Received-SPF: pass (google.com: domain of linux-kernel+bounces-202814-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Received-SPF: Pass (protection.outlook.com: domain of amd.com designates
 165.204.84.17 as permitted sender) receiver=protection.outlook.com;
 client-ip=165.204.84.17; helo=SATLEXMB04.amd.com; pr=C
From: Tom Lendacky <thomas.lendacky@amd.com>
To: <linux-kernel@vger.kernel.org>, <x86@kernel.org>,
	<linux-coco@lists.linux.dev>, <svsm-devel@coconut-svsm.dev>
CC: Thomas Gleixner <tglx@linutronix.de>, Ingo Molnar <mingo@redhat.com>,
	Borislav Petkov <bp@alien8.de>, Dave Hansen <dave.hansen@linux.intel.com>,
	"H. Peter Anvin" <hpa@zytor.com>, Andy Lutomirski <luto@kernel.org>, "Peter
 Zijlstra" <peterz@infradead.org>, Dan Williams <dan.j.williams@intel.com>,
	Michael Roth <michael.roth@amd.com>, Ashish Kalra <ashish.kalra@amd.com>
Subject: [PATCH v5 04/13] x86/sev: Perform PVALIDATE using the SVSM when not at VMPL0
Date: Wed, 5 Jun 2024 10:18:47 -0500
Message-ID: <4c4017d8b94512d565de9ccb555b1a9f8983c69c.1717600736.git.thomas.lendacky@amd.com>
In-Reply-To: <cover.1717600736.git.thomas.lendacky@amd.com>
References: <cover.1717600736.git.thomas.lendacky@amd.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Content-Type: text/plain
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 05 Jun 2024 15:19:37.2891
 (UTC)
X-MS-Exchange-CrossTenant-Network-Message-Id: 6383bc25-84d7-45ef-35bc-08dc8572e9ba
X-MS-Exchange-CrossTenant-Id: 3dd8961f-e488-4e60-8e11-a82d994e183d
X-MS-Exchange-CrossTenant-OriginalAttributedTenantConnectingIp: TenantId=3dd8961f-e488-4e60-8e11-a82d994e183d;Ip=[165.204.84.17];Helo=[SATLEXMB04.amd.com]
X-MS-Exchange-CrossTenant-AuthSource:
	CH1PEPF0000A34A.namprd04.prod.outlook.com
X-MS-Exchange-CrossTenant-AuthAs: Anonymous
X-MS-Exchange-CrossTenant-FromEntityHeader: HybridOnPrem
X-MS-Exchange-Transport-CrossTenantHeadersStamped: DS0PR12MB7873

The PVALIDATE instruction can only be performed at VMPL0. An SVSM will
be present when running at VMPL1 or a lower privilege level.

When an SVSM is present, 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.

Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
---
 arch/x86/boot/compressed/sev.c |  45 +++++-
 arch/x86/include/asm/sev.h     |  26 ++++
 arch/x86/kernel/sev-shared.c   | 250 +++++++++++++++++++++++++++++++--
 arch/x86/kernel/sev.c          |  30 ++--
 4 files changed, 325 insertions(+), 26 deletions(-)

diff --git a/arch/x86/boot/compressed/sev.c b/arch/x86/boot/compressed/sev.c
index 927b71495122..1cc3106a3ba7 100644
--- a/arch/x86/boot/compressed/sev.c
+++ b/arch/x86/boot/compressed/sev.c
@@ -129,6 +129,34 @@ static bool fault_in_kernel_space(unsigned long address)
 /* Include code for early handlers */
 #include "../../kernel/sev-shared.c"
 
+static struct svsm_ca *svsm_get_caa(void)
+{
+	return boot_svsm_caa;
+}
+
+static u64 svsm_get_caa_pa(void)
+{
+	return boot_svsm_caa_pa;
+}
+
+static int svsm_perform_call_protocol(struct svsm_call *call)
+{
+	struct ghcb *ghcb;
+	int ret;
+
+	if (boot_ghcb)
+		ghcb = boot_ghcb;
+	else
+		ghcb = NULL;
+
+	do {
+		ret = ghcb ? svsm_perform_ghcb_protocol(ghcb, call)
+			   : svsm_perform_msr_protocol(call);
+	} while (ret == -EAGAIN);
+
+	return ret;
+}
+
 bool sev_snp_enabled(void)
 {
 	return sev_status & MSR_AMD64_SEV_SNP_ENABLED;
@@ -145,8 +173,8 @@ static void __page_state_change(unsigned long paddr, enum psc_op op)
 	 * If private -> shared then invalidate the page before requesting the
 	 * state change in the RMP table.
 	 */
-	if (op == SNP_PAGE_STATE_SHARED && pvalidate(paddr, RMP_PG_SIZE_4K, 0))
-		sev_es_terminate(SEV_TERM_SET_LINUX, GHCB_TERM_PVALIDATE);
+	if (op == SNP_PAGE_STATE_SHARED)
+		pvalidate_4k_page(paddr, paddr, false);
 
 	/* Issue VMGEXIT to change the page state in RMP table. */
 	sev_es_wr_ghcb_msr(GHCB_MSR_PSC_REQ_GFN(paddr >> PAGE_SHIFT, op));
@@ -161,8 +189,8 @@ static void __page_state_change(unsigned long paddr, enum psc_op op)
 	 * Now that page state is changed in the RMP table, validate it so that it is
 	 * consistent with the RMP entry.
 	 */
-	if (op == SNP_PAGE_STATE_PRIVATE && pvalidate(paddr, RMP_PG_SIZE_4K, 1))
-		sev_es_terminate(SEV_TERM_SET_LINUX, GHCB_TERM_PVALIDATE);
+	if (op == SNP_PAGE_STATE_PRIVATE)
+		pvalidate_4k_page(paddr, paddr, true);
 }
 
 void snp_set_page_private(unsigned long paddr)
@@ -255,6 +283,15 @@ void sev_es_shutdown_ghcb(void)
 	if (!sev_es_check_cpu_features())
 		error("SEV-ES CPU Features missing.");
 
+	/*
+	 * This is used to determine whether to use the GHCB MSR protocol or
+	 * the GHCB shared page to perform a GHCB request. Since the GHCB page
+	 * is being changed to encrypted, it can't be used to perform GHCB
+	 * requests. Clear the boot_ghcb variable so that the GHCB MSR protocol
+	 * is used to change the GHCB page over to an encrypted page.
+	 */
+	boot_ghcb = NULL;
+
 	/*
 	 * GHCB Page must be flushed from the cache and mapped encrypted again.
 	 * Otherwise the running kernel will see strange cache effects when
diff --git a/arch/x86/include/asm/sev.h b/arch/x86/include/asm/sev.h
index 3abc2d759db7..01e3866c4d61 100644
--- a/arch/x86/include/asm/sev.h
+++ b/arch/x86/include/asm/sev.h
@@ -187,6 +187,31 @@ struct svsm_ca {
 #define SVSM_ERR_INVALID_PARAMETER		0x80000005
 #define SVSM_ERR_INVALID_REQUEST		0x80000006
 #define SVSM_ERR_BUSY				0x80000007
+#define SVSM_PVALIDATE_FAIL_SIZEMISMATCH	0x80001006
+
+/*
+ * The SVSM PVALIDATE related structures
+ */
+struct svsm_pvalidate_entry {
+	u64 page_size		: 2,
+	    action		: 1,
+	    ignore_cf		: 1,
+	    rsvd		: 8,
+	    pfn			: 52;
+};
+
+struct svsm_pvalidate_call {
+	u16 num_entries;
+	u16 cur_index;
+
+	u8 rsvd1[4];
+
+	struct svsm_pvalidate_entry entry[];
+};
+
+#define SVSM_PVALIDATE_MAX_COUNT	((sizeof_field(struct svsm_ca, svsm_buffer) -		\
+					  offsetof(struct svsm_pvalidate_call, entry)) /	\
+					 sizeof(struct svsm_pvalidate_entry))
 
 /*
  * SVSM protocol structure
@@ -207,6 +232,7 @@ struct svsm_call {
 
 #define SVSM_CORE_CALL(x)		((0ULL << 32) | (x))
 #define SVSM_CORE_REMAP_CA		0
+#define SVSM_CORE_PVALIDATE		1
 
 #ifdef CONFIG_AMD_MEM_ENCRYPT
 extern void __sev_es_ist_enter(struct pt_regs *regs);
diff --git a/arch/x86/kernel/sev-shared.c b/arch/x86/kernel/sev-shared.c
index 00173deefc46..c274fa826ef0 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;
@@ -1216,19 +1220,97 @@ static void __head setup_cpuid_table(const struct cc_blob_sev_info *cc_info)
 	}
 }
 
-static void pvalidate_pages(struct snp_psc_desc *desc)
+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;
+
+	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 pval_terminate(u64 pfn, bool action, unsigned int page_size, int ret)
+{
+	WARN(1, "PVALIDATE failure: pfn 0x%llx, action=%u, size=%u - ret=%d\n",
+	     pfn, action, page_size, ret);
+	sev_es_terminate(SEV_TERM_SET_LINUX, GHCB_TERM_PVALIDATE);
+}
+
+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 in the 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);
+	}
+}
+
+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;
 
@@ -1236,20 +1318,170 @@ static void pvalidate_pages(struct snp_psc_desc *desc)
 		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);
 			}
-		}
-
-		if (rc) {
-			WARN(1, "Failed to validate address 0x%lx ret %d", vaddr, rc);
-			sev_es_terminate(SEV_TERM_SET_LINUX, GHCB_TERM_PVALIDATE);
+		} else if (rc) {
+			pval_terminate(pfn, validate, size, rc);
 		}
 	}
 }
 
+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 void svsm_build_ca_from_psc_desc(struct snp_psc_desc *desc,
+					struct svsm_pvalidate_call *pc)
+{
+	struct svsm_pvalidate_entry *pe;
+	unsigned int desc_entry;
+	struct psc_entry *e;
+
+	desc_entry = desc->hdr.cur_entry;
+
+	/* 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;
+	}
+
+	desc->hdr.cur_entry = 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;
+
+	while (desc->hdr.cur_entry <= desc->hdr.end_entry) {
+		svsm_build_ca_from_psc_desc(desc, pc);
+
+		do {
+			ret = svsm_perform_call_protocol(&call);
+			if (ret) {
+				/*
+				 * 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)
 {
 	int cur_entry, end_entry, ret = 0;
diff --git a/arch/x86/kernel/sev.c b/arch/x86/kernel/sev.c
index 6bab3244a3b9..b5c18ed4c572 100644
--- a/arch/x86/kernel/sev.c
+++ b/arch/x86/kernel/sev.c
@@ -626,6 +626,17 @@ static struct svsm_ca *svsm_get_caa(void)
 						    : RIP_REL_REF(boot_svsm_caa);
 }
 
+static u64 svsm_get_caa_pa(void)
+{
+	/*
+	 * Use rip-relative references when called early in the boot. If
+	 * cas_initialized is set, then it is late in the boot and no need
+	 * to worry about rip-relative references.
+	 */
+	return RIP_REL_REF(sev_cfg).cas_initialized ? this_cpu_read(svsm_caa_pa)
+						    : RIP_REL_REF(boot_svsm_caa_pa);
+}
+
 static noinstr void __sev_put_ghcb(struct ghcb_state *state)
 {
 	struct sev_es_runtime_data *data;
@@ -798,7 +809,6 @@ early_set_pages_state(unsigned long vaddr, unsigned long paddr,
 {
 	unsigned long paddr_end;
 	u64 val;
-	int ret;
 
 	vaddr = vaddr & PAGE_MASK;
 
@@ -806,12 +816,9 @@ 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
@@ -833,12 +840,9 @@ 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;
-- 
2.43.2