https://bugs.documentfoundation.org/show_bug.cgi?id=114939
--- Comment #5 from JimF <[email protected]> ---
The legacy busted code on import sounds correct. As long as it is only used
upon a hash failure and upon any write the correct code is used to update the
file.
but that is way beyond the scope of my knowledge of this project (which is zero
knowledge).
All I did was told about the problem, and investigated. I first looked at the
code and got an idea about what 'might' be wrong, and was able to quickly
convert a stock 'working' sha1 implementation to provide the same (wrong)
results as the LibreOffice was returning. I then cut the code for sha1 out of
digest.cxx, converted to C, put in a main that would sha1 a command line
argument or contents of a file, and was then able to detect and make sure this
bug only was happening when the length of the last sha1 limb was 52, 53, 54,
55. It was then very easy to simply find the change that corrected the
results.
Btw, discussion of the bug can be seen here:
https://github.com/magnumripper/JohnTheRipper/issues/3089
--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Libreoffice-bugs mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/libreoffice-bugs
