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
There has been some concern that the atomic extension (A) is too burdensome for certain soft CPU designs. In lieu of the A extension, it is possible to only implement the LR/SC instructions (Zalrsc). However, at the moment, some target software (e.g., Linux, glibc) make active use of AMO instructions. Thus, that software would need to be modified to use LR/SC sequences in place of AMO instructions, or the AMO instructions would need to be provided by trap and emulate in supervisor or machine mode.
In any case, I think it is important to show a concrete proof of concept working system with Zalrsc instead of A before the specification reflects the reduced requirement.
The text was updated successfully, but these errors were encountered:
There has been some concern that the atomic extension (A) is too burdensome for certain soft CPU designs. In lieu of the A extension, it is possible to only implement the LR/SC instructions (Zalrsc). However, at the moment, some target software (e.g., Linux, glibc) make active use of AMO instructions. Thus, that software would need to be modified to use LR/SC sequences in place of AMO instructions, or the AMO instructions would need to be provided by trap and emulate in supervisor or machine mode.
In any case, I think it is important to show a concrete proof of concept working system with Zalrsc instead of A before the specification reflects the reduced requirement.
The text was updated successfully, but these errors were encountered: