00:54
tilpner has quit [Ping timeout: 240 seconds]
01:13
tilpner has joined #nixos-aarch64
01:44
chiefgoat has quit [Ping timeout: 252 seconds]
01:52
chiefgoat has joined #nixos-aarch64
01:55
chiefgoat has quit [Client Quit]
01:56
chiefgoat has joined #nixos-aarch64
01:59
chiefgoat has quit [Client Quit]
02:10
orivej has quit [Ping timeout: 276 seconds]
02:16
h0m1 has quit [Ping timeout: 246 seconds]
02:18
h0m1 has joined #nixos-aarch64
02:19
chiefgoat has joined #nixos-aarch64
03:46
andi- has quit [Remote host closed the connection]
03:53
andi- has joined #nixos-aarch64
06:04
Acou_Bass has quit [Ping timeout: 268 seconds]
06:09
Acou_Bass has joined #nixos-aarch64
06:12
zupo has joined #nixos-aarch64
06:21
Acou_Bass has quit [Ping timeout: 265 seconds]
06:28
Acou_Bass has joined #nixos-aarch64
07:16
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
07:24
Acou_Bass has quit [Ping timeout: 240 seconds]
07:31
Acou_Bass has joined #nixos-aarch64
08:00
Acou_Bass has joined #nixos-aarch64
08:13
kolbycrouch has quit [Remote host closed the connection]
08:18
zupo has joined #nixos-aarch64
08:28
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
09:17
zupo has joined #nixos-aarch64
09:36
Thra11_ has quit [Ping timeout: 268 seconds]
09:49
Thra11_ has joined #nixos-aarch64
10:20
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
10:39
h0m1 has quit [Quit: WeeChat 2.6]
10:40
h0m1 has joined #nixos-aarch64
10:51
h0m1 has quit [Quit: WeeChat 2.6]
10:52
h0m1 has joined #nixos-aarch64
10:59
FRidh has joined #nixos-aarch64
11:22
orivej has joined #nixos-aarch64
11:30
lopsided98_ has joined #nixos-aarch64
11:31
lopsided98 has quit [Ping timeout: 245 seconds]
11:48
Thra11_ has quit [Ping timeout: 240 seconds]
12:01
Thra11_ has joined #nixos-aarch64
12:50
Thra11_ has quit [Ping timeout: 240 seconds]
13:04
Thra11_ has joined #nixos-aarch64
13:11
hexa- has quit [Quit: WeeChat 2.6]
13:13
hexa- has joined #nixos-aarch64
13:39
kai_w has joined #nixos-aarch64
15:04
h0m1 has quit [Quit: WeeChat 2.6]
15:05
h0m1 has joined #nixos-aarch64
15:12
kai_w has quit [Quit: Konversation terminated!]
15:31
zupo has joined #nixos-aarch64
15:37
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
16:25
Thra11_ has quit [Ping timeout: 240 seconds]
16:38
Thra11_ has joined #nixos-aarch64
16:40
pbb has joined #nixos-aarch64
16:43
pbb has quit [Remote host closed the connection]
16:44
h0m1 has quit [Quit: WeeChat 2.6]
16:45
pbb has joined #nixos-aarch64
16:45
h0m1 has joined #nixos-aarch64
17:18
zmacs has joined #nixos-aarch64
17:20
zmacs has quit [Client Quit]
17:20
zmacs has joined #nixos-aarch64
18:16
Thra11_ has quit [Ping timeout: 240 seconds]
18:30
Thra11_ has joined #nixos-aarch64
18:33
adisbladis has joined #nixos-aarch64
19:11
zupo has joined #nixos-aarch64
19:12
jslight90 has quit [Quit: Leaving]
19:15
Thra11_ has quit [Ping timeout: 240 seconds]
19:31
Thra11_ has joined #nixos-aarch64
19:45
FRidh has quit [Quit: Konversation terminated!]
19:57
tilpner_ has joined #nixos-aarch64
19:57
tilpner has quit [Ping timeout: 265 seconds]
20:09
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
20:17
zupo has joined #nixos-aarch64
20:40
tilpner_ is now known as tilpner
20:44
Thra11_ has quit [Ping timeout: 265 seconds]
21:17
<
samueldr >
nope, didn't know about that
21:22
<
samueldr >
looks like it's new
21:23
<
samueldr >
>> scrapes websites and wikis of known mobile Linux distributions
21:23
<
samueldr >
ah, then I guess it's coming soon :)
21:58
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
22:17
Thra11_ has joined #nixos-aarch64
22:46
THFKA4 has quit [Ping timeout: 246 seconds]
22:47
THFKA4 has joined #nixos-aarch64
23:02
THFKA4 has joined #nixos-aarch64
23:02
THFKA4 has quit [Changing host]
23:02
tilpner_ has joined #nixos-aarch64
23:04
tilpner has quit [Ping timeout: 240 seconds]
23:08
tilpner_ has quit [Ping timeout: 268 seconds]
23:08
tilpner_ has joined #nixos-aarch64
23:35
<
clever >
samueldr: currently, ~154kb of the 512kb of flash is used
23:35
<
samueldr >
memsys** could those be DDR training data?
23:35
<
clever >
maybe, they are all 21kb
23:35
<
clever >
-rw-r--r-- 1 clever users 21K Nov 11 19:29 memsys00.bin
23:35
<
clever >
-rw-r--r-- 1 clever users 21K Nov 11 19:29 memsys01.bin
23:35
<
clever >
-rw-r--r-- 1 clever users 21K Nov 11 19:29 memsys02.bin
23:36
<
clever >
google only found those filenames in a single place, a chat log from a user in #raspberrypi , who got zero replies, and his username hasnt been seen since
23:36
<
samueldr >
someone asked on IRC in the past, but had no replies it seems
23:36
<
samueldr >
[18:08] <laptop2> LPDDR4 initialization & calibration?
23:36
<
samueldr >
[18:08] <laptop2> what's the role of memsys00 to memsys07 on RPi4?
23:36
<
clever >
thats the one :P
23:37
<
samueldr >
I wonder if the strings memsys are in the bootcode
23:37
<
samueldr >
and if it's something that "obviously looks at an identifier and loads the right memsys"
23:37
<
clever >
they are present in the blob with magic 55aaf00f
23:37
<
clever >
i suspect 55aaf00f is the old bootcode.bin file
23:37
<
clever >
another thing of note, is that there is ~358kb of 0xff's present
23:38
<
clever >
and now that we know the structure, we could just add more files, with regular (16 byte max) filenames
23:38
<
samueldr >
could it be simply to always write a full image?
23:38
<
clever >
and the firmware would just ignore the extra files
23:38
<
clever >
i believe the official tools already write full images
23:38
<
clever >
you dump the flash->bin, then use a python tool to mutate the bootconf.txt at the end, then burn bin->flash
23:38
<
samueldr >
hmm, the current rpi4 build for u-boot is ~432K
23:39
<
samueldr >
wondering how much slimming can happen
23:39
<
samueldr >
while still having it be useful
23:39
<
clever >
samueldr: edited the gist to add more filesize info
23:39
<
samueldr >
we need to add 40K for a dtb
23:40
<
samueldr >
though that might also be something that can be reduced
23:40
<
clever >
assuming you only need a single memsys, for the chip currently on the board
23:40
<
samueldr >
hmmm... if we drop the whole network stuff
23:40
<
clever >
you could delete the others, and trim it to 85kb
23:40
<
clever >
which gives you 426kb free
23:40
<
samueldr >
also drop extlinux, to be left with uef
23:40
<
clever >
the bigger problem, is that bootcode.bin can only load start.elf
23:40
<
clever >
(which is 2mb)
23:40
<
clever >
start.elf is then responsible for bringing the ARM online
23:40
<
samueldr >
ah, forgot about that part
23:41
<
clever >
so, you would have to start over, with the open-source firmware
23:41
<
