Hacking Sigpatches for Atmosphere (Hekate, fss0, fusee & package3)

  • Thread starter Thread starter ShadowOne333
  • Start date Start date
  • Views Views 5,447,360
  • Replies Replies 7,443
  • Likes Likes 266
no patches need updating (aside from buildid and offset changes for .ips patches)
Post automatically merged:

Hello! Maybe a stupid question but just to be clear: when you say that hekate boot / fusee boot both are patched for everything, does that mean that Tinfoil also works on your latest release? https://github.com/borntohonk/Atmosphere/releases#release-1.11.2+v2
i am a tinfoil contributor, but no, it doesn't work that way, tinfoil needs to be recompiled with latest libnx and some tweaks before it can work with the latest version of atmosphere. (the lead developer doesn't have the time to do it)
 
Last edited by bth,
  • Love
Reactions: impeeza
Thanks mate for the update, so no new SYS-Patch Sysmodule is requiered?
to my knowledge, no, there is not a need for a new version for firmware 23.0.0

there is however a new IPC service interface in FS which seem related to file handling, which both qlaunch and auth (system-applets) call for;

but none of the existing patches need any changes.
A super minor tweak would be changing am patch to only apply to 22.0.0, but not important or necessary
 
Last edited by bth,
to my knowledge, no, there is not a need for a new version for firmware 23.0.0

there is however a new IPC service interface in FS which seem related to file handling, which both qlaunch and auth (system-applets) call for;

but none of the existing patches need any changes.
A super minor tweak would be changing am patch to only apply to 22.0.0, but not important or necessary
Cool, mate thanks for sharing your knowledge. regards.
 
hey. So i actually downloaded* ldn_mitm but it just kept crashing and even resetted the sigpatch smh. I fixed the sigpatch thing thanks to BTH but the issue is that LDN_MITM is still not working and crashes the whole atmosphere smh. (I'm sorry it's out of the context and in the wrong community but i think if someone could help me on it or stop it from crashing and make ldn work it would be so helpful and thank you.
 
hey. So i actually downloaded* ldn_mitm but it just kept crashing and even resetted the sigpatch smh. I fixed the sigpatch thing thanks to BTH but the issue is that LDN_MITM is still not working and crashes the whole atmosphere smh. (I'm sorry it's out of the context and in the wrong community but i think if someone could help me on it or stop it from crashing and make ldn work it would be so helpful and thank you.
my understanding is that for 22.0.0+ that is an atmosphere libs / libnx compatibility issue.

https://github.com/DefenderOfHyrule/ldn_mitm/releases

this fork seems to claim to have a release addressing 22.5.0 support
 
here's .ips sigpatches for 23.0.0 and atmosphere 1.12.0;

sys-patch does not need an update for 23.0.0 or atmosphere 1.12.0 (no need to ask)
 

Attachments

Thanks for your info.
Is there anyone tried v23.0.0 firmware with the latest syspatch, atmosphere and hekate?
If so, have you find any problems?
 
  • Like
Reactions: impeeza
here's .ips sigpatches for 23.0.0 and atmosphere 1.12.0;

sys-patch does not need an update for 23.0.0 or atmosphere 1.12.0 (no need to ask)
do homebrew apps need to recompile for ams 1.12.0?
all homebrews (dbi, goldleaf, vgedit, breeze, edizon se, tinwoo) crashes when I run them on fw 22.1 using your build of ams, haven't try the official

update: official ams 1.12.0 works fine
 
Last edited by dsrules,
All titles can not be launched after upgraded to v23.0.0 firmware with the latest available syspatch, atmosphere and hekate.
And it can connect to the Nintendo server to update the titles in emuNAND although it is blocked before.
 
Last edited by johnliu,
All titles can not be launched after upgraded to v23.0.0 firmware with the latest available syspatch, atmosphere and hekate.
And it can connect to the Nintendo server to update the titles in emuNAND although it is blocked before.
Yeah, it is a known problem. Look at atmosphere github and/or atmosphere here on gba.
 
