You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Midi implementation in win2k kernel mode + user-provided file is a rather dangerous combination.
Mostly because the midi implementation doesn't sanitize the bank format. At least, I've gotten the NT4 extracted implementation to crash program with garbage data(reading bank out of a naked malloc, lol).
Adition to suggestion, user mode component should send FM writes one buffer at a time(flushing) to reduce context switches.
Of course, as a bonus, this lets the user diddle with the implementation.
The text was updated successfully, but these errors were encountered:
Not much we can do about it, Win2k driver architecture is like it is.
NT4 did a lot of stuff in Usermode, even GDI (via CSRSS), which sucked big time, because it's insanely slow, that's why i.e. GDI also was moved to kernel mode in Win2k (Win32k.sys) and the stupid concept with running all the stuff in usermode was given up.
I guess the same applies to sound architecture.
Midi implementation in win2k kernel mode + user-provided file is a rather dangerous combination.
Mostly because the midi implementation doesn't sanitize the bank format. At least, I've gotten the NT4 extracted implementation to crash program with garbage data(reading bank out of a naked malloc, lol).
Adition to suggestion, user mode component should send FM writes one buffer at a time(flushing) to reduce context switches.
Of course, as a bonus, this lets the user diddle with the implementation.
The text was updated successfully, but these errors were encountered: