On 4/6/20 3:55 PM, Simon Smith wrote:
This commit adds new unit tests for commit a4d956b93904 ("KVM: nVMX:
vmread should not set rflags to specify success in case of #PF")
The two new tests force a vmread and a vmwrite on an unmapped
address to cause a #PF and verify that the low byte of %rflags is
preserved and that %rip is not advanced. The cherry-pick fixed a
bug in vmread, but we include a test for vmwrite as well for
completeness.
Before the aforementioned commit, the ALU flags would be incorrectly
cleared and %rip would be advanced (for vmread).
Reviewed-by: Jim Mattson <jmattson@xxxxxxxxxx>
Signed-off-by: Simon Smith <brigidsmith@xxxxxxxxxx>
---
x86/vmx.c | 121 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 121 insertions(+)
diff --git a/x86/vmx.c b/x86/vmx.c
index 647ab49408876..e9235ec4fcad9 100644
--- a/x86/vmx.c
+++ b/x86/vmx.c
@@ -32,6 +32,7 @@
#include "processor.h"
#include "alloc_page.h"
#include "vm.h"
+#include "vmalloc.h"
#include "desc.h"
#include "vmx.h"
#include "msr.h"
@@ -368,6 +369,122 @@ static void test_vmwrite_vmread(void)
free_page(vmcs);
}
+ulong finish_fault;
+u8 sentinel;
+bool handler_called;
+static void pf_handler(struct ex_regs *regs)
+{
+ // check that RIP was not improperly advanced and that the
+ // flags value was preserved.
+ report("RIP has not been advanced!",
+ regs->rip < finish_fault);
+ report("The low byte of RFLAGS was preserved!",
+ ((u8)regs->rflags == ((sentinel | 2) & 0xd7)));
+
+ regs->rip = finish_fault;
+ handler_called = true;
+
+}
+
+static void prep_flags_test_env(void **vpage, struct vmcs **vmcs, handler *old)
+{
+ // get an unbacked address that will cause a #PF
+ *vpage = alloc_vpage();
+
+ // set up VMCS so we have something to read from
+ *vmcs = alloc_page();
+
+ memset(*vmcs, 0, PAGE_SIZE);
+ (*vmcs)->hdr.revision_id = basic.revision;
+ assert(!vmcs_clear(*vmcs));
+ assert(!make_vmcs_current(*vmcs));
+
+ *old = handle_exception(PF_VECTOR, &pf_handler);
+}
+
+static void test_read_sentinel(void)
+{
+ void *vpage;
+ struct vmcs *vmcs;
+ handler old;
+
+ prep_flags_test_env(&vpage, &vmcs, &old);
+
+ // set the proper label
+ extern char finish_read_fault;
+
+ finish_fault = (ulong)&finish_read_fault;
+
+ // execute the vmread instruction that will cause a #PF
+ handler_called = false;
+ asm volatile ("movb %[byte], %%ah\n\t"
+ "sahf\n\t"
+ "vmread %[enc], %[val]; finish_read_fault:"
+ : [val] "=m" (*(u64 *)vpage)
+ : [byte] "Krm" (sentinel),
+ [enc] "r" ((u64)GUEST_SEL_SS)
+ : "cc", "ah"
+ );
+ report("The #PF handler was invoked", handler_called);
+
+ // restore old #PF handler
+ handle_exception(PF_VECTOR, old);
+}
+
+static void test_vmread_flags_touch(void)
+{
+ // set up the sentinel value in the flags register. we
+ // choose these two values because they candy-stripe
+ // the 5 flags that sahf sets.
+ sentinel = 0x91;
+ test_read_sentinel();
+
+ sentinel = 0x45;
+ test_read_sentinel();
+}
+
+static void test_write_sentinel(void)
+{
+ void *vpage;
+ struct vmcs *vmcs;
+ handler old;
+
+ prep_flags_test_env(&vpage, &vmcs, &old);
+
+ // set the proper label
+ extern char finish_write_fault;
+
+ finish_fault = (ulong)&finish_write_fault;
+
+ // execute the vmwrite instruction that will cause a #PF
+ handler_called = false;
+ asm volatile ("movb %[byte], %%ah\n\t"
+ "sahf\n\t"
+ "vmwrite %[val], %[enc]; finish_write_fault:"
+ : [val] "=m" (*(u64 *)vpage)
+ : [byte] "Krm" (sentinel),
+ [enc] "r" ((u64)GUEST_SEL_SS)
+ : "cc", "ah"
+ );
+ report("The #PF handler was invoked", handler_called);
+
+ // restore old #PF handler
+ handle_exception(PF_VECTOR, old);
+}
+
+static void test_vmwrite_flags_touch(void)
+{
+ // set up the sentinel value in the flags register. we
+ // choose these two values because they candy-stripe
+ // the 5 flags that sahf sets.
+ sentinel = 0x91;
+ test_write_sentinel();
+
+ sentinel = 0x45;
+ test_write_sentinel();
+}
+
+
static void test_vmcs_high(void)
{
struct vmcs *vmcs = alloc_page();
@@ -1994,6 +2111,10 @@ int main(int argc, const char *argv[])
test_vmcs_lifecycle();
if (test_wanted("test_vmx_caps", argv, argc))
test_vmx_caps();
+ if (test_wanted("test_vmread_flags_touch", argv, argc))
+ test_vmread_flags_touch();
+ if (test_wanted("test_vmwrite_flags_touch", argv, argc))
+ test_vmwrite_flags_touch();
/* Balance vmxon from test_vmxon. */
vmx_off();
Not related to your patch, but just thought of mentioning it here. I
find the name 'handle_exception' odd, because we really don't handle an
exception in there, we just set the handler passed in and return the old
one. May be, we should call it set_exception_handler ?
Reviewed-by: Krish Sadhukhan <krish.sadhukhan@xxxxxxxxxx>