All titles can not be launched after upgraded to v23.0.0 firmware with the latest available syspatch, atmosphere and hekate.
And it can connect to the Nintendo server to update the titles in emuNAND although it is blocked before.
next version of atmosphere fixes that, dns.mitm wasn't filtering all the dns requests in 23.0.0

anyway damage already done, you done sent telemetry to nintendo and updated eveyrthing that was in your emummc
 
The real serial number is also shown in atmosphere after update to firmware v23.0.0.
Perhaps alot of consoles will be banned later once upgraded.
 
  • Like
Reactions: impeeza
So I updated my emuMMC to 23.0.0 the other day and everything went sideways. Hekate would hang on a blank yellow screen, injecting fusee.bin directly did the exact same thing, and the one time I did get something to boot, half my library threw "corrupted data detected" or "unable to start software."

For context: V1 unpatched, file-based emuMMC, was running the HATS pack totally fine on 22.5 before this.

Stuff I tried first that did NOT help (saving you the time):
- re-downloading atmosphere and hekate like five times
- rebuilding hekate_ipl.ini from scratch
- deleting/renaming the contents folder (this did get it booting, but then nothing would launch - more on that below)
- running the sd card through checkdisk. card was fine, fat32, zero errors

What was actually going on, pieced together from the github issues and posts in this sub:

1. CORRECTED (per bth, Hpshout, YoshiMK): 23.0.0 actually FREES ~9mb for sysmodules on stock AMS 1.12 (embedded memory-saving patches). The yellow/blank hang was outdated sysmodules/overlays not ready for 1.12/23.0.0 - Horizon OC confirmed (HOC #114; Plane-Article2870 isolated it module-by-module). Pull HOC and any not-yet-updated sysmodules/overlays FIRST. Full detail in my reply to Hpshout.
2. Vanilla atmosphere ships without sigpatches. So once you strip the HATS stuff out to get it booting, your installed games fail signature checks and horizon flags them as corrupted. The files are fine. It's just missing patches.
3. dns mitm is broken on 23.0.0. do not go online on cfw right now until 1.12.1

What actually fixed it, in order:

1. Backed up saves first. Copied emuMMC/RAW1/nintendo/save to my pc (plus album if you care about screenshots). It's tiny, takes two minutes, don't skip it.
2. Deleted atmosphere/ and bootloader/ off the sd root. Did NOT touch nintendo/, emuMMC/ or switch/.
3. Dropped in vanilla atmosphere 1.12.0 from the official github release. Just the atmosphere folder from the zip, nothing else.
4. Wrote a fresh bootloader/hekate_ipl.ini and kept it stupid simple:

[config]
autoboot=0
bootwait=3

[cfw emummc]
fss0=atmosphere/package3
emummcforce=1

Saved it as "all files" + ANSI encoding. If your editor sneaks in a BOM or saves it as utf-16, hekate just hangs on a blank screen. That one cost me hours, honestly.

5. Merged bth's sigpatches zip (the one for atmosphere 1.12.0 / cfw 23.0.0, from this thread) - atmosphere and bootloader folders onto the card.
6. Added impeza's sys-patch v1.6.2.3, put the 420000000000000B folder in atmosphere/contents/. That is the ONLY thing in my contents folder right now.
7. Did NOT put tesla, ovlloader, horizon oc or any overlay back. Losing the overlays sucks, I know, but they're what's killing boot on 23.0.0. Add them back once each project ships a 1.12/23.0.0-compatible release (HOC is frozen until the AMS side settles, per their #114).
8. Booted hekate -> launch -> emummc, airplane mode on immediately, stayed offline the whole time.

After that, the games that were complaining made me run the "check for corrupted data" scan once, it came back clean, and they've launched fine ever since. Big first-party titles included.

Extra notes:
- If hekate still hangs on you after all this: rename the config folder at the sd root to config.bak, rename emuMMC/emummc.ini to emummc.ini.bak, then inject fusee.bin straight from tegrarcmgui. That boots sysnand clean and proves your card and files are fine. Then put emummc.ini back and launch from hekate like normal.

