Mirrors/m1n1
2
0
Fork 0
mirror of https://github.com/AsahiLinux/m1n1 synced 2024-11-23 07:03:08 +00:00
Code Issues Projects Releases Packages Wiki Activity
1455 commits 13 branches 49 tags 8.6 MiB
Commit graph

906 commits

Author SHA1 Message Date
Asahi Lina
cb5e8dbf3f trace/agx.py: Update to new channel stuff names 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:46:01 +09:00
Asahi Lina
033bf7a130 m1n1.fw.agx.channels: Fix a bunch of structs, s/Channels/ChannelInfoSet/ 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:40:46 +09:00
Asahi Lina
ebe18e1fac m1n1.agx: Add .work() and .log() 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:40:46 +09:00
Asahi Lina
3bcc2243b5 m1n1.agx: Fix allocator mapping flags 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:40:46 +09:00
Asahi Lina
bf73b7f671 m1n1.agx.object: GPUAllocator: Support configurable page sizes & context 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:40:46 +09:00
Asahi Lina
c78e92e9ad m1n1.agx: Add global object tracker 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:40:46 +09:00
Asahi Lina
b7da00f761 m1n1.agx.object: Improve GPUObject 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:40:46 +09:00
Asahi Lina
d531af3a6d m1n1.fw.agx: Add address decoding & iotranslate 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:40:46 +09:00
Asahi Lina
8a5ab0047a m1n1.fw.agx: Rename endpoints & fix/add proper messages 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:40:46 +09:00
Asahi Lina
0fcaf69728 m1n1.hw.uat: Add RegMap accessor 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:28:54 +09:00
Asahi Lina
d7a4f30043 m1n1.agx: Use new UAT init API 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:28:54 +09:00
Asahi Lina
8d808904a0 m1n1.hw.uat: Handle multiple contexts 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:21:18 +09:00
Asahi Lina
3a773c8a6e m1n1.fw.agx and friends: Add __all__ to lots of modules 
This unborks reloading, since otherwise non-reloaded modules can clobber
symbols from reloaded modules with their old versions

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:17:33 +09:00
Asahi Lina
6175739132 m1n1.hw.asc: Add work_for(time) 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:12:31 +09:00
Asahi Lina
1dad2dad6a m1n1.fw.asc.crash: Decode registers and stacktraces 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:12:23 +09:00
Asahi Lina
eb15ba0102 m1n1.adt: Decode more GPU properties 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:11:23 +09:00
Asahi Lina
4dcdb608a7 m1n1.trace: Fix PrintTracer address decoding 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:10:57 +09:00
Asahi Lina
682333e520 m1n1.proxyutils: Add 512-bit read/write accessors 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:04:51 +09:00
Asahi Lina
f13489f516 m1n1.hv: Fix atomic instruction mask 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:04:31 +09:00
Asahi Lina
0a25a3ef0d m1n1.hv: Fix trace_range() 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:04:09 +09:00
Asahi Lina
e1e8d05ead m1n1.constructutils: Fix building ConstructValueClass 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:03:21 +09:00
Asahi Lina
a9ae6b81fa m1n1.constructutils: Add a regmap() to make a RegMap out of a Construct 
Handy for ringbuf pointer type structures where we need to be able to
read/write individual members using atomic primitives, without going
through Construct.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:03:21 +09:00
Asahi Lina
892ee0e76b m1n1.utils: Split RegMap into BaseRegMap (without metaclass) 
This allows alternative BaseRegMap subclasses to dynamically build the
map instead of doing the metaclass stuff.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:03:19 +09:00
Asahi Lina
28fdcb4284 m1n1.constructutils: Improve str representation, add dump() for Python 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:03:19 +09:00
Asahi Lina
4ad733493b m1n1.constructutils: Add a global struct address tracer 
Currently unconditional, kind of hacky. Good for HV use.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:03:19 +09:00
Asahi Lina
de4f024ced m1n1.constructutils: Make the Pointer<->addr association two-way 
Also add Lazy support

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 23:03:17 +09:00
Asahi Lina
570ea8c4b8 m1n1.constructutils: Add .sizeof() 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 22:57:32 +09:00
Asahi Lina
83dfa1ef70 m1n1.constructutils: Move _off to class, improve addr stuff 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 22:56:52 +09:00
Asahi Lina
029ba73f44 m1n1.constructutils: Make Dec buildable 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 22:54:21 +09:00
Asahi Lina
1c35ce60c7 m1n1.constructutils: Better string representations 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 22:53:57 +09:00
Asahi Lina
3fe8f3e72f m1n1.constructutils: Improve recursive reload 
Reload more things, but also avoid reloading the same class multiple
times.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-06-02 22:53:17 +09:00
Hector Martin
714420a694 display: Put DCP to sleep if display is external 
This stops DCP from killing our modeset if the connection cycles.

Also force a (potential) configure cycle if the display is external;
this makes sure updated stage2s will have a chance at fixing issues of
old stage1s. Modesetting is fast when it's the same mode as before.

Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-06-01 01:54:11 +09:00
Hector Martin
2bd6060f43 experiments/dcp_iboot.py: Fully shut down DCP when done 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-06-01 00:57:17 +09:00
Hector Martin
e9f383aa6b m1n1.fw.dcp.dcpav: Add some DCPAV EPIC stuff 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-06-01 00:57:17 +09:00
Hector Martin
5c69011fc7 m1n1.fw.dcp.iboot: Fix padding 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-06-01 00:57:17 +09:00
Hector Martin
58db65db63 experiments/dcp_iboot.py: Shut down display before doing anything 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-06-01 00:57:17 +09:00
Hector Martin
7282f30527 m1n1.fw.afk: Fix a bunch of brokenness 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-06-01 00:57:17 +09:00
Hector Martin
eb4483f8e1 hv/trace_dcp.py: Log hexdumps to the HV log 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-06-01 00:57:17 +09:00
Hector Martin
06ec66e2a9 proxy: Export pmgr_reset() 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-06-01 00:57:17 +09:00
Hector Martin
0472bc05cd proxy: Export display functions 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-06-01 00:57:17 +09:00
Hector Martin
9548ddd75d fw.asc: Autodetect whether to boot the ASC or wake it up 
Technically both can be done unconditionally, but let's be explicit in
the python side to see if we ever find this algorithm doesn't work.

Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-06-01 00:57:17 +09:00
Martin Povišer
60254a38ed experiments/ohmmeter.py: Make CS42L84 your ohmmeter 
Add script which prints out the (rough) measured impedance of connected
headphones on machines with the CS42L84 jack codec (2021 Macbook Pros).

  $ proxyclient/experiments/ohmmeter.py
  Waiting for plug... measuring... 64.3 ohms... yanked
  Waiting for plug... measuring... 697.1 ohms...

Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-31 21:51:53 +09:00
Martin Povišer
4d306bb0d8 m1n1.hw.i2c: Fix & extend register access 
Fix reading of registers (there have been STOPs issued between the two
messages which surprisingly went unnoticed until now). Also add support
for multibyte register addresses.

Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-31 21:51:53 +09:00
Martin Povišer
e1b50f9ced experiments/aop.py: Bring up AOP 
Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:27:04 +09:00
Martin Povišer
d924c97a0c tools/second_proxy.py: Add setup of second proxy stream 
Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Martin Povišer
e78cd14fee experiments/mmio_sweep.py: Add MMIO sweep tool 
Tool can do blanket sweep or isolate ranges associated with a power
domain.

Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Martin Povišer
96d4b5dd07 m1n1.proxyutils: Fortify against non-working disassembler 
If not caught it can lead to an ugly crash...

Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Martin Povišer
d040e61aa8 m1n1.hw.admac: Add UNK4 report flag 
Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Martin Povišer
f0ff2e9eaf m1n1.hw.admac: Do without an ADT node 
Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Martin Povišer
38ca45816f m1n1.hw.admac: Support early poll() exit 
Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Martin Povišer
f353017ac6 m1n1.hw.loadobjs: Accept queer call methods 
Let user specify a call method to use on symbols.

Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Martin Povišer
dcc75ab4f1 m1n1.hw.mca: Document the feedback disable 
Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Martin Povišer
9e32a75bc6 experiments/speaker_amp.py: Remove dead power-domain call 
/arm-io/mca0 doesn't have a PD (instead /arm-io/mca-switch has the MCA0
domain).

Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Martin Povišer
2579a97fde m1n1.loadobjs: Avoid pasting paths into cmdline strings 
Do not go through a formatted cmdline string when invoking tools. There
are user-controlled paths involved which can contain spaces (and other
foul characters).

Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Martin Povišer
72c6170ee8 m1n1.loadobjs: Support loading of inline C snippets 
Signed-off-by: Martin Povišer <povik@protonmail.com>
 2022-05-30 23:26:31 +09:00
Hector Martin
9212ac2036 m1n1.hw.asc: Add more mailbox control fields 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-05-30 23:23:37 +09:00
Janne Grunau
12db5c3910 tools/reset_panic_counter.py: Use ADT to support t600x 
In addition make it executable and add python shebang as for the other
tools.

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-05-30 22:56:43 +09:00
Janne Grunau
8a90ced4b4 tools/reset_panic_counter.py: minor fixes 
- add SPDX-License-Identifier
 - make it executable
 - add python shebang
 - remove unused import

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-05-30 22:56:43 +09:00
Sven Peter
afe9c9bb16 proxyclient: add AES experiment 
Signed-off-by: Sven Peter <sven@svenpeter.dev>
 2022-05-30 22:54:50 +09:00
Sven Peter
ee045af977 proxyclient: add AES registers 
Signed-off-by: Sven Peter <sven@svenpeter.dev>
 2022-05-30 22:54:50 +09:00
Janne Grunau
8c82a7cb92 m1n1.fw.dcp.ipc: Add Call/Callbacks observed while tracing macos 12.3 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-05-30 22:49:00 +09:00
Janne Grunau
5ef9197edc m1n1.fw.dcp.ipc: Annotate 'swap_complete_ap_gated' arguments 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-05-30 22:49:00 +09:00
Janne Grunau
8119130e88 hv/trace_dcp.py: update to mac OS 12.3 API 
Verified calls A000 through A358 due to inconsistencies in the trace
log for A104 and A105.

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-05-30 22:49:00 +09:00
Janne Grunau
8366bcd859 m1n1.fw.dcp.ipc: Update struct sizes to macos 12.3 firmware 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-05-30 22:49:00 +09:00
Akihiko Odaki
4a67fbe10e m1n1.shell: Don't save history if init failed 
Signed-off-by: Akihiko Odaki <akihiko.odaki@gmail.com>
 2022-05-30 22:48:40 +09:00
Akihiko Odaki
3440aadc3c proxy: Specify coordinates in fb_fill method 
Signed-off-by: Akihiko Odaki <akihiko.odaki@gmail.com>
 2022-05-30 22:47:26 +09:00
Hector Martin
45ac8633bd m1n1.shell: Don't print bools as hex 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-05-30 22:38:30 +09:00
Hector Martin
e9080b8139 hv: Don't allow lower() on non-fault states 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-05-30 22:38:11 +09:00
Hector Martin
984b22cd51 m1n1.hv,proxyutils: Refactor how faults are printed re: context 
Instead of directly taking a proxy entry code to determine what to
decode, just take an is_fault argument and only show and decode ESR/FAR
if true.

Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-05-30 22:36:37 +09:00
Hector Martin
7c5d47ff85 m1n1.hv: Use hv.run_shell() everywhere 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-05-30 22:35:49 +09:00
Hector Martin
8eeb7966aa hv: Refactor CPU switch logic, make hv.cpu() not exit shell 
Get rid of the hv_rearm() thing (which was always a bit dodgy) and
instead properly make sure that all CPUs rendezvous when needed and
switch the active proxy thread without ever exiting exception context.

The Python side can now switch proxy context (by waiting directly for
a proxy boot) without having to exit out of the hypervisor callback,
so cpu() now works as a normal Python method.

Add a cpus() iterator so you can do things like:

>>> for i in cpus(): bt()

Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-05-30 22:33:44 +09:00
Hector Martin
01a1a0f597 hv: Make time stealing run-time configurable 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-05-30 19:37:23 +09:00
R
dd111b22b8 prores: test 36-bit iova; fix tracer 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
8406d44db3 prores: some cleanups 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
fb968ef3fe prores: figure out output size 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
2cc5c9488f prores: rename some variables 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
200bdf9ce6 prores: figure out alpha 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
270da3849b prores: input format / chroma flags 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
b95baad365 prores: figure out offset 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
846b1cb37b prores: get data into block the correct way 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
fbd63c4bc7 prores: legible image with wrong colors 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
6986aa9c81 prores: sizes are less huge now 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
da6c8379fd prores: make container hack slightly smarter 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
479fe52128 prores: play with sizes 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
95ff15f2bd prores: load in some luma stuff 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
2b6e1ed218 prores: this bug got perturbed away 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
6fff4af32f prores: add a big fat hack so we can actually look at the output 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
baf5ebcde9 prores: clarify 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
0328517051 prores: picture header fields 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
af8ebc03ef prores: figure out frame header stuff 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
014edb643c prores: ack IRQ 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
7af2335cda prores: initial work on an experiment 
it appears to succeed, but no data is written to the output

Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
66c019cfec prores: trace enough to dump some buffers 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
3970b25add prores: start tracing descriptors 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
3f60769b39 prores: add results of register bruteforce 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
R
6a8a816bfa prores: initial commit 
Signed-off-by: R <rqou@berkeley.edu>
 2022-05-30 18:05:16 +09:00
Janne Grunau
d5ceafe41d m1n1.hv: fix argument names in HV.trace_range() 
event_mmio() uses 'start' kwarg since 'zone' conflicts with
add_tracer()'s argument of the same name.

Fixes
TypeError: HV.add_tracer() got multiple values for argument 'zone'

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-05-25 15:24:05 +09:00
Asahi Lina
d44306a1c7 m1n1.agx: Initial AGX client scaffolding 
Moves things from m1n1.fw.agx to here; the fw object will only do the
basic ASC stuff, GPU management gets its own object.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
ae1a8ad504 m1n1.fw.agx.initdata: Remove complex creation code 
Moving this to its own place

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
891bdd3101 m1n1.fw.agx.channels: Remove creation code 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
3133dcfa15 m1n1.trace.agx: Many improvements to tracer 
Also fix how cmdqueues are interpreted

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
2af56cbc0a m1n1.fw.agx.cmdqueue: Lots of struct improvements 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
9893219052 m1n1.fw.agx.controllist: Add lots more structs/fields 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
7fb99adf10 m1n1.fw.agx.channels: Add meta_fn arg to get_message() 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
8df6f6d060 m1n1.fw.agx.channels: s/DeviceControl_17/DC_DestroyContext/ 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
01d7c5f6cd m1n1.fw.agx.channels.NotifyCmdQueueWork: Show more info 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
a993f35874 hv/trace_agx.py: Add hypercalls for pause/resume, disable bulk tracers 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
f5ba42e3bb tools/reset_panic_counter.py: New tool 
Also adds m1n1.hw.spmi

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
25472a4373 m1n1.hw.uat: Remove init stuff, single allocator, make flags kwargs 
Let's use this allocator only for ASC stuff. GPU stuff will have its own
va space allocators.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
4de4fdfb27 m1n1.fw.asc.mgmt: Allow being less verbose 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
cbe0e98f4c m1n1.fw.asc: Add boot_or_start() 
This tries both ways so it'll work regardless of how the ASC was shut
down.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
f3830c034f m1n1.constructutils: Add debug (commented out) 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
797e58568b m1n1.constructutils: Make Pointers useful for builds 
- Add ROPointer subclass which is not built through
- Add build hook to detect (RO)Pointers and assign the address field

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
b681de37c5 m1n1.constructutils: Add colors, offsets, support additional metadata 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
4320d065c0 m1n1.constructutils: Support Switch() properly 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
5b6afed509 m1n1.constructutils: Add Dec(), improve printing 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
3ec82d989d m1n1.hw.uat: Always print through translation failures 
Make sure it's loud in case exceptions get eaten by Construct

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
87e0b69fc6 m1n1.trace: Propagate show_cpu in log() 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
9201ce3a13 m1n1.utils: Work around Register reloads breaking Constants 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
88c443008f m1n1.hv: Handle atomic instructions by disabling those pages 
Ugly hack, but enough for now

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
2f2794508a m1n1.hv: Make SIGQUIT dump stacks and exit 
This is useful in case a tracer is stuck in an infinite loop

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
802b37aaf3 m1n1.hv: Add support for "hypercalls" via BRK 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
c746549c2e hv: Map RAM via tracer infra, handle carveouts in Python 
Previously RAM was mapped ad-hoc, but this can end up interacting
poorly with the tracer infrastructure which we are now using for RAM
too. Move to mapping guest RAM via the tracer infra, and also unmap the
TZ carveouts in the Python side so it knows about them.

