Move KVM's PKRU handling code in fpu_copy_uabi_to_guest_fpstate() to copy_uabi_to_xstate() so that it is shared with other APIs that write the XSTATE such as PTRACE_SETREGSET with NT_X86_XSTATE.
This restores the pre-5.14 behavior of ptrace. The regression can be seen by running gdb and executing `p $pkru`, `set $pkru = 42`, and `p $pkru`. On affected kernels (5.14+) the write to the PKRU register (which gdb performs through ptrace) is ignored.
Fixes: e84ba47e313d ("x86/fpu: Hook up PKRU into ptrace()") Signed-off-by: Kyle Huey me@kylehuey.com Cc: Dave Hansen dave.hansen@linux.intel.com Cc: Thomas Gleixner tglx@linutronix.de Cc: Borislav Petkov bp@suse.de Cc: stable@vger.kernel.org # 5.14+ --- arch/x86/kernel/fpu/core.c | 13 +------------ arch/x86/kernel/fpu/xstate.c | 21 ++++++++++++++++++++- 2 files changed, 21 insertions(+), 13 deletions(-)
diff --git a/arch/x86/kernel/fpu/core.c b/arch/x86/kernel/fpu/core.c index 550157686323..46b935bc87c8 100644 --- a/arch/x86/kernel/fpu/core.c +++ b/arch/x86/kernel/fpu/core.c @@ -391,8 +391,6 @@ int fpu_copy_uabi_to_guest_fpstate(struct fpu_guest *gfpu, const void *buf, { struct fpstate *kstate = gfpu->fpstate; const union fpregs_state *ustate = buf; - struct pkru_state *xpkru; - int ret;
if (!cpu_feature_enabled(X86_FEATURE_XSAVE)) { if (ustate->xsave.header.xfeatures & ~XFEATURE_MASK_FPSSE) @@ -406,16 +404,7 @@ int fpu_copy_uabi_to_guest_fpstate(struct fpu_guest *gfpu, const void *buf, if (ustate->xsave.header.xfeatures & ~xcr0) return -EINVAL;
- ret = copy_uabi_from_kernel_to_xstate(kstate, ustate, vpkru); - if (ret) - return ret; - - /* Retrieve PKRU if not in init state */ - if (kstate->regs.xsave.header.xfeatures & XFEATURE_MASK_PKRU) { - xpkru = get_xsave_addr(&kstate->regs.xsave, XFEATURE_PKRU); - *vpkru = xpkru->pkru; - } - return 0; + return copy_uabi_from_kernel_to_xstate(kstate, ustate, vpkru); } EXPORT_SYMBOL_GPL(fpu_copy_uabi_to_guest_fpstate); #endif /* CONFIG_KVM */ diff --git a/arch/x86/kernel/fpu/xstate.c b/arch/x86/kernel/fpu/xstate.c index 3a6ced76e932..bebc30c29ed3 100644 --- a/arch/x86/kernel/fpu/xstate.c +++ b/arch/x86/kernel/fpu/xstate.c @@ -1205,10 +1205,22 @@ static int copy_from_buffer(void *dst, unsigned int offset, unsigned int size, * @fpstate: The fpstate buffer to copy to * @kbuf: The UABI format buffer, if it comes from the kernel * @ubuf: The UABI format buffer, if it comes from userspace - * @pkru: unused + * @pkru: The location to write the PKRU value to * * Converts from the UABI format into the kernel internal hardware * dependent format. + * + * This function ultimately has three different callers with distinct PKRU + * behavior. + * 1. When called from sigreturn the PKRU register will be restored from + * @fpstate via an XRSTOR. Correctly copying the UABI format buffer to + * @fpstate is sufficient to cover this case, but the caller will also + * pass a pointer to the thread_struct's pkru field in @pkru and updating + * it is harmless. + * 2. When called from ptrace the PKRU register will be restored from the + * thread_struct's pkru field. A pointer to that is passed in @pkru. + * 3. When called from KVM the PKRU register will be restored from the vcpu's + * pkru field. A pointer to that is passed in @pkru. */ static int copy_uabi_to_xstate(struct fpstate *fpstate, const void *kbuf, const void __user *ubuf, u32 *pkru) @@ -1260,6 +1272,13 @@ static int copy_uabi_to_xstate(struct fpstate *fpstate, const void *kbuf, } }
+ if (hdr.xfeatures & XFEATURE_MASK_PKRU) { + struct pkru_state *xpkru; + + xpkru = __raw_xsave_addr(xsave, XFEATURE_PKRU); + *pkru = xpkru->pkru; + } + /* * The state that came in from userspace was user-state only. * Mask all the user states out of 'xfeatures':