- Stay offline until dns mitm lands in a minor atmosphere release. Airplane mode, every single time.

That's it. No formatting the card, no reinstalling games, no lost saves. Huge thanks to bth and impeeza for keeping patches current, you two carry this scene. Hope this saves someone the time I just lost.

I also know that just because this worked for me, it may not work for other people - but I figured I'd write this up just in case it does help other people!

UPDATE: impeeza (sys-patch dev) reacted to this with a sad face, which tells me I got some technical details wrong. SUPERSEDED - per bth, post #7,438: sys-patch and the .ips apply the SAME patch, and 23.0.0 is ready except the dns.mitm gap until AMS 1.12.1. Full record in my post #7,440. This guide is just what worked for my specific setup, not a universal fix. I also realized I mistyped the folder name - 4200000000000000D - when IT SHOULD BE 420000000000000B - and have edited it above in the reply. My apologies on some of the inaccuracies - I dont want to mislead anyone at all!
 
Last edited by cbreezy210,
The real serial number is also shown in atmosphere after update to firmware v23.0.0.
Perhaps alot of consoles will be banned later once upgraded.
the serial number was always a cosmetic thing, the real effect is the ssl certificate being blanked.
it's your certificate they block from AWS/dauth, not your serial number.


3. dns mitm is broken on 23.0.0 because nintendo moved to an async resolver. do not go online on cfw right now.
anyway it turns out the issue for dns.mitm was just that some, not all official sysmodules, used new ipc commands under sfdnsres instead of the old ones, and hexkyz already said he has verified a fix for it, and it will be pushed with atmosphere 1.12.1

the async resolver stuff is LLM hallucinations for most part. (this is why to not blindly accept what your precious AI/LLM tells you, and to question it's logic, and hopefully not let it do tasks that are beyond your understanding, as then you cannot call it out on it)


UPDATE: impeeza (sys-patch dev) reacted to this with a sad face, which tells me I got some technical details wrong. The sigpatches (.ips files) are what actually fix the signature checks, not sys-patch itself. Also, the scene consensus is that 23.0.0 isn't ready yet - if you're not already on it, don't update. If you are, consider downgrading to 22.5.0. This guide is just what worked for my specific setup, not a universal fix. I also realized I mistyped the folder name - 420000000000000D - when IT SHOULD BE 420000000000000B - and have edited it above in the reply. My apologies on some of the inaccuracies - I dont want to mislead anyone at all!

i am the one who maintains both, and sys-patch does fix it. the .ips sigpatches are a byproduct i make verifying the patterns for sys-patch. They do the exact same patch, just at hardcoded offsets.

(the .ips patch bundle actually contains more patches than sys-patch current, as i added the new memory saving patches there too, not that they are needed for regular atmosphere, as atmosphere hardcode embeds them, but i made patterns for them anyway, as my fork of atmosphere doesn't support the old type embedded patches)

23.0.0 is ready, minus atmosphere's dns.mitm not covering the new parts of sfdnsres just yet (already being worked on and was identified in full)


in future, don't update to a firmware revision and boot without active ES patch, that is what corrupts your games.

sys-patch didn't need an update for 23.0.0.
 
Last edited by bth,
1. 23.0.0 cut the memory budget for custom sysmodules way down. The HATS extras (tesla/ovlloader, horizon oc, status monitor, all of it) push you over the limit and boot just dies silently. That blank/yellow hang a lot of people are hitting? That's it.
Would have thought that if you did not have memory issues on 22.5.0 using a ton of HATS bloat then with the same setup on 23.0.0 you'd actually gain more memory (+9mb~)

For me I now have around 17mb free (was previously 8mb free)
 
the serial number was always a cosmetic thing, the real effect is the ssl certificate being blanked.
it's your certificate they block from AWS/dauth, not your serial number.



anyway it turns out the issue for dns.mitm was just that some, not all official sysmodules, used new ipc commands under sfdnsres instead of the old ones, and hexkyz already said he has verified a fix for it, and it will be pushed with atmosphere 1.12.1

the async resolver stuff is LLM hallucinations for most part. (this is why to not blindly accept what your precious AI/LLM tells you, and to question it's logic, and hopefully not let it do tasks that are beyond your understanding, as then you cannot call it out on it)




i am the one who maintains both, and sys-patch does fix it. the .ips sigpatches are a byproduct i make verifying the patterns for sys-patch. They do the exact same patch, just at hardcoded offsets.

