Merge tag 'hyperv-fixes-signed-20240303' of...

Merge tag 'hyperv-fixes-signed-20240303' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux

Pull hyperv fixes from Wei Liu:

 - Multiple fixes, cleanups and documentations for Hyper-V core code and
   drivers

* tag 'hyperv-fixes-signed-20240303' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux:
  Drivers: hv: vmbus: make hv_bus const
  x86/hyperv: Allow 15-bit APIC IDs for VTL platforms
  x86/hyperv: Make encrypted/decrypted changes safe for load_unaligned_zeropad()
  x86/mm: Regularize set_memory_p() parameters and make non-static
  x86/hyperv: Use slow_virt_to_phys() in page transition hypervisor callback
  Documentation: hyperv: Add overview of PCI pass-thru device support
  Drivers: hv: vmbus: Update indentation in create_gpadl_header()
  Drivers: hv: vmbus: Remove duplication and cleanup code in create_gpadl_header()
  fbdev/hyperv_fb: Fix logic error for Gen2 VMs in hvfb_getmem()
  Drivers: hv: vmbus: Calculate ring buffer size for more efficient use of memory
  hv_utils: Allow implicit ICTIMESYNCFLAG_SYNC

Documentation/virt/hyperv/index.rst

@@ -10,3 +10,4 @@ Hyper-V Enlightenments
    overview
    vmbus
    clocks
+   vpci

arch/x86/hyperv/hv_vtl.c

@@ -16,6 +16,11 @@
 extern struct boot_params boot_params;
 static struct real_mode_header hv_vtl_real_mode_header;
 
+static bool __init hv_vtl_msi_ext_dest_id(void)
+{
+	return true;
+}
+
 void __init hv_vtl_init_platform(void)
 {
 	pr_info("Linux runs in Hyper-V Virtual Trust Level\n");
@@ -38,6 +43,8 @@ void __init hv_vtl_init_platform(void)
 	x86_platform.legacy.warm_reset = 0;
 	x86_platform.legacy.reserve_bios_regions = 0;
 	x86_platform.legacy.devices.pnpbios = 0;
+
+	x86_init.hyper.msi_ext_dest_id = hv_vtl_msi_ext_dest_id;
 }
 
 static inline u64 hv_vtl_system_desc_base(struct ldttss_desc *desc)

arch/x86/hyperv/ivm.c

@@ -15,6 +15,7 @@
 #include <asm/io.h>
 #include <asm/coco.h>
 #include <asm/mem_encrypt.h>
+#include <asm/set_memory.h>
 #include <asm/mshyperv.h>
 #include <asm/hypervisor.h>
 #include <asm/mtrr.h>
@@ -502,6 +503,31 @@ static int hv_mark_gpa_visibility(u16 count, const u64 pfn[],
 		return -EFAULT;
 }
 
+/*
+ * When transitioning memory between encrypted and decrypted, the caller
+ * of set_memory_encrypted() or set_memory_decrypted() is responsible for
+ * ensuring that the memory isn't in use and isn't referenced while the
+ * transition is in progress.  The transition has multiple steps, and the
+ * memory is in an inconsistent state until all steps are complete. A
+ * reference while the state is inconsistent could result in an exception
+ * that can't be cleanly fixed up.
+ *
+ * But the Linux kernel load_unaligned_zeropad() mechanism could cause a
+ * stray reference that can't be prevented by the caller, so Linux has
+ * specific code to handle this case. But when the #VC and #VE exceptions
+ * routed to a paravisor, the specific code doesn't work. To avoid this
+ * problem, mark the pages as "not present" while the transition is in
+ * progress. If load_unaligned_zeropad() causes a stray reference, a normal
+ * page fault is generated instead of #VC or #VE, and the page-fault-based
+ * handlers for load_unaligned_zeropad() resolve the reference.  When the
+ * transition is complete, hv_vtom_set_host_visibility() marks the pages
+ * as "present" again.
+ */
+static bool hv_vtom_clear_present(unsigned long kbuffer, int pagecount, bool enc)
+{
+	return !set_memory_np(kbuffer, pagecount);
+}
+
 /*
  * hv_vtom_set_host_visibility - Set specified memory visible to host.
  *
@@ -515,16 +541,28 @@ static bool hv_vtom_set_host_visibility(unsigned long kbuffer, int pagecount, bo
 	enum hv_mem_host_visibility visibility = enc ?
 			VMBUS_PAGE_NOT_VISIBLE : VMBUS_PAGE_VISIBLE_READ_WRITE;
 	u64 *pfn_array;
+	phys_addr_t paddr;
+	void *vaddr;
 	int ret = 0;
 	bool result = true;
 	int i, pfn;
 
 	pfn_array = kmalloc(HV_HYP_PAGE_SIZE, GFP_KERNEL);
-	if (!pfn_array)
-		return false;
+	if (!pfn_array) {
+		result = false;
+		goto err_set_memory_p;
+	}
 
