BDK will allow developers to use the full collection of drivers,
with limited editing, if any, for making payloads for Nintendo Switch.
Using a single source for everything will also help decoupling
Switch specific code and easily port it to other Tegra X1/X1+ platforms.
And maybe even to lower targets.
Everything is now centrilized into bdk folder.
Every module or project can utilize it by simply including it.
This is just the start and it will continue to improve.
gcc 10 no longer lets you get away with not externing global variables in header files. This adds the necessary extern and adds defines in appropriate c files
exosphere.ini will always be loaded and the values set by it.
User can still choose to override them with the `cal0blank` and `cal0writesys` in a boot entry.
Override keys get a 0 or 1.
By using `BOOT_CFG_AUTOBOOT_EN` in `boot_cfg` and `EXTRA_CFG_NYX_UMS` in `extra_cfg` you can launch UMS from boot.
The `ums` variable must be populated with one of the following:
```
NYX_UMS_SD_CARD 0
NYX_UMS_EMMC_BOOT0 1
NYX_UMS_EMMC_BOOT1 2
NYX_UMS_EMMC_GPP 3
NYX_UMS_EMUMMC_BOOT0 4
NYX_UMS_EMUMMC_BOOT1 5
NYX_UMS_EMUMMC_GPP 6
```
Use the exact same clocks with HOS and utilize low jitter clock parents.
Add back our compatibility mode and the missing timeout clock parent.
Hekate main will continue to use PLLP clock parent for all.
hekate main always runs in compatibility mode (SDR82).
This ensures speed on boot process.
Nyx will first try SDR104.
If the sd card is a sandisk U1 and fails, it will try the compatibility mode.
After that it fallbacks to lower bus speeds.
Both support 1bit mode for broken sd card readers.
Having the new error checking in the sdmmc driver, allows for all that to work.
It can now fail instead of continuing, like how HOS reacts.
Allows more kip names to be added. To use it on them you will need to dump the kips via the Dump pkg1/2 function and then sha256 to get the hash.
The name must be used as is, otherwise the name compare will fail.
Because of that, also allow unordered parsing and gluing.
Using the key `emupath` on a boot entry will load the selected emuMMC.
This can also be forced by using the correct boot cfg storage bit and writing the path at the emummc path offset. Check readme for these.
This can only be used if the emuMMC was created via Nyx. because of the raw_based and file_based files that have emuMMC info.
(emupath=emuMMC/RAW1, emupath=emuMMC/SD00, etc)
This disables low battery monitor shutdown (LBM shutdown) on boot and checks if battery is enough.
The logic is as follows:
If battery is not enough:
- If not charging and 15s pass, it will re enable LBM shutdown and power off.
- If charging, it will wait until it is charged above the limit.
Screen will auto turn off to save power. A press on Power button or a change on charger, will enable it for another 15s.
If battery is enough:
- Enables LBM shutdown and continues with the boot process.
Use key `fss0experimental=1` in a boot entry with `fss0=` defined to enable experimental content.
(Older versions of hekate will just skip any experimental content.)
Additionally disable archive bit fix tool in hekate main to shave binary size.
The archive bit fix in Nyx is the only maintained one and so the only one that should be used.
There is an edge case fixed where the whole would be freed and this would make use of a nullptr.
Additionally, remove usage of reserved names for vars and add comments on how it works.
This fixes the following case:
When force is 0 and the payload is not hekate, the function jumps to the 'is old' check.
This skips the is magic value check and only does the version check.
In case the foreign payload has a low value on that offset, it causes an overwrite which is unneeded.
Fixes a hang caused when rebooting 2 payload from L4T with old hekate in vendor partition.
L4T does not overwrite the nyx storage where the Minerva configuration is stored.
This makes new Minerva parse the wrong tables from old hekate and eventually hang the RAM, which causes an exception on BPMP.
If `fss0=` key is detected, sept will be loaded from fusee-secondary.bin instead of `sept/sept_*`.
This will negate missing sept and failed to decrypt pkg2 errors, when booting HOS, for users that forget to update sept folder.
It will now show erros for the following:
- Missing or old libsys_minerva.bso (DRAM training).
- Missing libsys_lp0.bso (LP0 sleep mode).
- Missing or old Nyx version
This is conditional:
In order to not cancel booting while trying to enter fastboot in android, the combo MUST be initiated with `VOL+` (hold) -> `VOL-` (hold)
3 users had issues with 602MHz.
This will probably bring the SoC binning compatibility to 100%.
Additionally, make it easy to change default boost frequency.
The tiny loss in perf, will be mitigated in Nyx. (It's actually even faster)
There's an increasing ammount of users that kill their batteries when forgetting their devices into AutoRCM / RCM mode.
This will now force a shutdown the moment the battery reaches 2.8V. Even if device is inside RCM mode.
Notice: We might need to increase the limit.
