x86/sev: Perform PVALIDATE using the SVSM when not at VMPL0

The PVALIDATE instruction can only be performed at VMPL0. If an SVSM is
present, it will be running at VMPL0 while the guest itself is then
running at VMPL1 or a lower privilege level.

In that case, use the SVSM_CORE_PVALIDATE call to perform memory
validation instead of issuing the PVALIDATE instruction directly.

The validation of a single 4K page is now explicitly identified as such
in the function name, pvalidate_4k_page(). The pvalidate_pages()
function is used for validating 1 or more pages at either 4K or 2M in
size. Each function, however, determines whether it can issue the
PVALIDATE directly or whether the SVSM needs to be invoked.

  [ bp: Touchups. ]
  [ Tom: fold in a fix for Coconut SVSM:
    https://lore.kernel.org/r/234bb23c-d295-76e5-a690-7ea68dc1118b@amd.com  ]

Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/4c4017d8b94512d565de9ccb555b1a9f8983c69c.1717600736.git.thomas.lendacky@amd.com

2 files changed, 260 insertions, 20 deletions

diff --git a/arch/x86/kernel/sev-shared.c b/arch/x86/kernel/sev-shared.c
index b5110c68d241..7933c1203b63 100644
--- a/arch/x86/kernel/sev-shared.c
+++ b/arch/x86/kernel/sev-shared.c
@@ -40,6 +40,10 @@ static u8 snp_vmpl __ro_after_init;
 static struct svsm_ca *boot_svsm_caa __ro_after_init;
 static u64 boot_svsm_caa_pa __ro_after_init;
 
+static struct svsm_ca *svsm_get_caa(void);
+static u64 svsm_get_caa_pa(void);
+static int svsm_perform_call_protocol(struct svsm_call *call);
+
 /* I/O parameters for CPUID-related helpers */
 struct cpuid_leaf {
 	u32 fn;
@@ -1216,38 +1220,268 @@ static void __head setup_cpuid_table(const struct cc_blob_sev_info *cc_info)
 	}
 }
 
-static void pvalidate_pages(struct snp_psc_desc *desc)
+static inline void __pval_terminate(u64 pfn, bool action, unsigned int page_size,
+				    int ret, u64 svsm_ret)
+{
+	WARN(1, "PVALIDATE failure: pfn: 0x%llx, action: %u, size: %u, ret: %d, svsm_ret: 0x%llx\n",
+	     pfn, action, page_size, ret, svsm_ret);
+
+	sev_es_terminate(SEV_TERM_SET_LINUX, GHCB_TERM_PVALIDATE);
+}
+
+static void svsm_pval_terminate(struct svsm_pvalidate_call *pc, int ret, u64 svsm_ret)
+{
+	unsigned int page_size;
+	bool action;
+	u64 pfn;
+
+	pfn = pc->entry[pc->cur_index].pfn;
+	action = pc->entry[pc->cur_index].action;
+	page_size = pc->entry[pc->cur_index].page_size;
+
+	__pval_terminate(pfn, action, page_size, ret, svsm_ret);
+}
+
+static void svsm_pval_4k_page(unsigned long paddr, bool validate)
+{
+	struct svsm_pvalidate_call *pc;
+	struct svsm_call call = {};
+	unsigned long flags;
+	u64 pc_pa;
+	int ret;
+
+	/*
+	 * This can be called very early in the boot, use native functions in
+	 * order to avoid paravirt issues.
+	 */
+	flags = native_local_irq_save();
+
+	call.caa = svsm_get_caa();
+
+	pc = (struct svsm_pvalidate_call *)call.caa->svsm_buffer;
+	pc_pa = svsm_get_caa_pa() + offsetof(struct svsm_ca, svsm_buffer);
+
+	pc->num_entries = 1;
+	pc->cur_index   = 0;
+	pc->entry[0].page_size = RMP_PG_SIZE_4K;
+	pc->entry[0].action    = validate;
+	pc->entry[0].ignore_cf = 0;
+	pc->entry[0].pfn       = paddr >> PAGE_SHIFT;
+
+	/* Protocol 0, Call ID 1 */
+	call.rax = SVSM_CORE_CALL(SVSM_CORE_PVALIDATE);
+	call.rcx = pc_pa;
+
+	ret = svsm_perform_call_protocol(&call);
+	if (ret)
+		svsm_pval_terminate(pc, ret, call.rax_out);
+
+	native_local_irq_restore(flags);
+}
+
+static void pvalidate_4k_page(unsigned long vaddr, unsigned long paddr, bool validate)
+{
+	int ret;
+
+	/*
+	 * This can be called very early during boot, so use rIP-relative
+	 * references as needed.
+	 */
+	if (RIP_REL_REF(snp_vmpl)) {
+		svsm_pval_4k_page(paddr, validate);
+	} else {
+		ret = pvalidate(vaddr, RMP_PG_SIZE_4K, validate);
+		if (ret)
+			__pval_terminate(PHYS_PFN(paddr), validate, RMP_PG_SIZE_4K, ret, 0);
+	}
+}
+
+static void pval_pages(struct snp_psc_desc *desc)
 {
 	struct psc_entry *e;
 	unsigned long vaddr;
 	unsigned int size;
 	unsigned int i;
 	bool validate;
+	u64 pfn;
 	int rc;
 
 	for (i = 0; i <= desc->hdr.end_entry; i++) {
 		e = &desc->entries[i];
 
-		vaddr = (unsigned long)pfn_to_kaddr(e->gfn);
+		pfn = e->gfn;
+		vaddr = (unsigned long)pfn_to_kaddr(pfn);
 		size = e->pagesize ? RMP_PG_SIZE_2M : RMP_PG_SIZE_4K;
 		validate = e->operation == SNP_PAGE_STATE_PRIVATE;
 
 		rc = pvalidate(vaddr, size, validate);
+		if (!rc)
+			continue;
+
 		if (rc == PVALIDATE_FAIL_SIZEMISMATCH && size == RMP_PG_SIZE_2M) {
 			unsigned long vaddr_end = vaddr + PMD_SIZE;
 
-			for (; vaddr < vaddr_end; vaddr += PAGE_SIZE) {
+			for (; vaddr < vaddr_end; vaddr += PAGE_SIZE, pfn++) {
 				rc = pvalidate(vaddr, RMP_PG_SIZE_4K, validate);
 				if (rc)
-					break;
+					__pval_terminate(pfn, validate, RMP_PG_SIZE_4K, rc, 0);
 			}
+		} else {
+			__pval_terminate(pfn, validate, size, rc, 0);
 		}
+	}
+}
+
+static u64 svsm_build_ca_from_pfn_range(u64 pfn, u64 pfn_end, bool action,
+					struct svsm_pvalidate_call *pc)
+{
+	struct svsm_pvalidate_entry *pe;
+
+	/* Nothing in the CA yet */
+	pc->num_entries = 0;
+	pc->cur_index   = 0;
+
+	pe = &pc->entry[0];
+
+	while (pfn < pfn_end) {
+		pe->page_size = RMP_PG_SIZE_4K;
+		pe->action    = action;
+		pe->ignore_cf = 0;
+		pe->pfn       = pfn;
+
+		pe++;
+		pfn++;
+
+		pc->num_entries++;
+		if (pc->num_entries == SVSM_PVALIDATE_MAX_COUNT)
+			break;
+	}
+
+	return pfn;
+}
+
+static int svsm_build_ca_from_psc_desc(struct snp_psc_desc *desc, unsigned int desc_entry,
+				       struct svsm_pvalidate_call *pc)
+{
+	struct svsm_pvalidate_entry *pe;
+	struct psc_entry *e;
+
+	/* Nothing in the CA yet */
+	pc->num_entries = 0;
+	pc->cur_index   = 0;
+
+	pe = &pc->entry[0];
+	e  = &desc->entries[desc_entry];
+
+	while (desc_entry <= desc->hdr.end_entry) {
+		pe->page_size = e->pagesize ? RMP_PG_SIZE_2M : RMP_PG_SIZE_4K;
+		pe->action    = e->operation == SNP_PAGE_STATE_PRIVATE;
+		pe->ignore_cf = 0;
+		pe->pfn       = e->gfn;
+
+		pe++;
+		e++;
+
+		desc_entry++;
+		pc->num_entries++;
+		if (pc->num_entries == SVSM_PVALIDATE_MAX_COUNT)
+			break;
+	}
+
+	return desc_entry;
+}
+
+static void svsm_pval_pages(struct snp_psc_desc *desc)
+{
+	struct svsm_pvalidate_entry pv_4k[VMGEXIT_PSC_MAX_ENTRY];
+	unsigned int i, pv_4k_count = 0;
+	struct svsm_pvalidate_call *pc;
+	struct svsm_call call = {};
+	unsigned long flags;
+	bool action;
+	u64 pc_pa;
+	int ret;
+
+	/*
+	 * This can be called very early in the boot, use native functions in
+	 * order to avoid paravirt issues.
+	 */
+	flags = native_local_irq_save();
+
+	/*
+	 * The SVSM calling area (CA) can support processing 510 entries at a
+	 * time. Loop through the Page State Change descriptor until the CA is
+	 * full or the last entry in the descriptor is reached, at which time
+	 * the SVSM is invoked. This repeats until all entries in the descriptor
+	 * are processed.
+	 */
+	call.caa = svsm_get_caa();
+
+	pc = (struct svsm_pvalidate_call *)call.caa->svsm_buffer;
+	pc_pa = svsm_get_caa_pa() + offsetof(struct svsm_ca, svsm_buffer);
+
+	/* Protocol 0, Call ID 1 */
+	call.rax = SVSM_CORE_CALL(SVSM_CORE_PVALIDATE);
+	call.rcx = pc_pa;
+
+	for (i = 0; i <= desc->hdr.end_entry;) {
+		i = svsm_build_ca_from_psc_desc(desc, i, pc);
+
+		do {
+			ret = svsm_perform_call_protocol(&call);
+			if (!ret)
+				continue;
 