 	for (i = 0, pfn = 0; i < pagecount; i++) {
-		pfn_array[pfn] = virt_to_hvpfn((void *)kbuffer + i * HV_HYP_PAGE_SIZE);
+		/*
+		 * Use slow_virt_to_phys() because the PRESENT bit has been
+		 * temporarily cleared in the PTEs.  slow_virt_to_phys() works
+		 * without the PRESENT bit while virt_to_hvpfn() or similar
+		 * does not.
+		 */
+		vaddr = (void *)kbuffer + (i * HV_HYP_PAGE_SIZE);
+		paddr = slow_virt_to_phys(vaddr);
+		pfn_array[pfn] = paddr >> HV_HYP_PAGE_SHIFT;
 		pfn++;
 
 		if (pfn == HV_MAX_MODIFY_GPA_REP_COUNT || i == pagecount - 1) {
@@ -538,14 +576,30 @@ static bool hv_vtom_set_host_visibility(unsigned long kbuffer, int pagecount, bo
 		}
 	}
 
- err_free_pfn_array:
+err_free_pfn_array:
 	kfree(pfn_array);
+
+err_set_memory_p:
+	/*
+	 * Set the PTE PRESENT bits again to revert what hv_vtom_clear_present()
+	 * did. Do this even if there is an error earlier in this function in
+	 * order to avoid leaving the memory range in a "broken" state. Setting
+	 * the PRESENT bits shouldn't fail, but return an error if it does.
+	 */
+	if (set_memory_p(kbuffer, pagecount))
+		result = false;
+
 	return result;
 }
 
 static bool hv_vtom_tlb_flush_required(bool private)
 {
-	return true;
+	/*
+	 * Since hv_vtom_clear_present() marks the PTEs as "not present"
+	 * and flushes the TLB, they can't be in the TLB. That makes the
+	 * flush controlled by this function redundant, so return "false".
+	 */
+	return false;
 }
 
 static bool hv_vtom_cache_flush_required(void)
@@ -608,6 +662,7 @@ void __init hv_vtom_init(void)
 	x86_platform.hyper.is_private_mmio = hv_is_private_mmio;
 	x86_platform.guest.enc_cache_flush_required = hv_vtom_cache_flush_required;
 	x86_platform.guest.enc_tlb_flush_required = hv_vtom_tlb_flush_required;
+	x86_platform.guest.enc_status_change_prepare = hv_vtom_clear_present;
 	x86_platform.guest.enc_status_change_finish = hv_vtom_set_host_visibility;
 
 	/* Set WB as the default cache mode. */

arch/x86/include/asm/set_memory.h

@@ -47,6 +47,7 @@ int set_memory_uc(unsigned long addr, int numpages);
 int set_memory_wc(unsigned long addr, int numpages);
 int set_memory_wb(unsigned long addr, int numpages);
 int set_memory_np(unsigned long addr, int numpages);
+int set_memory_p(unsigned long addr, int numpages);
 int set_memory_4k(unsigned long addr, int numpages);
 int set_memory_encrypted(unsigned long addr, int numpages);
 int set_memory_decrypted(unsigned long addr, int numpages);

arch/x86/mm/pat/set_memory.c

@@ -755,10 +755,14 @@ pmd_t *lookup_pmd_address(unsigned long address)
  * areas on 32-bit NUMA systems.  The percpu areas can
  * end up in this kind of memory, for instance.
  *
- * This could be optimized, but it is only intended to be
- * used at initialization time, and keeping it
- * unoptimized should increase the testing coverage for
- * the more obscure platforms.
+ * Note that as long as the PTEs are well-formed with correct PFNs, this
+ * works without checking the PRESENT bit in the leaf PTE.  This is unlike
+ * the similar vmalloc_to_page() and derivatives.  Callers may depend on
+ * this behavior.
+ *
+ * This could be optimized, but it is only used in paths that are not perf
+ * sensitive, and keeping it unoptimized should increase the testing coverage
+ * for the more obscure platforms.
  */
 phys_addr_t slow_virt_to_phys(void *__virt_addr)
 {
@@ -2041,17 +2045,12 @@ int set_mce_nospec(unsigned long pfn)
 	return rc;
 }
 
-static int set_memory_p(unsigned long *addr, int numpages)
-{
-	return change_page_attr_set(addr, numpages, __pgprot(_PAGE_PRESENT), 0);
-}
-
 /* Restore full speculative operation to the pfn. */
 int clear_mce_nospec(unsigned long pfn)
 {
 	unsigned long addr = (unsigned long) pfn_to_kaddr(pfn);
 
-	return set_memory_p(&addr, 1);
+	return set_memory_p(addr, 1);
 }
 EXPORT_SYMBOL_GPL(clear_mce_nospec);
 #endif /* CONFIG_X86_64 */
@@ -2104,6 +2103,11 @@ int set_memory_np_noalias(unsigned long addr, int numpages)
 					CPA_NO_CHECK_ALIAS, NULL);
 }
 
+int set_memory_p(unsigned long addr, int numpages)
+{
+	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_PRESENT), 0);
+}
+
 int set_memory_4k(unsigned long addr, int numpages)
 {
 	return change_page_attr_set_clr(&addr, numpages, __pgprot(0),

drivers/hv/channel.c

@@ -322,21 +322,17 @@ static int create_gpadl_header(enum hv_gpadl_type type, void *kbuffer,
 
 	pagecount = hv_gpadl_size(type, size) >> HV_HYP_PAGE_SHIFT;
 
-	/* do we need a gpadl body msg */
 	pfnsize = MAX_SIZE_CHANNEL_MESSAGE -
 		  sizeof(struct vmbus_channel_gpadl_header) -
 		  sizeof(struct gpa_range);
-	pfncount = pfnsize / sizeof(u64);
+	pfncount = umin(pagecount, pfnsize / sizeof(u64));
 
-	if (pagecount > pfncount) {
-		/* we need a gpadl body */
-		/* fill in the header */
 	msgsize = sizeof(struct vmbus_channel_msginfo) +
 		  sizeof(struct vmbus_channel_gpadl_header) +
 		  sizeof(struct gpa_range) + pfncount * sizeof(u64);
 	msgheader =  kzalloc(msgsize, GFP_KERNEL);
 	if (!msgheader)
-			goto nomem;
+		return -ENOMEM;
 
 	INIT_LIST_HEAD(&msgheader->submsglist);
 	msgheader->msgsize = msgsize;
@@ -356,18 +352,17 @@ static int create_gpadl_header(enum hv_gpadl_type type, void *kbuffer,
 	pfnsum = pfncount;
 	pfnleft = pagecount - pfncount;
 
-		/* how many pfns can we fit */
+	/* how many pfns can we fit in a body message */
 	pfnsize = MAX_SIZE_CHANNEL_MESSAGE -
 		  sizeof(struct vmbus_channel_gpadl_body);
 	pfncount = pfnsize / sizeof(u64);
 
-		/* fill in the body */
+	/*
+	 * If pfnleft is zero, everything fits in the header and no body
+	 * messages are needed
+	 */
 	while (pfnleft) {
-			if (pfnleft > pfncount)
-				pfncurr = pfncount;
-			else
-				pfncurr = pfnleft;
-
+		pfncurr = umin(pfncount, pfnleft);
 		msgsize = sizeof(struct vmbus_channel_msginfo) +
 			  sizeof(struct vmbus_channel_gpadl_body) +
 			  pfncurr * sizeof(u64);
@@ -386,13 +381,12 @@ static int create_gpadl_header(enum hv_gpadl_type type, void *kbuffer,
 				list_del(&pos->msglistentry);
 				kfree(pos);
 			}
-
-				goto nomem;
+			kfree(msgheader);
+			return -ENOMEM;
 		}
 
 		msgbody->msgsize = msgsize;
-			gpadl_body =
-				(struct vmbus_channel_gpadl_body *)msgbody->msg;
+		gpadl_body = (struct vmbus_channel_gpadl_body *)msgbody->msg;
 
 		/*
 		 * Gpadl is u32 and we are using a pointer which could
@@ -405,42 +399,12 @@ static int create_gpadl_header(enum hv_gpadl_type type, void *kbuffer,
 				kbuffer, size, send_offset, pfnsum + i);
 
 		/* add to msg header */
-			list_add_tail(&msgbody->msglistentry,
-				      &msgheader->submsglist);
+		list_add_tail(&msgbody->msglistentry, &msgheader->submsglist);
 		pfnsum += pfncurr;
 		pfnleft -= pfncurr;
 	}
-	} else {
-		/* everything fits in a header */
-		msgsize = sizeof(struct vmbus_channel_msginfo) +
-			  sizeof(struct vmbus_channel_gpadl_header) +
-			  sizeof(struct gpa_range) + pagecount * sizeof(u64);
-		msgheader = kzalloc(msgsize, GFP_KERNEL);
-		if (msgheader == NULL)
-			goto nomem;
-
-		INIT_LIST_HEAD(&msgheader->submsglist);
-		msgheader->msgsize = msgsize;
-
-		gpadl_header = (struct vmbus_channel_gpadl_header *)
-			msgheader->msg;
-		gpadl_header->rangecount = 1;
-		gpadl_header->range_buflen = sizeof(struct gpa_range) +
-					 pagecount * sizeof(u64);
-		gpadl_header->range[0].byte_offset = 0;
-		gpadl_header->range[0].byte_count = hv_gpadl_size(type, size);
-		for (i = 0; i < pagecount; i++)
-			gpadl_header->range[0].pfn_array[i] = hv_gpadl_hvpfn(
-				type, kbuffer, size, send_offset, i);
-
-		*msginfo = msgheader;
-	}
 
 	return 0;
-nomem:
-	kfree(msgheader);
-	kfree(msgbody);
-	return -ENOMEM;
 }
 
 /*

drivers/hv/hv_util.c

@@ -296,6 +296,11 @@ static struct {
 	spinlock_t			lock;
 } host_ts;
 
