From: Lai Jiangshan jiangshan.ljs@antgroup.com

commit 520a7e80c96d655fbe4650d9cc985bd9d0443389 upstream.

error_entry() calls fixup_bad_iret() before sync_regs() if it is a fault from a bad IRET, to copy pt_regs to the kernel stack. It switches to the kernel stack directly after sync_regs().

But error_entry() itself is also a function call, so it has to stash the address it is going to return to, in %r12 which is unnecessarily complicated.

Move the stack switching after error_entry() and get rid of the need to handle the return address.

[ bp: Massage commit message. ]

Signed-off-by: Lai Jiangshan jiangshan.ljs@antgroup.com Signed-off-by: Borislav Petkov bp@suse.de Link: https://lore.kernel.org/r/20220503032107.680190-3-jiangshanlai@gmail.com Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/entry/entry_64.S | 16 ++++++---------- 1 file changed, 6 insertions(+), 10 deletions(-)

--- a/arch/x86/entry/entry_64.S +++ b/arch/x86/entry/entry_64.S @@ -326,6 +326,8 @@ SYM_CODE_END(ret_from_fork) .macro idtentry_body cfunc has_error_code:req

call error_entry + movq %rax, %rsp /* switch to the task stack if from userspace */ + ENCODE_FRAME_POINTER UNWIND_HINT_REGS

movq %rsp, %rdi /* pt_regs pointer into 1st argument*/ @@ -1003,14 +1005,10 @@ SYM_CODE_START_LOCAL(error_entry) /* We have user CR3. Change to kernel CR3. */ SWITCH_TO_KERNEL_CR3 scratch_reg=%rax

+ leaq 8(%rsp), %rdi /* arg0 = pt_regs pointer */ .Lerror_entry_from_usermode_after_swapgs: /* Put us onto the real thread stack. */ - popq %r12 /* save return addr in %12 */ - movq %rsp, %rdi /* arg0 = pt_regs pointer */ call sync_regs - movq %rax, %rsp /* switch stack */ - ENCODE_FRAME_POINTER - pushq %r12 RET

/* @@ -1042,6 +1040,7 @@ SYM_CODE_START_LOCAL(error_entry) */ .Lerror_entry_done_lfence: FENCE_SWAPGS_KERNEL_ENTRY + leaq 8(%rsp), %rax /* return pt_regs pointer */ RET

.Lbstep_iret: @@ -1062,12 +1061,9 @@ SYM_CODE_START_LOCAL(error_entry) * Pretend that the exception came from user mode: set up pt_regs * as if we faulted immediately after IRET. */ - popq %r12 /* save return addr in %12 */ - movq %rsp, %rdi /* arg0 = pt_regs pointer */ + leaq 8(%rsp), %rdi /* arg0 = pt_regs pointer */ call fixup_bad_iret - mov %rax, %rsp - ENCODE_FRAME_POINTER - pushq %r12 + mov %rax, %rdi jmp .Lerror_entry_from_usermode_after_swapgs SYM_CODE_END(error_entry)

From: Lai Jiangshan jiangshan.ljs@antgroup.com

commit 64cbd0acb58203fb769ed2f4eab526d43e243847 upstream.

XENPV guests enter already on the task stack and they can't fault for native_iret() nor native_load_gs_index() since they use their own pvop for IRET and load_gs_index(). A CR3 switch is not needed either.

So there is no reason to call error_entry() in XENPV.

[ bp: Massage commit message. ]

Signed-off-by: Lai Jiangshan jiangshan.ljs@antgroup.com Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Juergen Gross jgross@suse.com Link: https://lore.kernel.org/r/20220503032107.680190-6-jiangshanlai@gmail.com Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/entry/entry_64.S | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-)

--- a/arch/x86/entry/entry_64.S +++ b/arch/x86/entry/entry_64.S @@ -336,8 +336,17 @@ SYM_CODE_END(push_and_clear_regs) call push_and_clear_regs UNWIND_HINT_REGS

- call error_entry - movq %rax, %rsp /* switch to the task stack if from userspace */ + /* + * Call error_entry() and switch to the task stack if from userspace. + * + * When in XENPV, it is already in the task stack, and it can't fault + * for native_iret() nor native_load_gs_index() since XENPV uses its + * own pvops for IRET and load_gs_index(). And it doesn't need to + * switch the CR3. So it can skip invoking error_entry(). + */ + ALTERNATIVE "call error_entry; movq %rax, %rsp", \ + "", X86_FEATURE_XENPV + ENCODE_FRAME_POINTER UNWIND_HINT_REGS

From: Peter Zijlstra peterz@infradead.org

commit 742ab6df974ae8384a2dd213db1a3a06cf6d8936 upstream.

The recent mmio_stale_data fixes broke the noinstr constraints:

vmlinux.o: warning: objtool: vmx_vcpu_enter_exit+0x15b: call to wrmsrl.constprop.0() leaves .noinstr.text section vmlinux.o: warning: objtool: vmx_vcpu_enter_exit+0x1bf: call to kvm_arch_has_assigned_device() leaves .noinstr.text section

make it all happy again.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kvm/vmx/vmx.c | 6 +++--- arch/x86/kvm/x86.c | 4 ++-- include/linux/kvm_host.h | 2 +- 3 files changed, 6 insertions(+), 6 deletions(-)

--- a/arch/x86/kvm/vmx/vmx.c +++ b/arch/x86/kvm/vmx/vmx.c @@ -383,9 +383,9 @@ static __always_inline void vmx_disable_ if (!vmx->disable_fb_clear) return;

- rdmsrl(MSR_IA32_MCU_OPT_CTRL, msr); + msr = __rdmsr(MSR_IA32_MCU_OPT_CTRL); msr |= FB_CLEAR_DIS; - wrmsrl(MSR_IA32_MCU_OPT_CTRL, msr); + native_wrmsrl(MSR_IA32_MCU_OPT_CTRL, msr); /* Cache the MSR value to avoid reading it later */ vmx->msr_ia32_mcu_opt_ctrl = msr; } @@ -396,7 +396,7 @@ static __always_inline void vmx_enable_f return;

vmx->msr_ia32_mcu_opt_ctrl &= ~FB_CLEAR_DIS; - wrmsrl(MSR_IA32_MCU_OPT_CTRL, vmx->msr_ia32_mcu_opt_ctrl); + native_wrmsrl(MSR_IA32_MCU_OPT_CTRL, vmx->msr_ia32_mcu_opt_ctrl); }

static void vmx_update_fb_clear_dis(struct kvm_vcpu *vcpu, struct vcpu_vmx *vmx) --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -12533,9 +12533,9 @@ void kvm_arch_end_assignment(struct kvm } EXPORT_SYMBOL_GPL(kvm_arch_end_assignment);

-bool kvm_arch_has_assigned_device(struct kvm *kvm) +bool noinstr kvm_arch_has_assigned_device(struct kvm *kvm) { - return atomic_read(&kvm->arch.assigned_device_count); + return arch_atomic_read(&kvm->arch.assigned_device_count); } EXPORT_SYMBOL_GPL(kvm_arch_has_assigned_device);

--- a/include/linux/kvm_host.h +++ b/include/linux/kvm_host.h @@ -1511,7 +1511,7 @@ static inline void kvm_arch_end_assignme { }

-static inline bool kvm_arch_has_assigned_device(struct kvm *kvm) +static __always_inline bool kvm_arch_has_assigned_device(struct kvm *kvm) { return false; }

From: Peter Zijlstra peterz@infradead.org

commit 369ae6ffc41a3c1137cab697635a84d0cc7cdcea upstream.

On it's own not much of a cleanup but it prepares for more/similar code.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/include/asm/disabled-features.h | 9 ++++++++- arch/x86/include/asm/nospec-branch.h | 7 +++---- arch/x86/net/bpf_jit_comp.c | 7 +++---- 3 files changed, 14 insertions(+), 9 deletions(-)

--- a/arch/x86/include/asm/disabled-features.h +++ b/arch/x86/include/asm/disabled-features.h @@ -56,6 +56,13 @@ # define DISABLE_PTI (1 << (X86_FEATURE_PTI & 31)) #endif

+#ifdef CONFIG_RETPOLINE +# define DISABLE_RETPOLINE 0 +#else +# define DISABLE_RETPOLINE ((1 << (X86_FEATURE_RETPOLINE & 31)) | \ + (1 << (X86_FEATURE_RETPOLINE_LFENCE & 31))) +#endif + #ifdef CONFIG_INTEL_IOMMU_SVM # define DISABLE_ENQCMD 0 #else @@ -82,7 +89,7 @@ #define DISABLED_MASK8 0 #define DISABLED_MASK9 (DISABLE_SMAP|DISABLE_SGX) #define DISABLED_MASK10 0 -#define DISABLED_MASK11 0 +#define DISABLED_MASK11 (DISABLE_RETPOLINE) #define DISABLED_MASK12 0 #define DISABLED_MASK13 0 #define DISABLED_MASK14 0 --- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -120,17 +120,16 @@ _ASM_PTR " 999b\n\t" \ ".popsection\n\t"

-#ifdef CONFIG_RETPOLINE - typedef u8 retpoline_thunk_t[RETPOLINE_THUNK_SIZE]; +extern retpoline_thunk_t __x86_indirect_thunk_array[]; + +#ifdef CONFIG_RETPOLINE

#define GEN(reg) \ extern retpoline_thunk_t __x86_indirect_thunk_ ## reg; #include <asm/GEN-for-each-reg.h> #undef GEN

-extern retpoline_thunk_t __x86_indirect_thunk_array[]; - #ifdef CONFIG_X86_64

/* --- a/arch/x86/net/bpf_jit_comp.c +++ b/arch/x86/net/bpf_jit_comp.c @@ -407,16 +407,15 @@ static void emit_indirect_jump(u8 **ppro { u8 *prog = *pprog;

-#ifdef CONFIG_RETPOLINE if (cpu_feature_enabled(X86_FEATURE_RETPOLINE_LFENCE)) { EMIT_LFENCE(); EMIT2(0xFF, 0xE0 + reg); } else if (cpu_feature_enabled(X86_FEATURE_RETPOLINE)) { OPTIMIZER_HIDE_VAR(reg); emit_jump(&prog, &__x86_indirect_thunk_array[reg], ip); - } else -#endif - EMIT2(0xFF, 0xE0 + reg); + } else { + EMIT2(0xFF, 0xE0 + reg); + }

*pprog = prog; }

From: Peter Zijlstra peterz@infradead.org

commit 0b53c374b9eff2255a386f1f1cfb9a928e52a5ae upstream.

Utilize -mfunction-return=thunk-extern when available to have the compiler replace RET instructions with direct JMPs to the symbol __x86_return_thunk. This does not affect assembler (.S) sources, only C sources.

-mfunction-return=thunk-extern has been available since gcc 7.3 and clang 15.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Nick Desaulniers ndesaulniers@google.com Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Tested-by: Nick Desaulniers ndesaulniers@google.com Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/Makefile | 2 ++ arch/x86/include/asm/nospec-branch.h | 2 ++ arch/x86/lib/retpoline.S | 13 +++++++++++++ 3 files changed, 17 insertions(+)

--- a/arch/x86/Makefile +++ b/arch/x86/Makefile @@ -15,11 +15,13 @@ endif ifdef CONFIG_CC_IS_GCC RETPOLINE_CFLAGS := $(call cc-option,-mindirect-branch=thunk-extern -mindirect-branch-register) RETPOLINE_CFLAGS += $(call cc-option,-mindirect-branch-cs-prefix) +RETPOLINE_CFLAGS += $(call cc-option,-mfunction-return=thunk-extern) RETPOLINE_VDSO_CFLAGS := $(call cc-option,-mindirect-branch=thunk-inline -mindirect-branch-register) endif ifdef CONFIG_CC_IS_CLANG RETPOLINE_CFLAGS := -mretpoline-external-thunk RETPOLINE_VDSO_CFLAGS := -mretpoline +RETPOLINE_CFLAGS += $(call cc-option,-mfunction-return=thunk-extern) endif export RETPOLINE_CFLAGS export RETPOLINE_VDSO_CFLAGS --- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -123,6 +123,8 @@ typedef u8 retpoline_thunk_t[RETPOLINE_THUNK_SIZE]; extern retpoline_thunk_t __x86_indirect_thunk_array[];

+extern void __x86_return_thunk(void); + #ifdef CONFIG_RETPOLINE

#define GEN(reg) \ --- a/arch/x86/lib/retpoline.S +++ b/arch/x86/lib/retpoline.S @@ -67,3 +67,16 @@ SYM_CODE_END(__x86_indirect_thunk_array) #define GEN(reg) EXPORT_THUNK(reg) #include <asm/GEN-for-each-reg.h> #undef GEN + +/* + * This function name is magical and is used by -mfunction-return=thunk-extern + * for the compiler to generate JMPs to it. + */ +SYM_CODE_START(__x86_return_thunk) + UNWIND_HINT_EMPTY + ANNOTATE_NOENDBR + ret + int3 +SYM_CODE_END(__x86_return_thunk) + +__EXPORT_THUNK(__x86_return_thunk)

From: Peter Zijlstra peterz@infradead.org

commit d9e9d2300681d68a775c28de6aa6e5290ae17796 upstream.

Find all the return-thunk sites and record them in a .return_sites section such that the kernel can undo this.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- tools/objtool/arch/x86/decode.c | 5 ++ tools/objtool/check.c | 74 ++++++++++++++++++++++++++++++++ tools/objtool/include/objtool/arch.h | 1 tools/objtool/include/objtool/elf.h | 1 tools/objtool/include/objtool/objtool.h | 1 tools/objtool/objtool.c | 1 6 files changed, 83 insertions(+)

--- a/tools/objtool/arch/x86/decode.c +++ b/tools/objtool/arch/x86/decode.c @@ -787,3 +787,8 @@ bool arch_is_retpoline(struct symbol *sy { return !strncmp(sym->name, "__x86_indirect_", 15); } + +bool arch_is_rethunk(struct symbol *sym) +{ + return !strcmp(sym->name, "__x86_return_thunk"); +} --- a/tools/objtool/check.c +++ b/tools/objtool/check.c @@ -747,6 +747,52 @@ static int create_retpoline_sites_sectio return 0; }

+static int create_return_sites_sections(struct objtool_file *file) +{ + struct instruction *insn; + struct section *sec; + int idx; + + sec = find_section_by_name(file->elf, ".return_sites"); + if (sec) { + WARN("file already has .return_sites, skipping"); + return 0; + } + + idx = 0; + list_for_each_entry(insn, &file->return_thunk_list, call_node) + idx++; + + if (!idx) + return 0; + + sec = elf_create_section(file->elf, ".return_sites", 0, + sizeof(int), idx); + if (!sec) { + WARN("elf_create_section: .return_sites"); + return -1; + } + + idx = 0; + list_for_each_entry(insn, &file->return_thunk_list, call_node) { + + int *site = (int *)sec->data->d_buf + idx; + *site = 0; + + if (elf_add_reloc_to_insn(file->elf, sec, + idx * sizeof(int), + R_X86_64_PC32, + insn->sec, insn->offset)) { + WARN("elf_add_reloc_to_insn: .return_sites"); + return -1; + } + + idx++; + } + + return 0; +} + static int create_ibt_endbr_seal_sections(struct objtool_file *file) { struct instruction *insn; @@ -1081,6 +1127,11 @@ __weak bool arch_is_retpoline(struct sym return false; }

+__weak bool arch_is_rethunk(struct symbol *sym) +{ + return false; +} + #define NEGATIVE_RELOC ((void *)-1L)

static struct reloc *insn_reloc(struct objtool_file *file, struct instruction *insn) @@ -1248,6 +1299,18 @@ static void add_retpoline_call(struct ob annotate_call_site(file, insn, false); }

+static void add_return_call(struct objtool_file *file, struct instruction *insn) +{ + /* + * Return thunk tail calls are really just returns in disguise, + * so convert them accordingly. + */ + insn->type = INSN_RETURN; + insn->retpoline_safe = true; + + list_add_tail(&insn->call_node, &file->return_thunk_list); +} + static bool same_function(struct instruction *insn1, struct instruction *insn2) { return insn1->func->pfunc == insn2->func->pfunc; @@ -1300,6 +1363,9 @@ static int add_jump_destinations(struct } else if (reloc->sym->retpoline_thunk) { add_retpoline_call(file, insn); continue; + } else if (reloc->sym->return_thunk) { + add_return_call(file, insn); + continue; } else if (insn->func) { /* * External sibling call or internal sibling call with @@ -2182,6 +2248,9 @@ static int classify_symbols(struct objto if (arch_is_retpoline(func)) func->retpoline_thunk = true;

+ if (arch_is_rethunk(func)) + func->return_thunk = true; + if (!strcmp(func->name, "__fentry__")) func->fentry = true;

@@ -3935,6 +4004,11 @@ int check(struct objtool_file *file) if (ret < 0) goto out; warnings += ret; + + ret = create_return_sites_sections(file); + if (ret < 0) + goto out; + warnings += ret; }

if (mcount) { --- a/tools/objtool/include/objtool/arch.h +++ b/tools/objtool/include/objtool/arch.h @@ -89,6 +89,7 @@ const char *arch_ret_insn(int len); int arch_decode_hint_reg(u8 sp_reg, int *base);

bool arch_is_retpoline(struct symbol *sym); +bool arch_is_rethunk(struct symbol *sym);

int arch_rewrite_retpolines(struct objtool_file *file);

--- a/tools/objtool/include/objtool/elf.h +++ b/tools/objtool/include/objtool/elf.h @@ -57,6 +57,7 @@ struct symbol { u8 uaccess_safe : 1; u8 static_call_tramp : 1; u8 retpoline_thunk : 1; + u8 return_thunk : 1; u8 fentry : 1; u8 profiling_func : 1; struct list_head pv_target; --- a/tools/objtool/include/objtool/objtool.h +++ b/tools/objtool/include/objtool/objtool.h @@ -24,6 +24,7 @@ struct objtool_file { struct list_head insn_list; DECLARE_HASHTABLE(insn_hash, 20); struct list_head retpoline_call_list; + struct list_head return_thunk_list; struct list_head static_call_list; struct list_head mcount_loc_list; struct list_head endbr_list; --- a/tools/objtool/objtool.c +++ b/tools/objtool/objtool.c @@ -126,6 +126,7 @@ struct objtool_file *objtool_open_read(c INIT_LIST_HEAD(&file.insn_list); hash_init(file.insn_hash); INIT_LIST_HEAD(&file.retpoline_call_list); + INIT_LIST_HEAD(&file.return_thunk_list); INIT_LIST_HEAD(&file.static_call_list); INIT_LIST_HEAD(&file.mcount_loc_list); INIT_LIST_HEAD(&file.endbr_list);

From: Peter Zijlstra peterz@infradead.org

commit 1f001e9da6bbf482311e45e48f53c2bd2179e59c upstream.

Use the return thunk in ftrace trampolines, if needed.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kernel/ftrace.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)

--- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -303,7 +303,7 @@ union ftrace_op_code_union { } __attribute__((packed)); };

-#define RET_SIZE 1 + IS_ENABLED(CONFIG_SLS) +#define RET_SIZE (IS_ENABLED(CONFIG_RETPOLINE) ? 5 : 1 + IS_ENABLED(CONFIG_SLS))

static unsigned long create_trampoline(struct ftrace_ops *ops, unsigned int *tramp_size) @@ -359,7 +359,10 @@ create_trampoline(struct ftrace_ops *ops goto fail;

ip = trampoline + size; - memcpy(ip, retq, RET_SIZE); + if (cpu_feature_enabled(X86_FEATURE_RETHUNK)) + __text_gen_insn(ip, JMP32_INSN_OPCODE, ip, &__x86_return_thunk, JMP32_INSN_SIZE); + else + memcpy(ip, retq, sizeof(retq));

/* No need to test direct calls on created trampolines */ if (ops->flags & FTRACE_OPS_FL_SAVE_REGS) {

From: Peter Zijlstra peterz@infradead.org

commit af2e140f34208a5dfb6b7a8ad2d56bda88f0524d upstream.

Prepare the SETcc fastop stuff for when RET can be larger still.

The tricky bit here is that the expressions should not only be constant C expressions, but also absolute GAS expressions. This means no ?: and 'true' is ~0.

Also ensure em_setcc() has the same alignment as the actual FOP_SETCC() ops, this ensures there cannot be an alignment hole between em_setcc() and the first op.

Additionally, add a .skip directive to the FOP_SETCC() macro to fill any remaining space with INT3 traps; however the primary purpose of this directive is to generate AS warnings when the remaining space goes negative. Which is a very good indication the alignment magic went side-ways.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kvm/emulate.c | 28 +++++++++++++++------------- 1 file changed, 15 insertions(+), 13 deletions(-)

--- a/arch/x86/kvm/emulate.c +++ b/arch/x86/kvm/emulate.c @@ -325,13 +325,15 @@ static int fastop(struct x86_emulate_ctx #define FOP_RET(name) \ __FOP_RET(#name)

-#define FOP_START(op) \ +#define __FOP_START(op, align) \ extern void em_##op(struct fastop *fake); \ asm(".pushsection .text, "ax" \n\t" \ ".global em_" #op " \n\t" \ - ".align " __stringify(FASTOP_SIZE) " \n\t" \ + ".align " __stringify(align) " \n\t" \ "em_" #op ":\n\t"

+#define FOP_START(op) __FOP_START(op, FASTOP_SIZE) + #define FOP_END \ ".popsection")

@@ -435,16 +437,15 @@ static int fastop(struct x86_emulate_ctx /* * Depending on .config the SETcc functions look like: * - * ENDBR [4 bytes; CONFIG_X86_KERNEL_IBT] - * SETcc %al [3 bytes] - * RET [1 byte] - * INT3 [1 byte; CONFIG_SLS] - * - * Which gives possible sizes 4, 5, 8 or 9. When rounded up to the - * next power-of-two alignment they become 4, 8 or 16 resp. + * ENDBR [4 bytes; CONFIG_X86_KERNEL_IBT] + * SETcc %al [3 bytes] + * RET | JMP __x86_return_thunk [1,5 bytes; CONFIG_RETPOLINE] + * INT3 [1 byte; CONFIG_SLS] */ -#define SETCC_LENGTH (ENDBR_INSN_SIZE + 4 + IS_ENABLED(CONFIG_SLS)) -#define SETCC_ALIGN (4 << IS_ENABLED(CONFIG_SLS) << HAS_KERNEL_IBT) +#define RET_LENGTH (1 + (4 * IS_ENABLED(CONFIG_RETPOLINE)) + \ + IS_ENABLED(CONFIG_SLS)) +#define SETCC_LENGTH (ENDBR_INSN_SIZE + 3 + RET_LENGTH) +#define SETCC_ALIGN (4 << ((SETCC_LENGTH > 4) & 1) << ((SETCC_LENGTH > 8) & 1)) static_assert(SETCC_LENGTH <= SETCC_ALIGN);

#define FOP_SETCC(op) \ @@ -453,9 +454,10 @@ static_assert(SETCC_LENGTH <= SETCC_ALIG #op ": \n\t" \ ASM_ENDBR \ #op " %al \n\t" \ - __FOP_RET(#op) + __FOP_RET(#op) \ + ".skip " __stringify(SETCC_ALIGN) " - (.-" #op "), 0xcc \n\t"

-FOP_START(setcc) +__FOP_START(setcc, SETCC_ALIGN) FOP_SETCC(seto) FOP_SETCC(setno) FOP_SETCC(setc)

From: Kim Phillips kim.phillips@amd.com

commit 0ee9073000e8791f8b134a8ded31bcc767f7f232 upstream.

