Re: [Xen-devel] [PATCH for-4.8] x86/svm: Don't clobber eax and edx if an RDMSR intercept fails

2016-11-10 Thread Wei Liu
On Wed, Nov 09, 2016 at 12:28:27PM +, Andrew Cooper wrote: > The original code has a bug; eax and edx get unconditionally updated even when > hvm_msr_read_intercept() doesn't return X86EMUL_OKAY. > > It is only by blind luck (vmce_rdmsr() eagerly initialising its msr_content > pointer) that

Re: [Xen-devel] [PATCH for-4.8] x86/svm: Don't clobber eax and edx if an RDMSR intercept fails

2016-11-09 Thread Andrew Cooper
On 09/11/16 14:14, Jan Beulich wrote: On 09.11.16 at 13:28, wrote: >> The original code has a bug; eax and edx get unconditionally updated even >> when >> hvm_msr_read_intercept() doesn't return X86EMUL_OKAY. >> >> It is only by blind luck (vmce_rdmsr() eagerly

Re: [Xen-devel] [PATCH for-4.8] x86/svm: Don't clobber eax and edx if an RDMSR intercept fails

2016-11-09 Thread Boris Ostrovsky
On 11/09/2016 07:28 AM, Andrew Cooper wrote: > The original code has a bug; eax and edx get unconditionally updated even when > hvm_msr_read_intercept() doesn't return X86EMUL_OKAY. > > It is only by blind luck (vmce_rdmsr() eagerly initialising its msr_content > pointer) that this isn't an

Re: [Xen-devel] [PATCH for-4.8] x86/svm: Don't clobber eax and edx if an RDMSR intercept fails

2016-11-09 Thread Jan Beulich
>>> On 09.11.16 at 13:28, wrote: > The original code has a bug; eax and edx get unconditionally updated even when > hvm_msr_read_intercept() doesn't return X86EMUL_OKAY. > > It is only by blind luck (vmce_rdmsr() eagerly initialising its msr_content > pointer) that

[Xen-devel] [PATCH for-4.8] x86/svm: Don't clobber eax and edx if an RDMSR intercept fails

2016-11-09 Thread Andrew Cooper
The original code has a bug; eax and edx get unconditionally updated even when hvm_msr_read_intercept() doesn't return X86EMUL_OKAY. It is only by blind luck (vmce_rdmsr() eagerly initialising its msr_content pointer) that this isn't an information leak into guests. While fixing this bug, reduce