Hello all, I have a quick question that I hope somebody can shed some
light on. Inside of helper functions, can you have a page fault that
must be serviced by the OS? I ask because I cannot see how the cpu
state is restored inside of a helper function. Inside tlb_fill, when
invoked not from a helper but from the translated code, the retaddr is
correct and the cpu_restore_state will be successful. However, when
invoked from a helper function, the retaddr will be that of ldq_kernel
or whatever function calls __ld version of the function in
softmmu_template.h and softmmu_header.h. So surely tb will not be
found and the cpu state will not be restored. Can somebody shed some
more light on this, or should accesses that may cause a page fault be
avoided in helper code.
void tlb_fill(target_ulong addr, int is_write, int mmu_idx, void *retaddr)
{
TranslationBlock *tb;
int ret;
unsigned long pc;
CPUX86State *saved_env;
/* XXX: hack to restore env in all cases, even if not called from
generated code */
saved_env = env;
env = cpu_single_env;
ret = cpu_x86_handle_mmu_fault(env, addr, is_write, mmu_idx, 1);
if (ret) {
if (retaddr) {
/* now we have a real cpu fault */
pc = (unsigned long)retaddr;
tb = tb_find_pc(pc);
if (tb) {
/* the PC is inside the translated code. It means that we have
a virtual CPU fault */
cpu_restore_state(tb, env, pc, NULL);
}
}
raise_exception_err(env->exception_index, env->error_code);
}
env = saved_env;
}
Thanks, -Jim