Specifically, it's because __enc_copy() encrypts the kernel after being relocated outside the kernel in sme_encrypt_execute(), and the RET macro's jmp offset isn't amended prior to execution.

Signed-off-by: Kim Phillips kim.phillips@amd.com Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/mm/mem_encrypt_boot.S | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)

--- a/arch/x86/mm/mem_encrypt_boot.S +++ b/arch/x86/mm/mem_encrypt_boot.S @@ -65,7 +65,9 @@ SYM_FUNC_START(sme_encrypt_execute) movq %rbp, %rsp /* Restore original stack pointer */ pop %rbp

- RET + /* Offset to __x86_return_thunk would be wrong here */ + ret + int3 SYM_FUNC_END(sme_encrypt_execute)

SYM_FUNC_START(__enc_copy) @@ -151,6 +153,8 @@ SYM_FUNC_START(__enc_copy) pop %r12 pop %r15

- RET + /* Offset to __x86_return_thunk would be wrong here */ + ret + int3 .L__enc_copy_end: SYM_FUNC_END(__enc_copy)

From: Peter Zijlstra peterz@infradead.org

commit 7c81c0c9210c9bfab2bae76aab2999de5bad27db upstream.

Commit

ee774dac0da1 ("x86/entry: Move PUSH_AND_CLEAR_REGS out of error_entry()")

manages to introduce a CALL/RET pair that is before SWITCH_TO_KERNEL_CR3, which means it is before RETBleed can be mitigated.

Revert to an earlier version of the commit in Fixes. Down side is that this will bloat .text size somewhat. The alternative is fully reverting it.

The purpose of this patch was to allow migrating error_entry() to C, including the whole of kPTI. Much care needs to be taken moving that forward to not re-introduce this problem of early RETs.

Fixes: ee774dac0da1 ("x86/entry: Move PUSH_AND_CLEAR_REGS out of error_entry()") Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/entry/entry_64.S | 12 ++---------- 1 file changed, 2 insertions(+), 10 deletions(-)

--- a/arch/x86/entry/entry_64.S +++ b/arch/x86/entry/entry_64.S @@ -317,14 +317,6 @@ SYM_CODE_END(ret_from_fork) #endif .endm

-/* Save all registers in pt_regs */ -SYM_CODE_START_LOCAL(push_and_clear_regs) - UNWIND_HINT_FUNC - PUSH_AND_CLEAR_REGS save_ret=1 - ENCODE_FRAME_POINTER 8 - RET -SYM_CODE_END(push_and_clear_regs) - /** * idtentry_body - Macro to emit code calling the C function * @cfunc: C function to be called @@ -332,8 +324,8 @@ SYM_CODE_END(push_and_clear_regs) */ .macro idtentry_body cfunc has_error_code:req

- call push_and_clear_regs - UNWIND_HINT_REGS + PUSH_AND_CLEAR_REGS + ENCODE_FRAME_POINTER

/* * Call error_entry() and switch to the task stack if from userspace.

From: Thadeu Lima de Souza Cascardo cascardo@canonical.com

commit a149180fbcf336e97ce4eb2cdc13672727feb94d upstream.

Note: needs to be in a section distinct from Retpolines such that the Retpoline RET substitution cannot possibly use immediate jumps.

ORC unwinding for zen_untrain_ret() and __x86_return_thunk() is a little tricky but works due to the fact that zen_untrain_ret() doesn't have any stack ops and as such will emit a single ORC entry at the start (+0x3f).

Meanwhile, unwinding an IP, including the __x86_return_thunk() one (+0x40) will search for the largest ORC entry smaller or equal to the IP, these will find the one ORC entry (+0x3f) and all works.

[ Alexandre: SVM part. ] [ bp: Build fix, massages. ]

Suggested-by: Andrew Cooper Andrew.Cooper3@citrix.com Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de [cascardo: conflicts at arch/x86/entry/entry_64_compat.S] Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/entry/entry_64.S | 6 ++ arch/x86/entry/entry_64_compat.S | 4 + arch/x86/include/asm/cpufeatures.h | 1 arch/x86/include/asm/disabled-features.h | 3 - arch/x86/include/asm/nospec-branch.h | 17 ++++++++ arch/x86/kernel/vmlinux.lds.S | 2 arch/x86/kvm/svm/vmenter.S | 18 ++++++++ arch/x86/lib/retpoline.S | 64 +++++++++++++++++++++++++++++-- tools/objtool/check.c | 21 ++++++++-- 9 files changed, 127 insertions(+), 9 deletions(-)

--- a/arch/x86/entry/entry_64.S +++ b/arch/x86/entry/entry_64.S @@ -96,6 +96,7 @@ SYM_CODE_START(entry_SYSCALL_64)

SYM_INNER_LABEL(entry_SYSCALL_64_safe_stack, SYM_L_GLOBAL) ANNOTATE_NOENDBR + UNTRAIN_RET

/* Construct struct pt_regs on stack */ pushq $__USER_DS /* pt_regs->ss */ @@ -708,6 +709,7 @@ native_irq_return_ldt: pushq %rdi /* Stash user RDI */ swapgs /* to kernel GS */ SWITCH_TO_KERNEL_CR3 scratch_reg=%rdi /* to kernel CR3 */ + UNTRAIN_RET

movq PER_CPU_VAR(espfix_waddr), %rdi movq %rax, (0*8)(%rdi) /* user RAX */ @@ -903,6 +905,7 @@ SYM_CODE_START_LOCAL(paranoid_entry) * be retrieved from a kernel internal table. */ SAVE_AND_SWITCH_TO_KERNEL_CR3 scratch_reg=%rax save_reg=%r14 + UNTRAIN_RET

/* * Handling GSBASE depends on the availability of FSGSBASE. @@ -1013,6 +1016,7 @@ SYM_CODE_START_LOCAL(error_entry) FENCE_SWAPGS_USER_ENTRY /* We have user CR3. Change to kernel CR3. */ SWITCH_TO_KERNEL_CR3 scratch_reg=%rax + UNTRAIN_RET

leaq 8(%rsp), %rdi /* arg0 = pt_regs pointer */ .Lerror_entry_from_usermode_after_swapgs: @@ -1065,6 +1069,7 @@ SYM_CODE_START_LOCAL(error_entry) SWAPGS FENCE_SWAPGS_USER_ENTRY SWITCH_TO_KERNEL_CR3 scratch_reg=%rax + UNTRAIN_RET

/* * Pretend that the exception came from user mode: set up pt_regs @@ -1160,6 +1165,7 @@ SYM_CODE_START(asm_exc_nmi) movq %rsp, %rdx movq PER_CPU_VAR(cpu_current_top_of_stack), %rsp UNWIND_HINT_IRET_REGS base=%rdx offset=8 + UNTRAIN_RET pushq 5*8(%rdx) /* pt_regs->ss */ pushq 4*8(%rdx) /* pt_regs->rsp */ pushq 3*8(%rdx) /* pt_regs->flags */ --- a/arch/x86/entry/entry_64_compat.S +++ b/arch/x86/entry/entry_64_compat.S @@ -14,6 +14,7 @@ #include <asm/irqflags.h> #include <asm/asm.h> #include <asm/smap.h> +#include <asm/nospec-branch.h> #include <linux/linkage.h> #include <linux/err.h>

@@ -72,6 +73,7 @@ SYM_CODE_START(entry_SYSENTER_compat) pushq $__USER32_CS /* pt_regs->cs */ pushq $0 /* pt_regs->ip = 0 (placeholder) */ SYM_INNER_LABEL(entry_SYSENTER_compat_after_hwframe, SYM_L_GLOBAL) + UNTRAIN_RET

/* * User tracing code (ptrace or signal handlers) might assume that @@ -215,6 +217,7 @@ SYM_CODE_START(entry_SYSCALL_compat)

SYM_INNER_LABEL(entry_SYSCALL_compat_safe_stack, SYM_L_GLOBAL) ANNOTATE_NOENDBR + UNTRAIN_RET

/* Construct struct pt_regs on stack */ pushq $__USER32_DS /* pt_regs->ss */ @@ -382,6 +385,7 @@ SYM_CODE_START(entry_INT80_compat) pushq (%rdi) /* pt_regs->di */ .Lint80_keep_stack:

+ UNTRAIN_RET pushq %rsi /* pt_regs->si */ xorl %esi, %esi /* nospec si */ pushq %rdx /* pt_regs->dx */ --- a/arch/x86/include/asm/cpufeatures.h +++ b/arch/x86/include/asm/cpufeatures.h @@ -300,6 +300,7 @@ #define X86_FEATURE_RETPOLINE (11*32+12) /* "" Generic Retpoline mitigation for Spectre variant 2 */ #define X86_FEATURE_RETPOLINE_LFENCE (11*32+13) /* "" Use LFENCE for Spectre variant 2 */ #define X86_FEATURE_RETHUNK (11*32+14) /* "" Use REturn THUNK */ +#define X86_FEATURE_UNRET (11*32+15) /* "" AMD BTB untrain return */

/* Intel-defined CPU features, CPUID level 0x00000007:1 (EAX), word 12 */ #define X86_FEATURE_AVX_VNNI (12*32+ 4) /* AVX VNNI instructions */ --- a/arch/x86/include/asm/disabled-features.h +++ b/arch/x86/include/asm/disabled-features.h @@ -61,7 +61,8 @@ #else # define DISABLE_RETPOLINE ((1 << (X86_FEATURE_RETPOLINE & 31)) | \ (1 << (X86_FEATURE_RETPOLINE_LFENCE & 31)) | \ - (1 << (X86_FEATURE_RETHUNK & 31))) + (1 << (X86_FEATURE_RETHUNK & 31)) | \ + (1 << (X86_FEATURE_UNRET & 31))) #endif

#ifdef CONFIG_INTEL_IOMMU_SVM --- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -112,6 +112,22 @@ #endif .endm

+/* + * Mitigate RETBleed for AMD/Hygon Zen uarch. Requires KERNEL CR3 because the + * return thunk isn't mapped into the userspace tables (then again, AMD + * typically has NO_MELTDOWN). + * + * Doesn't clobber any registers but does require a stable stack. + * + * As such, this must be placed after every *SWITCH_TO_KERNEL_CR3 at a point + * where we have a stack but before any RET instruction. + */ +.macro UNTRAIN_RET +#ifdef CONFIG_RETPOLINE + ALTERNATIVE "", "call zen_untrain_ret", X86_FEATURE_UNRET +#endif +.endm + #else /* __ASSEMBLY__ */

#define ANNOTATE_RETPOLINE_SAFE \ @@ -124,6 +140,7 @@ typedef u8 retpoline_thunk_t[RETPOLINE_T extern retpoline_thunk_t __x86_indirect_thunk_array[];

extern void __x86_return_thunk(void); +extern void zen_untrain_ret(void);

#ifdef CONFIG_RETPOLINE

--- a/arch/x86/kernel/vmlinux.lds.S +++ b/arch/x86/kernel/vmlinux.lds.S @@ -141,7 +141,7 @@ SECTIONS

#ifdef CONFIG_RETPOLINE __indirect_thunk_start = .; - *(.text.__x86.indirect_thunk) + *(.text.__x86.*) __indirect_thunk_end = .; #endif } :text =0xcccc --- a/arch/x86/kvm/svm/vmenter.S +++ b/arch/x86/kvm/svm/vmenter.S @@ -111,6 +111,15 @@ SYM_FUNC_START(__svm_vcpu_run) #endif

/* + * Mitigate RETBleed for AMD/Hygon Zen uarch. RET should be + * untrained as soon as we exit the VM and are back to the + * kernel. This should be done before re-enabling interrupts + * because interrupt handlers won't sanitize 'ret' if the return is + * from the kernel. + */ + UNTRAIN_RET + + /* * Clear all general purpose registers except RSP and RAX to prevent * speculative use of the guest's values, even those that are reloaded * via the stack. In theory, an L1 cache miss when restoring registers @@ -190,6 +199,15 @@ SYM_FUNC_START(__svm_sev_es_vcpu_run) FILL_RETURN_BUFFER %_ASM_AX, RSB_CLEAR_LOOPS, X86_FEATURE_RETPOLINE #endif

+ /* + * Mitigate RETBleed for AMD/Hygon Zen uarch. RET should be + * untrained as soon as we exit the VM and are back to the + * kernel. This should be done before re-enabling interrupts + * because interrupt handlers won't sanitize RET if the return is + * from the kernel. + */ + UNTRAIN_RET + pop %_ASM_BX

#ifdef CONFIG_X86_64 --- a/arch/x86/lib/retpoline.S +++ b/arch/x86/lib/retpoline.S @@ -72,11 +72,67 @@ SYM_CODE_END(__x86_indirect_thunk_array) * This function name is magical and is used by -mfunction-return=thunk-extern * for the compiler to generate JMPs to it. */ -SYM_CODE_START(__x86_return_thunk) - UNWIND_HINT_EMPTY - ANNOTATE_NOENDBR + .section .text.__x86.return_thunk + +/* + * Safety details here pertain to the AMD Zen{1,2} microarchitecture: + * 1) The RET at __x86_return_thunk must be on a 64 byte boundary, for + * alignment within the BTB. + * 2) The instruction at zen_untrain_ret must contain, and not + * end with, the 0xc3 byte of the RET. + * 3) STIBP must be enabled, or SMT disabled, to prevent the sibling thread + * from re-poisioning the BTB prediction. + */ + .align 64 + .skip 63, 0xcc +SYM_FUNC_START_NOALIGN(zen_untrain_ret); + + /* + * As executed from zen_untrain_ret, this is: + * + * TEST $0xcc, %bl + * LFENCE + * JMP __x86_return_thunk + * + * Executing the TEST instruction has a side effect of evicting any BTB + * prediction (potentially attacker controlled) attached to the RET, as + * __x86_return_thunk + 1 isn't an instruction boundary at the moment. + */ + .byte 0xf6 + + /* + * As executed from __x86_return_thunk, this is a plain RET. + * + * As part of the TEST above, RET is the ModRM byte, and INT3 the imm8. + * + * We subsequently jump backwards and architecturally execute the RET. + * This creates a correct BTB prediction (type=ret), but in the + * meantime we suffer Straight Line Speculation (because the type was + * no branch) which is halted by the INT3. + * + * With SMT enabled and STIBP active, a sibling thread cannot poison + * RET's prediction to a type of its choice, but can evict the + * prediction due to competitive sharing. If the prediction is + * evicted, __x86_return_thunk will suffer Straight Line Speculation + * which will be contained safely by the INT3. + */ +SYM_INNER_LABEL(__x86_return_thunk, SYM_L_GLOBAL) ret int3 SYM_CODE_END(__x86_return_thunk)

-__EXPORT_THUNK(__x86_return_thunk) + /* + * Ensure the TEST decoding / BTB invalidation is complete. + */ + lfence + + /* + * Jump back and execute the RET in the middle of the TEST instruction. + * INT3 is for SLS protection. + */ + jmp __x86_return_thunk + int3 +SYM_FUNC_END(zen_untrain_ret) +__EXPORT_THUNK(zen_untrain_ret) + +EXPORT_SYMBOL(__x86_return_thunk) --- a/tools/objtool/check.c +++ b/tools/objtool/check.c @@ -1300,7 +1300,7 @@ static void add_retpoline_call(struct ob annotate_call_site(file, insn, false); }

-static void add_return_call(struct objtool_file *file, struct instruction *insn) +static void add_return_call(struct objtool_file *file, struct instruction *insn, bool add) { /* * Return thunk tail calls are really just returns in disguise, @@ -1310,7 +1310,7 @@ static void add_return_call(struct objto insn->retpoline_safe = true;

/* Skip the non-text sections, specially .discard ones */ - if (insn->sec->text) + if (add && insn->sec->text) list_add_tail(&insn->call_node, &file->return_thunk_list); }

@@ -1367,7 +1367,7 @@ static int add_jump_destinations(struct add_retpoline_call(file, insn); continue; } else if (reloc->sym->return_thunk) { - add_return_call(file, insn); + add_return_call(file, insn, true); continue; } else if (insn->func) { /* @@ -1387,6 +1387,21 @@ static int add_jump_destinations(struct

jump_dest = find_insn(file, dest_sec, dest_off); if (!jump_dest) { + struct symbol *sym = find_symbol_by_offset(dest_sec, dest_off); + + /* + * This is a special case for zen_untrain_ret(). + * It jumps to __x86_return_thunk(), but objtool + * can't find the thunk's starting RET + * instruction, because the RET is also in the + * middle of another instruction. Objtool only + * knows about the outer instruction. + */ + if (sym && sym->return_thunk) { + add_return_call(file, insn, false); + continue; + } + WARN_FUNC("can't find jump dest instruction at %s+0x%lx", insn->sec, insn->offset, dest_sec->name, dest_off);

From: Alexandre Chartre alexandre.chartre@oracle.com

commit 7fbf47c7ce50b38a64576b150e7011ae73d54669 upstream.

Add the "retbleed=<value>" boot parameter to select a mitigation for RETBleed. Possible values are "off", "auto" and "unret" (JMP2RET mitigation). The default value is "auto".

Currently, "retbleed=auto" will select the unret mitigation on AMD and Hygon and no mitigation on Intel (JMP2RET is not effective on Intel).

[peterz: rebase; add hygon] [jpoimboe: cleanups]

Signed-off-by: Alexandre Chartre alexandre.chartre@oracle.com Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- Documentation/admin-guide/kernel-parameters.txt | 15 +++ arch/x86/Kconfig | 3 arch/x86/kernel/cpu/bugs.c | 108 +++++++++++++++++++++++- 3 files changed, 125 insertions(+), 1 deletion(-)

--- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -5124,6 +5124,21 @@

retain_initrd [RAM] Keep initrd memory after extraction

+ retbleed= [X86] Control mitigation of RETBleed (Arbitrary + Speculative Code Execution with Return Instructions) + vulnerability. + + off - unconditionally disable + auto - automatically select a migitation + unret - force enable untrained return thunks, + only effective on AMD Zen {1,2} + based systems. + + Selecting 'auto' will choose a mitigation method at run + time according to the CPU. + + Not specifying this option is equivalent to retbleed=auto. + rfkill.default_state= 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm, etc. communication is blocked by default. --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -469,6 +469,9 @@ config RETPOLINE config CC_HAS_SLS def_bool $(cc-option,-mharden-sls=all)

+config CC_HAS_RETURN_THUNK + def_bool $(cc-option,-mfunction-return=thunk-extern) + config SLS bool "Mitigate Straight-Line-Speculation" depends on CC_HAS_SLS && X86_64 --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -37,6 +37,7 @@ #include "cpu.h"

static void __init spectre_v1_select_mitigation(void); +static void __init retbleed_select_mitigation(void); static void __init spectre_v2_select_mitigation(void); static void __init ssb_select_mitigation(void); static void __init l1tf_select_mitigation(void); @@ -120,6 +121,12 @@ void __init check_bugs(void)

/* Select the proper CPU mitigations before patching alternatives: */ spectre_v1_select_mitigation(); + retbleed_select_mitigation(); + /* + * spectre_v2_select_mitigation() relies on the state set by + * retbleed_select_mitigation(); specifically the STIBP selection is + * forced for UNRET. + */ spectre_v2_select_mitigation(); ssb_select_mitigation(); l1tf_select_mitigation(); @@ -746,6 +753,100 @@ static int __init nospectre_v1_cmdline(c early_param("nospectre_v1", nospectre_v1_cmdline);

#undef pr_fmt +#define pr_fmt(fmt) "RETBleed: " fmt + +enum retbleed_mitigation { + RETBLEED_MITIGATION_NONE, + RETBLEED_MITIGATION_UNRET, +}; + +enum retbleed_mitigation_cmd { + RETBLEED_CMD_OFF, + RETBLEED_CMD_AUTO, + RETBLEED_CMD_UNRET, +}; + +const char * const retbleed_strings[] = { + [RETBLEED_MITIGATION_NONE] = "Vulnerable", + [RETBLEED_MITIGATION_UNRET] = "Mitigation: untrained return thunk", +}; + +static enum retbleed_mitigation retbleed_mitigation __ro_after_init = + RETBLEED_MITIGATION_NONE; +static enum retbleed_mitigation_cmd retbleed_cmd __ro_after_init = + RETBLEED_CMD_AUTO; + +static int __init retbleed_parse_cmdline(char *str) +{ + if (!str) + return -EINVAL; + + if (!strcmp(str, "off")) + retbleed_cmd = RETBLEED_CMD_OFF; + else if (!strcmp(str, "auto")) + retbleed_cmd = RETBLEED_CMD_AUTO; + else if (!strcmp(str, "unret")) + retbleed_cmd = RETBLEED_CMD_UNRET; + else + pr_err("Unknown retbleed option (%s). Defaulting to 'auto'\n", str); + + return 0; +} +early_param("retbleed", retbleed_parse_cmdline); + +#define RETBLEED_UNTRAIN_MSG "WARNING: BTB untrained return thunk mitigation is only effective on AMD/Hygon!\n" +#define RETBLEED_COMPILER_MSG "WARNING: kernel not compiled with RETPOLINE or -mfunction-return capable compiler!\n" + +static void __init retbleed_select_mitigation(void) +{ + if (!boot_cpu_has_bug(X86_BUG_RETBLEED) || cpu_mitigations_off()) + return; + + switch (retbleed_cmd) { + case RETBLEED_CMD_OFF: + return; + + case RETBLEED_CMD_UNRET: + retbleed_mitigation = RETBLEED_MITIGATION_UNRET; + break; + + case RETBLEED_CMD_AUTO: + default: + if (!boot_cpu_has_bug(X86_BUG_RETBLEED)) + break; + + if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD || + boot_cpu_data.x86_vendor == X86_VENDOR_HYGON) + retbleed_mitigation = RETBLEED_MITIGATION_UNRET; + break; + } + + switch (retbleed_mitigation) { + case RETBLEED_MITIGATION_UNRET: + + if (!IS_ENABLED(CONFIG_RETPOLINE) || + !IS_ENABLED(CONFIG_CC_HAS_RETURN_THUNK)) { + pr_err(RETBLEED_COMPILER_MSG); + retbleed_mitigation = RETBLEED_MITIGATION_NONE; + break; + } + + setup_force_cpu_cap(X86_FEATURE_RETHUNK); + setup_force_cpu_cap(X86_FEATURE_UNRET); + + if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD && + boot_cpu_data.x86_vendor != X86_VENDOR_HYGON) + pr_err(RETBLEED_UNTRAIN_MSG); + break; + + default: + break; + } + + pr_info("%s\n", retbleed_strings[retbleed_mitigation]); +} + +#undef pr_fmt #define pr_fmt(fmt) "Spectre V2 : " fmt

static enum spectre_v2_mitigation spectre_v2_enabled __ro_after_init = @@ -1989,7 +2090,12 @@ static ssize_t srbds_show_state(char *bu

static ssize_t retbleed_show_state(char *buf) { - return sprintf(buf, "Vulnerable\n"); + if (retbleed_mitigation == RETBLEED_MITIGATION_UNRET && + (boot_cpu_data.x86_vendor != X86_VENDOR_AMD && + boot_cpu_data.x86_vendor != X86_VENDOR_HYGON)) + return sprintf(buf, "Vulnerable: untrained return thunk on non-Zen uarch\n"); + + return sprintf(buf, "%s\n", retbleed_strings[retbleed_mitigation]); }

static ssize_t cpu_show_common(struct device *dev, struct device_attribute *attr,

From: Peter Zijlstra peterz@infradead.org

commit caa0ff24d5d0e02abce5e65c3d2b7f20a6617be5 upstream.

Due to TIF_SSBD and TIF_SPEC_IB the actual IA32_SPEC_CTRL value can differ from x86_spec_ctrl_base. As such, keep a per-CPU value reflecting the current task's MSR content.

[jpoimboe: rename]

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/include/asm/nospec-branch.h | 1 + arch/x86/kernel/cpu/bugs.c | 28 +++++++++++++++++++++++----- arch/x86/kernel/process.c | 2 +- 3 files changed, 25 insertions(+), 6 deletions(-)

--- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -253,6 +253,7 @@ static inline void indirect_branch_predi

/* The Intel SPEC CTRL MSR base value cache */ extern u64 x86_spec_ctrl_base; +extern void write_spec_ctrl_current(u64 val);

/* * With retpoline, we must use IBRS to restrict branch prediction --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -49,12 +49,30 @@ static void __init mmio_select_mitigatio static void __init srbds_select_mitigation(void); static void __init l1d_flush_select_mitigation(void);

-/* The base value of the SPEC_CTRL MSR that always has to be preserved. */ +/* The base value of the SPEC_CTRL MSR without task-specific bits set */ u64 x86_spec_ctrl_base; EXPORT_SYMBOL_GPL(x86_spec_ctrl_base); + +/* The current value of the SPEC_CTRL MSR with task-specific bits set */ +DEFINE_PER_CPU(u64, x86_spec_ctrl_current); +EXPORT_SYMBOL_GPL(x86_spec_ctrl_current); + static DEFINE_MUTEX(spec_ctrl_mutex);

/* + * Keep track of the SPEC_CTRL MSR value for the current task, which may differ + * from x86_spec_ctrl_base due to STIBP/SSB in __speculation_ctrl_update(). + */ +void write_spec_ctrl_current(u64 val) +{ + if (this_cpu_read(x86_spec_ctrl_current) == val) + return; + + this_cpu_write(x86_spec_ctrl_current, val); + wrmsrl(MSR_IA32_SPEC_CTRL, val); +} + +/* * The vendor and possibly platform specific bits which can be modified in * x86_spec_ctrl_base. */ @@ -1272,7 +1290,7 @@ static void __init spectre_v2_select_mit if (spectre_v2_in_eibrs_mode(mode)) { /* Force it so VMEXIT will restore correctly */ x86_spec_ctrl_base |= SPEC_CTRL_IBRS; - wrmsrl(MSR_IA32_SPEC_CTRL, x86_spec_ctrl_base); + write_spec_ctrl_current(x86_spec_ctrl_base); }

switch (mode) { @@ -1327,7 +1345,7 @@ static void __init spectre_v2_select_mit

static void update_stibp_msr(void * __unused) { - wrmsrl(MSR_IA32_SPEC_CTRL, x86_spec_ctrl_base); + write_spec_ctrl_current(x86_spec_ctrl_base); }

/* Update x86_spec_ctrl_base in case SMT state changed. */ @@ -1570,7 +1588,7 @@ static enum ssb_mitigation __init __ssb_ x86_amd_ssb_disable(); } else { x86_spec_ctrl_base |= SPEC_CTRL_SSBD; - wrmsrl(MSR_IA32_SPEC_CTRL, x86_spec_ctrl_base); + write_spec_ctrl_current(x86_spec_ctrl_base); } }

@@ -1821,7 +1839,7 @@ int arch_prctl_spec_ctrl_get(struct task void x86_spec_ctrl_setup_ap(void) { if (boot_cpu_has(X86_FEATURE_MSR_SPEC_CTRL)) - wrmsrl(MSR_IA32_SPEC_CTRL, x86_spec_ctrl_base); + write_spec_ctrl_current(x86_spec_ctrl_base);

if (ssb_mode == SPEC_STORE_BYPASS_DISABLE) x86_amd_ssb_disable(); --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -600,7 +600,7 @@ static __always_inline void __speculatio }

if (updmsr) - wrmsrl(MSR_IA32_SPEC_CTRL, msr); + write_spec_ctrl_current(msr); }

static unsigned long speculation_ctrl_update_tif(struct task_struct *tsk)

From: Peter Zijlstra peterz@infradead.org

commit c779bc1a9002fa474175b80e72b85c9bf628abb0 upstream.

When changing SPEC_CTRL for user control, the WRMSR can be delayed until return-to-user when KERNEL_IBRS has been enabled.

This avoids an MSR write during context switch.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/include/asm/nospec-branch.h | 2 +- arch/x86/kernel/cpu/bugs.c | 18 ++++++++++++------ arch/x86/kernel/process.c | 2 +- 3 files changed, 14 insertions(+), 8 deletions(-)

--- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -253,7 +253,7 @@ static inline void indirect_branch_predi

/* The Intel SPEC CTRL MSR base value cache */ extern u64 x86_spec_ctrl_base; -extern void write_spec_ctrl_current(u64 val); +extern void write_spec_ctrl_current(u64 val, bool force);

/* * With retpoline, we must use IBRS to restrict branch prediction --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -63,13 +63,19 @@ static DEFINE_MUTEX(spec_ctrl_mutex); * Keep track of the SPEC_CTRL MSR value for the current task, which may differ * from x86_spec_ctrl_base due to STIBP/SSB in __speculation_ctrl_update(). */ -void write_spec_ctrl_current(u64 val) +void write_spec_ctrl_current(u64 val, bool force) { if (this_cpu_read(x86_spec_ctrl_current) == val) return;

this_cpu_write(x86_spec_ctrl_current, val); - wrmsrl(MSR_IA32_SPEC_CTRL, val); + + /* + * When KERNEL_IBRS this MSR is written on return-to-user, unless + * forced the update can be delayed until that time. + */ + if (force || !cpu_feature_enabled(X86_FEATURE_KERNEL_IBRS)) + wrmsrl(MSR_IA32_SPEC_CTRL, val); }

/* @@ -1290,7 +1296,7 @@ static void __init spectre_v2_select_mit if (spectre_v2_in_eibrs_mode(mode)) { /* Force it so VMEXIT will restore correctly */ x86_spec_ctrl_base |= SPEC_CTRL_IBRS; - write_spec_ctrl_current(x86_spec_ctrl_base); + write_spec_ctrl_current(x86_spec_ctrl_base, true); }

switch (mode) { @@ -1345,7 +1351,7 @@ static void __init spectre_v2_select_mit

static void update_stibp_msr(void * __unused) { - write_spec_ctrl_current(x86_spec_ctrl_base); + write_spec_ctrl_current(x86_spec_ctrl_base, true); }

/* Update x86_spec_ctrl_base in case SMT state changed. */ @@ -1588,7 +1594,7 @@ static enum ssb_mitigation __init __ssb_ x86_amd_ssb_disable(); } else { x86_spec_ctrl_base |= SPEC_CTRL_SSBD; - write_spec_ctrl_current(x86_spec_ctrl_base); + write_spec_ctrl_current(x86_spec_ctrl_base, true); } }

@@ -1839,7 +1845,7 @@ int arch_prctl_spec_ctrl_get(struct task void x86_spec_ctrl_setup_ap(void) { if (boot_cpu_has(X86_FEATURE_MSR_SPEC_CTRL)) - write_spec_ctrl_current(x86_spec_ctrl_base); + write_spec_ctrl_current(x86_spec_ctrl_base, true);

if (ssb_mode == SPEC_STORE_BYPASS_DISABLE) x86_amd_ssb_disable(); --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -600,7 +600,7 @@ static __always_inline void __speculatio }

if (updmsr) - write_spec_ctrl_current(msr); + write_spec_ctrl_current(msr, false); }

static unsigned long speculation_ctrl_update_tif(struct task_struct *tsk)

From: Peter Zijlstra peterz@infradead.org

commit 166115c08a9b0b846b783088808a27d739be6e8d upstream.

retbleed will depend on spectre_v2, while spectre_v2_user depends on retbleed. Break this cycle.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kernel/cpu/bugs.c | 25 +++++++++++++++++-------- 1 file changed, 17 insertions(+), 8 deletions(-)

--- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -37,8 +37,9 @@ #include "cpu.h"

static void __init spectre_v1_select_mitigation(void); -static void __init retbleed_select_mitigation(void); static void __init spectre_v2_select_mitigation(void); +static void __init retbleed_select_mitigation(void); +static void __init spectre_v2_user_select_mitigation(void); static void __init ssb_select_mitigation(void); static void __init l1tf_select_mitigation(void); static void __init mds_select_mitigation(void); @@ -145,13 +146,19 @@ void __init check_bugs(void)

/* Select the proper CPU mitigations before patching alternatives: */ spectre_v1_select_mitigation(); + spectre_v2_select_mitigation(); + /* + * retbleed_select_mitigation() relies on the state set by + * spectre_v2_select_mitigation(); specifically it wants to know about + * spectre_v2=ibrs. + */ retbleed_select_mitigation(); /* - * spectre_v2_select_mitigation() relies on the state set by + * spectre_v2_user_select_mitigation() relies on the state set by * retbleed_select_mitigation(); specifically the STIBP selection is * forced for UNRET. */ - spectre_v2_select_mitigation(); + spectre_v2_user_select_mitigation(); ssb_select_mitigation(); l1tf_select_mitigation(); md_clear_select_mitigation(); @@ -1006,13 +1013,15 @@ static void __init spec_v2_user_print_co pr_info("spectre_v2_user=%s forced on command line.\n", reason); }

+static __ro_after_init enum spectre_v2_mitigation_cmd spectre_v2_cmd; + static enum spectre_v2_user_cmd __init -spectre_v2_parse_user_cmdline(enum spectre_v2_mitigation_cmd v2_cmd) +spectre_v2_parse_user_cmdline(void) { char arg[20]; int ret, i;

- switch (v2_cmd) { + switch (spectre_v2_cmd) { case SPECTRE_V2_CMD_NONE: return SPECTRE_V2_USER_CMD_NONE; case SPECTRE_V2_CMD_FORCE: @@ -1047,7 +1056,7 @@ static inline bool spectre_v2_in_ibrs_mo }

static void __init -spectre_v2_user_select_mitigation(enum spectre_v2_mitigation_cmd v2_cmd) +spectre_v2_user_select_mitigation(void) { enum spectre_v2_user_mitigation mode = SPECTRE_V2_USER_NONE; bool smt_possible = IS_ENABLED(CONFIG_SMP); @@ -1060,7 +1069,7 @@ spectre_v2_user_select_mitigation(enum s cpu_smt_control == CPU_SMT_NOT_SUPPORTED) smt_possible = false;

- cmd = spectre_v2_parse_user_cmdline(v2_cmd); + cmd = spectre_v2_parse_user_cmdline(); switch (cmd) { case SPECTRE_V2_USER_CMD_NONE: goto set_mode; @@ -1384,7 +1393,7 @@ static void __init spectre_v2_select_mit }

/* Set up IBPB and STIBP depending on the general spectre V2 command */ - spectre_v2_user_select_mitigation(cmd); + spectre_v2_cmd = cmd; }

static void update_stibp_msr(void * __unused)

From: Peter Zijlstra peterz@infradead.org

commit bf5835bcdb9635c97f85120dba9bfa21e111130f upstream.

Having IBRS enabled while the SMT sibling is idle unnecessarily slows down the running sibling. OTOH, disabling IBRS around idle takes two MSR writes, which will increase the idle latency.

Therefore, only disable IBRS around deeper idle states. Shallow idle states are bounded by the tick in duration, since NOHZ is not allowed for them by virtue of their short target residency.

Only do this for mwait-driven idle, since that keeps interrupts disabled across idle, which makes disabling IBRS vs IRQ-entry a non-issue.

Note: C6 is a random threshold, most importantly C1 probably shouldn't disable IBRS, benchmarking needed.

Suggested-by: Tim Chen tim.c.chen@linux.intel.com Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/include/asm/nospec-branch.h | 1 arch/x86/kernel/cpu/bugs.c | 6 ++++ drivers/idle/intel_idle.c | 44 ++++++++++++++++++++++++++++++----- 3 files changed, 45 insertions(+), 6 deletions(-)

--- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -255,6 +255,7 @@ static inline void indirect_branch_predi /* The Intel SPEC CTRL MSR base value cache */ extern u64 x86_spec_ctrl_base; extern void write_spec_ctrl_current(u64 val, bool force); +extern u64 spec_ctrl_current(void);

/* * With retpoline, we must use IBRS to restrict branch prediction --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -79,6 +79,12 @@ void write_spec_ctrl_current(u64 val, bo wrmsrl(MSR_IA32_SPEC_CTRL, val); }

+u64 spec_ctrl_current(void) +{ + return this_cpu_read(x86_spec_ctrl_current); +} +EXPORT_SYMBOL_GPL(spec_ctrl_current); + /* * The vendor and possibly platform specific bits which can be modified in * x86_spec_ctrl_base. --- a/drivers/idle/intel_idle.c +++ b/drivers/idle/intel_idle.c @@ -47,11 +47,13 @@ #include <linux/tick.h> #include <trace/events/power.h> #include <linux/sched.h> +#include <linux/sched/smt.h> #include <linux/notifier.h> #include <linux/cpu.h> #include <linux/moduleparam.h> #include <asm/cpu_device_id.h> #include <asm/intel-family.h> +#include <asm/nospec-branch.h> #include <asm/mwait.h> #include <asm/msr.h>

@@ -106,6 +108,12 @@ static unsigned int mwait_substates __in #define CPUIDLE_FLAG_ALWAYS_ENABLE BIT(15)

/* + * Disable IBRS across idle (when KERNEL_IBRS), is exclusive vs IRQ_ENABLE + * above. + */ +#define CPUIDLE_FLAG_IBRS BIT(16) + +/* * MWAIT takes an 8-bit "hint" in EAX "suggesting" * the C-state (top nibble) and sub-state (bottom nibble) * 0x00 means "MWAIT(C1)", 0x10 means "MWAIT(C2)" etc. @@ -159,6 +167,24 @@ static __cpuidle int intel_idle_irq(stru return ret; }

+static __cpuidle int intel_idle_ibrs(struct cpuidle_device *dev, + struct cpuidle_driver *drv, int index) +{ + bool smt_active = sched_smt_active(); + u64 spec_ctrl = spec_ctrl_current(); + int ret; + + if (smt_active) + wrmsrl(MSR_IA32_SPEC_CTRL, 0); + + ret = __intel_idle(dev, drv, index); + + if (smt_active) + wrmsrl(MSR_IA32_SPEC_CTRL, spec_ctrl); + + return ret; +} + /** * intel_idle_s2idle - Ask the processor to enter the given idle state. * @dev: cpuidle device of the target CPU. @@ -680,7 +706,7 @@ static struct cpuidle_state skl_cstates[ { .name = "C6", .desc = "MWAIT 0x20", - .flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED, + .flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED | CPUIDLE_FLAG_IBRS, .exit_latency = 85, .target_residency = 200, .enter = &intel_idle, @@ -688,7 +714,7 @@ static struct cpuidle_state skl_cstates[ { .name = "C7s", .desc = "MWAIT 0x33", - .flags = MWAIT2flg(0x33) | CPUIDLE_FLAG_TLB_FLUSHED, + .flags = MWAIT2flg(0x33) | CPUIDLE_FLAG_TLB_FLUSHED | CPUIDLE_FLAG_IBRS, .exit_latency = 124, .target_residency = 800, .enter = &intel_idle, @@ -696,7 +722,7 @@ static struct cpuidle_state skl_cstates[ { .name = "C8", .desc = "MWAIT 0x40", - .flags = MWAIT2flg(0x40) | CPUIDLE_FLAG_TLB_FLUSHED, + .flags = MWAIT2flg(0x40) | CPUIDLE_FLAG_TLB_FLUSHED | CPUIDLE_FLAG_IBRS, .exit_latency = 200, .target_residency = 800, .enter = &intel_idle, @@ -704,7 +730,7 @@ static struct cpuidle_state skl_cstates[ { .name = "C9", .desc = "MWAIT 0x50", - .flags = MWAIT2flg(0x50) | CPUIDLE_FLAG_TLB_FLUSHED, + .flags = MWAIT2flg(0x50) | CPUIDLE_FLAG_TLB_FLUSHED | CPUIDLE_FLAG_IBRS, .exit_latency = 480, .target_residency = 5000, .enter = &intel_idle, @@ -712,7 +738,7 @@ static struct cpuidle_state skl_cstates[ { .name = "C10", .desc = "MWAIT 0x60", - .flags = MWAIT2flg(0x60) | CPUIDLE_FLAG_TLB_FLUSHED, + .flags = MWAIT2flg(0x60) | CPUIDLE_FLAG_TLB_FLUSHED | CPUIDLE_FLAG_IBRS, .exit_latency = 890, .target_residency = 5000, .enter = &intel_idle, @@ -741,7 +767,7 @@ static struct cpuidle_state skx_cstates[ { .name = "C6", .desc = "MWAIT 0x20", - .flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED, + .flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED | CPUIDLE_FLAG_IBRS, .exit_latency = 133, .target_residency = 600, .enter = &intel_idle, @@ -1686,6 +1712,12 @@ static void __init intel_idle_init_cstat if (cpuidle_state_table[cstate].flags & CPUIDLE_FLAG_IRQ_ENABLE) drv->states[drv->state_count].enter = intel_idle_irq;

+ if (cpu_feature_enabled(X86_FEATURE_KERNEL_IBRS) && + cpuidle_state_table[cstate].flags & CPUIDLE_FLAG_IBRS) { + WARN_ON_ONCE(cpuidle_state_table[cstate].flags & CPUIDLE_FLAG_IRQ_ENABLE); + drv->states[drv->state_count].enter = intel_idle_ibrs; + } + if ((disabled_states_mask & BIT(drv->state_count)) || ((icpu->use_acpi || force_use_acpi) && intel_idle_off_by_default(mwait_hint) &&

From: Peter Zijlstra peterz@infradead.org

commit b75b7f8ef1148be1b9321ffc2f6c19238904b438 upstream.

Native SYS{CALL,ENTER} entry points are called entry_SYS{CALL,ENTER}_{64,compat}, make sure the Xen versions are named consistently.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/xen/setup.c | 6 +++--- arch/x86/xen/xen-asm.S | 20 ++++++++++---------- arch/x86/xen/xen-ops.h | 6 +++--- 3 files changed, 16 insertions(+), 16 deletions(-)

--- a/arch/x86/xen/setup.c +++ b/arch/x86/xen/setup.c @@ -918,7 +918,7 @@ void xen_enable_sysenter(void) if (!boot_cpu_has(sysenter_feature)) return;

- ret = register_callback(CALLBACKTYPE_sysenter, xen_sysenter_target); + ret = register_callback(CALLBACKTYPE_sysenter, xen_entry_SYSENTER_compat); if(ret != 0) setup_clear_cpu_cap(sysenter_feature); } @@ -927,7 +927,7 @@ void xen_enable_syscall(void) { int ret;

- ret = register_callback(CALLBACKTYPE_syscall, xen_syscall_target); + ret = register_callback(CALLBACKTYPE_syscall, xen_entry_SYSCALL_64); if (ret != 0) { printk(KERN_ERR "Failed to set syscall callback: %d\n", ret); /* Pretty fatal; 64-bit userspace has no other @@ -936,7 +936,7 @@ void xen_enable_syscall(void)

if (boot_cpu_has(X86_FEATURE_SYSCALL32)) { ret = register_callback(CALLBACKTYPE_syscall32, - xen_syscall32_target); + xen_entry_SYSCALL_compat); if (ret != 0) setup_clear_cpu_cap(X86_FEATURE_SYSCALL32); } --- a/arch/x86/xen/xen-asm.S +++ b/arch/x86/xen/xen-asm.S @@ -234,7 +234,7 @@ SYM_CODE_END(xenpv_restore_regs_and_retu */

/* Normal 64-bit system call target */ -SYM_CODE_START(xen_syscall_target) +SYM_CODE_START(xen_entry_SYSCALL_64) UNWIND_HINT_EMPTY ENDBR popq %rcx @@ -249,12 +249,12 @@ SYM_CODE_START(xen_syscall_target) movq $__USER_CS, 1*8(%rsp)

jmp entry_SYSCALL_64_after_hwframe -SYM_CODE_END(xen_syscall_target) +SYM_CODE_END(xen_entry_SYSCALL_64)

#ifdef CONFIG_IA32_EMULATION

/* 32-bit compat syscall target */ -SYM_CODE_START(xen_syscall32_target) +SYM_CODE_START(xen_entry_SYSCALL_compat) UNWIND_HINT_EMPTY ENDBR popq %rcx @@ -269,10 +269,10 @@ SYM_CODE_START(xen_syscall32_target) movq $__USER32_CS, 1*8(%rsp)

jmp entry_SYSCALL_compat_after_hwframe -SYM_CODE_END(xen_syscall32_target) +SYM_CODE_END(xen_entry_SYSCALL_compat)

/* 32-bit compat sysenter target */ -SYM_CODE_START(xen_sysenter_target) +SYM_CODE_START(xen_entry_SYSENTER_compat) UNWIND_HINT_EMPTY ENDBR /* @@ -291,19 +291,19 @@ SYM_CODE_START(xen_sysenter_target) movq $__USER32_CS, 1*8(%rsp)

jmp entry_SYSENTER_compat_after_hwframe -SYM_CODE_END(xen_sysenter_target) +SYM_CODE_END(xen_entry_SYSENTER_compat)

#else /* !CONFIG_IA32_EMULATION */

-SYM_CODE_START(xen_syscall32_target) -SYM_CODE_START(xen_sysenter_target) +SYM_CODE_START(xen_entry_SYSCALL_compat) +SYM_CODE_START(xen_entry_SYSENTER_compat) UNWIND_HINT_EMPTY ENDBR lea 16(%rsp), %rsp /* strip %rcx, %r11 */ mov $-ENOSYS, %rax pushq $0 jmp hypercall_iret -SYM_CODE_END(xen_sysenter_target) -SYM_CODE_END(xen_syscall32_target) +SYM_CODE_END(xen_entry_SYSENTER_compat) +SYM_CODE_END(xen_entry_SYSCALL_compat)

#endif /* CONFIG_IA32_EMULATION */ --- a/arch/x86/xen/xen-ops.h +++ b/arch/x86/xen/xen-ops.h @@ -10,10 +10,10 @@ /* These are code, but not functions. Defined in entry.S */ extern const char xen_failsafe_callback[];

-void xen_sysenter_target(void); +void xen_entry_SYSENTER_compat(void); #ifdef CONFIG_X86_64 -void xen_syscall_target(void); -void xen_syscall32_target(void); +void xen_entry_SYSCALL_64(void); +void xen_entry_SYSCALL_compat(void); #endif

extern void *xen_initial_gdt;

From: Peter Zijlstra peterz@infradead.org

commit 3ebc170068885b6fc7bedda6c667bb2c4d533159 upstream.

jmp2ret mitigates the easy-to-attack case at relatively low overhead. It mitigates the long speculation windows after a mispredicted RET, but it does not mitigate the short speculation window from arbitrary instruction boundaries.

