[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [Xen-devel] [PATCH] viridian: fix cpuid leaf 0x40000003
The response to viridian leaf 3 needs to split a 64-bit mask across EAX and EBX, with the low order 32 bits in EAX and the high order 32 bits in EBX. To facilitate this a union of two uint32_t values and the mask (type HV_PARTITION_PRIVILEGE_MASK) is allocated on stack as follows: union { HV_PARTITION_PRIVILEGE_MASK mask; uint32_t lo, hi; } u; This, of course, is incorrect as both lo and hi will alias the low order 32 bits of the mask. This patch wraps lo and hi in an anonmymous struct to achieve the desired effect. NOTE: Fixing this also stops Windows making the HvGetPartitionId hypercall which was previously considered erroneous behaviour. Thus the hypercall handler is also modified to stop squashing the 'unimplemented' warning for this hypercall. Signed-off-by: Paul Durrant <paul.durrant@xxxxxxxxxx> --- Cc: Jan Beulich <jbeulich@xxxxxxxx> Cc: Andrew Cooper <andrew.cooper3@xxxxxxxxxx> --- xen/arch/x86/hvm/viridian.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) diff --git a/xen/arch/x86/hvm/viridian.c b/xen/arch/x86/hvm/viridian.c index d6aa89d..694eae6 100644 --- a/xen/arch/x86/hvm/viridian.c +++ b/xen/arch/x86/hvm/viridian.c @@ -245,7 +245,7 @@ void cpuid_viridian_leaves(const struct vcpu *v, uint32_t leaf, }; union { HV_PARTITION_PRIVILEGE_MASK mask; - uint32_t lo, hi; + struct { uint32_t lo, hi; }; } u; if ( !(viridian_feature_mask(d) & HVMPV_no_freq) ) @@ -964,12 +964,10 @@ int viridian_hypercall(struct cpu_user_regs *regs) gprintk(XENLOG_WARNING, "unimplemented hypercall %04x\n", input.call_code); /* Fallthrough. */ - case HvGetPartitionId: case HvExtCallQueryCapabilities: /* - * These hypercalls seem to be erroneously issued by Windows - * despite neither AccessPartitionId nor EnableExtendedHypercalls - * being set in CPUID leaf 2. + * This hypercall seems to be erroneously issued by Windows + * despite EnableExtendedHypercalls not being set in CPUID leaf 2. * Given that return a status of 'invalid code' has not so far * caused any problems it's not worth logging. */ -- 2.1.4 _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxxxxxxxxx https://lists.xenproject.org/mailman/listinfo/xen-devel
|
Lists.xenproject.org is hosted with RackSpace, monitoring our |