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
This is a reminder to myself to investigate this issue.
The reddit user 1800lilith contacted me, observing full-system freezes when loading ableton via wine while having LEETMOUSE loaded. If LEETMOUSE is unloaded, everything works fine.
1800lilith also observed debug messages ("Float traps") spewing kernel messages here and there. Maybe this bug is related to #4 and #3.
I could imagine, that on 1800lilith's system configuration, the FPU unit on the CPU is having problems due to an improper use by LEETMOUSE. Keep in mind: Kernel code is generally not supposed to use any floating point arithmetic. However LEETMOUSE relies on this and tries to circumvent all the pitfalls when using FPU in the kernel according to the book. While I think, the current implementation is "almost safe" now, I observed some pitfalls here and there (#4), where I had to specifically test for "invalid data" (Hence the "Float traps"). 1800lilith observations might indicate, that the FPU can still get messed up and my "workaround" is just trying to fix the symptoms, rather than fixing the cause.
Maybe I have overseen something?
The text was updated successfully, but these errors were encountered:
This is a reminder to myself to investigate this issue.
The reddit user 1800lilith contacted me, observing full-system freezes when loading ableton via wine while having LEETMOUSE loaded. If LEETMOUSE is unloaded, everything works fine.
1800lilith also observed debug messages ("Float traps") spewing kernel messages here and there. Maybe this bug is related to #4 and #3.
I could imagine, that on 1800lilith's system configuration, the FPU unit on the CPU is having problems due to an improper use by LEETMOUSE. Keep in mind: Kernel code is generally not supposed to use any floating point arithmetic. However LEETMOUSE relies on this and tries to circumvent all the pitfalls when using FPU in the kernel according to the book. While I think, the current implementation is "almost safe" now, I observed some pitfalls here and there (#4), where I had to specifically test for "invalid data" (Hence the "Float traps"). 1800lilith observations might indicate, that the FPU can still get messed up and my "workaround" is just trying to fix the symptoms, rather than fixing the cause.
Maybe I have overseen something?
The text was updated successfully, but these errors were encountered: