6.1.4-rc1 review

@@ -1288,9 +1288,8 @@ void pdc_io_reset_devices(void)
 
 #endif /* defined(BOOTLOADER) */
 
-/* locked by pdc_console_lock */
-static int __attribute__((aligned(8)))   iodc_retbuf[32];
-static char __attribute__((aligned(64))) iodc_dbuf[4096];
+/* locked by pdc_lock */
+static char iodc_dbuf[4096] __page_aligned_bss;
 
 /**
  * pdc_iodc_print - Console print using IODC.

@@ -1307,6 +1306,9 @@ int pdc_iodc_print(const unsigned char *
 	unsigned int i;
 	unsigned long flags;
 
+	count = min_t(unsigned int, count, sizeof(iodc_dbuf));
+
+	spin_lock_irqsave(&pdc_lock, flags);
 	for (i = 0; i < count;) {
 		switch(str[i]) {
 		case '\n':

@@ -1322,12 +1324,11 @@ int pdc_iodc_print(const unsigned char *
 	}
 
 print:
-        spin_lock_irqsave(&pdc_lock, flags);
-        real32_call(PAGE0->mem_cons.iodc_io,
-                    (unsigned long)PAGE0->mem_cons.hpa, ENTRY_IO_COUT,
-                    PAGE0->mem_cons.spa, __pa(PAGE0->mem_cons.dp.layers),
-                    __pa(iodc_retbuf), 0, __pa(iodc_dbuf), i, 0);
-        spin_unlock_irqrestore(&pdc_lock, flags);
+	real32_call(PAGE0->mem_cons.iodc_io,
+		(unsigned long)PAGE0->mem_cons.hpa, ENTRY_IO_COUT,
+		PAGE0->mem_cons.spa, __pa(PAGE0->mem_cons.dp.layers),
+		__pa(pdc_result), 0, __pa(iodc_dbuf), i, 0);
+	spin_unlock_irqrestore(&pdc_lock, flags);
 
 	return i;
 }

@@ -1354,10 +1355,11 @@ int pdc_iodc_getc(void)
 	real32_call(PAGE0->mem_kbd.iodc_io,
 		    (unsigned long)PAGE0->mem_kbd.hpa, ENTRY_IO_CIN,
 		    PAGE0->mem_kbd.spa, __pa(PAGE0->mem_kbd.dp.layers), 
-		    __pa(iodc_retbuf), 0, __pa(iodc_dbuf), 1, 0);
+		    __pa(pdc_result), 0, __pa(iodc_dbuf), 1, 0);
 
 	ch = *iodc_dbuf;
-	status = *iodc_retbuf;
+	/* like convert_to_wide() but for first return value only: */
+	status = *(int *)&pdc_result;
 	spin_unlock_irqrestore(&pdc_lock, flags);
 
 	if (status == 0)

index ab7620f695be..b16fa9bac5f4 100644
@@ -208,23 +208,3 @@ int kgdb_arch_handle_exception(int trap, int signo,
 	}
 	return -1;
 }
-
-/* KGDB console driver which uses PDC to read chars from keyboard */
-
-static void kgdb_pdc_write_char(u8 chr)
-{
-	/* no need to print char. kgdb will do it. */
-}
-
-static struct kgdb_io kgdb_pdc_io_ops = {
-	.name		= "kgdb_pdc",
-	.read_char	= pdc_iodc_getc,
-	.write_char	= kgdb_pdc_write_char,
-};
-
-static int __init kgdb_pdc_init(void)
-{
-	kgdb_register_io_module(&kgdb_pdc_io_ops);
-	return 0;
-}
-early_initcall(kgdb_pdc_init);

@@ -166,8 +166,8 @@ extern void __update_cache(pte_t pte);
 
 /* This calculates the number of initial pages we need for the initial
  * page tables */
-#if (KERNEL_INITIAL_ORDER) >= (PMD_SHIFT)
-# define PT_INITIAL	(1 << (KERNEL_INITIAL_ORDER - PMD_SHIFT))
+#if (KERNEL_INITIAL_ORDER) >= (PLD_SHIFT + BITS_PER_PTE)
+# define PT_INITIAL	(1 << (KERNEL_INITIAL_ORDER - PLD_SHIFT - BITS_PER_PTE))
 #else
 # define PT_INITIAL	(1)  /* all initial PTEs fit into one page */
 #endif

@@ -82,6 +82,17 @@ free_and_exit:
 	return -ENOMEM;
 }
 
+static void dev_exceptions_move(struct list_head *dest, struct list_head *orig)
+{
+	struct dev_exception_item *ex, *tmp;
+
+	lockdep_assert_held(&devcgroup_mutex);
+
+	list_for_each_entry_safe(ex, tmp, orig, list) {
+		list_move_tail(&ex->list, dest);
+	}
+}
+
 /*
  * called under devcgroup_mutex
  */

@@ -604,11 +615,13 @@ static int devcgroup_update_access(struc
 	int count, rc = 0;
 	struct dev_exception_item ex;
 	struct dev_cgroup *parent = css_to_devcgroup(devcgroup->css.parent);
+	struct dev_cgroup tmp_devcgrp;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
 	memset(&ex, 0, sizeof(ex));
+	memset(&tmp_devcgrp, 0, sizeof(tmp_devcgrp));
 	b = buffer;
 
 	switch (*b) {

@@ -620,15 +633,27 @@ static int devcgroup_update_access(struc
 
 			if (!may_allow_all(parent))
 				return -EPERM;
-			dev_exception_clean(devcgroup);
-			devcgroup->behavior = DEVCG_DEFAULT_ALLOW;
-			if (!parent)
+			if (!parent) {
+				devcgroup->behavior = DEVCG_DEFAULT_ALLOW;
+				dev_exception_clean(devcgroup);
 				break;
+			}
 
+			INIT_LIST_HEAD(&tmp_devcgrp.exceptions);
+			rc = dev_exceptions_copy(&tmp_devcgrp.exceptions,
+						 &devcgroup->exceptions);
+			if (rc)
+				return rc;
+			dev_exception_clean(devcgroup);
 			rc = dev_exceptions_copy(&devcgroup->exceptions,
 						 &parent->exceptions);
-			if (rc)
+			if (rc) {
+				dev_exceptions_move(&devcgroup->exceptions,
+						    &tmp_devcgrp.exceptions);
 				return rc;
+			}
+			devcgroup->behavior = DEVCG_DEFAULT_ALLOW;
+			dev_exception_clean(&tmp_devcgrp);
 			break;
 		case DEVCG_DENY:
 			if (css_has_online_children(&devcgroup->css))

@@ -582,6 +582,9 @@ void drm_connector_cleanup(struct drm_co
 	mutex_destroy(&connector->mutex);
 
 	memset(connector, 0, sizeof(*connector));
+
+	if (dev->registered)
+		drm_sysfs_hotplug_event(dev);
 }
 EXPORT_SYMBOL(drm_connector_cleanup);
 

@@ -308,7 +308,8 @@ void vmw_kms_cursor_snoop(struct vmw_sur
 	if (cmd->dma.guest.ptr.offset % PAGE_SIZE ||
 	    box->x != 0    || box->y != 0    || box->z != 0    ||
 	    box->srcx != 0 || box->srcy != 0 || box->srcz != 0 ||
-	    box->d != 1    || box_count != 1) {
+	    box->d != 1    || box_count != 1 ||
+	    box->w > 64 || box->h > 64) {
 		/* TODO handle none page aligned offsets */
 		/* TODO handle more dst & src != 0 */
 		/* TODO handle more then one copy */

@@ -284,7 +284,8 @@ static void mgag200_g200se_04_pixpllc_at
 	pixpllcp = pixpllc->p - 1;
 	pixpllcs = pixpllc->s;
 
-	xpixpllcm = pixpllcm | ((pixpllcn & BIT(8)) >> 1);
+	// For G200SE A, BIT(7) should be set unconditionally.
+	xpixpllcm = BIT(7) | pixpllcm;
 	xpixpllcn = pixpllcn;
 	xpixpllcp = (pixpllcs << 3) | pixpllcp;
 

@@ -135,6 +135,19 @@ static void etnaviv_iommu_remove_mapping
 	drm_mm_remove_node(&mapping->vram_node);
 }
 
+void etnaviv_iommu_reap_mapping(struct etnaviv_vram_mapping *mapping)
+{
+	struct etnaviv_iommu_context *context = mapping->context;
+
+	lockdep_assert_held(&context->lock);
+	WARN_ON(mapping->use);
+
+	etnaviv_iommu_remove_mapping(context, mapping);
+	etnaviv_iommu_context_put(mapping->context);
+	mapping->context = NULL;
+	list_del_init(&mapping->mmu_node);
+}
+
 static int etnaviv_iommu_find_iova(struct etnaviv_iommu_context *context,
 				   struct drm_mm_node *node, size_t size)
 {

@@ -202,10 +215,7 @@ static int etnaviv_iommu_find_iova(struc
 		 * this mapping.
 		 */
 		list_for_each_entry_safe(m, n, &list, scan_node) {
-			etnaviv_iommu_remove_mapping(context, m);
-			etnaviv_iommu_context_put(m->context);
-			m->context = NULL;
-			list_del_init(&m->mmu_node);
+			etnaviv_iommu_reap_mapping(m);
 			list_del_init(&m->scan_node);
 		}
 

@@ -257,10 +267,7 @@ static int etnaviv_iommu_insert_exact(st
 	}
 
 	list_for_each_entry_safe(m, n, &scan_list, scan_node) {
-		etnaviv_iommu_remove_mapping(context, m);
-		etnaviv_iommu_context_put(m->context);
-		m->context = NULL;
-		list_del_init(&m->mmu_node);
+		etnaviv_iommu_reap_mapping(m);
 		list_del_init(&m->scan_node);
 	}
 

@@ -91,6 +91,7 @@ int etnaviv_iommu_map_gem(struct etnaviv
 	struct etnaviv_vram_mapping *mapping, u64 va);
 void etnaviv_iommu_unmap_gem(struct etnaviv_iommu_context *context,
 	struct etnaviv_vram_mapping *mapping);
+void etnaviv_iommu_reap_mapping(struct etnaviv_vram_mapping *mapping);
 
 int etnaviv_iommu_get_suballoc_va(struct etnaviv_iommu_context *ctx,
 				  struct etnaviv_vram_mapping *mapping,

@@ -137,9 +137,9 @@ static enum port intel_dsi_seq_port_to_p
 		return ffs(intel_dsi->ports) - 1;
 
 	if (seq_port) {
-		if (intel_dsi->ports & PORT_B)
+		if (intel_dsi->ports & BIT(PORT_B))
 			return PORT_B;
-		else if (intel_dsi->ports & PORT_C)
+		else if (intel_dsi->ports & BIT(PORT_C))
 			return PORT_C;
 	}
 

@@ -1629,7 +1629,11 @@ static int ingenic_drm_init(void)
 			return err;
 	}
 
-	return platform_driver_register(&ingenic_drm_driver);
+	err = platform_driver_register(&ingenic_drm_driver);
+	if (IS_ENABLED(CONFIG_DRM_INGENIC_IPU) && err)
+		platform_driver_unregister(ingenic_ipu_driver_ptr);
+
+	return err;
 }
 module_init(ingenic_drm_init);
 

@@ -258,7 +258,12 @@ struct etnaviv_vram_mapping *etnaviv_gem
 		if (mapping->use == 0) {
 			mutex_lock(&mmu_context->lock);
 			if (mapping->context == mmu_context)
-				mapping->use += 1;
+				if (va && mapping->iova != va) {
+					etnaviv_iommu_reap_mapping(mapping);
+					mapping = NULL;
+				} else {
+					mapping->use += 1;
+				}
 			else
 				mapping = NULL;
 			mutex_unlock(&mmu_context->lock);

@@ -25,6 +25,7 @@ int efi_set_mapping_permissions(struct m
 ({									\
 	efi_virtmap_load();						\
 	__efi_fpsimd_begin();						\
+	spin_lock(&efi_rt_lock);					\
 })
 
 #undef arch_efi_call_virt

@@ -33,10 +34,12 @@ int efi_set_mapping_permissions(struct m
 
 #define arch_efi_call_virt_teardown()					\
 ({									\
+	spin_unlock(&efi_rt_lock);					\
 	__efi_fpsimd_end();						\
 	efi_virtmap_unload();						\
 })
 
+extern spinlock_t efi_rt_lock;
 efi_status_t __efi_rt_asm_wrapper(void *, const char *, ...);
 
 #define ARCH_EFI_IRQ_FLAGS_MASK (PSR_D_BIT | PSR_A_BIT | PSR_I_BIT | PSR_F_BIT)

@@ -144,3 +144,30 @@ asmlinkage efi_status_t efi_handle_corru
 	pr_err_ratelimited(FW_BUG "register x18 corrupted by EFI %s\n", f);
 	return s;
 }
+
+DEFINE_SPINLOCK(efi_rt_lock);
+
+asmlinkage u64 *efi_rt_stack_top __ro_after_init;
+
+/* EFI requires 8 KiB of stack space for runtime services */
+static_assert(THREAD_SIZE >= SZ_8K);
+
+static int __init arm64_efi_rt_init(void)
+{
+	void *p;
+
+	if (!efi_enabled(EFI_RUNTIME_SERVICES))
+		return 0;
+
+	p = __vmalloc_node(THREAD_SIZE, THREAD_ALIGN, GFP_KERNEL,
+			   NUMA_NO_NODE, &&l);
+l:	if (!p) {
+		pr_warn("Failed to allocate EFI runtime stack\n");
+		clear_bit(EFI_RUNTIME_SERVICES, &efi.flags);
+		return -ENOMEM;
+	}
+
+	efi_rt_stack_top = p + THREAD_SIZE;
+	return 0;
+}
+core_initcall(arm64_efi_rt_init);

@@ -222,13 +222,13 @@ void ext4_evict_inode(struct inode *inod
 
 	/*
 	 * For inodes with journalled data, transaction commit could have
-	 * dirtied the inode. Flush worker is ignoring it because of I_FREEING
-	 * flag but we still need to remove the inode from the writeback lists.
+	 * dirtied the inode. And for inodes with dioread_nolock, unwritten
+	 * extents converting worker could merge extents and also have dirtied
+	 * the inode. Flush worker is ignoring it because of I_FREEING flag but
+	 * we still need to remove the inode from the writeback lists.
 	 */
-	if (!list_empty_careful(&inode->i_io_list)) {
-		WARN_ON_ONCE(!ext4_should_journal_data(inode));
+	if (!list_empty_careful(&inode->i_io_list))
 		inode_io_list_del(inode);
-	}
 
