Re: [PATCH kvm-unit-tests v2] x86: Test task-switch with cs.rpl != cs.dpl

[Paolo Bonzini <pbonzini@redhat.com>]

Il 19/08/2014 15:34, Nadav Amit ha scritto:
> I know, but all the other ‘call' tests did. If there is an error, at
> least it should be consistent. ;-)

Right, let's fix it up:

diff --git a/x86/taskswitch2.c b/x86/taskswitch2.c
index fd9a404..3cfb467 100644
--- a/x86/taskswitch2.c
+++ b/x86/taskswitch2.c
@@ -190,7 +190,7 @@ void test_kernel_mode_int()

 	/* test that calling a task by lcall works */
 	test_count = 0;
-	set_intr_task_gate(0, irq_tss);
+	tss_intr.eip = (u32)irq_tss;
 	printf("Calling task by lcall\n");
 	/* hlt opcode is 0xf4 I use destination IP 0xf4f4f4f4 to catch
 	   incorrect instruction length calculation */
@@ -205,7 +205,7 @@ void test_kernel_mode_int()

 	/* test that calling a task by ljmp works */
 	test_count = 0;
-	set_intr_task_gate(0, jmp_tss);
+	tss_intr.eip = (u32)jmp_tss;
 	printf("Jumping to a task by ljmp\n");
 	asm volatile ("ljmp $" xstr(TSS_INTR) ", $0xf4f4f4f4");
 	printf("Jump back succeeded\n");


Paolo