-		if (rc) {
-			WARN(1, "Failed to validate address 0x%lx ret %d", vaddr, rc);
-			sev_es_terminate(SEV_TERM_SET_LINUX, GHCB_TERM_PVALIDATE);
+			/*
+			 * Check if the entry failed because of an RMP mismatch (a
+			 * PVALIDATE at 2M was requested, but the page is mapped in
+			 * the RMP as 4K).
+			 */
+
+			if (call.rax_out == SVSM_PVALIDATE_FAIL_SIZEMISMATCH &&
+			    pc->entry[pc->cur_index].page_size == RMP_PG_SIZE_2M) {
+				/* Save this entry for post-processing at 4K */
+				pv_4k[pv_4k_count++] = pc->entry[pc->cur_index];
+
+				/* Skip to the next one unless at the end of the list */
+				pc->cur_index++;
+				if (pc->cur_index < pc->num_entries)
+					ret = -EAGAIN;
+				else
+					ret = 0;
+			}
+		} while (ret == -EAGAIN);
+
+		if (ret)
+			svsm_pval_terminate(pc, ret, call.rax_out);
+	}
+
+	/* Process any entries that failed to be validated at 2M and validate them at 4K */
+	for (i = 0; i < pv_4k_count; i++) {
+		u64 pfn, pfn_end;
+
+		action  = pv_4k[i].action;
+		pfn     = pv_4k[i].pfn;
+		pfn_end = pfn + 512;
+
+		while (pfn < pfn_end) {
+			pfn = svsm_build_ca_from_pfn_range(pfn, pfn_end, action, pc);
+
+			ret = svsm_perform_call_protocol(&call);
+			if (ret)
+				svsm_pval_terminate(pc, ret, call.rax_out);
 		}
 	}
+
+	native_local_irq_restore(flags);
+}
+
+static void pvalidate_pages(struct snp_psc_desc *desc)
+{
+	if (snp_vmpl)
+		svsm_pval_pages(desc);
+	else
+		pval_pages(desc);
 }
 
 static int vmgexit_psc(struct ghcb *ghcb, struct snp_psc_desc *desc)
diff --git a/arch/x86/kernel/sev.c b/arch/x86/kernel/sev.c
index 51a0984b422c..f1d11e723f66 100644
--- a/arch/x86/kernel/sev.c
+++ b/arch/x86/kernel/sev.c
@@ -628,6 +628,19 @@ static inline struct svsm_ca *svsm_get_caa(void)
 		return RIP_REL_REF(boot_svsm_caa);
 }
 
+static u64 svsm_get_caa_pa(void)
+{
+	/*
+	 * Use rIP-relative references when called early in the boot. If
+	 * ->use_cas is set, then it is late in the boot and no need
+	 * to worry about rIP-relative references.
+	 */
+	if (RIP_REL_REF(sev_cfg).use_cas)
+		return this_cpu_read(svsm_caa_pa);
+	else
+		return RIP_REL_REF(boot_svsm_caa_pa);
+}
+
 static noinstr void __sev_put_ghcb(struct ghcb_state *state)
 {
 	struct sev_es_runtime_data *data;
@@ -800,7 +813,6 @@ early_set_pages_state(unsigned long vaddr, unsigned long paddr,
 {
 	unsigned long paddr_end;
 	u64 val;
-	int ret;
 
 	vaddr = vaddr & PAGE_MASK;
 
@@ -808,12 +820,9 @@ early_set_pages_state(unsigned long vaddr, unsigned long paddr,
 	paddr_end = paddr + (npages << PAGE_SHIFT);
 
 	while (paddr < paddr_end) {
-		if (op == SNP_PAGE_STATE_SHARED) {
-			/* Page validation must be rescinded before changing to shared */
-			ret = pvalidate(vaddr, RMP_PG_SIZE_4K, false);
-			if (WARN(ret, "Failed to validate address 0x%lx ret %d", paddr, ret))
-				goto e_term;
-		}
+		/* Page validation must be rescinded before changing to shared */
+		if (op == SNP_PAGE_STATE_SHARED)
+			pvalidate_4k_page(vaddr, paddr, false);
 
 		/*
 		 * Use the MSR protocol because this function can be called before
@@ -835,12 +844,9 @@ early_set_pages_state(unsigned long vaddr, unsigned long paddr,
 			 paddr, GHCB_MSR_PSC_RESP_VAL(val)))
 			goto e_term;
 
-		if (op == SNP_PAGE_STATE_PRIVATE) {
-			/* Page validation must be performed after changing to private */
-			ret = pvalidate(vaddr, RMP_PG_SIZE_4K, true);
-			if (WARN(ret, "Failed to validate address 0x%lx ret %d", paddr, ret))
-				goto e_term;
-		}
+		/* Page validation must be performed after changing to private */
+		if (op == SNP_PAGE_STATE_PRIVATE)
+			pvalidate_4k_page(vaddr, paddr, true);
 
 		vaddr += PAGE_SIZE;
 		paddr += PAGE_SIZE;