 	/*
 	 * Protect us against freezing - iput() caller didn't have to have any

@@ -4473,9 +4473,17 @@ static int __ext4_get_inode_loc(struct s
 	inodes_per_block = EXT4_SB(sb)->s_inodes_per_block;
 	inode_offset = ((ino - 1) %
 			EXT4_INODES_PER_GROUP(sb));
-	block = ext4_inode_table(sb, gdp) + (inode_offset / inodes_per_block);
 	iloc->offset = (inode_offset % inodes_per_block) * EXT4_INODE_SIZE(sb);
 
+	block = ext4_inode_table(sb, gdp);
+	if ((block <= le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block)) ||
+	    (block >= ext4_blocks_count(EXT4_SB(sb)->s_es))) {
+		ext4_error(sb, "Invalid inode table block %llu in "
+			   "block_group %u", block, iloc->block_group);
+		return -EFSCORRUPTED;
+	}
+	block += (inode_offset / inodes_per_block);
+
 	bh = sb_getblk(sb, block);
 	if (unlikely(!bh))
 		return -ENOMEM;

@@ -412,7 +412,7 @@ void ext4_orphan_cleanup(struct super_bl
 		/* don't clear list on RO mount w/ errors */
 		if (es->s_last_orphan && !(s_flags & SB_RDONLY)) {
 			ext4_msg(sb, KERN_INFO, "Errors on filesystem, "
-				  "clearing orphan list.\n");
+				  "clearing orphan list.");
 			es->s_last_orphan = 0;
 		}
 		ext4_debug("Skipping orphan recovery on fs with errors.\n");

@@ -5287,14 +5287,15 @@ static int __ext4_fill_super(struct fs_c
 		goto failed_mount3a;
 	} else {
 		/* Nojournal mode, all journal mount options are illegal */
-		if (test_opt2(sb, EXPLICIT_JOURNAL_CHECKSUM)) {
+		if (test_opt(sb, JOURNAL_ASYNC_COMMIT)) {
 			ext4_msg(sb, KERN_ERR, "can't mount with "
-				 "journal_checksum, fs mounted w/o journal");
+				 "journal_async_commit, fs mounted w/o journal");
 			goto failed_mount3a;
 		}
-		if (test_opt(sb, JOURNAL_ASYNC_COMMIT)) {
+
+		if (test_opt2(sb, EXPLICIT_JOURNAL_CHECKSUM)) {
 			ext4_msg(sb, KERN_ERR, "can't mount with "
-				 "journal_async_commit, fs mounted w/o journal");
+				 "journal_checksum, fs mounted w/o journal");
 			goto failed_mount3a;
 		}
 		if (sbi->s_commit_interval != JBD2_DEFAULT_MAX_COMMIT_AGE*HZ) {

@@ -79,7 +79,7 @@ static int pagecache_write(struct inode
 		size_t n = min_t(size_t, count,
 				 PAGE_SIZE - offset_in_page(pos));
 		struct page *page;
-		void *fsdata;
+		void *fsdata = NULL;
 		int res;
 
 		res = aops->write_begin(NULL, mapping, pos, n, &page, &fsdata);

@@ -4225,7 +4225,8 @@ int ext4_truncate(struct inode *inode)
 
 	/* If we zero-out tail of the page, we have to create jinode for jbd2 */
 	if (inode->i_size & (inode->i_sb->s_blocksize - 1)) {
-		if (ext4_inode_attach_jinode(inode) < 0)
+		err = ext4_inode_attach_jinode(inode);
+		if (err)
 			goto out_trace;
 	}
 

@@ -558,7 +558,7 @@ enum {
  *
  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
  */
-#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
+#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
 
 static inline void ext4_check_flag_values(void)

@@ -2964,7 +2964,8 @@ int do_journal_get_write_access(handle_t
 typedef enum {
 	EXT4_IGET_NORMAL =	0,
 	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
-	EXT4_IGET_HANDLE = 	0x0002	/* Inode # is from a handle */
+	EXT4_IGET_HANDLE = 	0x0002,	/* Inode # is from a handle */
+	EXT4_IGET_BAD =		0x0004  /* Allow to iget a bad inode */
 } ext4_iget_flags;
 
 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,

@@ -5053,8 +5053,14 @@ struct inode *__ext4_iget(struct super_b
 	if (IS_CASEFOLDED(inode) && !ext4_has_feature_casefold(inode->i_sb))
 		ext4_error_inode(inode, function, line, 0,
 				 "casefold flag without casefold feature");
-	brelse(iloc.bh);
+	if (is_bad_inode(inode) && !(flags & EXT4_IGET_BAD)) {
+		ext4_error_inode(inode, function, line, 0,
+				 "bad inode without EXT4_IGET_BAD flag");
+		ret = -EUCLEAN;
+		goto bad_inode;
+	}
 
+	brelse(iloc.bh);
 	unlock_new_inode(inode);
 	return inode;
 

@@ -374,7 +374,8 @@ static long swap_inode_boot_loader(struc
 	blkcnt_t blocks;
 	unsigned short bytes;
 
-	inode_bl = ext4_iget(sb, EXT4_BOOT_LOADER_INO, EXT4_IGET_SPECIAL);
+	inode_bl = ext4_iget(sb, EXT4_BOOT_LOADER_INO,
+			EXT4_IGET_SPECIAL | EXT4_IGET_BAD);
 	if (IS_ERR(inode_bl))
 		return PTR_ERR(inode_bl);
 	ei_bl = EXT4_I(inode_bl);

@@ -6887,6 +6887,20 @@ static int ext4_quota_on(struct super_bl
 	return err;
 }
 
+static inline bool ext4_check_quota_inum(int type, unsigned long qf_inum)
+{
+	switch (type) {
+	case USRQUOTA:
+		return qf_inum == EXT4_USR_QUOTA_INO;
+	case GRPQUOTA:
+		return qf_inum == EXT4_GRP_QUOTA_INO;
+	case PRJQUOTA:
+		return qf_inum >= EXT4_GOOD_OLD_FIRST_INO;
+	default:
+		BUG();
+	}
+}
+
 static int ext4_quota_enable(struct super_block *sb, int type, int format_id,
 			     unsigned int flags)
 {

@@ -6903,9 +6917,16 @@ static int ext4_quota_enable(struct supe
 	if (!qf_inums[type])
 		return -EPERM;
 
+	if (!ext4_check_quota_inum(type, qf_inums[type])) {
+		ext4_error(sb, "Bad quota inum: %lu, type: %d",
+				qf_inums[type], type);
+		return -EUCLEAN;
+	}
+
 	qf_inode = ext4_iget(sb, qf_inums[type], EXT4_IGET_SPECIAL);
 	if (IS_ERR(qf_inode)) {
-		ext4_error(sb, "Bad quota inode # %lu", qf_inums[type]);
+		ext4_error(sb, "Bad quota inode: %lu, type: %d",
+				qf_inums[type], type);
 		return PTR_ERR(qf_inode);
 	}
 

