2 configuration environment properly break open sti
kpcr - a lot of important data thread switching, structure
Into the kernel when fs no longer teb / tib; is KPCR .
While observing kifastcallentry, and found that indeed set fs is the core kpcr:
That we manually modified to kpcr;
Note: If not used directly fs; then check the relevant hard-coded instead.
The system does not effect the card, and any way can not kill, not kill off:
The current system is not stuck, and can thread a scheduling; but the test program can not be killed, can not end, because the end of the thread is inserted into the queue apc request, the process is asynchronous ; because the end of the thread is not immediately end, certainly you need to perform to a specific environment, or to execute a statement on the cut, then the environment is wrong, may cause a blue screen, resulting in 100% wrong.
Thread the end of the actual process:
Before returning from ring 0 ring 3; apc system checks its queue to see if there is a request to end. There kill (if it is the last thread of a process, then the correspondence process to clean out)
But here:
0 has an infinite loop in the ring; not lead to switching to the instant before the check ring 3; not kill it.
How to kill:
---- only modify the disruption of the endless loop.
-
!process 0 0
-
.process corresponding _eprocess
-
u 0xaddr
-
ew 401005 9090 changes. . . nop out. And can kill off the