On Zen2, there is a chicken bit which needs setting, which mitigates "arbitrary instruction boundaries" down to just "basic block boundaries".

But there is no fix for the short speculation window on basic block boundaries, other than to flush the entire BTB to evict all attacker predictions.

On the spectrum of "fast & blurry" -> "safe", there is (on top of STIBP or no-SMT):

1) Nothing System wide open 2) jmp2ret May stop a script kiddy 3) jmp2ret+chickenbit Raises the bar rather further 4) IBPB Only thing which can count as "safe".

Tentative numbers put IBPB-on-entry at a 2.5x hit on Zen2, and a 10x hit on Zen1 according to lmbench.

[ bp: Fixup feature bit comments, document option, 32-bit build fix. ]

Suggested-by: Andrew Cooper Andrew.Cooper3@citrix.com Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- Documentation/admin-guide/kernel-parameters.txt | 3 + arch/x86/entry/Makefile | 2 - arch/x86/entry/entry.S | 22 ++++++++++++ arch/x86/include/asm/cpufeatures.h | 2 - arch/x86/include/asm/nospec-branch.h | 8 +++- arch/x86/kernel/cpu/bugs.c | 43 ++++++++++++++++++------ 6 files changed, 67 insertions(+), 13 deletions(-) create mode 100644 arch/x86/entry/entry.S

--- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -5134,6 +5134,9 @@ disabling SMT if necessary for the full mitigation (only on Zen1 and older without STIBP). + ibpb - mitigate short speculation windows on + basic block boundaries too. Safe, highest + perf impact. unret - force enable untrained return thunks, only effective on AMD f15h-f17h based systems. --- a/arch/x86/entry/Makefile +++ b/arch/x86/entry/Makefile @@ -11,7 +11,7 @@ CFLAGS_REMOVE_common.o = $(CC_FLAGS_FTR

CFLAGS_common.o += -fno-stack-protector

-obj-y := entry_$(BITS).o thunk_$(BITS).o syscall_$(BITS).o +obj-y := entry.o entry_$(BITS).o thunk_$(BITS).o syscall_$(BITS).o obj-y += common.o

obj-y += vdso/ --- /dev/null +++ b/arch/x86/entry/entry.S @@ -0,0 +1,22 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * Common place for both 32- and 64-bit entry routines. + */ + +#include <linux/linkage.h> +#include <asm/export.h> +#include <asm/msr-index.h> + +.pushsection .noinstr.text, "ax" + +SYM_FUNC_START(entry_ibpb) + movl $MSR_IA32_PRED_CMD, %ecx + movl $PRED_CMD_IBPB, %eax + xorl %edx, %edx + wrmsr + RET +SYM_FUNC_END(entry_ibpb) +/* For KVM */ +EXPORT_SYMBOL_GPL(entry_ibpb); + +.popsection --- a/arch/x86/include/asm/cpufeatures.h +++ b/arch/x86/include/asm/cpufeatures.h @@ -295,7 +295,7 @@ #define X86_FEATURE_PER_THREAD_MBA (11*32+ 7) /* "" Per-thread Memory Bandwidth Allocation */ #define X86_FEATURE_SGX1 (11*32+ 8) /* "" Basic SGX */ #define X86_FEATURE_SGX2 (11*32+ 9) /* "" SGX Enclave Dynamic Memory Management (EDMM) */ -/* FREE! (11*32+10) */ +#define X86_FEATURE_ENTRY_IBPB (11*32+10) /* "" Issue an IBPB on kernel entry */ /* FREE! (11*32+11) */ #define X86_FEATURE_RETPOLINE (11*32+12) /* "" Generic Retpoline mitigation for Spectre variant 2 */ #define X86_FEATURE_RETPOLINE_LFENCE (11*32+13) /* "" Use LFENCE for Spectre variant 2 */ --- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -123,14 +123,17 @@ * return thunk isn't mapped into the userspace tables (then again, AMD * typically has NO_MELTDOWN). * - * Doesn't clobber any registers but does require a stable stack. + * While zen_untrain_ret() doesn't clobber anything but requires stack, + * entry_ibpb() will clobber AX, CX, DX. * * As such, this must be placed after every *SWITCH_TO_KERNEL_CR3 at a point * where we have a stack but before any RET instruction. */ .macro UNTRAIN_RET #ifdef CONFIG_RETPOLINE - ALTERNATIVE "", "call zen_untrain_ret", X86_FEATURE_UNRET + ALTERNATIVE_2 "", \ + "call zen_untrain_ret", X86_FEATURE_UNRET, \ + "call entry_ibpb", X86_FEATURE_ENTRY_IBPB #endif .endm

@@ -147,6 +150,7 @@ extern retpoline_thunk_t __x86_indirect_

extern void __x86_return_thunk(void); extern void zen_untrain_ret(void); +extern void entry_ibpb(void);

#ifdef CONFIG_RETPOLINE

--- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -798,6 +798,7 @@ static enum spectre_v2_mitigation spectr enum retbleed_mitigation { RETBLEED_MITIGATION_NONE, RETBLEED_MITIGATION_UNRET, + RETBLEED_MITIGATION_IBPB, RETBLEED_MITIGATION_IBRS, RETBLEED_MITIGATION_EIBRS, }; @@ -806,11 +807,13 @@ enum retbleed_mitigation_cmd { RETBLEED_CMD_OFF, RETBLEED_CMD_AUTO, RETBLEED_CMD_UNRET, + RETBLEED_CMD_IBPB, };

const char * const retbleed_strings[] = { [RETBLEED_MITIGATION_NONE] = "Vulnerable", [RETBLEED_MITIGATION_UNRET] = "Mitigation: untrained return thunk", + [RETBLEED_MITIGATION_IBPB] = "Mitigation: IBPB", [RETBLEED_MITIGATION_IBRS] = "Mitigation: IBRS", [RETBLEED_MITIGATION_EIBRS] = "Mitigation: Enhanced IBRS", }; @@ -840,6 +843,8 @@ static int __init retbleed_parse_cmdline retbleed_cmd = RETBLEED_CMD_AUTO; } else if (!strcmp(str, "unret")) { retbleed_cmd = RETBLEED_CMD_UNRET; + } else if (!strcmp(str, "ibpb")) { + retbleed_cmd = RETBLEED_CMD_IBPB; } else if (!strcmp(str, "nosmt")) { retbleed_nosmt = true; } else { @@ -854,11 +859,13 @@ static int __init retbleed_parse_cmdline early_param("retbleed", retbleed_parse_cmdline);

#define RETBLEED_UNTRAIN_MSG "WARNING: BTB untrained return thunk mitigation is only effective on AMD/Hygon!\n" -#define RETBLEED_COMPILER_MSG "WARNING: kernel not compiled with RETPOLINE or -mfunction-return capable compiler!\n" +#define RETBLEED_COMPILER_MSG "WARNING: kernel not compiled with RETPOLINE or -mfunction-return capable compiler; falling back to IBPB!\n" #define RETBLEED_INTEL_MSG "WARNING: Spectre v2 mitigation leaves CPU vulnerable to RETBleed attacks, data leaks possible!\n"

static void __init retbleed_select_mitigation(void) { + bool mitigate_smt = false; + if (!boot_cpu_has_bug(X86_BUG_RETBLEED) || cpu_mitigations_off()) return;

@@ -870,11 +877,21 @@ static void __init retbleed_select_mitig retbleed_mitigation = RETBLEED_MITIGATION_UNRET; break;

+ case RETBLEED_CMD_IBPB: + retbleed_mitigation = RETBLEED_MITIGATION_IBPB; + break; + case RETBLEED_CMD_AUTO: default: if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD || - boot_cpu_data.x86_vendor == X86_VENDOR_HYGON) - retbleed_mitigation = RETBLEED_MITIGATION_UNRET; + boot_cpu_data.x86_vendor == X86_VENDOR_HYGON) { + + if (IS_ENABLED(CONFIG_RETPOLINE) && + IS_ENABLED(CONFIG_CC_HAS_RETURN_THUNK)) + retbleed_mitigation = RETBLEED_MITIGATION_UNRET; + else + retbleed_mitigation = RETBLEED_MITIGATION_IBPB; + }

/* * The Intel mitigation (IBRS) was already selected in @@ -890,26 +907,34 @@ static void __init retbleed_select_mitig if (!IS_ENABLED(CONFIG_RETPOLINE) || !IS_ENABLED(CONFIG_CC_HAS_RETURN_THUNK)) { pr_err(RETBLEED_COMPILER_MSG); - retbleed_mitigation = RETBLEED_MITIGATION_NONE; - break; + retbleed_mitigation = RETBLEED_MITIGATION_IBPB; + goto retbleed_force_ibpb; }

setup_force_cpu_cap(X86_FEATURE_RETHUNK); setup_force_cpu_cap(X86_FEATURE_UNRET);

- if (!boot_cpu_has(X86_FEATURE_STIBP) && - (retbleed_nosmt || cpu_mitigations_auto_nosmt())) - cpu_smt_disable(false); - if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD && boot_cpu_data.x86_vendor != X86_VENDOR_HYGON) pr_err(RETBLEED_UNTRAIN_MSG); + + mitigate_smt = true; + break; + + case RETBLEED_MITIGATION_IBPB: +retbleed_force_ibpb: + setup_force_cpu_cap(X86_FEATURE_ENTRY_IBPB); + mitigate_smt = true; break;

default: break; }

+ if (mitigate_smt && !boot_cpu_has(X86_FEATURE_STIBP) && + (retbleed_nosmt || cpu_mitigations_auto_nosmt())) + cpu_smt_disable(false); + /* * Let IBRS trump all on Intel without affecting the effects of the * retbleed= cmdline option.

From: Thadeu Lima de Souza Cascardo cascardo@canonical.com

commit a09a6e2399ba0595c3042b3164f3ca68a3cff33e upstream.

Since entry asm is tricky, add a validation pass that ensures the retbleed mitigation has been done before the first actual RET instruction.

Entry points are those that either have UNWIND_HINT_ENTRY, which acts as UNWIND_HINT_EMPTY but marks the instruction as an entry point, or those that have UWIND_HINT_IRET_REGS at +0.

This is basically a variant of validate_branch() that is intra-function and it will simply follow all branches from marked entry points and ensures that all paths lead to ANNOTATE_UNRET_END.

If a path hits RET or an indirection the path is a fail and will be reported.

There are 3 ANNOTATE_UNRET_END instances:

- UNTRAIN_RET itself - exception from-kernel; this path doesn't need UNTRAIN_RET - all early exceptions; these also don't need UNTRAIN_RET

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de [cascardo: tools/objtool/builtin-check.c no link option validation] [cascardo: tools/objtool/check.c opts.ibt is ibt] [cascardo: tools/objtool/include/objtool/builtin.h leave unret option as bool, no struct opts] [cascardo: objtool is still called from scripts/link-vmlinux.sh] Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/entry/entry_64.S | 3 arch/x86/entry/entry_64_compat.S | 6 - arch/x86/include/asm/nospec-branch.h | 12 ++ arch/x86/include/asm/unwind_hints.h | 4 arch/x86/kernel/head_64.S | 5 arch/x86/xen/xen-asm.S | 10 - include/linux/objtool.h | 3 scripts/link-vmlinux.sh | 3 tools/include/linux/objtool.h | 3 tools/objtool/builtin-check.c | 3 tools/objtool/check.c | 177 ++++++++++++++++++++++++++++++-- tools/objtool/include/objtool/builtin.h | 2 tools/objtool/include/objtool/check.h | 11 + 13 files changed, 220 insertions(+), 22 deletions(-)

--- a/arch/x86/entry/entry_64.S +++ b/arch/x86/entry/entry_64.S @@ -85,7 +85,7 @@ */

SYM_CODE_START(entry_SYSCALL_64) - UNWIND_HINT_EMPTY + UNWIND_HINT_ENTRY ENDBR

swapgs @@ -1088,6 +1088,7 @@ SYM_CODE_START_LOCAL(error_entry) .Lerror_entry_done_lfence: FENCE_SWAPGS_KERNEL_ENTRY leaq 8(%rsp), %rax /* return pt_regs pointer */ + ANNOTATE_UNRET_END RET

.Lbstep_iret: --- a/arch/x86/entry/entry_64_compat.S +++ b/arch/x86/entry/entry_64_compat.S @@ -49,7 +49,7 @@ * 0(%ebp) arg6 */ SYM_CODE_START(entry_SYSENTER_compat) - UNWIND_HINT_EMPTY + UNWIND_HINT_ENTRY ENDBR /* Interrupts are off on entry. */ SWAPGS @@ -204,7 +204,7 @@ SYM_CODE_END(entry_SYSENTER_compat) * 0(%esp) arg6 */ SYM_CODE_START(entry_SYSCALL_compat) - UNWIND_HINT_EMPTY + UNWIND_HINT_ENTRY ENDBR /* Interrupts are off on entry. */ swapgs @@ -353,7 +353,7 @@ SYM_CODE_END(entry_SYSCALL_compat) * ebp arg6 */ SYM_CODE_START(entry_INT80_compat) - UNWIND_HINT_EMPTY + UNWIND_HINT_ENTRY ENDBR /* * Interrupts are off on entry. --- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -82,6 +82,17 @@ #define ANNOTATE_UNRET_SAFE ANNOTATE_RETPOLINE_SAFE

/* + * Abuse ANNOTATE_RETPOLINE_SAFE on a NOP to indicate UNRET_END, should + * eventually turn into it's own annotation. + */ +.macro ANNOTATE_UNRET_END +#ifdef CONFIG_DEBUG_ENTRY + ANNOTATE_RETPOLINE_SAFE + nop +#endif +.endm + +/* * JMP_NOSPEC and CALL_NOSPEC macros can be used instead of a simple * indirect jmp/call which may be susceptible to the Spectre variant 2 * attack. @@ -131,6 +142,7 @@ */ .macro UNTRAIN_RET #ifdef CONFIG_RETPOLINE + ANNOTATE_UNRET_END ALTERNATIVE_2 "", \ "call zen_untrain_ret", X86_FEATURE_UNRET, \ "call entry_ibpb", X86_FEATURE_ENTRY_IBPB --- a/arch/x86/include/asm/unwind_hints.h +++ b/arch/x86/include/asm/unwind_hints.h @@ -11,6 +11,10 @@ UNWIND_HINT sp_reg=ORC_REG_UNDEFINED type=UNWIND_HINT_TYPE_CALL end=1 .endm

+.macro UNWIND_HINT_ENTRY + UNWIND_HINT sp_reg=ORC_REG_UNDEFINED type=UNWIND_HINT_TYPE_ENTRY end=1 +.endm + .macro UNWIND_HINT_REGS base=%rsp offset=0 indirect=0 extra=1 partial=0 .if \base == %rsp .if \indirect --- a/arch/x86/kernel/head_64.S +++ b/arch/x86/kernel/head_64.S @@ -334,6 +334,8 @@ SYM_CODE_START_NOALIGN(vc_boot_ghcb) UNWIND_HINT_IRET_REGS offset=8 ENDBR

+ ANNOTATE_UNRET_END + /* Build pt_regs */ PUSH_AND_CLEAR_REGS

@@ -393,6 +395,7 @@ SYM_CODE_END(early_idt_handler_array)

SYM_CODE_START_LOCAL(early_idt_handler_common) UNWIND_HINT_IRET_REGS offset=16 + ANNOTATE_UNRET_END /* * The stack is the hardware frame, an error code or zero, and the * vector number. @@ -442,6 +445,8 @@ SYM_CODE_START_NOALIGN(vc_no_ghcb) UNWIND_HINT_IRET_REGS offset=8 ENDBR

+ ANNOTATE_UNRET_END + /* Build pt_regs */ PUSH_AND_CLEAR_REGS

--- a/arch/x86/xen/xen-asm.S +++ b/arch/x86/xen/xen-asm.S @@ -121,7 +121,7 @@ SYM_FUNC_END(xen_read_cr2_direct);

.macro xen_pv_trap name SYM_CODE_START(xen_\name) - UNWIND_HINT_EMPTY + UNWIND_HINT_ENTRY ENDBR pop %rcx pop %r11 @@ -235,7 +235,7 @@ SYM_CODE_END(xenpv_restore_regs_and_retu

/* Normal 64-bit system call target */ SYM_CODE_START(xen_entry_SYSCALL_64) - UNWIND_HINT_EMPTY + UNWIND_HINT_ENTRY ENDBR popq %rcx popq %r11 @@ -255,7 +255,7 @@ SYM_CODE_END(xen_entry_SYSCALL_64)

/* 32-bit compat syscall target */ SYM_CODE_START(xen_entry_SYSCALL_compat) - UNWIND_HINT_EMPTY + UNWIND_HINT_ENTRY ENDBR popq %rcx popq %r11 @@ -273,7 +273,7 @@ SYM_CODE_END(xen_entry_SYSCALL_compat)

/* 32-bit compat sysenter target */ SYM_CODE_START(xen_entry_SYSENTER_compat) - UNWIND_HINT_EMPTY + UNWIND_HINT_ENTRY ENDBR /* * NB: Xen is polite and clears TF from EFLAGS for us. This means @@ -297,7 +297,7 @@ SYM_CODE_END(xen_entry_SYSENTER_compat)

SYM_CODE_START(xen_entry_SYSCALL_compat) SYM_CODE_START(xen_entry_SYSENTER_compat) - UNWIND_HINT_EMPTY + UNWIND_HINT_ENTRY ENDBR lea 16(%rsp), %rsp /* strip %rcx, %r11 */ mov $-ENOSYS, %rax --- a/include/linux/objtool.h +++ b/include/linux/objtool.h @@ -32,11 +32,14 @@ struct unwind_hint { * * UNWIND_HINT_FUNC: Generate the unwind metadata of a callable function. * Useful for code which doesn't have an ELF function annotation. + * + * UNWIND_HINT_ENTRY: machine entry without stack, SYSCALL/SYSENTER etc. */ #define UNWIND_HINT_TYPE_CALL 0 #define UNWIND_HINT_TYPE_REGS 1 #define UNWIND_HINT_TYPE_REGS_PARTIAL 2 #define UNWIND_HINT_TYPE_FUNC 3 +#define UNWIND_HINT_TYPE_ENTRY 4

#ifdef CONFIG_STACK_VALIDATION

--- a/scripts/link-vmlinux.sh +++ b/scripts/link-vmlinux.sh @@ -130,6 +130,9 @@ objtool_link()

if is_enabled CONFIG_VMLINUX_VALIDATION; then objtoolopt="${objtoolopt} --noinstr" + if is_enabled CONFIG_RETPOLINE; then + objtoolopt="${objtoolopt} --unret" + fi fi

if [ -n "${objtoolopt}" ]; then --- a/tools/include/linux/objtool.h +++ b/tools/include/linux/objtool.h @@ -32,11 +32,14 @@ struct unwind_hint { * * UNWIND_HINT_FUNC: Generate the unwind metadata of a callable function. * Useful for code which doesn't have an ELF function annotation. + * + * UNWIND_HINT_ENTRY: machine entry without stack, SYSCALL/SYSENTER etc. */ #define UNWIND_HINT_TYPE_CALL 0 #define UNWIND_HINT_TYPE_REGS 1 #define UNWIND_HINT_TYPE_REGS_PARTIAL 2 #define UNWIND_HINT_TYPE_FUNC 3 +#define UNWIND_HINT_TYPE_ENTRY 4

#ifdef CONFIG_STACK_VALIDATION

--- a/tools/objtool/builtin-check.c +++ b/tools/objtool/builtin-check.c @@ -21,7 +21,7 @@

bool no_fp, no_unreachable, retpoline, module, backtrace, uaccess, stats, lto, vmlinux, mcount, noinstr, backup, sls, dryrun, - ibt; + ibt, unret;

static const char * const check_usage[] = { "objtool check [<options>] file.o", @@ -37,6 +37,7 @@ const struct option check_options[] = { OPT_BOOLEAN('f', "no-fp", &no_fp, "Skip frame pointer validation"), OPT_BOOLEAN('u', "no-unreachable", &no_unreachable, "Skip 'unreachable instruction' warnings"), OPT_BOOLEAN('r', "retpoline", &retpoline, "Validate retpoline assumptions"), + OPT_BOOLEAN(0, "unret", &unret, "validate entry unret placement"), OPT_BOOLEAN('m', "module", &module, "Indicates the object will be part of a kernel module"), OPT_BOOLEAN('b', "backtrace", &backtrace, "unwind on error"), OPT_BOOLEAN('a', "uaccess", &uaccess, "enable uaccess checking"), --- a/tools/objtool/check.c +++ b/tools/objtool/check.c @@ -2031,16 +2031,24 @@ static int read_unwind_hints(struct objt

insn->hint = true;

- if (ibt && hint->type == UNWIND_HINT_TYPE_REGS_PARTIAL) { + if (hint->type == UNWIND_HINT_TYPE_REGS_PARTIAL) { struct symbol *sym = find_symbol_by_offset(insn->sec, insn->offset);

- if (sym && sym->bind == STB_GLOBAL && - insn->type != INSN_ENDBR && !insn->noendbr) { - WARN_FUNC("UNWIND_HINT_IRET_REGS without ENDBR", - insn->sec, insn->offset); + if (sym && sym->bind == STB_GLOBAL) { + if (ibt && insn->type != INSN_ENDBR && !insn->noendbr) { + WARN_FUNC("UNWIND_HINT_IRET_REGS without ENDBR", + insn->sec, insn->offset); + } + + insn->entry = 1; } }

+ if (hint->type == UNWIND_HINT_TYPE_ENTRY) { + hint->type = UNWIND_HINT_TYPE_CALL; + insn->entry = 1; + } + if (hint->type == UNWIND_HINT_TYPE_FUNC) { insn->cfi = &func_cfi; continue; @@ -2115,8 +2123,9 @@ static int read_retpoline_hints(struct o

if (insn->type != INSN_JUMP_DYNAMIC && insn->type != INSN_CALL_DYNAMIC && - insn->type != INSN_RETURN) { - WARN_FUNC("retpoline_safe hint not an indirect jump/call/ret", + insn->type != INSN_RETURN && + insn->type != INSN_NOP) { + WARN_FUNC("retpoline_safe hint not an indirect jump/call/ret/nop", insn->sec, insn->offset); return -1; } @@ -3412,8 +3421,8 @@ static int validate_branch(struct objtoo return 1; }

- visited = 1 << state.uaccess; - if (insn->visited) { + visited = VISITED_BRANCH << state.uaccess; + if (insn->visited & VISITED_BRANCH_MASK) { if (!insn->hint && !insn_cfi_match(insn, &state.cfi)) return 1;

@@ -3642,6 +3651,145 @@ static int validate_unwind_hints(struct return warnings; }

+/* + * Validate rethunk entry constraint: must untrain RET before the first RET. + * + * Follow every branch (intra-function) and ensure ANNOTATE_UNRET_END comes + * before an actual RET instruction. + */ +static int validate_entry(struct objtool_file *file, struct instruction *insn) +{ + struct instruction *next, *dest; + int ret, warnings = 0; + + for (;;) { + next = next_insn_to_validate(file, insn); + + if (insn->visited & VISITED_ENTRY) + return 0; + + insn->visited |= VISITED_ENTRY; + + if (!insn->ignore_alts && !list_empty(&insn->alts)) { + struct alternative *alt; + bool skip_orig = false; + + list_for_each_entry(alt, &insn->alts, list) { + if (alt->skip_orig) + skip_orig = true; + + ret = validate_entry(file, alt->insn); + if (ret) { + if (backtrace) + BT_FUNC("(alt)", insn); + return ret; + } + } + + if (skip_orig) + return 0; + } + + switch (insn->type) { + + case INSN_CALL_DYNAMIC: + case INSN_JUMP_DYNAMIC: + case INSN_JUMP_DYNAMIC_CONDITIONAL: + WARN_FUNC("early indirect call", insn->sec, insn->offset); + return 1; + + case INSN_JUMP_UNCONDITIONAL: + case INSN_JUMP_CONDITIONAL: + if (!is_sibling_call(insn)) { + if (!insn->jump_dest) { + WARN_FUNC("unresolved jump target after linking?!?", + insn->sec, insn->offset); + return -1; + } + ret = validate_entry(file, insn->jump_dest); + if (ret) { + if (backtrace) { + BT_FUNC("(branch%s)", insn, + insn->type == INSN_JUMP_CONDITIONAL ? "-cond" : ""); + } + return ret; + } + + if (insn->type == INSN_JUMP_UNCONDITIONAL) + return 0; + + break; + } + + /* fallthrough */ + case INSN_CALL: + dest = find_insn(file, insn->call_dest->sec, + insn->call_dest->offset); + if (!dest) { + WARN("Unresolved function after linking!?: %s", + insn->call_dest->name); + return -1; + } + + ret = validate_entry(file, dest); + if (ret) { + if (backtrace) + BT_FUNC("(call)", insn); + return ret; + } + /* + * If a call returns without error, it must have seen UNTRAIN_RET. + * Therefore any non-error return is a success. + */ + return 0; + + case INSN_RETURN: + WARN_FUNC("RET before UNTRAIN", insn->sec, insn->offset); + return 1; + + case INSN_NOP: + if (insn->retpoline_safe) + return 0; + break; + + default: + break; + } + + if (!next) { + WARN_FUNC("teh end!", insn->sec, insn->offset); + return -1; + } + insn = next; + } + + return warnings; +} + +/* + * Validate that all branches starting at 'insn->entry' encounter UNRET_END + * before RET. + */ +static int validate_unret(struct objtool_file *file) +{ + struct instruction *insn; + int ret, warnings = 0; + + for_each_insn(file, insn) { + if (!insn->entry) + continue; + + ret = validate_entry(file, insn); + if (ret < 0) { + WARN_FUNC("Failed UNRET validation", insn->sec, insn->offset); + return ret; + } + warnings += ret; + } + + return warnings; +} + static int validate_retpoline(struct objtool_file *file) { struct instruction *insn; @@ -4005,6 +4153,17 @@ int check(struct objtool_file *file) goto out; warnings += ret;

+ if (unret) { + /* + * Must be after validate_branch() and friends, it plays + * further games with insn->visited. + */ + ret = validate_unret(file); + if (ret < 0) + return ret; + warnings += ret; + } + if (ibt) { ret = validate_ibt(file); if (ret < 0) --- a/tools/objtool/include/objtool/builtin.h +++ b/tools/objtool/include/objtool/builtin.h @@ -10,7 +10,7 @@ extern const struct option check_options[]; extern bool no_fp, no_unreachable, retpoline, module, backtrace, uaccess, stats, lto, vmlinux, mcount, noinstr, backup, sls, dryrun, - ibt; + ibt, unret;

extern int cmd_parse_options(int argc, const char **argv, const char * const usage[]);

--- a/tools/objtool/include/objtool/check.h +++ b/tools/objtool/include/objtool/check.h @@ -51,8 +51,10 @@ struct instruction { ignore_alts : 1, hint : 1, retpoline_safe : 1, - noendbr : 1; - /* 2 bit hole */ + noendbr : 1, + entry : 1; + /* 1 bit hole */ + s8 instr; u8 visited; /* u8 hole */ @@ -69,6 +71,11 @@ struct instruction { struct cfi_state *cfi; };

+#define VISITED_BRANCH 0x01 +#define VISITED_BRANCH_UACCESS 0x02 +#define VISITED_BRANCH_MASK 0x03 +#define VISITED_ENTRY 0x04 + static inline bool is_static_jump(struct instruction *insn) { return insn->type == INSN_JUMP_CONDITIONAL ||

From: Josh Poimboeuf jpoimboe@kernel.org

commit b2620facef4889fefcbf2e87284f34dcd4189bce upstream.

If a kernel is built with CONFIG_RETPOLINE=n, but the user still wants to mitigate Spectre v2 using IBRS or eIBRS, the RSB filling will be silently disabled.

There's nothing retpoline-specific about RSB buffer filling. Remove the CONFIG_RETPOLINE guards around it.

Signed-off-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/entry/entry_32.S | 2 -- arch/x86/entry/entry_64.S | 2 -- arch/x86/include/asm/nospec-branch.h | 2 -- 3 files changed, 6 deletions(-)

--- a/arch/x86/entry/entry_32.S +++ b/arch/x86/entry/entry_32.S @@ -698,7 +698,6 @@ SYM_CODE_START(__switch_to_asm) movl %ebx, PER_CPU_VAR(__stack_chk_guard) #endif

-#ifdef CONFIG_RETPOLINE /* * When switching from a shallower to a deeper call stack * the RSB may either underflow or use entries populated @@ -707,7 +706,6 @@ SYM_CODE_START(__switch_to_asm) * speculative execution to prevent attack. */ FILL_RETURN_BUFFER %ebx, RSB_CLEAR_LOOPS, X86_FEATURE_RSB_CTXSW -#endif

/* Restore flags or the incoming task to restore AC state. */ popfl --- a/arch/x86/entry/entry_64.S +++ b/arch/x86/entry/entry_64.S @@ -250,7 +250,6 @@ SYM_FUNC_START(__switch_to_asm) movq %rbx, PER_CPU_VAR(fixed_percpu_data) + stack_canary_offset #endif

-#ifdef CONFIG_RETPOLINE /* * When switching from a shallower to a deeper call stack * the RSB may either underflow or use entries populated @@ -259,7 +258,6 @@ SYM_FUNC_START(__switch_to_asm) * speculative execution to prevent attack. */ FILL_RETURN_BUFFER %r12, RSB_CLEAR_LOOPS, X86_FEATURE_RSB_CTXSW -#endif

/* restore callee-saved registers */ popq %r15 --- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -122,11 +122,9 @@ * monstrosity above, manually. */ .macro FILL_RETURN_BUFFER reg:req nr:req ftr:req -#ifdef CONFIG_RETPOLINE ALTERNATIVE "jmp .Lskip_rsb_@", "", \ftr __FILL_RETURN_BUFFER(\reg,\nr,%_ASM_SP) .Lskip_rsb_@: -#endif .endm

/*

From: Josh Poimboeuf jpoimboe@kernel.org

commit 56aa4d221f1ee2c3a49b45b800778ec6e0ab73c5 upstream.

If the SMT state changes, SSBD might get accidentally disabled. Fix that.

Signed-off-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kernel/cpu/bugs.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)

--- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -1451,7 +1451,8 @@ static void __init spectre_v2_select_mit

static void update_stibp_msr(void * __unused) { - write_spec_ctrl_current(x86_spec_ctrl_base, true); + u64 val = spec_ctrl_current() | (x86_spec_ctrl_base & SPEC_CTRL_STIBP); + write_spec_ctrl_current(val, true); }

/* Update x86_spec_ctrl_base in case SMT state changed. */

From: Josh Poimboeuf jpoimboe@kernel.org

commit acac5e98ef8d638a411cfa2ee676c87e1973f126 upstream.

This mask has been made redundant by kvm_spec_ctrl_test_value(). And it doesn't even work when MSR interception is disabled, as the guest can just write to SPEC_CTRL directly.

Signed-off-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Borislav Petkov bp@suse.de Reviewed-by: Paolo Bonzini pbonzini@redhat.com Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kernel/cpu/bugs.c | 31 +------------------------------ 1 file changed, 1 insertion(+), 30 deletions(-)

--- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -86,12 +86,6 @@ u64 spec_ctrl_current(void) EXPORT_SYMBOL_GPL(spec_ctrl_current);

/* - * The vendor and possibly platform specific bits which can be modified in - * x86_spec_ctrl_base. - */ -static u64 __ro_after_init x86_spec_ctrl_mask = SPEC_CTRL_IBRS; - -/* * AMD specific MSR info for Speculative Store Bypass control. * x86_amd_ls_cfg_ssbd_mask is initialized in identify_boot_cpu(). */ @@ -146,10 +140,6 @@ void __init check_bugs(void) if (boot_cpu_has(X86_FEATURE_MSR_SPEC_CTRL)) rdmsrl(MSR_IA32_SPEC_CTRL, x86_spec_ctrl_base);

- /* Allow STIBP in MSR_SPEC_CTRL if supported */ - if (boot_cpu_has(X86_FEATURE_STIBP)) - x86_spec_ctrl_mask |= SPEC_CTRL_STIBP; - /* Select the proper CPU mitigations before patching alternatives: */ spectre_v1_select_mitigation(); spectre_v2_select_mitigation(); @@ -208,19 +198,10 @@ void __init check_bugs(void) void x86_virt_spec_ctrl(u64 guest_spec_ctrl, u64 guest_virt_spec_ctrl, bool setguest) { - u64 msrval, guestval, hostval = spec_ctrl_current(); + u64 msrval, guestval = guest_spec_ctrl, hostval = spec_ctrl_current(); struct thread_info *ti = current_thread_info();

- /* Is MSR_SPEC_CTRL implemented ? */ if (static_cpu_has(X86_FEATURE_MSR_SPEC_CTRL)) { - /* - * Restrict guest_spec_ctrl to supported values. Clear the - * modifiable bits in the host base value and or the - * modifiable bits from the guest value. - */ - guestval = hostval & ~x86_spec_ctrl_mask; - guestval |= guest_spec_ctrl & x86_spec_ctrl_mask; - if (hostval != guestval) { msrval = setguest ? guestval : hostval; wrmsrl(MSR_IA32_SPEC_CTRL, msrval); @@ -1659,16 +1640,6 @@ static enum ssb_mitigation __init __ssb_ }

/* - * If SSBD is controlled by the SPEC_CTRL MSR, then set the proper - * bit in the mask to allow guests to use the mitigation even in the - * case where the host does not enable it. - */ - if (static_cpu_has(X86_FEATURE_SPEC_CTRL_SSBD) || - static_cpu_has(X86_FEATURE_AMD_SSBD)) { - x86_spec_ctrl_mask |= SPEC_CTRL_SSBD; - } - - /* * We have three CPU feature flags that are in play here: * - X86_BUG_SPEC_STORE_BYPASS - CPU is susceptible. * - X86_FEATURE_SSBD - CPU is able to turn off speculative store bypass

From: Josh Poimboeuf jpoimboe@kernel.org

commit 8bd200d23ec42d66ccd517a72dd0b9cc6132d2fd upstream.

Move the vmx_vm{enter,exit}() functionality into __vmx_vcpu_run(). This will make it easier to do the spec_ctrl handling before the first RET.

Signed-off-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kvm/vmx/vmenter.S | 119 +++++++++++++++++---------------------------- 1 file changed, 46 insertions(+), 73 deletions(-)

--- a/arch/x86/kvm/vmx/vmenter.S +++ b/arch/x86/kvm/vmx/vmenter.S @@ -31,68 +31,6 @@ .section .noinstr.text, "ax"

/** - * vmx_vmenter - VM-Enter the current loaded VMCS - * - * %RFLAGS.ZF: !VMCS.LAUNCHED, i.e. controls VMLAUNCH vs. VMRESUME - * - * Returns: - * %RFLAGS.CF is set on VM-Fail Invalid - * %RFLAGS.ZF is set on VM-Fail Valid - * %RFLAGS.{CF,ZF} are cleared on VM-Success, i.e. VM-Exit - * - * Note that VMRESUME/VMLAUNCH fall-through and return directly if - * they VM-Fail, whereas a successful VM-Enter + VM-Exit will jump - * to vmx_vmexit. - */ -SYM_FUNC_START_LOCAL(vmx_vmenter) - /* EFLAGS.ZF is set if VMCS.LAUNCHED == 0 */ - je 2f - -1: vmresume - RET - -2: vmlaunch - RET - -3: cmpb $0, kvm_rebooting - je 4f - RET -4: ud2 - - _ASM_EXTABLE(1b, 3b) - _ASM_EXTABLE(2b, 3b) - -SYM_FUNC_END(vmx_vmenter) - -/** - * vmx_vmexit - Handle a VMX VM-Exit - * - * Returns: - * %RFLAGS.{CF,ZF} are cleared on VM-Success, i.e. VM-Exit - * - * This is vmx_vmenter's partner in crime. On a VM-Exit, control will jump - * here after hardware loads the host's state, i.e. this is the destination - * referred to by VMCS.HOST_RIP. - */ -SYM_FUNC_START(vmx_vmexit) -#ifdef CONFIG_RETPOLINE - ALTERNATIVE "jmp .Lvmexit_skip_rsb", "", X86_FEATURE_RETPOLINE - /* Preserve guest's RAX, it's used to stuff the RSB. */ - push %_ASM_AX - - /* IMPORTANT: Stuff the RSB immediately after VM-Exit, before RET! */ - FILL_RETURN_BUFFER %_ASM_AX, RSB_CLEAR_LOOPS, X86_FEATURE_RETPOLINE - - /* Clear RFLAGS.CF and RFLAGS.ZF to preserve VM-Exit, i.e. !VM-Fail. */ - or $1, %_ASM_AX - - pop %_ASM_AX -.Lvmexit_skip_rsb: -#endif - RET -SYM_FUNC_END(vmx_vmexit) - -/** * __vmx_vcpu_run - Run a vCPU via a transition to VMX guest mode * @vmx: struct vcpu_vmx * (forwarded to vmx_update_host_rsp) * @regs: unsigned long * (to guest registers) @@ -124,8 +62,7 @@ SYM_FUNC_START(__vmx_vcpu_run) /* Copy @launched to BL, _ASM_ARG3 is volatile. */ mov %_ASM_ARG3B, %bl

- /* Adjust RSP to account for the CALL to vmx_vmenter(). */ - lea -WORD_SIZE(%_ASM_SP), %_ASM_ARG2 + lea (%_ASM_SP), %_ASM_ARG2 call vmx_update_host_rsp

/* Load @regs to RAX. */ @@ -154,11 +91,37 @@ SYM_FUNC_START(__vmx_vcpu_run) /* Load guest RAX. This kills the @regs pointer! */ mov VCPU_RAX(%_ASM_AX), %_ASM_AX

- /* Enter guest mode */ - call vmx_vmenter + /* Check EFLAGS.ZF from 'testb' above */ + je .Lvmlaunch

- /* Jump on VM-Fail. */ - jbe 2f + /* + * After a successful VMRESUME/VMLAUNCH, control flow "magically" + * resumes below at 'vmx_vmexit' due to the VMCS HOST_RIP setting. + * So this isn't a typical function and objtool needs to be told to + * save the unwind state here and restore it below. + */ + UNWIND_HINT_SAVE + +/* + * If VMRESUME/VMLAUNCH and corresponding vmexit succeed, execution resumes at + * the 'vmx_vmexit' label below. + */ +.Lvmresume: + vmresume + jmp .Lvmfail + +.Lvmlaunch: + vmlaunch + jmp .Lvmfail + + _ASM_EXTABLE(.Lvmresume, .Lfixup) + _ASM_EXTABLE(.Lvmlaunch, .Lfixup) + +SYM_INNER_LABEL(vmx_vmexit, SYM_L_GLOBAL) + + /* Restore unwind state from before the VMRESUME/VMLAUNCH. */ + UNWIND_HINT_RESTORE + ENDBR

/* Temporarily save guest's RAX. */ push %_ASM_AX @@ -185,9 +148,13 @@ SYM_FUNC_START(__vmx_vcpu_run) mov %r15, VCPU_R15(%_ASM_AX) #endif

+ /* IMPORTANT: RSB must be stuffed before the first return. */ + FILL_RETURN_BUFFER %_ASM_BX, RSB_CLEAR_LOOPS, X86_FEATURE_RETPOLINE + /* Clear RAX to indicate VM-Exit (as opposed to VM-Fail). */ xor %eax, %eax

+.Lclear_regs: /* * Clear all general purpose registers except RSP and RAX to prevent * speculative use of the guest's values, even those that are reloaded @@ -197,7 +164,7 @@ SYM_FUNC_START(__vmx_vcpu_run) * free. RSP and RAX are exempt as RSP is restored by hardware during * VM-Exit and RAX is explicitly loaded with 0 or 1 to return VM-Fail. */ -1: xor %ecx, %ecx + xor %ecx, %ecx xor %edx, %edx xor %ebx, %ebx xor %ebp, %ebp @@ -216,8 +183,8 @@ SYM_FUNC_START(__vmx_vcpu_run)

/* "POP" @regs. */ add $WORD_SIZE, %_ASM_SP - pop %_ASM_BX

+ pop %_ASM_BX #ifdef CONFIG_X86_64 pop %r12 pop %r13 @@ -230,9 +197,15 @@ SYM_FUNC_START(__vmx_vcpu_run) pop %_ASM_BP RET

- /* VM-Fail. Out-of-line to avoid a taken Jcc after VM-Exit. */ -2: mov $1, %eax - jmp 1b +.Lfixup: + cmpb $0, kvm_rebooting + jne .Lvmfail + ud2 +.Lvmfail: + /* VM-Fail: set return value to 1 */ + mov $1, %eax + jmp .Lclear_regs + SYM_FUNC_END(__vmx_vcpu_run)

From: Josh Poimboeuf jpoimboe@kernel.org

commit fc02735b14fff8c6678b521d324ade27b1a3d4cf upstream.

On eIBRS systems, the returns in the vmexit return path from __vmx_vcpu_run() to vmx_vcpu_run() are exposed to RSB poisoning attacks.

Fix that by moving the post-vmexit spec_ctrl handling to immediately after the vmexit.