This is a HV ABI break.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
30c4503966 m1n1.hv: Support names and bases for PrintTracers 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
22228934bd m1n1.fw.agx.channels: EventMsg structure 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:12 +09:00
Asahi Lina
6b6dfde814 m1n1.fw.agx: Lots of GPU work 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:54:10 +09:00
Asahi Lina
dc21c15537 m1n1.fw.agx.channels/initdata: Use __repr__ instead of __str__ 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:52:43 +09:00
Asahi Lina
08c34dbc37 hv/trace_agx.py: Reload some stuff 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:52:43 +09:00
Asahi Lina
dc4673c03b m1n1.constructutils: Propagate force arg in _reloadcls() 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:52:43 +09:00
Asahi Lina
57d23f8491 m1n1.constructutils: Use repr() instead of str() 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-05-21 03:52:43 +09:00
Scott Mansell
a17e9e4f5e m1n1.fw.agx: WIP generate initdata 
Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:52:43 +09:00
Scott Mansell
79b01a7bcb m1n1.fw.agx: enough of a driver to boot gfx-asc 
Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:52:32 +09:00
Scott Mansell
fe9c1f4bca m1n1.hw.uat: Implement iomap 
Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:52:32 +09:00
Scott Mansell
43e3e0f3c2 m1n1.hw.uat: Reflect ARMv8 nature 
Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:52:32 +09:00
Scott Mansell
f72417b78f m1n1.trace.agx: Add support for Work submission 
Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:52:32 +09:00
Scott Mansell
636ebe70d7 m1n1.hw.uat: Allow accessing as a stream 
This allows us to use construct's Pointer

Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:52:32 +09:00
Scott Mansell
804e944a1b m1n1.constructutils: Add ConstructClass wrapper around construct 
Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:52:32 +09:00
Scott Mansell
42edf41e37 m1n1.hw.uat: fix VA_MASK 
Was missing the lowest bits, which broke unaligned reads/writes

Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:52:32 +09:00
Scott Mansell
980b9241d2 m1n1.trace.agx: Trace AGX wip 
Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:46:36 +09:00
Scott Mansell
90eae66b33 m1n1.hw.uat: Intitial UAT implementaiton 
Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:46:26 +09:00
Scott Mansell
254f01b3ea m1n1.proxyutils.RegMon: Allow custom read lambda 
Useful for monitoring ranges beind an iommu

Signed-off-by: Scott Mansell <phiren@gmail.com>
 2022-05-21 03:46:14 +09:00
Hector Martin
144c3da78e experiments/amcc_err_handler.py: New old thing to dump AMCC errors 2022-04-19 23:32:11 +09:00
Janne Grunau
d4637513e0 WIP: m1n1.hv: add support for starting CPUs on the second M1 Ultra die 
notes:
 - macos guest only works if the display is not initialized
 - macos guest appears to be very slow

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-04-19 21:39:52 +09:00
R
90479ac755 jpeg: Add documentation 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
04fc26639b jpeg: Unbreak encode for RGB formats 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
d28ab8cc81 jpeg: Figured out how to activate tiling 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
b2a0352a8f jpeg: Implement encoding from planar formats 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
f2b8054309 jpeg: Support encoding linear YUV 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
c71d341ff1 jpeg: Implement YUV10 mode for encoding 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
4681053ea9 jpeg: Implement encoding other RGB formats 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
970b2789d0 jpeg: Implement encoding subsampling modes 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
45876d09a1 jpeg: Import working encode experiment 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
9fca773c8f jpeg: Implement decoding into planar formats 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
3a18075d89 jpeg: Decode linear YUV 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
4c8807fcca jpeg: Add RGB565 to decoder 
Note that dithering does not get configured

Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
7350baf58b jpeg: Implement decoding other subsampling modes 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
0c1f89d1f9 jpeg: Import working basic decode experiment 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
49cdf7fe22 jpeg: probe registers more thoroughly 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
c6169c1c15 jpeg: Import register definitions 
This comes from experiments that were performed out-of-tree

Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:40:44 +09:00
R
0f6991ccbc dart-t8110: fill in more details from poking 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:24:00 +09:00
R
96b2d584fe dart-t8110: Import more stuff pulled from symbols 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:24:00 +09:00
R
ce5e5fe476 dart-t8110: Add other seen registers; poke protect 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:24:00 +09:00
R
18535a2385 dart-t8110: uncomment helper functions 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:24:00 +09:00
R
25ebf8adb1 dart-t8110: map/translate 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:24:00 +09:00
R
5b07956e43 dart-t8110: flush/init 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:24:00 +09:00
R
ecb9d1210c dart-t8110: dumping appears to work completely 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:24:00 +09:00
R
16028224f9 dart-t8110: dump first-level translation tables 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:24:00 +09:00
R
4a09eca74a dart-t8110: Initial commit 
Signed-off-by: R <rqou@berkeley.edu>
 2022-04-16 19:24:00 +09:00
Kyle Evans
0ba2027e5e proxyclient: tools: add a FreeBSD loader 
The differences are documented in the script itself, but the main odd
bit is that we pass the provided kernel as an initramfs.  Our modified
loader will probe for the initramfs start/end props in FDT and put up
a memdisk that we load the kernel from.  This choice was made for two
reasons:

1.) Avoid ad-hoc interpretations of a memory region in m1n1, it doesn't
    care if we actually passed it an initramfs or not anyways.

2.) We already had some code on hand to do this in our loader, so I
    suspect we've done similar shenanigans elsewhere anyways.

This also requires a U-Boot change to allow bootefi of an arbitrary
location, as long the size is provided.

The alignment constraints for loader/kernel were kept arbitrarily.  The
kernel will be bounced anyways by loader. I don't recall what
alignment requirements UEFI payloads have, but 2M seems reasonable.

Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
 2022-04-16 19:22:55 +09:00
Kyle Evans
3b68165900 proxyclient: tools: fix linux U-Boot support 
-t is optional, we may not have a tty_dev, so we can avoid trying to
grab the baudrate from it if it's unspecified.

We don't want to use dtb_addr for booti; this is the unmodified payload.
m1n1 will load it and add bits to it at a separate address for passing
on to U-Boot.  For booti purposes, we actually want the augmented FDT
rather than the user-specified.

Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
 2022-04-16 19:18:06 +09:00
Davide Cavalca
eeebd9dad3 Make scripts executable 
Signed-off-by: Davide Cavalca <davide@cavalca.name>
 2022-04-16 19:17:33 +09:00
Asahi Lina
87a849287a m1n1.utils.Reloadable: Add force argument to _reloadcls() 
Useful to force reload classes that haven't changed when their
dependencies might have.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-04-16 19:06:22 +09:00
Asahi Lina
8b10324815 m1n1.trace: Add space after CPU number in PrintTracer logs 
This makes the logs align properly with other trace logs.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-04-16 19:06:22 +09:00
Asahi Lina
6c57d3ded9 m1n1.proxyutils.RegMonitor: Add support for custom log function 
Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-04-16 19:06:22 +09:00
Asahi Lina
e2f63a07b2 hv: Improve VM emulation for tracer 
Adds support for up to 64-byte ops and more SIMD/paired operations.
This is good enough to trace a lot of GPU VM address space.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-04-16 19:06:22 +09:00
Asahi Lina
cf8be42aa9 hv: Fix bug when unmapping IO ranges 
Sometimes ranges were not properly unmapped if they did not overlap the
currently defined range maps.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-04-16 19:06:22 +09:00
Asahi Lina
302a9ec4c0 m1n1.hv: Add BYPASS trace mode 
Forces SW mapping but does not do any hooking. This can be useful for
testing the emulation or to work around issues with HW/SW map
differences.

Signed-off-by: Asahi Lina <lina@asahilina.net>
 2022-04-16 19:06:22 +09:00
Hector Martin
9af9e834bf chainload.py: Pad image to terminate payloads 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-03-27 18:34:53 +09:00
Janne Grunau
df53f16f76 m1n1.hv: Copy /chosen/memory-map/preoslog if it exists 
preoslog was added in macos 12.3 and seems to contain iBootStage1 log
output. Since macos 12.3 accesses it it has to be copied to after the
guest image.

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-20 03:04:13 +09:00
Hector Martin
94d640aa20 tools/run_guest.py: Support restricting CPU list available to guest 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-03-14 21:08:23 +09:00
Hector Martin
be1421d2ad m1n1.hw.spi: Update register definitions 
This had been sitting in my worktree...

Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-03-11 22:14:46 +09:00
Hector Martin
822b3c3cca m1n1.trace.spi: Remove UNBUF mode 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-03-11 22:13:27 +09:00
Janne Grunau
c2a48da4c1 m1n1.hv: Add trace script keyboard/trackpad 
Input devices use HID over SPI. HID communication is obscured by "sio"
DMA.

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:06:01 +09:00
Janne Grunau
387fab6132 m1n1.utils: Add CRC-16 (polygon 0xA001) 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:06:01 +09:00
Janne Grunau
025fa41e90 m1n1.trace: Add SPI tracer 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:06:01 +09:00
Janne Grunau
45824d2676 m1n1: move SPI register to m1n1.hw 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:06:01 +09:00
Janne Grunau
e90be797a8 m1n1.trace.gpio: Update register description based on the Linux driver 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:06:01 +09:00
Janne Grunau
9c80a69266 m1n1.hv: Use dynamic pin maps in trace_gpio.py 
Hardcoded pin maps do not make sense with additional HW support.
Allows tracing of a single pin in another tracing module, for example
for tracing the SPI cs pin.

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:06:01 +09:00
Janne Grunau
5e8beca9fe experiments/dcp.py: Upload fb in 512k chunks for slow (v)UARTs 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:05:21 +09:00
Janne Grunau
f297decfb4 experiments/dcp.py: Use display size for frame buffers 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:05:21 +09:00
Janne Grunau
6a8016a8e3 experiments/dcp.py: set DVA offset from ADT 
This does not appear to be necessary. It worked on t8103 before and I
see no different behavior on t6001.

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:05:21 +09:00
Janne Grunau
66f4d625b8 experiments/dcp.py: Use t8103/t600x specific code paths 
MMIO space differes between t8103 and t600x so the hardcoded RegMonitor
addresses make only sens on the former.
set_display_device difference is taken from macOS trace on t6001.

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:05:21 +09:00
Janne Grunau
d1d93eb2b5 m1n1.fw.dcp: Add callbacks used by t600x-dcp 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:05:21 +09:00
Janne Grunau
7028f3e2d8 m1n1.fw.dcp: Zero dcpep's shared memeory 
Fixes a crash of DCP with experiments/dcp.py after chainloading m1n1 on
j314c.

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-03-11 12:05:21 +09:00
Hector Martin
99124a41fe tools/chainload.py: Support chainloading raw binaries 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-03-10 00:30:15 +09:00
Hector Martin
123d1df1a7 kboot: Replace kboot_set_bootargs with kboot_set_chosen 
This allows setting arbitrary /chosen string props

Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-03-09 20:55:34 +09:00
Hector Martin
adb1cb943e proxy: Export smp_set_wfe_mode 
Signed-off-by: Hector Martin <marcan@marcan.st>
 2022-03-08 00:02:06 +09:00
Janne Grunau
d76f3a6631 m1n1.hw.DART: coalesce continuos l2 ptes 
Signed-off-by: Janne Grunau <j@jannau.net>
 2022-02-24 23:55:46 +09:00
Janne Grunau
e1c5df5037 experiments/dart_dump.py: Use DART.from_adt() 
The label of the ADT node is easier to use. Uses the proper DART type on
t600x SoCs.

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-02-24 23:55:46 +09:00
Janne Grunau
a57696366f hv/trace_dcp.py: Update to macOS 12.x API 
DCP API seems to be mostly unchanged from 12 beta to 12.2.

Signed-off-by: Janne Grunau <j@jannau.net>
 2022-02-24 23:55:46 +09:00