Add support for FPU flag for PDs #404
Conversation
|
There is a good reason why I made the FPU enabled by default. If you do disable it by default, you better give a big warning that people should compile the code with -mgeneral-regs-only, otherwise they'll get crashes. |
JDuchniewicz
force-pushed
the
jakub/fpu_flag
branch
2 times, most recently
from
efe6d3d to
95044b3
Compare
Signed-off-by: Jakub Duchniewicz <[email protected]>
|
Yes. Now it's enabled by default. |
| affinity: Word(vcpu.cpu.0.into()), | ||
| prio: virtual_machine.priority, | ||
| max_prio: virtual_machine.priority, | ||
| fpu: true, |
There was a problem hiding this comment.
we should probably have a VM fpu attribute here too
There was a problem hiding this comment.
Very small chance anyone would want to disable the FPU for a VM though.
Also, this wouldn't work for x86, where the flag is used for the host task and the FPU is always enabled for the guest itself.
There was a problem hiding this comment.
As I understand it it will be refined at a later step during the system setup as mentioned below:
As discussed offline,
fpu: trueis fine here. As this function's purpose is to only create what can be inferred from just the ELF, while leaving everything else with reasonable defaults. Refining the TCB will happen at a later step in the caller atmicrokit/tool/microkit/src/capdl/builder.rs
Line 972 in 23f1f6d
Signed-off-by: Jakub Duchniewicz <[email protected]>
Signed-off-by: Jakub Duchniewicz <[email protected]>
4 participants
In conjunction with seL4/rust-sel4#316 allows for using the
fpuflag to specify that a PD can be using the FPU.