Signed-off-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/include/asm/nospec-branch.h | 1 arch/x86/kernel/cpu/bugs.c | 4 ++ arch/x86/kvm/vmx/run_flags.h | 1 arch/x86/kvm/vmx/vmenter.S | 49 +++++++++++++++++++++++++++-------- arch/x86/kvm/vmx/vmx.c | 48 ++++++++++++++++++++-------------- arch/x86/kvm/vmx/vmx.h | 1 6 files changed, 73 insertions(+), 31 deletions(-)

--- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -274,6 +274,7 @@ static inline void indirect_branch_predi

/* The Intel SPEC CTRL MSR base value cache */ extern u64 x86_spec_ctrl_base; +extern u64 x86_spec_ctrl_current; extern void write_spec_ctrl_current(u64 val, bool force); extern u64 spec_ctrl_current(void);

--- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -195,6 +195,10 @@ void __init check_bugs(void) #endif }

+/* + * NOTE: For VMX, this function is not called in the vmexit path. + * It uses vmx_spec_ctrl_restore_host() instead. + */ void x86_virt_spec_ctrl(u64 guest_spec_ctrl, u64 guest_virt_spec_ctrl, bool setguest) { --- a/arch/x86/kvm/vmx/run_flags.h +++ b/arch/x86/kvm/vmx/run_flags.h @@ -3,5 +3,6 @@ #define __KVM_X86_VMX_RUN_FLAGS_H

#define VMX_RUN_VMRESUME (1 << 0) +#define VMX_RUN_SAVE_SPEC_CTRL (1 << 1)

#endif /* __KVM_X86_VMX_RUN_FLAGS_H */ --- a/arch/x86/kvm/vmx/vmenter.S +++ b/arch/x86/kvm/vmx/vmenter.S @@ -33,9 +33,10 @@

/** * __vmx_vcpu_run - Run a vCPU via a transition to VMX guest mode - * @vmx: struct vcpu_vmx * (forwarded to vmx_update_host_rsp) + * @vmx: struct vcpu_vmx * * @regs: unsigned long * (to guest registers) - * @flags: VMX_RUN_VMRESUME: use VMRESUME instead of VMLAUNCH + * @flags: VMX_RUN_VMRESUME: use VMRESUME instead of VMLAUNCH + * VMX_RUN_SAVE_SPEC_CTRL: save guest SPEC_CTRL into vmx->spec_ctrl * * Returns: * 0 on VM-Exit, 1 on VM-Fail @@ -54,6 +55,12 @@ SYM_FUNC_START(__vmx_vcpu_run) #endif push %_ASM_BX

+ /* Save @vmx for SPEC_CTRL handling */ + push %_ASM_ARG1 + + /* Save @flags for SPEC_CTRL handling */ + push %_ASM_ARG3 + /* * Save @regs, _ASM_ARG2 may be modified by vmx_update_host_rsp() and * @regs is needed after VM-Exit to save the guest's register values. @@ -149,25 +156,23 @@ SYM_INNER_LABEL(vmx_vmexit, SYM_L_GLOBAL mov %r15, VCPU_R15(%_ASM_AX) #endif

- /* IMPORTANT: RSB must be stuffed before the first return. */ - FILL_RETURN_BUFFER %_ASM_BX, RSB_CLEAR_LOOPS, X86_FEATURE_RETPOLINE - - /* Clear RAX to indicate VM-Exit (as opposed to VM-Fail). */ - xor %eax, %eax + /* Clear return value to indicate VM-Exit (as opposed to VM-Fail). */ + xor %ebx, %ebx

.Lclear_regs: /* - * Clear all general purpose registers except RSP and RAX to prevent + * Clear all general purpose registers except RSP and RBX to prevent * speculative use of the guest's values, even those that are reloaded * via the stack. In theory, an L1 cache miss when restoring registers * could lead to speculative execution with the guest's values. * Zeroing XORs are dirt cheap, i.e. the extra paranoia is essentially * free. RSP and RAX are exempt as RSP is restored by hardware during - * VM-Exit and RAX is explicitly loaded with 0 or 1 to return VM-Fail. + * VM-Exit and RBX is explicitly loaded with 0 or 1 to hold the return + * value. */ + xor %eax, %eax xor %ecx, %ecx xor %edx, %edx - xor %ebx, %ebx xor %ebp, %ebp xor %esi, %esi xor %edi, %edi @@ -185,6 +190,28 @@ SYM_INNER_LABEL(vmx_vmexit, SYM_L_GLOBAL /* "POP" @regs. */ add $WORD_SIZE, %_ASM_SP

+ /* + * IMPORTANT: RSB filling and SPEC_CTRL handling must be done before + * the first unbalanced RET after vmexit! + * + * For retpoline, RSB filling is needed to prevent poisoned RSB entries + * and (in some cases) RSB underflow. + * + * eIBRS has its own protection against poisoned RSB, so it doesn't + * need the RSB filling sequence. But it does need to be enabled + * before the first unbalanced RET. + */ + + FILL_RETURN_BUFFER %_ASM_CX, RSB_CLEAR_LOOPS, X86_FEATURE_RETPOLINE + + pop %_ASM_ARG2 /* @flags */ + pop %_ASM_ARG1 /* @vmx */ + + call vmx_spec_ctrl_restore_host + + /* Put return value in AX */ + mov %_ASM_BX, %_ASM_AX + pop %_ASM_BX #ifdef CONFIG_X86_64 pop %r12 @@ -204,7 +231,7 @@ SYM_INNER_LABEL(vmx_vmexit, SYM_L_GLOBAL ud2 .Lvmfail: /* VM-Fail: set return value to 1 */ - mov $1, %eax + mov $1, %_ASM_BX jmp .Lclear_regs

SYM_FUNC_END(__vmx_vcpu_run) --- a/arch/x86/kvm/vmx/vmx.c +++ b/arch/x86/kvm/vmx/vmx.c @@ -846,6 +846,14 @@ unsigned int __vmx_vcpu_run_flags(struct if (vmx->loaded_vmcs->launched) flags |= VMX_RUN_VMRESUME;

+ /* + * If writes to the SPEC_CTRL MSR aren't intercepted, the guest is free + * to change it directly without causing a vmexit. In that case read + * it after vmexit and store it in vmx->spec_ctrl. + */ + if (unlikely(!msr_write_intercepted(vmx, MSR_IA32_SPEC_CTRL))) + flags |= VMX_RUN_SAVE_SPEC_CTRL; + return flags; }

@@ -6824,6 +6832,26 @@ void noinstr vmx_update_host_rsp(struct } }

+void noinstr vmx_spec_ctrl_restore_host(struct vcpu_vmx *vmx, + unsigned int flags) +{ + u64 hostval = this_cpu_read(x86_spec_ctrl_current); + + if (!cpu_feature_enabled(X86_FEATURE_MSR_SPEC_CTRL)) + return; + + if (flags & VMX_RUN_SAVE_SPEC_CTRL) + vmx->spec_ctrl = __rdmsr(MSR_IA32_SPEC_CTRL); + + /* + * If the guest/host SPEC_CTRL values differ, restore the host value. + */ + if (vmx->spec_ctrl != hostval) + native_wrmsrl(MSR_IA32_SPEC_CTRL, hostval); + + barrier_nospec(); +} + static fastpath_t vmx_exit_handlers_fastpath(struct kvm_vcpu *vcpu) { switch (to_vmx(vcpu)->exit_reason.basic) { @@ -6967,26 +6995,6 @@ static fastpath_t vmx_vcpu_run(struct kv /* The actual VMENTER/EXIT is in the .noinstr.text section. */ vmx_vcpu_enter_exit(vcpu, vmx, __vmx_vcpu_run_flags(vmx));

- /* - * We do not use IBRS in the kernel. If this vCPU has used the - * SPEC_CTRL MSR it may have left it on; save the value and - * turn it off. This is much more efficient than blindly adding - * it to the atomic save/restore list. Especially as the former - * (Saving guest MSRs on vmexit) doesn't even exist in KVM. - * - * For non-nested case: - * If the L01 MSR bitmap does not intercept the MSR, then we need to - * save it. - * - * For nested case: - * If the L02 MSR bitmap does not intercept the MSR, then we need to - * save it. - */ - if (unlikely(!msr_write_intercepted(vmx, MSR_IA32_SPEC_CTRL))) - vmx->spec_ctrl = native_read_msr(MSR_IA32_SPEC_CTRL); - - x86_spec_ctrl_restore_host(vmx->spec_ctrl, 0); - /* All fields are clean at this point */ if (static_branch_unlikely(&enable_evmcs)) { current_evmcs->hv_clean_fields |= --- a/arch/x86/kvm/vmx/vmx.h +++ b/arch/x86/kvm/vmx/vmx.h @@ -405,6 +405,7 @@ void vmx_set_virtual_apic_mode(struct kv struct vmx_uret_msr *vmx_find_uret_msr(struct vcpu_vmx *vmx, u32 msr); void pt_update_intercept_for_msr(struct kvm_vcpu *vcpu); void vmx_update_host_rsp(struct vcpu_vmx *vmx, unsigned long host_rsp); +void vmx_spec_ctrl_restore_host(struct vcpu_vmx *vmx, unsigned int flags); unsigned int __vmx_vcpu_run_flags(struct vcpu_vmx *vmx); bool __vmx_vcpu_run(struct vcpu_vmx *vmx, unsigned long *regs, unsigned int flags);

From: Josh Poimboeuf jpoimboe@kernel.org

commit 9756bba28470722dacb79ffce554336dd1f6a6cd upstream.

Prevent RSB underflow/poisoning attacks with RSB. While at it, add a bunch of comments to attempt to document the current state of tribal knowledge about RSB attacks and what exactly is being mitigated.

Signed-off-by: Josh Poimboeuf jpoimboe@kernel.org Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/include/asm/cpufeatures.h | 2 - arch/x86/kernel/cpu/bugs.c | 63 ++++++++++++++++++++++++++++++++++--- arch/x86/kvm/vmx/vmenter.S | 6 +-- 3 files changed, 62 insertions(+), 9 deletions(-)

--- a/arch/x86/include/asm/cpufeatures.h +++ b/arch/x86/include/asm/cpufeatures.h @@ -204,7 +204,7 @@ /* FREE! ( 7*32+10) */ #define X86_FEATURE_PTI ( 7*32+11) /* Kernel Page Table Isolation enabled */ #define X86_FEATURE_KERNEL_IBRS ( 7*32+12) /* "" Set/clear IBRS on kernel entry/exit */ -/* FREE! ( 7*32+13) */ +#define X86_FEATURE_RSB_VMEXIT ( 7*32+13) /* "" Fill RSB on VM-Exit */ #define X86_FEATURE_INTEL_PPIN ( 7*32+14) /* Intel Processor Inventory Number */ #define X86_FEATURE_CDP_L2 ( 7*32+15) /* Code and Data Prioritization L2 */ #define X86_FEATURE_MSR_SPEC_CTRL ( 7*32+16) /* "" MSR SPEC_CTRL is implemented */ --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -1394,17 +1394,70 @@ static void __init spectre_v2_select_mit pr_info("%s\n", spectre_v2_strings[mode]);

/* - * If spectre v2 protection has been enabled, unconditionally fill - * RSB during a context switch; this protects against two independent - * issues: + * If Spectre v2 protection has been enabled, fill the RSB during a + * context switch. In general there are two types of RSB attacks + * across context switches, for which the CALLs/RETs may be unbalanced. * - * - RSB underflow (and switch to BTB) on Skylake+ - * - SpectreRSB variant of spectre v2 on X86_BUG_SPECTRE_V2 CPUs + * 1) RSB underflow + * + * Some Intel parts have "bottomless RSB". When the RSB is empty, + * speculated return targets may come from the branch predictor, + * which could have a user-poisoned BTB or BHB entry. + * + * AMD has it even worse: *all* returns are speculated from the BTB, + * regardless of the state of the RSB. + * + * When IBRS or eIBRS is enabled, the "user -> kernel" attack + * scenario is mitigated by the IBRS branch prediction isolation + * properties, so the RSB buffer filling wouldn't be necessary to + * protect against this type of attack. + * + * The "user -> user" attack scenario is mitigated by RSB filling. + * + * 2) Poisoned RSB entry + * + * If the 'next' in-kernel return stack is shorter than 'prev', + * 'next' could be tricked into speculating with a user-poisoned RSB + * entry. + * + * The "user -> kernel" attack scenario is mitigated by SMEP and + * eIBRS. + * + * The "user -> user" scenario, also known as SpectreBHB, requires + * RSB clearing. + * + * So to mitigate all cases, unconditionally fill RSB on context + * switches. + * + * FIXME: Is this pointless for retbleed-affected AMD? */ setup_force_cpu_cap(X86_FEATURE_RSB_CTXSW); pr_info("Spectre v2 / SpectreRSB mitigation: Filling RSB on context switch\n");

/* + * Similar to context switches, there are two types of RSB attacks + * after vmexit: + * + * 1) RSB underflow + * + * 2) Poisoned RSB entry + * + * When retpoline is enabled, both are mitigated by filling/clearing + * the RSB. + * + * When IBRS is enabled, while #1 would be mitigated by the IBRS branch + * prediction isolation protections, RSB still needs to be cleared + * because of #2. Note that SMEP provides no protection here, unlike + * user-space-poisoned RSB entries. + * + * eIBRS, on the other hand, has RSB-poisoning protections, so it + * doesn't need RSB clearing after vmexit. + */ + if (boot_cpu_has(X86_FEATURE_RETPOLINE) || + boot_cpu_has(X86_FEATURE_KERNEL_IBRS)) + setup_force_cpu_cap(X86_FEATURE_RSB_VMEXIT); + + /* * Retpoline protects the kernel, but doesn't protect firmware. IBRS * and Enhanced IBRS protect firmware too, so enable IBRS around * firmware calls only when IBRS / Enhanced IBRS aren't otherwise --- a/arch/x86/kvm/vmx/vmenter.S +++ b/arch/x86/kvm/vmx/vmenter.S @@ -194,15 +194,15 @@ SYM_INNER_LABEL(vmx_vmexit, SYM_L_GLOBAL * IMPORTANT: RSB filling and SPEC_CTRL handling must be done before * the first unbalanced RET after vmexit! * - * For retpoline, RSB filling is needed to prevent poisoned RSB entries - * and (in some cases) RSB underflow. + * For retpoline or IBRS, RSB filling is needed to prevent poisoned RSB + * entries and (in some cases) RSB underflow. * * eIBRS has its own protection against poisoned RSB, so it doesn't * need the RSB filling sequence. But it does need to be enabled * before the first unbalanced RET. */

- FILL_RETURN_BUFFER %_ASM_CX, RSB_CLEAR_LOOPS, X86_FEATURE_RETPOLINE + FILL_RETURN_BUFFER %_ASM_CX, RSB_CLEAR_LOOPS, X86_FEATURE_RSB_VMEXIT

pop %_ASM_ARG2 /* @flags */ pop %_ASM_ARG1 /* @vmx */

From: Peter Zijlstra peterz@infradead.org

commit 7a05bc95ed1c5a59e47aaade9fb4083c27de9e62 upstream.

The whole MMIO/RETBLEED enumeration went overboard on steppings. Get rid of all that and simply use ANY.

If a future stepping of these models would not be affected, it had better set the relevant ARCH_CAP_$FOO_NO bit in IA32_ARCH_CAPABILITIES.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de Acked-by: Dave Hansen dave.hansen@linux.intel.com Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kernel/cpu/common.c | 37 ++++++++++++++++--------------------- 1 file changed, 16 insertions(+), 21 deletions(-)

--- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -1258,32 +1258,27 @@ static const struct x86_cpu_id cpu_vuln_ VULNBL_INTEL_STEPPINGS(HASWELL, X86_STEPPING_ANY, SRBDS), VULNBL_INTEL_STEPPINGS(HASWELL_L, X86_STEPPING_ANY, SRBDS), VULNBL_INTEL_STEPPINGS(HASWELL_G, X86_STEPPING_ANY, SRBDS), - VULNBL_INTEL_STEPPINGS(HASWELL_X, BIT(2) | BIT(4), MMIO), - VULNBL_INTEL_STEPPINGS(BROADWELL_D, X86_STEPPINGS(0x3, 0x5), MMIO), + VULNBL_INTEL_STEPPINGS(HASWELL_X, X86_STEPPING_ANY, MMIO), + VULNBL_INTEL_STEPPINGS(BROADWELL_D, X86_STEPPING_ANY, MMIO), VULNBL_INTEL_STEPPINGS(BROADWELL_G, X86_STEPPING_ANY, SRBDS), VULNBL_INTEL_STEPPINGS(BROADWELL_X, X86_STEPPING_ANY, MMIO), VULNBL_INTEL_STEPPINGS(BROADWELL, X86_STEPPING_ANY, SRBDS), - VULNBL_INTEL_STEPPINGS(SKYLAKE_L, X86_STEPPINGS(0x3, 0x3), SRBDS | MMIO | RETBLEED), - VULNBL_INTEL_STEPPINGS(SKYLAKE_L, X86_STEPPING_ANY, SRBDS), - VULNBL_INTEL_STEPPINGS(SKYLAKE_X, BIT(3) | BIT(4) | BIT(6) | - BIT(7) | BIT(0xB), MMIO | RETBLEED), - VULNBL_INTEL_STEPPINGS(SKYLAKE, X86_STEPPINGS(0x3, 0x3), SRBDS | MMIO | RETBLEED), - VULNBL_INTEL_STEPPINGS(SKYLAKE, X86_STEPPING_ANY, SRBDS), - VULNBL_INTEL_STEPPINGS(KABYLAKE_L, X86_STEPPINGS(0x9, 0xC), SRBDS | MMIO | RETBLEED), - VULNBL_INTEL_STEPPINGS(KABYLAKE_L, X86_STEPPINGS(0x0, 0x8), SRBDS), - VULNBL_INTEL_STEPPINGS(KABYLAKE, X86_STEPPINGS(0x9, 0xD), SRBDS | MMIO | RETBLEED), - VULNBL_INTEL_STEPPINGS(KABYLAKE, X86_STEPPINGS(0x0, 0x8), SRBDS), - VULNBL_INTEL_STEPPINGS(ICELAKE_L, X86_STEPPINGS(0x5, 0x5), MMIO | MMIO_SBDS | RETBLEED), - VULNBL_INTEL_STEPPINGS(ICELAKE_D, X86_STEPPINGS(0x1, 0x1), MMIO), - VULNBL_INTEL_STEPPINGS(ICELAKE_X, X86_STEPPINGS(0x4, 0x6), MMIO), - VULNBL_INTEL_STEPPINGS(COMETLAKE, BIT(2) | BIT(3) | BIT(5), MMIO | MMIO_SBDS | RETBLEED), - VULNBL_INTEL_STEPPINGS(COMETLAKE_L, X86_STEPPINGS(0x1, 0x1), MMIO | MMIO_SBDS | RETBLEED), + VULNBL_INTEL_STEPPINGS(SKYLAKE_L, X86_STEPPING_ANY, SRBDS | MMIO | RETBLEED), + VULNBL_INTEL_STEPPINGS(SKYLAKE_X, X86_STEPPING_ANY, MMIO | RETBLEED), + VULNBL_INTEL_STEPPINGS(SKYLAKE, X86_STEPPING_ANY, SRBDS | MMIO | RETBLEED), + VULNBL_INTEL_STEPPINGS(KABYLAKE_L, X86_STEPPING_ANY, SRBDS | MMIO | RETBLEED), + VULNBL_INTEL_STEPPINGS(KABYLAKE, X86_STEPPING_ANY, SRBDS | MMIO | RETBLEED), + VULNBL_INTEL_STEPPINGS(ICELAKE_L, X86_STEPPING_ANY, MMIO | MMIO_SBDS | RETBLEED), + VULNBL_INTEL_STEPPINGS(ICELAKE_D, X86_STEPPING_ANY, MMIO), + VULNBL_INTEL_STEPPINGS(ICELAKE_X, X86_STEPPING_ANY, MMIO), + VULNBL_INTEL_STEPPINGS(COMETLAKE, X86_STEPPING_ANY, MMIO | MMIO_SBDS | RETBLEED), VULNBL_INTEL_STEPPINGS(COMETLAKE_L, X86_STEPPINGS(0x0, 0x0), MMIO | RETBLEED), - VULNBL_INTEL_STEPPINGS(LAKEFIELD, X86_STEPPINGS(0x1, 0x1), MMIO | MMIO_SBDS | RETBLEED), - VULNBL_INTEL_STEPPINGS(ROCKETLAKE, X86_STEPPINGS(0x1, 0x1), MMIO | RETBLEED), - VULNBL_INTEL_STEPPINGS(ATOM_TREMONT, X86_STEPPINGS(0x1, 0x1), MMIO | MMIO_SBDS), + VULNBL_INTEL_STEPPINGS(COMETLAKE_L, X86_STEPPING_ANY, MMIO | MMIO_SBDS | RETBLEED), + VULNBL_INTEL_STEPPINGS(LAKEFIELD, X86_STEPPING_ANY, MMIO | MMIO_SBDS | RETBLEED), + VULNBL_INTEL_STEPPINGS(ROCKETLAKE, X86_STEPPING_ANY, MMIO | RETBLEED), + VULNBL_INTEL_STEPPINGS(ATOM_TREMONT, X86_STEPPING_ANY, MMIO | MMIO_SBDS), VULNBL_INTEL_STEPPINGS(ATOM_TREMONT_D, X86_STEPPING_ANY, MMIO), - VULNBL_INTEL_STEPPINGS(ATOM_TREMONT_L, X86_STEPPINGS(0x0, 0x0), MMIO | MMIO_SBDS), + VULNBL_INTEL_STEPPINGS(ATOM_TREMONT_L, X86_STEPPING_ANY, MMIO | MMIO_SBDS),

VULNBL_AMD(0x15, RETBLEED), VULNBL_AMD(0x16, RETBLEED),

From: Peter Zijlstra peterz@infradead.org

commit f43b9876e857c739d407bc56df288b0ebe1a9164 upstream.

Do fine-grained Kconfig for all the various retbleed parts.

NOTE: if your compiler doesn't support return thunks this will silently 'upgrade' your mitigation to IBPB, you might not like this.

