[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [XTF PATCH 05/16] vvmx: add a general error handler for VMX instructions



On 12/16/16 20:16 +0000, Andrew Cooper wrote:
On 16/12/16 13:43, Haozhong Zhang wrote:
handle_vmxinsn_err() is added to check and output the mismatch between
errors in the execution of a VMX instruction and the expected errors.

Signed-off-by: Haozhong Zhang <haozhong.zhang@xxxxxxxxx>
---
 tests/vvmx/util.c | 104 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/vvmx/util.h |  17 +++++++++
 2 files changed, 121 insertions(+)

diff --git a/tests/vvmx/util.c b/tests/vvmx/util.c
index 8cd35c5..74b4d01 100644
--- a/tests/vvmx/util.c
+++ b/tests/vvmx/util.c
@@ -2,6 +2,110 @@
 #include <arch/x86/hvm/vmx/vmcs.h>
 #include "util.h"

+#define vvmx_failure(prefix, fmt, ...)                       \
+    do {                                                     \
+        xtf_failure("Fail: %s: "fmt, prefix, ##__VA_ARGS__); \
+    } while (0)

I don't see this used anywhere else in the series.  Is it only for the
function below?


Yes, only for handle_vmxinsn_err().

+
+bool handle_vmxinsn_err(const char *name, uint8_t err, exinfo_t fault_info,
+                        uint8_t exp_err, exinfo_t exp_fault_info,
+                        enum vmx_insn_errno exp_insn_errno)
+{
+    bool passed = true;
+
+    bool has_fault = !!(err & VMXERR_FAULT);

C99 bools don't need !!, as they have that property guaranteed by the
standard.


I'll remove !! in all these patches.

It came as a shock to me when Xen's old bool_t's didn't have that
property, which is what prompted me to fix Xen.  In particular,

static inline bool_t (void) { return x & 0x1000; }

used to truncate to 0 regardless of the value of x, which was
desperately unhelpful.

+    unsigned int fault_vec = exinfo_vec(fault_info);
+    unsigned int fault_ec = exinfo_ec(fault_info);
+    bool exp_fault = !!(exp_err & VMXERR_FAULT);
+    unsigned int exp_fault_vec = exinfo_vec(exp_fault_info);
+    bool ec = !!(exp_fault_vec & X86_EXC_HAVE_EC);
+    unsigned int exp_fault_ec = ec ? exinfo_ec(exp_fault_info) : 0;
+
+    if ( !exp_fault && has_fault )
+    {
+        vvmx_failure(name,
+                     "unexpected fault #%u(%u), but no fault is expected\n",
+                     fault_vec, fault_ec);

Please use x86_decode_exinfo(), which formats an exinfo_t with mnemonics.


Sure, will change.

I already have half a patch series to introduce a %p for exinfo_t.  I
should clean the series up and post it.

~Andrew

Thanks,
Haozhong

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.