clever >
this gets you linux on the arm, without any gpu services (no audio, no hdmi, no composite video, no 3d, no hw accel h264)
23:42
<
clever >
in theory, it could be ported to the rpi4, and have a uboot image baked into it
23:42
<
clever >
then linux would be responsible for loading the "gpu firmware" later on
23:42
<
clever >
(once its even developed)
23:42
<
samueldr >
without any video output, that's not really useful in the end for many applications :)
23:43
<
samueldr >
like, selecting a boot option in grub
23:43
<
clever >
serial port is easy (ive already done it)
23:43
<
clever >
usb/ethernet/sd will probably work
23:44
<
clever >
also of note, the bootcode.bin half of things, must be under 128kb
23:44
<
clever >
because thats the size of the L2 cache
23:44
<
clever >
its main job is to just turn on dram, so the fat start.elf can move in
23:44
<
clever >
-rw-r--r-- 1 clever users 52K Nov 2 18:04 /home/clever/apps/rpi/firmware/boot/bootcode.bin
23:44
<
clever >
interesting
23:45
<
clever >
the rpi4 bootcode (if i'm looking at the right file) isnt much bigger
23:45
<
clever >
just 9kb bigger
23:46
<
clever >
[clever@amd-nixos:~/apps/rpi/rpi-eeprom/firmware/beta]$ /nix/store/rr7860kh97077r72mvrdlp8yp5yj7vnr-vc4-elf-binutils-2.31.1/bin/vc4-elf-objdump -b binary -D -m vc4 main-body.bin | less
23:46
<
clever >
208: e0 00 version r0
23:46
<
clever >
204: f0 81 00 c0 addcmpbne r0,-1,0x0,0x204
23:46
<
clever >
200: 00 b0 00 01 mov r0,256
23:47
<
clever >
samueldr: boom!, that looks extremely similar to the start of the real bootcode.bin!
23:47
<
clever >
[clever@amd-nixos:~/apps/rpi/rpi-eeprom/firmware/beta]$ /nix/store/rr7860kh97077r72mvrdlp8yp5yj7vnr-vc4-elf-binutils-2.31.1/bin/vc4-elf-objdump -b binary -D -m vc4 ~/apps/rpi/firmware/boot/bootcode.bin | less
23:47
<
clever >
200: e0 00 version r0
23:47
<
clever >
they just added another 2 opcodes at the start
23:48
<
clever >
that makes me wonder, how does recovery.bin un-brick the pi...
23:48
<
clever >
its only 60kb
23:48
<
clever >
thats 1kb smaller then the bootcode.bin in the actual spi flash....
23:49
<
clever >
Verify VL805 EEPROM
23:49
<
clever >
Updating VL805
23:49
<
clever >
samueldr: i see strings in recovery.bin, that deal with updating the usb controller firmware
23:50
<
clever >
File has Invalid/missing signature: '%s' (%u)
23:50
<
clever >
Writing EEPROM
23:50
<
clever >
and the rpi firmware
23:50
<
clever >
Failed to rename recovery.bin
23:50
<
clever >
and i think the recover sd card will "self destruct" (rename recovery.bin)
23:51
<
clever >
pieeprom.upd
23:51
<
clever >
pieeprom.bin
23:51
<
clever >
pieeprom.sig
23:51
<
clever >
it expects those files to be on the sd card
23:52
<
clever >
DHCP_TIMEOUT
23:53
<
clever >
NET_CONSOLE
23:53
<
clever >
thats fishy...
23:53
<
clever >
samueldr: most of the strings in this region, are bootconf.txt entries
23:53
<
clever >
NET_CONSOLE isnt documented...
23:53
<
samueldr >
ooh, fun
23:54
<
clever >
they have a zip here, with pieeprom.bin pieeprom.sig README.txt recovery.bin vl805.bin vl805.sig
23:54
<
clever >
and i can confirm, the sig files are only sha256 hashes of the matching bin's
23:55
<
clever >
so you can just reuse recovery.bin to flash your own custom firmware
23:55
<
clever >
no need to un-brick to foundation firmware, then flashrom back to custom
23:57
<
clever >
MEMSYS INIT failed: %d
23:57
<
clever >
MEMSYS DPFE failed: %d
23:57
<
clever >
samueldr: memsys also appears in other strings
23:58
<
clever >
Bad EEPROM capacity: 0x%x
23:58
<
clever >
samueldr: the blob in the spi flash, may abort if the spi flash is the wrong size??
23:59
<
clever >
PXEClient:Arch:00000:UNDI:002001