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de [cascardo: there is no CONFIG_OBJTOOL] [cascardo: objtool calling and option parsing has changed] Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/Kconfig | 106 +++++++++++++++++++++++-------- arch/x86/Makefile | 8 +- arch/x86/entry/calling.h | 4 + arch/x86/include/asm/disabled-features.h | 18 ++++- arch/x86/include/asm/linkage.h | 4 - arch/x86/include/asm/nospec-branch.h | 10 ++ arch/x86/include/asm/static_call.h | 2 arch/x86/kernel/alternative.c | 5 + arch/x86/kernel/cpu/amd.c | 2 arch/x86/kernel/cpu/bugs.c | 42 +++++++----- arch/x86/kernel/static_call.c | 2 arch/x86/kvm/emulate.c | 4 - arch/x86/lib/retpoline.S | 4 + scripts/Makefile.build | 1 scripts/link-vmlinux.sh | 2 security/Kconfig | 11 --- tools/objtool/builtin-check.c | 3 tools/objtool/check.c | 9 ++ tools/objtool/include/objtool/builtin.h | 2 19 files changed, 170 insertions(+), 69 deletions(-)

--- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -457,30 +457,6 @@ config GOLDFISH def_bool y depends on X86_GOLDFISH

-config RETPOLINE - bool "Avoid speculative indirect branches in kernel" - default y - help - Compile kernel with the retpoline compiler options to guard against - kernel-to-user data leaks by avoiding speculative indirect - branches. Requires a compiler with -mindirect-branch=thunk-extern - support for full protection. The kernel may run slower. - -config CC_HAS_SLS - def_bool $(cc-option,-mharden-sls=all) - -config CC_HAS_RETURN_THUNK - def_bool $(cc-option,-mfunction-return=thunk-extern) - -config SLS - bool "Mitigate Straight-Line-Speculation" - depends on CC_HAS_SLS && X86_64 - default n - help - Compile the kernel with straight-line-speculation options to guard - against straight line speculation. The kernel image might be slightly - larger. - config X86_CPU_RESCTRL bool "x86 CPU resource control support" depends on X86 && (CPU_SUP_INTEL || CPU_SUP_AMD) @@ -2452,6 +2428,88 @@ source "kernel/livepatch/Kconfig"

endmenu

+config CC_HAS_SLS + def_bool $(cc-option,-mharden-sls=all) + +config CC_HAS_RETURN_THUNK + def_bool $(cc-option,-mfunction-return=thunk-extern) + +menuconfig SPECULATION_MITIGATIONS + bool "Mitigations for speculative execution vulnerabilities" + default y + help + Say Y here to enable options which enable mitigations for + speculative execution hardware vulnerabilities. + + If you say N, all mitigations will be disabled. You really + should know what you are doing to say so. + +if SPECULATION_MITIGATIONS + +config PAGE_TABLE_ISOLATION + bool "Remove the kernel mapping in user mode" + default y + depends on (X86_64 || X86_PAE) + help + This feature reduces the number of hardware side channels by + ensuring that the majority of kernel addresses are not mapped + into userspace. + + See Documentation/x86/pti.rst for more details. + +config RETPOLINE + bool "Avoid speculative indirect branches in kernel" + default y + help + Compile kernel with the retpoline compiler options to guard against + kernel-to-user data leaks by avoiding speculative indirect + branches. Requires a compiler with -mindirect-branch=thunk-extern + support for full protection. The kernel may run slower. + +config RETHUNK + bool "Enable return-thunks" + depends on RETPOLINE && CC_HAS_RETURN_THUNK + default y + help + Compile the kernel with the return-thunks compiler option to guard + against kernel-to-user data leaks by avoiding return speculation. + Requires a compiler with -mfunction-return=thunk-extern + support for full protection. The kernel may run slower. + +config CPU_UNRET_ENTRY + bool "Enable UNRET on kernel entry" + depends on CPU_SUP_AMD && RETHUNK + default y + help + Compile the kernel with support for the retbleed=unret mitigation. + +config CPU_IBPB_ENTRY + bool "Enable IBPB on kernel entry" + depends on CPU_SUP_AMD + default y + help + Compile the kernel with support for the retbleed=ibpb mitigation. + +config CPU_IBRS_ENTRY + bool "Enable IBRS on kernel entry" + depends on CPU_SUP_INTEL + default y + help + Compile the kernel with support for the spectre_v2=ibrs mitigation. + This mitigates both spectre_v2 and retbleed at great cost to + performance. + +config SLS + bool "Mitigate Straight-Line-Speculation" + depends on CC_HAS_SLS && X86_64 + default n + help + Compile the kernel with straight-line-speculation options to guard + against straight line speculation. The kernel image might be slightly + larger. + +endif + config ARCH_HAS_ADD_PAGES def_bool y depends on ARCH_ENABLE_MEMORY_HOTPLUG --- a/arch/x86/Makefile +++ b/arch/x86/Makefile @@ -15,14 +15,18 @@ endif ifdef CONFIG_CC_IS_GCC RETPOLINE_CFLAGS := $(call cc-option,-mindirect-branch=thunk-extern -mindirect-branch-register) RETPOLINE_CFLAGS += $(call cc-option,-mindirect-branch-cs-prefix) -RETPOLINE_CFLAGS += $(call cc-option,-mfunction-return=thunk-extern) RETPOLINE_VDSO_CFLAGS := $(call cc-option,-mindirect-branch=thunk-inline -mindirect-branch-register) endif ifdef CONFIG_CC_IS_CLANG RETPOLINE_CFLAGS := -mretpoline-external-thunk RETPOLINE_VDSO_CFLAGS := -mretpoline -RETPOLINE_CFLAGS += $(call cc-option,-mfunction-return=thunk-extern) endif + +ifdef CONFIG_RETHUNK +RETHUNK_CFLAGS := -mfunction-return=thunk-extern +RETPOLINE_CFLAGS += $(RETHUNK_CFLAGS) +endif + export RETPOLINE_CFLAGS export RETPOLINE_VDSO_CFLAGS

--- a/arch/x86/entry/calling.h +++ b/arch/x86/entry/calling.h @@ -296,6 +296,7 @@ For 32-bit we have the following convent * Assumes x86_spec_ctrl_{base,current} to have SPEC_CTRL_IBRS set. */ .macro IBRS_ENTER save_reg +#ifdef CONFIG_CPU_IBRS_ENTRY ALTERNATIVE "jmp .Lend_@", "", X86_FEATURE_KERNEL_IBRS movl $MSR_IA32_SPEC_CTRL, %ecx

@@ -316,6 +317,7 @@ For 32-bit we have the following convent shr $32, %rdx wrmsr .Lend_@: +#endif .endm

/* @@ -323,6 +325,7 @@ For 32-bit we have the following convent * regs. Must be called after the last RET. */ .macro IBRS_EXIT save_reg +#ifdef CONFIG_CPU_IBRS_ENTRY ALTERNATIVE "jmp .Lend_@", "", X86_FEATURE_KERNEL_IBRS movl $MSR_IA32_SPEC_CTRL, %ecx

@@ -337,6 +340,7 @@ For 32-bit we have the following convent shr $32, %rdx wrmsr .Lend_@: +#endif .endm

/* --- a/arch/x86/include/asm/disabled-features.h +++ b/arch/x86/include/asm/disabled-features.h @@ -60,9 +60,19 @@ # define DISABLE_RETPOLINE 0 #else # define DISABLE_RETPOLINE ((1 << (X86_FEATURE_RETPOLINE & 31)) | \ - (1 << (X86_FEATURE_RETPOLINE_LFENCE & 31)) | \ - (1 << (X86_FEATURE_RETHUNK & 31)) | \ - (1 << (X86_FEATURE_UNRET & 31))) + (1 << (X86_FEATURE_RETPOLINE_LFENCE & 31))) +#endif + +#ifdef CONFIG_RETHUNK +# define DISABLE_RETHUNK 0 +#else +# define DISABLE_RETHUNK (1 << (X86_FEATURE_RETHUNK & 31)) +#endif + +#ifdef CONFIG_CPU_UNRET_ENTRY +# define DISABLE_UNRET 0 +#else +# define DISABLE_UNRET (1 << (X86_FEATURE_UNRET & 31)) #endif

#ifdef CONFIG_INTEL_IOMMU_SVM @@ -91,7 +101,7 @@ #define DISABLED_MASK8 0 #define DISABLED_MASK9 (DISABLE_SMAP|DISABLE_SGX) #define DISABLED_MASK10 0 -#define DISABLED_MASK11 (DISABLE_RETPOLINE) +#define DISABLED_MASK11 (DISABLE_RETPOLINE|DISABLE_RETHUNK|DISABLE_UNRET) #define DISABLED_MASK12 0 #define DISABLED_MASK13 0 #define DISABLED_MASK14 0 --- a/arch/x86/include/asm/linkage.h +++ b/arch/x86/include/asm/linkage.h @@ -19,7 +19,7 @@ #define __ALIGN_STR __stringify(__ALIGN) #endif

-#if defined(CONFIG_RETPOLINE) && !defined(__DISABLE_EXPORTS) && !defined(BUILD_VDSO) +#if defined(CONFIG_RETHUNK) && !defined(__DISABLE_EXPORTS) && !defined(BUILD_VDSO) #define RET jmp __x86_return_thunk #else /* CONFIG_RETPOLINE */ #ifdef CONFIG_SLS @@ -31,7 +31,7 @@

#else /* __ASSEMBLY__ */

-#if defined(CONFIG_RETPOLINE) && !defined(__DISABLE_EXPORTS) && !defined(BUILD_VDSO) +#if defined(CONFIG_RETHUNK) && !defined(__DISABLE_EXPORTS) && !defined(BUILD_VDSO) #define ASM_RET "jmp __x86_return_thunk\n\t" #else /* CONFIG_RETPOLINE */ #ifdef CONFIG_SLS --- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -127,6 +127,12 @@ .Lskip_rsb_@: .endm

+#ifdef CONFIG_CPU_UNRET_ENTRY +#define CALL_ZEN_UNTRAIN_RET "call zen_untrain_ret" +#else +#define CALL_ZEN_UNTRAIN_RET "" +#endif + /* * Mitigate RETBleed for AMD/Hygon Zen uarch. Requires KERNEL CR3 because the * return thunk isn't mapped into the userspace tables (then again, AMD @@ -139,10 +145,10 @@ * where we have a stack but before any RET instruction. */ .macro UNTRAIN_RET -#ifdef CONFIG_RETPOLINE +#if defined(CONFIG_CPU_UNRET_ENTRY) || defined(CONFIG_CPU_IBPB_ENTRY) ANNOTATE_UNRET_END ALTERNATIVE_2 "", \ - "call zen_untrain_ret", X86_FEATURE_UNRET, \ + CALL_ZEN_UNTRAIN_RET, X86_FEATURE_UNRET, \ "call entry_ibpb", X86_FEATURE_ENTRY_IBPB #endif .endm --- a/arch/x86/include/asm/static_call.h +++ b/arch/x86/include/asm/static_call.h @@ -46,7 +46,7 @@ #define ARCH_DEFINE_STATIC_CALL_TRAMP(name, func) \ __ARCH_DEFINE_STATIC_CALL_TRAMP(name, ".byte 0xe9; .long " #func " - (. + 4)")

-#ifdef CONFIG_RETPOLINE +#ifdef CONFIG_RETHUNK #define ARCH_DEFINE_STATIC_CALL_NULL_TRAMP(name) \ __ARCH_DEFINE_STATIC_CALL_TRAMP(name, "jmp __x86_return_thunk") #else --- a/arch/x86/kernel/alternative.c +++ b/arch/x86/kernel/alternative.c @@ -508,6 +508,7 @@ void __init_or_module noinline apply_ret } }

+#ifdef CONFIG_RETHUNK /* * Rewrite the compiler generated return thunk tail-calls. * @@ -569,6 +570,10 @@ void __init_or_module noinline apply_ret } } } +#else +void __init_or_module noinline apply_returns(s32 *start, s32 *end) { } +#endif /* CONFIG_RETHUNK */ + #else /* !RETPOLINES || !CONFIG_STACK_VALIDATION */

void __init_or_module noinline apply_retpolines(s32 *start, s32 *end) { } --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -864,6 +864,7 @@ static void init_amd_bd(struct cpuinfo_x

void init_spectral_chicken(struct cpuinfo_x86 *c) { +#ifdef CONFIG_CPU_UNRET_ENTRY u64 value;

/* @@ -880,6 +881,7 @@ void init_spectral_chicken(struct cpuinf wrmsrl_safe(MSR_ZEN2_SPECTRAL_CHICKEN, value); } } +#endif }

static void init_amd_zn(struct cpuinfo_x86 *c) --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -835,7 +835,6 @@ static int __init retbleed_parse_cmdline early_param("retbleed", retbleed_parse_cmdline);

#define RETBLEED_UNTRAIN_MSG "WARNING: BTB untrained return thunk mitigation is only effective on AMD/Hygon!\n" -#define RETBLEED_COMPILER_MSG "WARNING: kernel not compiled with RETPOLINE or -mfunction-return capable compiler; falling back to IBPB!\n" #define RETBLEED_INTEL_MSG "WARNING: Spectre v2 mitigation leaves CPU vulnerable to RETBleed attacks, data leaks possible!\n"

static void __init retbleed_select_mitigation(void) @@ -850,18 +849,33 @@ static void __init retbleed_select_mitig return;

case RETBLEED_CMD_UNRET: - retbleed_mitigation = RETBLEED_MITIGATION_UNRET; + if (IS_ENABLED(CONFIG_CPU_UNRET_ENTRY)) { + retbleed_mitigation = RETBLEED_MITIGATION_UNRET; + } else { + pr_err("WARNING: kernel not compiled with CPU_UNRET_ENTRY.\n"); + goto do_cmd_auto; + } break;

case RETBLEED_CMD_IBPB: - retbleed_mitigation = RETBLEED_MITIGATION_IBPB; + if (IS_ENABLED(CONFIG_CPU_IBPB_ENTRY)) { + retbleed_mitigation = RETBLEED_MITIGATION_IBPB; + } else { + pr_err("WARNING: kernel not compiled with CPU_IBPB_ENTRY.\n"); + goto do_cmd_auto; + } break;

+do_cmd_auto: case RETBLEED_CMD_AUTO: default: if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD || - boot_cpu_data.x86_vendor == X86_VENDOR_HYGON) - retbleed_mitigation = RETBLEED_MITIGATION_UNRET; + boot_cpu_data.x86_vendor == X86_VENDOR_HYGON) { + if (IS_ENABLED(CONFIG_CPU_UNRET_ENTRY)) + retbleed_mitigation = RETBLEED_MITIGATION_UNRET; + else if (IS_ENABLED(CONFIG_CPU_IBPB_ENTRY)) + retbleed_mitigation = RETBLEED_MITIGATION_IBPB; + }

/* * The Intel mitigation (IBRS or eIBRS) was already selected in @@ -874,14 +888,6 @@ static void __init retbleed_select_mitig

switch (retbleed_mitigation) { case RETBLEED_MITIGATION_UNRET: - - if (!IS_ENABLED(CONFIG_RETPOLINE) || - !IS_ENABLED(CONFIG_CC_HAS_RETURN_THUNK)) { - pr_err(RETBLEED_COMPILER_MSG); - retbleed_mitigation = RETBLEED_MITIGATION_IBPB; - goto retbleed_force_ibpb; - } - setup_force_cpu_cap(X86_FEATURE_RETHUNK); setup_force_cpu_cap(X86_FEATURE_UNRET);

@@ -893,7 +899,6 @@ static void __init retbleed_select_mitig break;

case RETBLEED_MITIGATION_IBPB: -retbleed_force_ibpb: setup_force_cpu_cap(X86_FEATURE_ENTRY_IBPB); mitigate_smt = true; break; @@ -1264,6 +1269,12 @@ static enum spectre_v2_mitigation_cmd __ return SPECTRE_V2_CMD_AUTO; }

+ if (cmd == SPECTRE_V2_CMD_IBRS && !IS_ENABLED(CONFIG_CPU_IBRS_ENTRY)) { + pr_err("%s selected but not compiled in. Switching to AUTO select\n", + mitigation_options[i].option); + return SPECTRE_V2_CMD_AUTO; + } + if (cmd == SPECTRE_V2_CMD_IBRS && boot_cpu_data.x86_vendor != X86_VENDOR_INTEL) { pr_err("%s selected but not Intel CPU. Switching to AUTO select\n", mitigation_options[i].option); @@ -1321,7 +1332,8 @@ static void __init spectre_v2_select_mit break; }

- if (boot_cpu_has_bug(X86_BUG_RETBLEED) && + if (IS_ENABLED(CONFIG_CPU_IBRS_ENTRY) && + boot_cpu_has_bug(X86_BUG_RETBLEED) && retbleed_cmd != RETBLEED_CMD_OFF && boot_cpu_has(X86_FEATURE_IBRS) && boot_cpu_data.x86_vendor == X86_VENDOR_INTEL) { --- a/arch/x86/kernel/static_call.c +++ b/arch/x86/kernel/static_call.c @@ -126,7 +126,7 @@ void arch_static_call_transform(void *si } EXPORT_SYMBOL_GPL(arch_static_call_transform);

-#ifdef CONFIG_RETPOLINE +#ifdef CONFIG_RETHUNK /* * This is called by apply_returns() to fix up static call trampolines, * specifically ARCH_DEFINE_STATIC_CALL_NULL_TRAMP which is recorded as --- a/arch/x86/kvm/emulate.c +++ b/arch/x86/kvm/emulate.c @@ -439,10 +439,10 @@ static int fastop(struct x86_emulate_ctx * * ENDBR [4 bytes; CONFIG_X86_KERNEL_IBT] * SETcc %al [3 bytes] - * RET | JMP __x86_return_thunk [1,5 bytes; CONFIG_RETPOLINE] + * RET | JMP __x86_return_thunk [1,5 bytes; CONFIG_RETHUNK] * INT3 [1 byte; CONFIG_SLS] */ -#define RET_LENGTH (1 + (4 * IS_ENABLED(CONFIG_RETPOLINE)) + \ +#define RET_LENGTH (1 + (4 * IS_ENABLED(CONFIG_RETHUNK)) + \ IS_ENABLED(CONFIG_SLS)) #define SETCC_LENGTH (ENDBR_INSN_SIZE + 3 + RET_LENGTH) #define SETCC_ALIGN (4 << ((SETCC_LENGTH > 4) & 1) << ((SETCC_LENGTH > 8) & 1)) --- a/arch/x86/lib/retpoline.S +++ b/arch/x86/lib/retpoline.S @@ -72,6 +72,8 @@ SYM_CODE_END(__x86_indirect_thunk_array) * This function name is magical and is used by -mfunction-return=thunk-extern * for the compiler to generate JMPs to it. */ +#ifdef CONFIG_RETHUNK + .section .text.__x86.return_thunk

/* @@ -136,3 +138,5 @@ SYM_FUNC_END(zen_untrain_ret) __EXPORT_THUNK(zen_untrain_ret)

EXPORT_SYMBOL(__x86_return_thunk) + +#endif /* CONFIG_RETHUNK */ --- a/scripts/Makefile.build +++ b/scripts/Makefile.build @@ -233,6 +233,7 @@ objtool_args = \ $(if $(CONFIG_FRAME_POINTER),, --no-fp) \ $(if $(CONFIG_GCOV_KERNEL), --no-unreachable) \ $(if $(CONFIG_RETPOLINE), --retpoline) \ + $(if $(CONFIG_RETHUNK), --rethunk) \ $(if $(CONFIG_X86_SMAP), --uaccess) \ $(if $(CONFIG_FTRACE_MCOUNT_USE_OBJTOOL), --mcount) \ $(if $(CONFIG_SLS), --sls) --- a/scripts/link-vmlinux.sh +++ b/scripts/link-vmlinux.sh @@ -130,7 +130,7 @@ objtool_link()

if is_enabled CONFIG_VMLINUX_VALIDATION; then objtoolopt="${objtoolopt} --noinstr" - if is_enabled CONFIG_RETPOLINE; then + if is_enabled CONFIG_CPU_UNRET_ENTRY; then objtoolopt="${objtoolopt} --unret" fi fi --- a/security/Kconfig +++ b/security/Kconfig @@ -54,17 +54,6 @@ config SECURITY_NETWORK implement socket and networking access controls. If you are unsure how to answer this question, answer N.

-config PAGE_TABLE_ISOLATION - bool "Remove the kernel mapping in user mode" - default y - depends on (X86_64 || X86_PAE) && !UML - help - This feature reduces the number of hardware side channels by - ensuring that the majority of kernel addresses are not mapped - into userspace. - - See Documentation/x86/pti.rst for more details. - config SECURITY_INFINIBAND bool "Infiniband Security Hooks" depends on SECURITY && INFINIBAND --- a/tools/objtool/builtin-check.c +++ b/tools/objtool/builtin-check.c @@ -21,7 +21,7 @@

bool no_fp, no_unreachable, retpoline, module, backtrace, uaccess, stats, lto, vmlinux, mcount, noinstr, backup, sls, dryrun, - ibt, unret; + ibt, unret, rethunk;

static const char * const check_usage[] = { "objtool check [<options>] file.o", @@ -37,6 +37,7 @@ const struct option check_options[] = { OPT_BOOLEAN('f', "no-fp", &no_fp, "Skip frame pointer validation"), OPT_BOOLEAN('u', "no-unreachable", &no_unreachable, "Skip 'unreachable instruction' warnings"), OPT_BOOLEAN('r', "retpoline", &retpoline, "Validate retpoline assumptions"), + OPT_BOOLEAN(0, "rethunk", &rethunk, "validate and annotate rethunk usage"), OPT_BOOLEAN(0, "unret", &unret, "validate entry unret placement"), OPT_BOOLEAN('m', "module", &module, "Indicates the object will be part of a kernel module"), OPT_BOOLEAN('b', "backtrace", &backtrace, "unwind on error"), --- a/tools/objtool/check.c +++ b/tools/objtool/check.c @@ -3854,8 +3854,11 @@ static int validate_retpoline(struct obj continue;

if (insn->type == INSN_RETURN) { - WARN_FUNC("'naked' return found in RETPOLINE build", - insn->sec, insn->offset); + if (rethunk) { + WARN_FUNC("'naked' return found in RETHUNK build", + insn->sec, insn->offset); + } else + continue; } else { WARN_FUNC("indirect %s found in RETPOLINE build", insn->sec, insn->offset, @@ -4228,7 +4231,9 @@ int check(struct objtool_file *file) if (ret < 0) goto out; warnings += ret; + }

+ if (rethunk) { ret = create_return_sites_sections(file); if (ret < 0) goto out; --- a/tools/objtool/include/objtool/builtin.h +++ b/tools/objtool/include/objtool/builtin.h @@ -10,7 +10,7 @@ extern const struct option check_options[]; extern bool no_fp, no_unreachable, retpoline, module, backtrace, uaccess, stats, lto, vmlinux, mcount, noinstr, backup, sls, dryrun, - ibt, unret; + ibt, unret, rethunk;

extern int cmd_parse_options(int argc, const char **argv, const char * const usage[]);

From: Peter Zijlstra peterz@infradead.org

commit 2c08b9b38f5b0f4a6c2d29be22b695e4ec4a556b upstream.

Commit

ee774dac0da1 ("x86/entry: Move PUSH_AND_CLEAR_REGS out of error_entry()")

moved PUSH_AND_CLEAR_REGS out of error_entry, into its own function, in part to avoid calling error_entry() for XenPV.

However, commit

7c81c0c9210c ("x86/entry: Avoid very early RET")

had to change that because the 'ret' was too early and moved it into idtentry, bloating the text size, since idtentry is expanded for every exception vector.

However, with the advent of xen_error_entry() in commit

d147553b64bad ("x86/xen: Add UNTRAIN_RET")

it became possible to remove PUSH_AND_CLEAR_REGS from idtentry, back into *error_entry().

Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Borislav Petkov bp@suse.de [cascardo: error_entry still does cld] Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/entry/entry_64.S | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)

--- a/arch/x86/entry/entry_64.S +++ b/arch/x86/entry/entry_64.S @@ -323,6 +323,8 @@ SYM_CODE_END(ret_from_fork)

SYM_CODE_START_LOCAL(xen_error_entry) UNWIND_HINT_FUNC + PUSH_AND_CLEAR_REGS save_ret=1 + ENCODE_FRAME_POINTER 8 UNTRAIN_RET RET SYM_CODE_END(xen_error_entry) @@ -334,9 +336,6 @@ SYM_CODE_END(xen_error_entry) */ .macro idtentry_body cfunc has_error_code:req

- PUSH_AND_CLEAR_REGS - ENCODE_FRAME_POINTER - /* * Call error_entry() and switch to the task stack if from userspace. * @@ -1035,6 +1034,10 @@ SYM_CODE_END(paranoid_exit) SYM_CODE_START_LOCAL(error_entry) UNWIND_HINT_FUNC cld + + PUSH_AND_CLEAR_REGS save_ret=1 + ENCODE_FRAME_POINTER 8 + testb $3, CS+8(%rsp) jz .Lerror_kernelspace

From: Konrad Rzeszutek Wilk konrad.wilk@oracle.com

commit 697977d8415d61f3acbc4ee6d564c9dcf0309507 upstream.

All the invocations unroll to __x86_return_thunk and this file must be PIC independent.

This fixes kexec on 64-bit AMD boxes.

