Skip to content

fix: correct incorrectly extract salt from tail merging optimization ic204_fw_2049023401#67

Merged
jglim merged 1 commit into
jglim:mainfrom
Mattwmaster58:main
Mar 16, 2026
Merged

fix: correct incorrectly extract salt from tail merging optimization ic204_fw_2049023401#67
jglim merged 1 commit into
jglim:mainfrom
Mattwmaster58:main

Conversation

@Mattwmaster58

@Mattwmaster58 Mattwmaster58 commented Mar 13, 2026

Copy link
Copy Markdown
Contributor

The last byte of some of the levels was the same, which then caused the compiler to do tail merging optimization, ie a single mov + a jump to that mov instead of having each access level have a self-contained block of movs (ie, a mov for each level cases).

This in turn made my naive script-based extraction fail.

This specific sw version salts have been verified "bad" in hardware, so we for sure aren't making anything worse. There is probably one other sw version that has this defect which I'll get to when I have the time, but I figured I would submit this PR for a known bad issue while I have someone available to test a specific SW version for me.

…ic204_fw_2049023401

The last byte of some of the levels was the same, which then caused the compiler to do tail merging optimization, ie a single store with a jump to that store instead of having each access level be self contained.

This in turn made naive script based extraction fail.
@Mattwmaster58

Copy link
Copy Markdown
Contributor Author

Checked other sw versions lightly - there were others with this tail merging optimization, but in inconsequential levels - ie not {1,3,5,7}. It just so happened that this was the only consequential case.

@jglim
jglim merged commit 2f6b972 into jglim:main Mar 16, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants