Merge tag 'kvm-x86-guest-6.18' of https://github.com/kvm-x86/linux into HEAD

x86/kvm guest side changes for 6.18

 - For the legacy PCI hole (memory between TOLUD and 4GiB) to UC when
   overriding guest MTRR for TDX/SNP to fix an issue where ACPI auto-mapping
   could map devices as WB and prevent the device drivers from mapping their
   devices with UC/UC-.

 - Make kvm_async_pf_task_wake() a local static helper and remove its
   export.

 - Use native qspinlocks when running in a VM with dedicated vCPU=>pCPU
   bindings even when PV_UNHALT is unsupported.

2 files changed, 30 insertions, 16 deletions

diff --git a/arch/x86/include/asm/kvm_para.h b/arch/x86/include/asm/kvm_para.h
index 57bc74e112f2..4a47c16e2df8 100644
--- a/arch/x86/include/asm/kvm_para.h
+++ b/arch/x86/include/asm/kvm_para.h
@@ -124,7 +124,6 @@ bool kvm_para_available(void);
 unsigned int kvm_arch_para_features(void);
 unsigned int kvm_arch_para_hints(void);
 void kvm_async_pf_task_wait_schedule(u32 token);
-void kvm_async_pf_task_wake(u32 token);
 u32 kvm_read_and_reset_apf_flags(void);
 bool __kvm_handle_async_pf(struct pt_regs *regs, u32 token);
 
@@ -148,7 +147,6 @@ static inline void kvm_spinlock_init(void)
 
 #else /* CONFIG_KVM_GUEST */
 #define kvm_async_pf_task_wait_schedule(T) do {} while(0)
-#define kvm_async_pf_task_wake(T) do {} while(0)
 
 static inline bool kvm_para_available(void)
 {
diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index 8ae750cde0c6..b67d7c59dca0 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -190,7 +190,7 @@ static void apf_task_wake_all(void)
 	}
 }
 
-void kvm_async_pf_task_wake(u32 token)
+static void kvm_async_pf_task_wake(u32 token)
 {
 	u32 key = hash_32(token, KVM_TASK_SLEEP_HASHBITS);
 	struct kvm_task_sleep_head *b = &async_pf_sleepers[key];
@@ -241,7 +241,6 @@ again:
 	/* A dummy token might be allocated and ultimately not used.  */
 	kfree(dummy);
 }
-EXPORT_SYMBOL_GPL(kvm_async_pf_task_wake);
 
 noinstr u32 kvm_read_and_reset_apf_flags(void)
 {
@@ -933,6 +932,19 @@ static void kvm_sev_hc_page_enc_status(unsigned long pfn, int npages, bool enc)
 
 static void __init kvm_init_platform(void)
 {
+	u64 tolud = PFN_PHYS(e820__end_of_low_ram_pfn());
+	/*
+	 * Note, hardware requires variable MTRR ranges to be power-of-2 sized
+	 * and naturally aligned.  But when forcing guest MTRR state, Linux
+	 * doesn't program the forced ranges into hardware.  Don't bother doing
+	 * the math to generate a technically-legal range.
+	 */
+	struct mtrr_var_range pci_hole = {
+		.base_lo = tolud | X86_MEMTYPE_UC,
+		.mask_lo = (u32)(~(SZ_4G - tolud - 1)) | MTRR_PHYSMASK_V,
+		.mask_hi = (BIT_ULL(boot_cpu_data.x86_phys_bits) - 1) >> 32,
+	};
+
 	if (cc_platform_has(CC_ATTR_GUEST_MEM_ENCRYPT) &&
 	    kvm_para_has_feature(KVM_FEATURE_MIGRATION_CONTROL)) {
 		unsigned long nr_pages;
@@ -982,8 +994,12 @@ static void __init kvm_init_platform(void)
 	kvmclock_init();
 	x86_platform.apic_post_init = kvm_apic_init;
 
-	/* Set WB as the default cache mode for SEV-SNP and TDX */
-	guest_force_mtrr_state(NULL, 0, MTRR_TYPE_WRBACK);
+	/*
+	 * Set WB as the default cache mode for SEV-SNP and TDX, with a single
+	 * UC range for the legacy PCI hole, e.g. so that devices that expect
+	 * to get UC/WC mappings don't get surprised with WB.
+	 */
+	guest_force_mtrr_state(&pci_hole, 1, MTRR_TYPE_WRBACK);
 }
 
 #if defined(CONFIG_AMD_MEM_ENCRYPT)
@@ -1073,16 +1089,6 @@ static void kvm_wait(u8 *ptr, u8 val)
 void __init kvm_spinlock_init(void)
 {
 	/*
-	 * In case host doesn't support KVM_FEATURE_PV_UNHALT there is still an
-	 * advantage of keeping virt_spin_lock_key enabled: virt_spin_lock() is
-	 * preferred over native qspinlock when vCPU is preempted.
-	 */
-	if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT)) {
-		pr_info("PV spinlocks disabled, no host support\n");
-		return;
-	}
-
-	/*
 	 * Disable PV spinlocks and use native qspinlock when dedicated pCPUs
 	 * are available.
 	 */
@@ -1101,6 +1107,16 @@ void __init kvm_spinlock_init(void)
 		goto out;
 	}
 
+	/*
+	 * In case host doesn't support KVM_FEATURE_PV_UNHALT there is still an
+	 * advantage of keeping virt_spin_lock_key enabled: virt_spin_lock() is
+	 * preferred over native qspinlock when vCPU is preempted.
+	 */
+	if (!kvm_para_has_feature(KVM_FEATURE_PV_UNHALT)) {
+		pr_info("PV spinlocks disabled, no host support\n");
+		return;
+	}
+
 	pr_info("PV spinlocks enabled\n");
 
 	__pv_init_lock_hash();