+static bool timesync_implicit;
+
+module_param(timesync_implicit, bool, 0644);
+MODULE_PARM_DESC(timesync_implicit, "If set treat SAMPLE as SYNC when clock is behind");
+
 static inline u64 reftime_to_ns(u64 reftime)
 {
 	return (reftime - WLTIMEDELTA) * 100;
@@ -344,6 +349,29 @@ static void hv_set_host_time(struct work_struct *work)
 		do_settimeofday64(&ts);
 }
 
+/*
+ * Due to a bug on Hyper-V hosts, the sync flag may not always be sent on resume.
+ * Force a sync if the guest is behind.
+ */
+static inline bool hv_implicit_sync(u64 host_time)
+{
+	struct timespec64 new_ts;
+	struct timespec64 threshold_ts;
+
+	new_ts = ns_to_timespec64(reftime_to_ns(host_time));
+	ktime_get_real_ts64(&threshold_ts);
+
+	threshold_ts.tv_sec += 5;
+
+	/*
+	 * If guest behind the host by 5 or more seconds.
+	 */
+	if (timespec64_compare(&new_ts, &threshold_ts) >= 0)
+		return true;
+
+	return false;
+}
+
 /*
  * Synchronize time with host after reboot, restore, etc.
  *
@@ -384,7 +412,8 @@ static inline void adj_guesttime(u64 hosttime, u64 reftime, u8 adj_flags)
 	spin_unlock_irqrestore(&host_ts.lock, flags);
 
 	/* Schedule work to do do_settimeofday64() */
-	if (adj_flags & ICTIMESYNCFLAG_SYNC)
+	if ((adj_flags & ICTIMESYNCFLAG_SYNC) ||
+	    (timesync_implicit && hv_implicit_sync(host_ts.host_time)))
 		schedule_work(&adj_time_work);
 }
 

drivers/hv/vmbus_drv.c

@@ -988,7 +988,7 @@ static const struct dev_pm_ops vmbus_pm = {
 };
 
 /* The one and only one */
-static struct bus_type  hv_bus = {
+static const struct bus_type  hv_bus = {
 	.name =		"vmbus",
 	.match =		vmbus_match,
 	.shutdown =		vmbus_shutdown,

drivers/video/fbdev/hyperv_fb.c

@@ -1010,8 +1010,6 @@ static int hvfb_getmem(struct hv_device *hdev, struct fb_info *info)
 			goto getmem_done;
 		}
 		pr_info("Unable to allocate enough contiguous physical memory on Gen 1 VM. Using MMIO instead.\n");
-	} else {
-		goto err1;
 	}
 
 	/*

include/linux/hyperv.h

@@ -164,8 +164,28 @@ struct hv_ring_buffer {
 	u8 buffer[];
 } __packed;
 
+
+/*
+ * If the requested ring buffer size is at least 8 times the size of the
+ * header, steal space from the ring buffer for the header. Otherwise, add
+ * space for the header so that is doesn't take too much of the ring buffer
+ * space.
+ *
+ * The factor of 8 is somewhat arbitrary. The goal is to prevent adding a
+ * relatively small header (4 Kbytes on x86) to a large-ish power-of-2 ring
+ * buffer size (such as 128 Kbytes) and so end up making a nearly twice as
+ * large allocation that will be almost half wasted. As a contrasting example,
+ * on ARM64 with 64 Kbyte page size, we don't want to take 64 Kbytes for the
+ * header from a 128 Kbyte allocation, leaving only 64 Kbytes for the ring.
+ * In this latter case, we must add 64 Kbytes for the header and not worry
+ * about what's wasted.
+ */
+#define VMBUS_HEADER_ADJ(payload_sz) \
+	((payload_sz) >=  8 * sizeof(struct hv_ring_buffer) ? \
+	0 : sizeof(struct hv_ring_buffer))
+
 /* Calculate the proper size of a ringbuffer, it must be page-aligned */
-#define VMBUS_RING_SIZE(payload_sz) PAGE_ALIGN(sizeof(struct hv_ring_buffer) + \
+#define VMBUS_RING_SIZE(payload_sz) PAGE_ALIGN(VMBUS_HEADER_ADJ(payload_sz) + \
 					       (payload_sz))
 
 struct hv_ring_buffer_info {