[ bp: Fix 32-bit build. ]

Reported-by: Edward Tran edward.tran@oracle.com Reported-by: Awais Tanveer awais.tanveer@oracle.com Suggested-by: Ankur Arora ankur.a.arora@oracle.com Signed-off-by: Konrad Rzeszutek Wilk konrad.wilk@oracle.com Signed-off-by: Alexandre Chartre alexandre.chartre@oracle.com Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kernel/relocate_kernel_32.S | 25 +++++++++++++++++++------ arch/x86/kernel/relocate_kernel_64.S | 23 +++++++++++++++++------ 2 files changed, 36 insertions(+), 12 deletions(-)

--- a/arch/x86/kernel/relocate_kernel_32.S +++ b/arch/x86/kernel/relocate_kernel_32.S @@ -7,10 +7,12 @@ #include <linux/linkage.h> #include <asm/page_types.h> #include <asm/kexec.h> +#include <asm/nospec-branch.h> #include <asm/processor-flags.h>

/* - * Must be relocatable PIC code callable as a C function + * Must be relocatable PIC code callable as a C function, in particular + * there must be a plain RET and not jump to return thunk. */

#define PTR(x) (x << 2) @@ -91,7 +93,9 @@ SYM_CODE_START_NOALIGN(relocate_kernel) movl %edi, %eax addl $(identity_mapped - relocate_kernel), %eax pushl %eax - RET + ANNOTATE_UNRET_SAFE + ret + int3 SYM_CODE_END(relocate_kernel)

SYM_CODE_START_LOCAL_NOALIGN(identity_mapped) @@ -159,12 +163,15 @@ SYM_CODE_START_LOCAL_NOALIGN(identity_ma xorl %edx, %edx xorl %esi, %esi xorl %ebp, %ebp - RET + ANNOTATE_UNRET_SAFE + ret + int3 1: popl %edx movl CP_PA_SWAP_PAGE(%edi), %esp addl $PAGE_SIZE, %esp 2: + ANNOTATE_RETPOLINE_SAFE call *%edx

/* get the re-entry point of the peer system */ @@ -190,7 +197,9 @@ SYM_CODE_START_LOCAL_NOALIGN(identity_ma movl %edi, %eax addl $(virtual_mapped - relocate_kernel), %eax pushl %eax - RET + ANNOTATE_UNRET_SAFE + ret + int3 SYM_CODE_END(identity_mapped)

SYM_CODE_START_LOCAL_NOALIGN(virtual_mapped) @@ -208,7 +217,9 @@ SYM_CODE_START_LOCAL_NOALIGN(virtual_map popl %edi popl %esi popl %ebx - RET + ANNOTATE_UNRET_SAFE + ret + int3 SYM_CODE_END(virtual_mapped)

/* Do the copies */ @@ -271,7 +282,9 @@ SYM_CODE_START_LOCAL_NOALIGN(swap_pages) popl %edi popl %ebx popl %ebp - RET + ANNOTATE_UNRET_SAFE + ret + int3 SYM_CODE_END(swap_pages)

.globl kexec_control_code_size --- a/arch/x86/kernel/relocate_kernel_64.S +++ b/arch/x86/kernel/relocate_kernel_64.S @@ -13,7 +13,8 @@ #include <asm/unwind_hints.h>

/* - * Must be relocatable PIC code callable as a C function + * Must be relocatable PIC code callable as a C function, in particular + * there must be a plain RET and not jump to return thunk. */

#define PTR(x) (x << 3) @@ -105,7 +106,9 @@ SYM_CODE_START_NOALIGN(relocate_kernel) /* jump to identity mapped page */ addq $(identity_mapped - relocate_kernel), %r8 pushq %r8 - RET + ANNOTATE_UNRET_SAFE + ret + int3 SYM_CODE_END(relocate_kernel)

SYM_CODE_START_LOCAL_NOALIGN(identity_mapped) @@ -200,7 +203,9 @@ SYM_CODE_START_LOCAL_NOALIGN(identity_ma xorl %r14d, %r14d xorl %r15d, %r15d

- RET + ANNOTATE_UNRET_SAFE + ret + int3

1: popq %rdx @@ -219,7 +224,9 @@ SYM_CODE_START_LOCAL_NOALIGN(identity_ma call swap_pages movq $virtual_mapped, %rax pushq %rax - RET + ANNOTATE_UNRET_SAFE + ret + int3 SYM_CODE_END(identity_mapped)

SYM_CODE_START_LOCAL_NOALIGN(virtual_mapped) @@ -241,7 +248,9 @@ SYM_CODE_START_LOCAL_NOALIGN(virtual_map popq %r12 popq %rbp popq %rbx - RET + ANNOTATE_UNRET_SAFE + ret + int3 SYM_CODE_END(virtual_mapped)

/* Do the copies */ @@ -298,7 +307,9 @@ SYM_CODE_START_LOCAL_NOALIGN(swap_pages) lea PAGE_SIZE(%rax), %rsi jmp 0b 3: - RET + ANNOTATE_UNRET_SAFE + ret + int3 SYM_CODE_END(swap_pages)

.globl kexec_control_code_size

From: Thomas Gleixner tglx@linutronix.de

commit c27c753ea6fd1237f4f96abf8b623d7bab505513 upstream.

__static_call_fixup() invokes __static_call_transform() without holding text_mutex, which causes lockdep to complain in text_poke_bp().

Adding the proper locking cures that, but as this is either used during early boot or during module finalizing, it's not required to use text_poke_bp(). Add an argument to __static_call_transform() which tells it to use text_poke_early() for it.

Fixes: ee88d363d156 ("x86,static_call: Use alternative RET encoding") Signed-off-by: Thomas Gleixner tglx@linutronix.de Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kernel/static_call.c | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)

--- a/arch/x86/kernel/static_call.c +++ b/arch/x86/kernel/static_call.c @@ -25,7 +25,8 @@ static const u8 xor5rax[] = { 0x2e, 0x2e

static const u8 retinsn[] = { RET_INSN_OPCODE, 0xcc, 0xcc, 0xcc, 0xcc };

-static void __ref __static_call_transform(void *insn, enum insn_type type, void *func) +static void __ref __static_call_transform(void *insn, enum insn_type type, + void *func, bool modinit) { const void *emulate = NULL; int size = CALL_INSN_SIZE; @@ -60,7 +61,7 @@ static void __ref __static_call_transfor if (memcmp(insn, code, size) == 0) return;

- if (unlikely(system_state == SYSTEM_BOOTING)) + if (system_state == SYSTEM_BOOTING || modinit) return text_poke_early(insn, code, size);

text_poke_bp(insn, code, size, emulate); @@ -114,12 +115,12 @@ void arch_static_call_transform(void *si

if (tramp) { __static_call_validate(tramp, true, true); - __static_call_transform(tramp, __sc_insn(!func, true), func); + __static_call_transform(tramp, __sc_insn(!func, true), func, false); }

if (IS_ENABLED(CONFIG_HAVE_STATIC_CALL_INLINE) && site) { __static_call_validate(site, tail, false); - __static_call_transform(site, __sc_insn(!func, tail), func); + __static_call_transform(site, __sc_insn(!func, tail), func, false); }

mutex_unlock(&text_mutex); @@ -145,8 +146,10 @@ bool __static_call_fixup(void *tramp, u8 return false; }

+ mutex_lock(&text_mutex); if (op == RET_INSN_OPCODE || dest == &__x86_return_thunk) - __static_call_transform(tramp, RET, NULL); + __static_call_transform(tramp, RET, NULL, true); + mutex_unlock(&text_mutex);

return true; }

From: Nathan Chancellor nathan@kernel.org

commit db886979683a8360ced9b24ab1125ad0c4d2cf76 upstream.

Clang warns:

arch/x86/kernel/cpu/bugs.c:58:21: error: section attribute is specified on redeclared variable [-Werror,-Wsection] DEFINE_PER_CPU(u64, x86_spec_ctrl_current); ^ arch/x86/include/asm/nospec-branch.h:283:12: note: previous declaration is here extern u64 x86_spec_ctrl_current; ^ 1 error generated.

The declaration should be using DECLARE_PER_CPU instead so all attributes stay in sync.

Cc: stable@vger.kernel.org Fixes: fc02735b14ff ("KVM: VMX: Prevent guest RSB poisoning attacks with eIBRS") Reported-by: kernel test robot lkp@intel.com Signed-off-by: Nathan Chancellor nathan@kernel.org Signed-off-by: Linus Torvalds torvalds@linux-foundation.org Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/include/asm/nospec-branch.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)

--- a/arch/x86/include/asm/nospec-branch.h +++ b/arch/x86/include/asm/nospec-branch.h @@ -11,6 +11,7 @@ #include <asm/cpufeatures.h> #include <asm/msr-index.h> #include <asm/unwind_hints.h> +#include <asm/percpu.h>

#define RETPOLINE_THUNK_SIZE 32

@@ -280,7 +281,7 @@ static inline void indirect_branch_predi

/* The Intel SPEC CTRL MSR base value cache */ extern u64 x86_spec_ctrl_base; -extern u64 x86_spec_ctrl_current; +DECLARE_PER_CPU(u64, x86_spec_ctrl_current); extern void write_spec_ctrl_current(u64 val, bool force); extern u64 spec_ctrl_current(void);

From: Thadeu Lima de Souza Cascardo cascardo@canonical.com

commit 84e7051c0bc1f2a13101553959b3a9d9a8e24939 upstream.

The return thunk call makes the fastop functions larger, just like IBT does. Consider a 16-byte FASTOP_SIZE when CONFIG_RETHUNK is enabled.

Otherwise, functions will be incorrectly aligned and when computing their position for differently sized operators, they will executed in the middle or end of a function, which may as well be an int3, leading to a crash like:

[ 36.091116] int3: 0000 [#1] SMP NOPTI [ 36.091119] CPU: 3 PID: 1371 Comm: qemu-system-x86 Not tainted 5.15.0-41-generic #44 [ 36.091120] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.15.0-1 04/01/2014 [ 36.091121] RIP: 0010:xaddw_ax_dx+0x9/0x10 [kvm] [ 36.091185] Code: 00 0f bb d0 c3 cc cc cc cc 48 0f bb d0 c3 cc cc cc cc 0f 1f 80 00 00 00 00 0f c0 d0 c3 cc cc cc cc 66 0f c1 d0 c3 cc cc cc cc <0f> 1f 80 00 00 00 00 0f c1 d0 c3 cc cc cc cc 48 0f c1 d0 c3 cc cc [ 36.091186] RSP: 0018:ffffb1f541143c98 EFLAGS: 00000202 [ 36.091188] RAX: 0000000089abcdef RBX: 0000000000000001 RCX: 0000000000000000 [ 36.091188] RDX: 0000000076543210 RSI: ffffffffc073c6d0 RDI: 0000000000000200 [ 36.091189] RBP: ffffb1f541143ca0 R08: ffff9f1803350a70 R09: 0000000000000002 [ 36.091190] R10: ffff9f1803350a70 R11: 0000000000000000 R12: ffff9f1803350a70 [ 36.091190] R13: ffffffffc077fee0 R14: 0000000000000000 R15: 0000000000000000 [ 36.091191] FS: 00007efdfce8d640(0000) GS:ffff9f187dd80000(0000) knlGS:0000000000000000 [ 36.091192] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 36.091192] CR2: 0000000000000000 CR3: 0000000009b62002 CR4: 0000000000772ee0 [ 36.091195] PKRU: 55555554 [ 36.091195] Call Trace: [ 36.091197] <TASK> [ 36.091198] ? fastop+0x5a/0xa0 [kvm] [ 36.091222] x86_emulate_insn+0x7b8/0xe90 [kvm] [ 36.091244] x86_emulate_instruction+0x2f4/0x630 [kvm] [ 36.091263] ? kvm_arch_vcpu_load+0x7c/0x230 [kvm] [ 36.091283] ? vmx_prepare_switch_to_host+0xf7/0x190 [kvm_intel] [ 36.091290] complete_emulated_mmio+0x297/0x320 [kvm] [ 36.091310] kvm_arch_vcpu_ioctl_run+0x32f/0x550 [kvm] [ 36.091330] kvm_vcpu_ioctl+0x29e/0x6d0 [kvm] [ 36.091344] ? kvm_vcpu_ioctl+0x120/0x6d0 [kvm] [ 36.091357] ? __fget_files+0x86/0xc0 [ 36.091362] ? __fget_files+0x86/0xc0 [ 36.091363] __x64_sys_ioctl+0x92/0xd0 [ 36.091366] do_syscall_64+0x59/0xc0 [ 36.091369] ? syscall_exit_to_user_mode+0x27/0x50 [ 36.091370] ? do_syscall_64+0x69/0xc0 [ 36.091371] ? syscall_exit_to_user_mode+0x27/0x50 [ 36.091372] ? __x64_sys_writev+0x1c/0x30 [ 36.091374] ? do_syscall_64+0x69/0xc0 [ 36.091374] ? exit_to_user_mode_prepare+0x37/0xb0 [ 36.091378] ? syscall_exit_to_user_mode+0x27/0x50 [ 36.091379] ? do_syscall_64+0x69/0xc0 [ 36.091379] ? do_syscall_64+0x69/0xc0 [ 36.091380] ? do_syscall_64+0x69/0xc0 [ 36.091381] ? do_syscall_64+0x69/0xc0 [ 36.091381] entry_SYSCALL_64_after_hwframe+0x61/0xcb [ 36.091384] RIP: 0033:0x7efdfe6d1aff [ 36.091390] Code: 00 48 89 44 24 18 31 c0 48 8d 44 24 60 c7 04 24 10 00 00 00 48 89 44 24 08 48 8d 44 24 20 48 89 44 24 10 b8 10 00 00 00 0f 05 <41> 89 c0 3d 00 f0 ff ff 77 1f 48 8b 44 24 18 64 48 2b 04 25 28 00 [ 36.091391] RSP: 002b:00007efdfce8c460 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 [ 36.091393] RAX: ffffffffffffffda RBX: 000000000000ae80 RCX: 00007efdfe6d1aff [ 36.091393] RDX: 0000000000000000 RSI: 000000000000ae80 RDI: 000000000000000c [ 36.091394] RBP: 0000558f1609e220 R08: 0000558f13fb8190 R09: 00000000ffffffff [ 36.091394] R10: 0000558f16b5e950 R11: 0000000000000246 R12: 0000000000000000 [ 36.091394] R13: 0000000000000001 R14: 0000000000000000 R15: 0000000000000000 [ 36.091396] </TASK> [ 36.091397] Modules linked in: isofs nls_iso8859_1 kvm_intel joydev kvm input_leds serio_raw sch_fq_codel dm_multipath scsi_dh_rdac scsi_dh_emc scsi_dh_alua ipmi_devintf ipmi_msghandler drm msr ip_tables x_tables autofs4 btrfs blake2b_generic zstd_compress raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c raid1 raid0 multipath linear crct10dif_pclmul crc32_pclmul ghash_clmulni_intel aesni_intel virtio_net net_failover crypto_simd ahci xhci_pci cryptd psmouse virtio_blk libahci xhci_pci_renesas failover [ 36.123271] ---[ end trace db3c0ab5a48fabcc ]--- [ 36.123272] RIP: 0010:xaddw_ax_dx+0x9/0x10 [kvm] [ 36.123319] Code: 00 0f bb d0 c3 cc cc cc cc 48 0f bb d0 c3 cc cc cc cc 0f 1f 80 00 00 00 00 0f c0 d0 c3 cc cc cc cc 66 0f c1 d0 c3 cc cc cc cc <0f> 1f 80 00 00 00 00 0f c1 d0 c3 cc cc cc cc 48 0f c1 d0 c3 cc cc [ 36.123320] RSP: 0018:ffffb1f541143c98 EFLAGS: 00000202 [ 36.123321] RAX: 0000000089abcdef RBX: 0000000000000001 RCX: 0000000000000000 [ 36.123321] RDX: 0000000076543210 RSI: ffffffffc073c6d0 RDI: 0000000000000200 [ 36.123322] RBP: ffffb1f541143ca0 R08: ffff9f1803350a70 R09: 0000000000000002 [ 36.123322] R10: ffff9f1803350a70 R11: 0000000000000000 R12: ffff9f1803350a70 [ 36.123323] R13: ffffffffc077fee0 R14: 0000000000000000 R15: 0000000000000000 [ 36.123323] FS: 00007efdfce8d640(0000) GS:ffff9f187dd80000(0000) knlGS:0000000000000000 [ 36.123324] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 36.123325] CR2: 0000000000000000 CR3: 0000000009b62002 CR4: 0000000000772ee0 [ 36.123327] PKRU: 55555554 [ 36.123328] Kernel panic - not syncing: Fatal exception in interrupt [ 36.123410] Kernel Offset: 0x1400000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff) [ 36.135305] ---[ end Kernel panic - not syncing: Fatal exception in interrupt ]---

Fixes: aa3d480315ba ("x86: Use return-thunk in asm code") Signed-off-by: Thadeu Lima de Souza Cascardo cascardo@canonical.com Co-developed-by: Peter Zijlstra (Intel) peterz@infradead.org Cc: Borislav Petkov bp@suse.de Cc: Josh Poimboeuf jpoimboe@kernel.org Cc: Paolo Bonzini pbonzini@redhat.com Reported-by: Linux Kernel Functional Testing lkft@linaro.org Message-Id: 20220713171241.184026-1-cascardo@canonical.com Tested-by: Jack Wang jinpu.wang@ionos.com Signed-off-by: Paolo Bonzini pbonzini@redhat.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kvm/emulate.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)

--- a/arch/x86/kvm/emulate.c +++ b/arch/x86/kvm/emulate.c @@ -189,8 +189,12 @@ #define X8(x...) X4(x), X4(x) #define X16(x...) X8(x), X8(x)

-#define NR_FASTOP (ilog2(sizeof(ulong)) + 1) -#define FASTOP_SIZE (8 * (1 + HAS_KERNEL_IBT)) +#define NR_FASTOP (ilog2(sizeof(ulong)) + 1) +#define RET_LENGTH (1 + (4 * IS_ENABLED(CONFIG_RETHUNK)) + \ + IS_ENABLED(CONFIG_SLS)) +#define FASTOP_LENGTH (ENDBR_INSN_SIZE + 7 + RET_LENGTH) +#define FASTOP_SIZE (8 << ((FASTOP_LENGTH > 8) & 1) << ((FASTOP_LENGTH > 16) & 1)) +static_assert(FASTOP_LENGTH <= FASTOP_SIZE);

struct opcode { u64 flags; @@ -442,8 +446,6 @@ static int fastop(struct x86_emulate_ctx * RET | JMP __x86_return_thunk [1,5 bytes; CONFIG_RETHUNK] * INT3 [1 byte; CONFIG_SLS] */ -#define RET_LENGTH (1 + (4 * IS_ENABLED(CONFIG_RETHUNK)) + \ - IS_ENABLED(CONFIG_SLS)) #define SETCC_LENGTH (ENDBR_INSN_SIZE + 3 + RET_LENGTH) #define SETCC_ALIGN (4 << ((SETCC_LENGTH > 4) & 1) << ((SETCC_LENGTH > 8) & 1)) static_assert(SETCC_LENGTH <= SETCC_ALIGN);

From: Arnaldo Carvalho de Melo acme@redhat.com

commit 91d248c3b903b46a58cbc7e8d38d684d3e4007c2 upstream.

To pick up the changes from these csets:

4ad3278df6fe2b08 ("x86/speculation: Disable RRSBA behavior") d7caac991feeef1b ("x86/cpu/amd: Add Spectral Chicken")

That cause no changes to tooling:

$ tools/perf/trace/beauty/tracepoints/x86_msr.sh > before $ cp arch/x86/include/asm/msr-index.h tools/arch/x86/include/asm/msr-index.h $ tools/perf/trace/beauty/tracepoints/x86_msr.sh > after $ diff -u before after $

Just silences this perf build warning:

Warning: Kernel ABI header at 'tools/arch/x86/include/asm/msr-index.h' differs from latest version at 'arch/x86/include/asm/msr-index.h' diff -u tools/arch/x86/include/asm/msr-index.h arch/x86/include/asm/msr-index.h

Cc: Adrian Hunter adrian.hunter@intel.com Cc: Borislav Petkov bp@suse.de Cc: Ian Rogers irogers@google.com Cc: Jiri Olsa jolsa@kernel.org Cc: Namhyung Kim namhyung@kernel.org Cc: Pawan Gupta pawan.kumar.gupta@linux.intel.com Cc: Peter Zijlstra peterz@infradead.org Link: https://lore.kernel.org/lkml/YtQTm9wsB3hxQWvy@kernel.org Signed-off-by: Arnaldo Carvalho de Melo acme@redhat.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- tools/arch/x86/include/asm/msr-index.h | 4 ++++ 1 file changed, 4 insertions(+)

--- a/tools/arch/x86/include/asm/msr-index.h +++ b/tools/arch/x86/include/asm/msr-index.h @@ -93,6 +93,7 @@ #define MSR_IA32_ARCH_CAPABILITIES 0x0000010a #define ARCH_CAP_RDCL_NO BIT(0) /* Not susceptible to Meltdown */ #define ARCH_CAP_IBRS_ALL BIT(1) /* Enhanced IBRS support */ +#define ARCH_CAP_RSBA BIT(2) /* RET may use alternative branch predictors */ #define ARCH_CAP_SKIP_VMENTRY_L1DFLUSH BIT(3) /* Skip L1D flush on vmentry */ #define ARCH_CAP_SSB_NO BIT(4) /* * Not susceptible to Speculative Store Bypass @@ -561,6 +562,9 @@ /* Fam 17h MSRs */ #define MSR_F17H_IRPERF 0xc00000e9

+#define MSR_ZEN2_SPECTRAL_CHICKEN 0xc00110e3 +#define MSR_ZEN2_SPECTRAL_CHICKEN_BIT BIT_ULL(1) + /* Fam 16h MSRs */ #define MSR_F16H_L2I_PERF_CTL 0xc0010230 #define MSR_F16H_L2I_PERF_CTR 0xc0010231

From: Kim Phillips kim.phillips@amd.com

commit bcf163150cd37348a0cb59e95c916a83a9344b0e upstream.

Remove a superfluous ' in the mitigation string.

Fixes: e8ec1b6e08a2 ("x86/bugs: Enable STIBP for JMP2RET") Signed-off-by: Kim Phillips kim.phillips@amd.com Signed-off-by: Borislav Petkov bp@suse.de Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- arch/x86/kernel/cpu/bugs.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)

--- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -1174,7 +1174,7 @@ spectre_v2_user_select_mitigation(void) if (retbleed_mitigation == RETBLEED_MITIGATION_UNRET) { if (mode != SPECTRE_V2_USER_STRICT && mode != SPECTRE_V2_USER_STRICT_PREFERRED) - pr_info("Selecting STIBP always-on mode to complement retbleed mitigation'\n"); + pr_info("Selecting STIBP always-on mode to complement retbleed mitigation\n"); mode = SPECTRE_V2_USER_STRICT_PREFERRED; }