(the .ips patch bundle actually contains more patches than sys-patch current, as i added the new memory saving patches there too, not that they are needed for regular atmosphere, as atmosphere hardcode embeds them, but i made patterns for them anyway, as my fork of atmosphere doesn't support the old type embedded patches)

23.0.0 is ready, minus atmosphere's dns.mitm not covering the new parts of sfdnsres just yet (already being worked on and was identified in full)


in future, don't update to a firmware revision and boot without active ES patch, that is what corrupts your games.

sys-patch didn't need an update for 23.0.0.

Thanks for the corrections, bth - and for maintaining both sides of this (sys-patch and the .ips bundle). The bit about the .ips being a byproduct of your pattern verification makes total sense, and good to know sys-patch needed no 23.0.0 update.

Setting the record straight on my own posts, for anyone following along:

1. My "async resolver" claim was wrong. I picked it up from the PR discussion around ams #2865 and repeated it without verifying it at the binary level myself. The accurate version, per you and hexkyz: some official sysmodules moved to new sfdnsres IPC commands that dns.mitm didn't hook yet. Fix is verified and lands in Atmosphere 1.12.1. Not a resolver rewrite, not a reason to panic.

2. The "stay offline on CFW for now" advice stands until 1.12.1 is out and confirmed - but the reason is the sfdnsres gap, not a resolver rewrite.

3. On sigpatches vs sys-patch: my update was half right. Per bth: both apply the same patches. sys-patch via patterns, the .ips at hardcoded offsets, and the .ips bundle carries extra memory-saving patches for forks that don't embed them.

4. On bans: serial is cosmetic. It's the blanked SSL cert that AWS/dauth keys on. Good to have that straight.

5. On corrupted games: bth's rule is the one that matters - never boot a new firmware revision without active ES patches. That's what corrupts your games.

My guide stays labeled "what worked on my setup." I'll update the OP once 1.12.1 drops. Until then: on 23.0.0, stay offline with CFW and keep ES patches active; on 22.5.0 and happy, stay put.

My bad for adding noise to a fast-moving situation. I hold my tools to a byte-verified standard - I should hold my posts to the same one.
Post automatically merged:

Would have thought that if you did not have memory issues on 22.5.0 using a ton of HATS bloat then with the same setup on 23.0.0 you'd actually gain more memory (+9mb~)

For me I now have around 17mb free (was previously 8mb free)

Good catch, and thanks for the hard numbers. 8mb to 17mb free matches YoshiMK's report on the Reddit thread (9mb to 17mb), and it lines up with what bth said: Atmosphere 1.12 hardcode-embeds the new memory-saving patches for 23.0.0, so on stock AMS you gain headroom, you don't lose it.

So my point 1 was wrong twice over: 23.0.0 did not cut the sysmodule memory budget, and the blank/yellow hang is not a memory ceiling. The evidence points at outdated sysmodules/overlays that aren't 1.12/23.0.0-ready - Horizon OC is the confirmed one (their issue #114, and Plane-Article2870 isolated it by moving modules one-by-one through the same yellow-screen boot). If you're hitting the hang: pull HOC and any other not-yet-updated sysmodules/overlays FIRST, before touching anything else.

I'll strike point 1 from my guide and replace it with: "23.0.0 frees ~9mb for sysmodules on stock AMS 1.12 (embedded memory-saving patches, per bth). Yellow/blank boot hangs trace to outdated sysmodules like HOC, not memory."

Second correction, same rule as the last one: data beats my inference, and your data is better than my inference was. Thanks for keeping the thread honest.
 
Last edited by cbreezy210,

Site & Scene News

New Hot Discussed User Submitted