@@ -6944,8 +6965,9 @@ int ext4_enable_quotas(struct super_bloc
 			if (err) {
 				ext4_warning(sb,
 					"Failed to enable quota tracking "
-					"(type=%d, err=%d). Please run "
-					"e2fsck to fix.", type, err);
+					"(type=%d, err=%d, ino=%lu). "
+					"Please run e2fsck to fix.", type,
+					err, qf_inums[type]);
 				for (type--; type >= 0; type--) {
 					struct inode *inode;
 

@@ -2324,6 +2324,8 @@ static int vfs_setup_quota_inode(struct
 	struct super_block *sb = inode->i_sb;
 	struct quota_info *dqopt = sb_dqopt(sb);
 
+	if (is_bad_inode(inode))
+		return -EUCLEAN;
 	if (!S_ISREG(inode->i_mode))
 		return -EACCES;
 	if (IS_RDONLY(inode))

@@ -1371,7 +1371,7 @@ retry:
 		if (count_reserved)
 			count_rsvd(inode, lblk, orig_es.es_len - len1 - len2,
 				   &orig_es, &rc);
-		goto out;
+		goto out_get_reserved;
 	}
 
 	if (len1 > 0) {

@@ -1413,6 +1413,7 @@ retry:
 		}
 	}
 
+out_get_reserved:
 	if (count_reserved)
 		*reserved = get_rsvd(inode, end, es, &rc);
 out:

@@ -2247,7 +2247,7 @@ static int ext4_parse_param(struct fs_co
 			return -EINVAL;
 		}
 
-		error = fs_lookup_param(fc, param, 1, &path);
+		error = fs_lookup_param(fc, param, 1, LOOKUP_FOLLOW, &path);
 		if (error) {
 			ext4_msg(NULL, KERN_ERR, "error: could not find "
 				 "journal device path");

@@ -138,15 +138,16 @@ EXPORT_SYMBOL(__fs_parse);
  * @fc: The filesystem context to log errors through.
  * @param: The parameter.
  * @want_bdev: T if want a blockdev
+ * @flags: Pathwalk flags passed to filename_lookup()
  * @_path: The result of the lookup
  */
 int fs_lookup_param(struct fs_context *fc,
 		    struct fs_parameter *param,
 		    bool want_bdev,
+		    unsigned int flags,
 		    struct path *_path)
 {
 	struct filename *f;
-	unsigned int flags = 0;
 	bool put_f;
 	int ret;
 

@@ -76,6 +76,7 @@ static inline int fs_parse(struct fs_con
 extern int fs_lookup_param(struct fs_context *fc,
 			   struct fs_parameter *param,
 			   bool want_bdev,
+			   unsigned int flags,
 			   struct path *_path);
 
 extern int lookup_constant(const struct constant_table tbl[], const char *name, int not_found);

@@ -335,6 +335,12 @@ stop_handle:
 	ext4_xattr_inode_array_free(ea_inode_array);
 	return;
 no_delete:
+	/*
+	 * Check out some where else accidentally dirty the evicting inode,
+	 * which may probably cause inode use-after-free issues later.
+	 */
+	WARN_ON_ONCE(!list_empty_careful(&inode->i_io_list));
+
 	if (!list_empty(&EXT4_I(inode)->i_fc_list))
 		ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_NOMEM, NULL);
 	ext4_clear_inode(inode);	/* We must guarantee clearing of inode... */

@@ -425,7 +425,7 @@ static long swap_inode_boot_loader(struc
 	/* Protect extent tree against block allocations via delalloc */
 	ext4_double_down_write_data_sem(inode, inode_bl);
 
-	if (inode_bl->i_nlink == 0) {
+	if (is_bad_inode(inode_bl) || !S_ISREG(inode_bl->i_mode)) {
 		/* this inode has never been used as a BOOT_LOADER */
 		set_nlink(inode_bl, 1);
 		i_uid_write(inode_bl, 0);

@@ -5724,7 +5724,7 @@ static struct inode *ext4_get_journal_in
 
 	ext4_debug("Journal inode found at %p: %lld bytes\n",
 		  journal_inode, journal_inode->i_size);
-	if (!S_ISREG(journal_inode->i_mode)) {
+	if (!S_ISREG(journal_inode->i_mode) || IS_ENCRYPTED(journal_inode)) {
 		ext4_msg(sb, KERN_ERR, "invalid journal inode");
 		iput(journal_inode);
 		return NULL;

@@ -420,25 +420,34 @@ static int __track_dentry_update(struct
 	struct __track_dentry_update_args *dentry_update =
 		(struct __track_dentry_update_args *)arg;
 	struct dentry *dentry = dentry_update->dentry;
-	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
+	struct inode *dir = dentry->d_parent->d_inode;
+	struct super_block *sb = inode->i_sb;
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
 
 	mutex_unlock(&ei->i_fc_lock);
+
+	if (IS_ENCRYPTED(dir)) {
+		ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_ENCRYPTED_FILENAME,
+					NULL);
+		mutex_lock(&ei->i_fc_lock);
+		return -EOPNOTSUPP;
+	}
+
 	node = kmem_cache_alloc(ext4_fc_dentry_cachep, GFP_NOFS);
 	if (!node) {
-		ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_NOMEM, NULL);
+		ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_NOMEM, NULL);
 		mutex_lock(&ei->i_fc_lock);
 		return -ENOMEM;
 	}
 
 	node->fcd_op = dentry_update->op;
-	node->fcd_parent = dentry->d_parent->d_inode->i_ino;
+	node->fcd_parent = dir->i_ino;
 	node->fcd_ino = inode->i_ino;
 	if (dentry->d_name.len > DNAME_INLINE_LEN) {
 		node->fcd_name.name = kmalloc(dentry->d_name.len, GFP_NOFS);
 		if (!node->fcd_name.name) {
 			kmem_cache_free(ext4_fc_dentry_cachep, node);
-			ext4_fc_mark_ineligible(inode->i_sb,
-				EXT4_FC_REASON_NOMEM, NULL);
+			ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_NOMEM, NULL);
 			mutex_lock(&ei->i_fc_lock);
 			return -ENOMEM;
 		}

@@ -2249,17 +2258,17 @@ void ext4_fc_init(struct super_block *sb
 	journal->j_fc_cleanup_callback = ext4_fc_cleanup;
 }
 
-static const char *fc_ineligible_reasons[] = {
-	"Extended attributes changed",
-	"Cross rename",
-	"Journal flag changed",
-	"Insufficient memory",
-	"Swap boot",
-	"Resize",
-	"Dir renamed",
-	"Falloc range op",
-	"Data journalling",
-	"FC Commit Failed"
+static const char * const fc_ineligible_reasons[] = {
+	[EXT4_FC_REASON_XATTR] = "Extended attributes changed",
+	[EXT4_FC_REASON_CROSS_RENAME] = "Cross rename",
+	[EXT4_FC_REASON_JOURNAL_FLAG_CHANGE] = "Journal flag changed",
+	[EXT4_FC_REASON_NOMEM] = "Insufficient memory",
+	[EXT4_FC_REASON_SWAP_BOOT] = "Swap boot",
+	[EXT4_FC_REASON_RESIZE] = "Resize",
+	[EXT4_FC_REASON_RENAME_DIR] = "Dir renamed",
+	[EXT4_FC_REASON_FALLOC_RANGE] = "Falloc range op",
+	[EXT4_FC_REASON_INODE_JOURNAL_DATA] = "Data journalling",
+	[EXT4_FC_REASON_ENCRYPTED_FILENAME] = "Encrypted filename",
 };
 
 int ext4_fc_info_show(struct seq_file *seq, void *v)

@@ -96,6 +96,7 @@ enum {
 	EXT4_FC_REASON_RENAME_DIR,
 	EXT4_FC_REASON_FALLOC_RANGE,
 	EXT4_FC_REASON_INODE_JOURNAL_DATA,
+	EXT4_FC_REASON_ENCRYPTED_FILENAME,
 	EXT4_FC_REASON_MAX
 };
 

@@ -104,6 +104,7 @@ TRACE_DEFINE_ENUM(EXT4_FC_REASON_RESIZE)
 TRACE_DEFINE_ENUM(EXT4_FC_REASON_RENAME_DIR);
 TRACE_DEFINE_ENUM(EXT4_FC_REASON_FALLOC_RANGE);
 TRACE_DEFINE_ENUM(EXT4_FC_REASON_INODE_JOURNAL_DATA);
+TRACE_DEFINE_ENUM(EXT4_FC_REASON_ENCRYPTED_FILENAME);
 TRACE_DEFINE_ENUM(EXT4_FC_REASON_MAX);
 
 #define show_fc_reason(reason)						\

@@ -116,7 +117,8 @@ TRACE_DEFINE_ENUM(EXT4_FC_REASON_MAX);
 		{ EXT4_FC_REASON_RESIZE,	"RESIZE"},		\
 		{ EXT4_FC_REASON_RENAME_DIR,	"RENAME_DIR"},		\
 		{ EXT4_FC_REASON_FALLOC_RANGE,	"FALLOC_RANGE"},	\
-		{ EXT4_FC_REASON_INODE_JOURNAL_DATA,	"INODE_JOURNAL_DATA"})
+		{ EXT4_FC_REASON_INODE_JOURNAL_DATA,	"INODE_JOURNAL_DATA"}, \
+		{ EXT4_FC_REASON_ENCRYPTED_FILENAME,	"ENCRYPTED_FILENAME"})
 
 TRACE_EVENT(ext4_other_inode_update_time,
 	TP_PROTO(struct inode *inode, ino_t orig_ino),

@@ -2764,7 +2766,7 @@ TRACE_EVENT(ext4_fc_stats,
 	),
 
 	TP_printk("dev %d,%d fc ineligible reasons:\n"
-		  "%s:%u, %s:%u, %s:%u, %s:%u, %s:%u, %s:%u, %s:%u, %s:%u, %s:%u "
+		  "%s:%u, %s:%u, %s:%u, %s:%u, %s:%u, %s:%u, %s:%u, %s:%u, %s:%u, %s:%u"
 		  "num_commits:%lu, ineligible: %lu, numblks: %lu",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
 		  FC_REASON_NAME_STAT(EXT4_FC_REASON_XATTR),

@@ -2776,6 +2778,7 @@ TRACE_EVENT(ext4_fc_stats,
 		  FC_REASON_NAME_STAT(EXT4_FC_REASON_RENAME_DIR),
 		  FC_REASON_NAME_STAT(EXT4_FC_REASON_FALLOC_RANGE),
 		  FC_REASON_NAME_STAT(EXT4_FC_REASON_INODE_JOURNAL_DATA),
+		  FC_REASON_NAME_STAT(EXT4_FC_REASON_ENCRYPTED_FILENAME),
 		  __entry->fc_commits, __entry->fc_ineligible_commits,
 		  __entry->fc_numblks)
 );

@@ -737,6 +737,9 @@ static u8 *ext4_fc_reserve_space(struct
 		*crc = ext4_chksum(sbi, *crc, tl, EXT4_FC_TAG_BASE_LEN);
 	if (pad_len > 0)
 		ext4_fc_memzero(sb, tl + 1, pad_len, crc);
+	/* Don't leak uninitialized memory in the unused last byte. */
+	*((u8 *)(tl + 1) + pad_len) = 0;
+
 	ext4_fc_submit_bh(sb, false);
 
 	ret = jbd2_fc_get_buf(EXT4_SB(sb)->s_journal, &bh);

@@ -793,6 +796,8 @@ static int ext4_fc_write_tail(struct sup
 	dst += sizeof(tail.fc_tid);
 	tail.fc_crc = cpu_to_le32(crc);
 	ext4_fc_memcpy(sb, dst, &tail.fc_crc, sizeof(tail.fc_crc), NULL);
+	dst += sizeof(tail.fc_crc);
+	memset(dst, 0, bsize - off); /* Don't leak uninitialized memory. */
 
 	ext4_fc_submit_bh(sb, true);
 

@@ -3620,8 +3620,8 @@ extern void ext4_initialize_dirent_tail(
 					unsigned int blocksize);
 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
 				      struct buffer_head *bh);
-extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name,
-			 struct inode *inode);
+extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
+			 struct inode *inode, struct dentry *dentry);
 extern int __ext4_link(struct inode *dir, struct inode *inode,
 		       struct dentry *dentry);
 

@@ -1402,7 +1402,7 @@ static int ext4_fc_replay_unlink(struct
 		return 0;
 	}
 
-	ret = __ext4_unlink(NULL, old_parent, &entry, inode);
+	ret = __ext4_unlink(old_parent, &entry, inode, NULL);
 	/* -ENOENT ok coz it might not exist anymore. */
 	if (ret == -ENOENT)
 		ret = 0;

96 lines not on this page · raw message

@@ -1991,32 +1991,31 @@ void ext4_fc_replay_cleanup(struct super
 	kfree(sbi->s_fc_replay_state.fc_modified_inodes);
 }
 
-static inline bool ext4_fc_tag_len_isvalid(struct ext4_fc_tl *tl,
-					   u8 *val, u8 *end)
+static bool ext4_fc_value_len_isvalid(struct ext4_sb_info *sbi,
+				      int tag, int len)
 {
-	if (val + tl->fc_len > end)
-		return false;
-
-	/* Here only check ADD_RANGE/TAIL/HEAD which will read data when do
-	 * journal rescan before do CRC check. Other tags length check will
-	 * rely on CRC check.
-	 */
-	switch (tl->fc_tag) {
+	switch (tag) {
 	case EXT4_FC_TAG_ADD_RANGE:
-		return (sizeof(struct ext4_fc_add_range) == tl->fc_len);
-	case EXT4_FC_TAG_TAIL:
-		return (sizeof(struct ext4_fc_tail) <= tl->fc_len);
-	case EXT4_FC_TAG_HEAD:
-		return (sizeof(struct ext4_fc_head) == tl->fc_len);
+		return len == sizeof(struct ext4_fc_add_range);
 	case EXT4_FC_TAG_DEL_RANGE:
+		return len == sizeof(struct ext4_fc_del_range);
+	case EXT4_FC_TAG_CREAT:
 	case EXT4_FC_TAG_LINK:
 	case EXT4_FC_TAG_UNLINK:
-	case EXT4_FC_TAG_CREAT:
+		len -= sizeof(struct ext4_fc_dentry_info);
+		return len >= 1 && len <= EXT4_NAME_LEN;
 	case EXT4_FC_TAG_INODE:
+		len -= sizeof(struct ext4_fc_inode);
+		return len >= EXT4_GOOD_OLD_INODE_SIZE &&
+			len <= sbi->s_inode_size;
 	case EXT4_FC_TAG_PAD:
-	default:
-		return true;
+		return true; /* padding can have any length */
+	case EXT4_FC_TAG_TAIL:
+		return len >= sizeof(struct ext4_fc_tail);
+	case EXT4_FC_TAG_HEAD:
+		return len == sizeof(struct ext4_fc_head);
 	}
+	return false;
 }
 
 /*

@@ -2079,7 +2078,8 @@ static int ext4_fc_replay_scan(journal_t
 	     cur = cur + EXT4_FC_TAG_BASE_LEN + tl.fc_len) {
 		ext4_fc_get_tl(&tl, cur);
 		val = cur + EXT4_FC_TAG_BASE_LEN;
-		if (!ext4_fc_tag_len_isvalid(&tl, val, end)) {
+		if (tl.fc_len > end - val ||
+		    !ext4_fc_value_len_isvalid(sbi, tl.fc_tag, tl.fc_len)) {
 			ret = state->fc_replay_num_tags ?
 				JBD2_FC_REPLAY_STOP : -ECANCELED;
 			goto out_err;

@@ -58,7 +58,7 @@ struct ext4_fc_dentry_info {
 	__u8 fc_dname[];
 };
 
-/* Value structure for EXT4_FC_TAG_INODE and EXT4_FC_TAG_INODE_PARTIAL. */
+/* Value structure for EXT4_FC_TAG_INODE. */
 struct ext4_fc_inode {
 	__le32 fc_ino;
 	__u8 fc_raw_inode[];

@@ -675,6 +675,15 @@ static void ext4_fc_submit_bh(struct sup
 
 /* Ext4 commit path routines */
 
+/* memcpy to fc reserved space and update CRC */
+static void *ext4_fc_memcpy(struct super_block *sb, void *dst, const void *src,
+				int len, u32 *crc)
+{
+	if (crc)
+		*crc = ext4_chksum(EXT4_SB(sb), *crc, src, len);
+	return memcpy(dst, src, len);
+}
+
 /* memzero and update CRC */
 static void *ext4_fc_memzero(struct super_block *sb, void *dst, int len,
 				u32 *crc)

@@ -700,12 +709,13 @@ static void *ext4_fc_memzero(struct supe
  */
 static u8 *ext4_fc_reserve_space(struct super_block *sb, int len, u32 *crc)
 {
-	struct ext4_fc_tl *tl;
+	struct ext4_fc_tl tl;
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	struct buffer_head *bh;
 	int bsize = sbi->s_journal->j_blocksize;
 	int ret, off = sbi->s_fc_bytes % bsize;
 	int pad_len;
+	u8 *dst;
 
 	/*
 	 * After allocating len, we should have space at least for a 0 byte

@@ -729,16 +739,18 @@ static u8 *ext4_fc_reserve_space(struct
 		return sbi->s_fc_bh->b_data + off;
 	}
 	/* Need to add PAD tag */
-	tl = (struct ext4_fc_tl *)(sbi->s_fc_bh->b_data + off);
-	tl->fc_tag = cpu_to_le16(EXT4_FC_TAG_PAD);
+	dst = sbi->s_fc_bh->b_data + off;
+	tl.fc_tag = cpu_to_le16(EXT4_FC_TAG_PAD);
 	pad_len = bsize - off - 1 - EXT4_FC_TAG_BASE_LEN;
-	tl->fc_len = cpu_to_le16(pad_len);
-	if (crc)
-		*crc = ext4_chksum(sbi, *crc, tl, EXT4_FC_TAG_BASE_LEN);
-	if (pad_len > 0)
-		ext4_fc_memzero(sb, tl + 1, pad_len, crc);
+	tl.fc_len = cpu_to_le16(pad_len);
+	ext4_fc_memcpy(sb, dst, &tl, EXT4_FC_TAG_BASE_LEN, crc);
+	dst += EXT4_FC_TAG_BASE_LEN;
+	if (pad_len > 0) {
+		ext4_fc_memzero(sb, dst, pad_len, crc);
+		dst += pad_len;
+	}
 	/* Don't leak uninitialized memory in the unused last byte. */
-	*((u8 *)(tl + 1) + pad_len) = 0;
+	*dst = 0;
 
 	ext4_fc_submit_bh(sb, false);
 

@@ -750,15 +762,6 @@ static u8 *ext4_fc_reserve_space(struct
 	return sbi->s_fc_bh->b_data;
 }
 
-/* memcpy to fc reserved space and update CRC */
-static void *ext4_fc_memcpy(struct super_block *sb, void *dst, const void *src,
-				int len, u32 *crc)
-{
-	if (crc)
-		*crc = ext4_chksum(EXT4_SB(sb), *crc, src, len);
-	return memcpy(dst, src, len);
-}
-
 /*
  * Complete a fast commit by writing tail tag.
  *

@@ -714,43 +714,43 @@ static u8 *ext4_fc_reserve_space(struct
 	struct buffer_head *bh;
 	int bsize = sbi->s_journal->j_blocksize;
 	int ret, off = sbi->s_fc_bytes % bsize;
-	int pad_len;
+	int remaining;
 	u8 *dst;
 
 	/*
-	 * After allocating len, we should have space at least for a 0 byte
-	 * padding.
+	 * If 'len' is too long to fit in any block alongside a PAD tlv, then we
+	 * cannot fulfill the request.
 	 */
-	if (len + EXT4_FC_TAG_BASE_LEN > bsize)
+	if (len > bsize - EXT4_FC_TAG_BASE_LEN)
 		return NULL;
 
-	if (bsize - off - 1 > len + EXT4_FC_TAG_BASE_LEN) {
-		/*
-		 * Only allocate from current buffer if we have enough space for
-		 * this request AND we have space to add a zero byte padding.
-		 */
-		if (!sbi->s_fc_bh) {
-			ret = jbd2_fc_get_buf(EXT4_SB(sb)->s_journal, &bh);
-			if (ret)
-				return NULL;
-			sbi->s_fc_bh = bh;
-		}
-		sbi->s_fc_bytes += len;
-		return sbi->s_fc_bh->b_data + off;
+	if (!sbi->s_fc_bh) {
+		ret = jbd2_fc_get_buf(EXT4_SB(sb)->s_journal, &bh);
+		if (ret)
+			return NULL;
+		sbi->s_fc_bh = bh;
 	}
-	/* Need to add PAD tag */
 	dst = sbi->s_fc_bh->b_data + off;
+
+	/*
+	 * Allocate the bytes in the current block if we can do so while still
+	 * leaving enough space for a PAD tlv.
+	 */
+	remaining = bsize - EXT4_FC_TAG_BASE_LEN - off;
+	if (len <= remaining) {
+		sbi->s_fc_bytes += len;
+		return dst;
+	}
+
+	/*
+	 * Else, terminate the current block with a PAD tlv, then allocate a new
+	 * block and allocate the bytes at the start of that new block.
+	 */
+
 	tl.fc_tag = cpu_to_le16(EXT4_FC_TAG_PAD);
-	pad_len = bsize - off - 1 - EXT4_FC_TAG_BASE_LEN;
-	tl.fc_len = cpu_to_le16(pad_len);
+	tl.fc_len = cpu_to_le16(remaining);
 	ext4_fc_memcpy(sb, dst, &tl, EXT4_FC_TAG_BASE_LEN, crc);
-	dst += EXT4_FC_TAG_BASE_LEN;
-	if (pad_len > 0) {
-		ext4_fc_memzero(sb, dst, pad_len, crc);
-		dst += pad_len;
-	}
-	/* Don't leak uninitialized memory in the unused last byte. */
-	*dst = 0;
+	ext4_fc_memzero(sb, dst + EXT4_FC_TAG_BASE_LEN, remaining, crc);
 
