrypi-internals
f_ has quit [Client Quit]
f_ has joined ##raspberrypi-internals
f_ has quit [Quit: To contact me, send a memo using MemoServ, PM f_[xmpp], or send an email. See https://vitali64.duckdns.org/.]
pavelow has quit [Quit: No Ping reply in 180 seconds.]
pavelow has joined ##raspberrypi-internals
jcea has joined ##raspberrypi-internals
Stromeko has quit [Ping timeout: 260 seconds]
Stromeko has joined ##raspberrypi-internals
vriska has quit [Read error: Connection reset by peer]
vriska has joined ##raspberrypi-internals
bonda_000 has joined ##raspberrypi-internals
bonda_000 has quit [Quit: Leaving]
f_ has joined ##raspberrypi-internals
bonda_000 has joined ##raspberrypi-internals
bonda_000 has quit [Client Quit]
f_ has quit [Quit: To contact me, send a memo using MemoServ, PM f_[xmpp], or send an email. See https://vitali64.duckdns.org/.]
bonda_000 has joined ##raspberrypi-internals
<bonda_000> clever: any new findings?
<clever> none yet
<clever> been busy today with groceries and fixing the washing machine
<bonda_000> clever: do you know what a hisr/lisr ir?
<bonda_000> is*
<bonda_000> there are plenty of functions named like this in different blocks
<clever> bonda_000: interrupt service routine
<bonda_000> and what is h/l?
<bonda_000> high/low?
<clever> no clue
<bonda_000> and what does vchiq stand for?
<bonda_000> i just realized something
<bonda_000> I always associated vchiq with mmal
<bonda_000> vchiq_mmal, the communication between linux and vpu
<bonda_000> but mmal appears to be a part of something called "vchiq"
<bonda_000> on the vpu
<bonda_000> vchi to be more general
<bonda_000> found it
<bonda_000> Videocore Host Interface
<clever> bonda_000: omx, mmal, broadcom gl, dispmanx, and many more, all run over vchiq
<bonda_000> ilcs = OpenMAX IL component service
<bonda_000> is also a part of it
<clever> there is also a shared memory service
<clever> which lets you allocate objects in the gpu mem (or import arm objects)
<clever> and then you can use those objects in the other services
<bonda_000> so this OS is built around the servicing requests from another processor
<bonda_000> threadx part is very thin and is just responsible for multitasking
<bonda_000> so it doesn't really do much on its' own until it gets a request from ARM
<clever> yeah, thats the current design
<bonda_000> and that's when a task is created, that allocates required components, ports
<clever> the old broadcom gl would RPC every bbl
<bonda_000> every what?
<clever> every opengl function call to the firmware
<bonda_000> I noticed
<bonda_000> some camera data labels are actually missing in the unstripped elf
<bonda_000> if you set the gp=0ef07e00 on unicam_open()
<bonda_000> does it look like this to you?
<bonda_000> all the &DAT's
bonda_000 has quit [Quit: Leaving]