- Training and switch is now faster
- Compatibility checks: New Minerva does not allow old binaries. New binaries do not allow old Minerva
- MTC table is now in a safe region
- Periodic training period increased to every 250ms
- Allow reuse of unused sections that fit exactly to selected allocation size. Decreases fragmentation dramatically.
- Always allocate and align mapped memory to selected alignment. Avoids having fragmented unused maps that are not aligned.
- Use a static alignment based on BPMP and generally average cache line size. Boosts performance when MMU is used.
Allow the patch loop to end via empty source data as per original spec and error out on 0 length if source data is not null.
emuMMC force decompression patch safely avoids that.
- Allow SHA to calculate sizes > 16MB and refactor sha function
- Name various registers and magic numbers
- Fix various key access bugs
In a later commit this new design will boost verification times significantly and also allow full SHA256 hashes.
Remove patches_template.ini load.
- It has useless patches, which some times, users confuse them for "bad" patches, even though they aren't.
- No one reads on how it works and it's constantly mishandled when there's no patches.ini (which completely overrides it).
- It was not supposed to be edited.
Also release will not include it anymore.
Template will still exist for demonstrating the syntax of `patches.ini`
This commit also corrects the patches.ini encoding note as ASCII.
Tsec keys function always disabled host1x clock after running.
This interferes with display interface and disables further window frame syncing.
Display_end code already handles disable and reset of said clock.
It also fixes an ancient bug that was mitigated by removing the 5 frame sync on HOST1X_SYNC_SYNCPT_9 at channel 0:
5fd9daa364 (diff-6b0c56eab8515465d559ff0ea73a22c3L152)
By using the key `id=` with a max 7 ASCII id, hekate will search all inis automatically and pinpoint the boot entry with that id.
After that it forces a boot from this one.
The format is described in patches.ini.
For now it only supports the kips defined in hekate's code.
Next versions will add support for defining other kips.
Add a configuration option "Full w/ hashfile" to
the "verification" option menu, to enable hashfile
generation when doing full verification of a backup.
When enabled, during full backup verification we save the
chunk's SHA256 digest in a hashfile next to the output file
we're currently verifying.
The performance impact is negligible between "Full verify"
and "Full verify w/ hashfile", because we already
compute the SHA256 of the chunks when verifying.
We save the SHA256 per chunks (4 MB) because due to
SE limitations, we can't compute the SHA256 of the
whole partition (or rawnand).
On the other hand a pure software implementation
is way too slow to be bearable, even asm-optimized:
between 15 and 90 seconds per 4 MB chunk for
crc32/sha1/sha256, depending on the optimizations
and the actual algorithm.
The output hash file format is as follows:
# chunksize: <CHUNKSIZE_IN_BYTES>
sha256_of_chunk_1
sha256_of_chunk_2
...
sha256_of_chunk_N
Some Sandisk U1 sd cards do not behave nicely if they power cycle too fast. A min 100ms wait, is enough to mitigate that.
Fortunately, because of how the code paths are structured, this was never hit.
[Stock]
fss0={sd path}
stock=1
Can now work for both older and new HOS versions.
- <= 6.2.0 loads nothing and removes kernel patching.
- >= 7.0.0 loads exo, wb and removes kernel patching.
This requires that fss0 {sd path} is valid. Otherwise it will fail on ini cfg parsing.
If <= 6.2.0 and no FSS0
[Stock]
stock=1
will provide a cleaner stock with no kernel patching.
Using "fss0={sd path}" allows you to parse kips, exosphere and warmboot from a fusee secondary binary.
Exosphere and warmboot are overridable if these entries are defined after that.
Additionally any extra kip can be loaded as before via
kip1={sd path}/* for many
or
kip1={sd path} for a sigle one
Warning: Don't double load core kips like loader, pm, sm and ams_mitm. The result will be a hang.
This is important, otherwise it can break tools like memloader.
Until a semi-usb init solution is found, hekate will show bootlogo and allow you to go to menu.
If >= 7.0.0 after that, it will run sept and then show logo again without a wait. For as long as rebuilding the boot process takes.
- Use always aligned buffers to up the speed for DMA usage. For verification, backup and restore.
- Rework verification a little bit and make default Sparse mode.
Sparse mode protects from fake sd cards, bad sectors and frequent I/O corruption. Aka, ~100% of cases.
- Make debugmode for exosphere mandatory
- Support dev RSA modulus for warmboot
- Fix a critical bug where it allowed free() to be used on a non-heap address.
- Better the makefile
- Still compatible with old hekate.
- Allows for hotfix control
- Sept is now copied from actual running payload, negating the need to check update.bin
- If a foreign payload is found in sept then it is renamed and hekate copies itself. After sept run, it renames it back and continues with boot.
This fixes a regression in chainloading update.bin where the added check for if it's update or not, didn't copy the reloc code. And the calling was not updated to reflect this.
Additionally:
- Delete the reloc.S relic
- Force running config_hw in case we change hw init again in the future.