 	ext4_fc_submit_bh(sb, false);
 

@@ -758,7 +758,7 @@ static u8 *ext4_fc_reserve_space(struct
 	if (ret)
 		return NULL;
 	sbi->s_fc_bh = bh;
-	sbi->s_fc_bytes = (sbi->s_fc_bytes / bsize + 1) * bsize + len;
+	sbi->s_fc_bytes += bsize - off + len;
 	return sbi->s_fc_bh->b_data;
 }
 

@@ -789,7 +789,7 @@ static int ext4_fc_write_tail(struct sup
 	off = sbi->s_fc_bytes % bsize;
 
 	tl.fc_tag = cpu_to_le16(EXT4_FC_TAG_TAIL);
-	tl.fc_len = cpu_to_le16(bsize - off - 1 + sizeof(struct ext4_fc_tail));
+	tl.fc_len = cpu_to_le16(bsize - off + sizeof(struct ext4_fc_tail));
 	sbi->s_fc_bytes = round_up(sbi->s_fc_bytes, bsize);
 
 	ext4_fc_memcpy(sb, dst, &tl, EXT4_FC_TAG_BASE_LEN, &crc);

@@ -2056,7 +2056,7 @@ static int ext4_fc_replay_scan(journal_t
 	state = &sbi->s_fc_replay_state;
 
 	start = (u8 *)bh->b_data;
-	end = (__u8 *)bh->b_data + journal->j_blocksize - 1;
+	end = start + journal->j_blocksize;
 
 	if (state->fc_replay_expected_off == 0) {
 		state->fc_cur_tag = 0;

@@ -2077,7 +2077,7 @@ static int ext4_fc_replay_scan(journal_t
 	}
 
 	state->fc_replay_expected_off++;
-	for (cur = start; cur < end - EXT4_FC_TAG_BASE_LEN;
+	for (cur = start; cur <= end - EXT4_FC_TAG_BASE_LEN;
 	     cur = cur + EXT4_FC_TAG_BASE_LEN + tl.fc_len) {
 		ext4_fc_get_tl(&tl, cur);
 		val = cur + EXT4_FC_TAG_BASE_LEN;

@@ -2195,9 +2195,9 @@ static int ext4_fc_replay(journal_t *jou
 #endif
 
 	start = (u8 *)bh->b_data;
-	end = (__u8 *)bh->b_data + journal->j_blocksize - 1;
+	end = start + journal->j_blocksize;
 
-	for (cur = start; cur < end - EXT4_FC_TAG_BASE_LEN;
+	for (cur = start; cur <= end - EXT4_FC_TAG_BASE_LEN;
 	     cur = cur + EXT4_FC_TAG_BASE_LEN + tl.fc_len) {
 		ext4_fc_get_tl(&tl, cur);
 		val = cur + EXT4_FC_TAG_BASE_LEN;

@@ -1323,6 +1323,7 @@ static struct inode *ext4_alloc_inode(st
 		return NULL;
 
 	inode_set_iversion(&ei->vfs_inode, 1);
+	ei->i_flags = 0;
 	spin_lock_init(&ei->i_raw_lock);
 	INIT_LIST_HEAD(&ei->i_prealloc_list);
 	atomic_set(&ei->i_prealloc_active, 0);

@@ -3798,6 +3798,9 @@ static int ext4_rename(struct user_names
 	retval = dquot_initialize(old.dir);
 	if (retval)
 		return retval;
+	retval = dquot_initialize(old.inode);
+	if (retval)
+		return retval;
 	retval = dquot_initialize(new.dir);
 	if (retval)
 		return retval;

@@ -1159,14 +1159,16 @@ static int ext4_ioctl_getuuid(struct ext
 		return -EINVAL;
 	}
 
-	if (fsuuid.fsu_len != UUID_SIZE || fsuuid.fsu_flags != 0)
+	if (fsuuid.fsu_len < UUID_SIZE || fsuuid.fsu_flags != 0)
 		return -EINVAL;
 
 	lock_buffer(sbi->s_sbh);
 	memcpy(uuid, sbi->s_es->s_uuid, UUID_SIZE);
 	unlock_buffer(sbi->s_sbh);
 
-	if (copy_to_user(&ufsuuid->fsu_uuid[0], uuid, UUID_SIZE))
+	fsuuid.fsu_len = UUID_SIZE;
+	if (copy_to_user(ufsuuid, &fsuuid, sizeof(fsuuid)) ||
+	    copy_to_user(&ufsuuid->fsu_uuid[0], uuid, UUID_SIZE))
 		return -EFAULT;
 	return 0;
 }

@@ -5799,6 +5799,14 @@ int ext4_clu_mapped(struct inode *inode,
 	struct ext4_extent *extent;
 	ext4_lblk_t first_lblk, first_lclu, last_lclu;
 
+	/*
+	 * if data can be stored inline, the logical cluster isn't
+	 * mapped - no physical clusters have been allocated, and the
+	 * file has no extents
+	 */
+	if (ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA))
+		return 0;
+
 	/* search for the extent closest to the first block in the cluster */
 	path = ext4_find_extent(inode, EXT4_C2B(sbi, lclu), NULL, 0);
 	if (IS_ERR(path)) {

@@ -1596,8 +1596,8 @@ exit_journal:
 		int meta_bg = ext4_has_feature_meta_bg(sb);
 		sector_t old_gdb = 0;
 
-		update_backups(sb, sbi->s_sbh->b_blocknr, (char *)es,
-			       sizeof(struct ext4_super_block), 0);
+		update_backups(sb, ext4_group_first_block_no(sb, 0),
+			       (char *)es, sizeof(struct ext4_super_block), 0);
 		for (; gdb_num <= gdb_num_end; gdb_num++) {
 			struct buffer_head *gdb_bh;
 

@@ -1808,7 +1808,7 @@ errout:
 		if (test_opt(sb, DEBUG))
 			printk(KERN_DEBUG "EXT4-fs: extended group to %llu "
 			       "blocks\n", ext4_blocks_count(es));
-		update_backups(sb, EXT4_SB(sb)->s_sbh->b_blocknr,
+		update_backups(sb, ext4_group_first_block_no(sb, 0),
 			       (char *)es, sizeof(struct ext4_super_block), 0);
 	}
 	return err;

@@ -148,6 +148,7 @@ static Indirect *ext4_get_branch(struct
 	struct super_block *sb = inode->i_sb;
 	Indirect *p = chain;
 	struct buffer_head *bh;
+	unsigned int key;
 	int ret = -EIO;
 
 	*err = 0;

@@ -156,7 +157,13 @@ static Indirect *ext4_get_branch(struct
 	if (!p->key)
 		goto no_block;
 	while (--depth) {
-		bh = sb_getblk(sb, le32_to_cpu(p->key));
+		key = le32_to_cpu(p->key);
+		if (key > ext4_blocks_count(EXT4_SB(sb)->s_es)) {
+			/* the block was out of range */
+			ret = -EFSCORRUPTED;
+			goto failure;
+		}
+		bh = sb_getblk(sb, key);
 		if (unlikely(!bh)) {
 			ret = -ENOMEM;
 			goto failure;

@@ -1476,8 +1476,6 @@ static void ext4_update_super(struct sup
 	 * active. */
 	ext4_r_blocks_count_set(es, ext4_r_blocks_count(es) +
 				reserved_blocks);
-	ext4_superblock_csum_set(sb);
-	unlock_buffer(sbi->s_sbh);
 
 	/* Update the free space counts */
 	percpu_counter_add(&sbi->s_freeclusters_counter,

@@ -1513,6 +1511,8 @@ static void ext4_update_super(struct sup
 		ext4_calculate_overhead(sb);
 	es->s_overhead_clusters = cpu_to_le32(sbi->s_overhead);
 
+	ext4_superblock_csum_set(sb);
+	unlock_buffer(sbi->s_sbh);
 	if (test_opt(sb, DEBUG))
 		printk(KERN_DEBUG "EXT4-fs: added group %u:"
 		       "%llu blocks(%llu free %llu reserved)\n", flex_gd->count,

@@ -1154,9 +1154,10 @@ static int ext4_ioctl_getuuid(struct ext
 
 	if (fsuuid.fsu_len == 0) {
 		fsuuid.fsu_len = UUID_SIZE;
-		if (copy_to_user(ufsuuid, &fsuuid, sizeof(fsuuid.fsu_len)))
+		if (copy_to_user(&ufsuuid->fsu_len, &fsuuid.fsu_len,
+					sizeof(fsuuid.fsu_len)))
 			return -EFAULT;
-		return -EINVAL;
+		return 0;
 	}
 
 	if (fsuuid.fsu_len < UUID_SIZE || fsuuid.fsu_flags != 0)

@@ -1110,6 +1110,16 @@ exit_free:
 	return err;
 }
 
+static inline void ext4_set_block_group_nr(struct super_block *sb, char *data,
+					   ext4_group_t group)
+{
+	struct ext4_super_block *es = (struct ext4_super_block *) data;
+
+	es->s_block_group_nr = cpu_to_le16(group);
+	if (ext4_has_metadata_csum(sb))
+		es->s_checksum = ext4_superblock_csum(sb, es);
+}
+
 /*
  * Update the backup copies of the ext4 metadata.  These don't need to be part
  * of the main resize transaction, because e2fsck will re-write them if there

@@ -1158,7 +1168,8 @@ static void update_backups(struct super_
 	while (group < sbi->s_groups_count) {
 		struct buffer_head *bh;
 		ext4_fsblk_t backup_block;
-		struct ext4_super_block *es;
+		int has_super = ext4_bg_has_super(sb, group);
+		ext4_fsblk_t first_block = ext4_group_first_block_no(sb, group);
 
 		/* Out of journal space, and can't get more - abort - so sad */
 		err = ext4_resize_ensure_credits_batch(handle, 1);

@@ -1168,8 +1179,7 @@ static void update_backups(struct super_
 		if (meta_bg == 0)
 			backup_block = ((ext4_fsblk_t)group) * bpg + blk_off;
 		else
-			backup_block = (ext4_group_first_block_no(sb, group) +
-					ext4_bg_has_super(sb, group));
+			backup_block = first_block + has_super;
 
 		bh = sb_getblk(sb, backup_block);
 		if (unlikely(!bh)) {

@@ -1187,10 +1197,8 @@ static void update_backups(struct super_
 		memcpy(bh->b_data, data, size);
 		if (rest)
 			memset(bh->b_data + size, 0, rest);
-		es = (struct ext4_super_block *) bh->b_data;
-		es->s_block_group_nr = cpu_to_le16(group);
-		if (ext4_has_metadata_csum(sb))
-			es->s_checksum = ext4_superblock_csum(sb, es);
+		if (has_super && (backup_block == first_block))
+			ext4_set_block_group_nr(sb, bh->b_data, group);
 		set_buffer_uptodate(bh);
 		unlock_buffer(bh);
 		err = ext4_handle_dirty_metadata(handle, NULL, bh);

@@ -2070,19 +2070,11 @@ inserted:
 
 			goal = ext4_group_first_block_no(sb,
 						EXT4_I(inode)->i_block_group);
-
-			/* non-extent files can't have physical blocks past 2^32 */
-			if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
-				goal = goal & EXT4_MAX_BLOCK_FILE_PHYS;
-
 			block = ext4_new_meta_blocks(handle, inode, goal, 0,
 						     NULL, &error);
 			if (error)
 				goto cleanup;
 
-			if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
-				BUG_ON(block > EXT4_MAX_BLOCK_FILE_PHYS);
-
 			ea_idebug(inode, "creating block %llu",
 				  (unsigned long long)block);
 

@@ -1281,7 +1281,7 @@ retry_ref:
 				ce = mb_cache_entry_get(ea_block_cache, hash,
 							bh->b_blocknr);
 				if (ce) {
-					ce->e_reusable = 1;
+					set_bit(MBE_REUSABLE_B, &ce->e_flags);
 					mb_cache_entry_put(ea_block_cache, ce);
 				}
 			}

@@ -2042,7 +2042,7 @@ inserted:
 				}
 				BHDR(new_bh)->h_refcount = cpu_to_le32(ref);
 				if (ref == EXT4_XATTR_REFCOUNT_MAX)
-					ce->e_reusable = 0;
+					clear_bit(MBE_REUSABLE_B, &ce->e_flags);
 				ea_bdebug(new_bh, "reusing; refcount now=%d",
 					  ref);
 				ext4_xattr_block_csum_set(inode, new_bh);

@@ -100,8 +100,9 @@ int mb_cache_entry_create(struct mb_cach
 	atomic_set(&entry->e_refcnt, 2);
 	entry->e_key = key;
 	entry->e_value = value;
-	entry->e_reusable = reusable;
-	entry->e_referenced = 0;
+	entry->e_flags = 0;
+	if (reusable)
+		set_bit(MBE_REUSABLE_B, &entry->e_flags);
 	head = mb_cache_entry_head(cache, key);
 	hlist_bl_lock(head);
 	hlist_bl_for_each_entry(dup, dup_node, head, e_hash_list) {

@@ -165,7 +166,8 @@ static struct mb_cache_entry *__entry_fi
 	while (node) {
 		entry = hlist_bl_entry(node, struct mb_cache_entry,
 				       e_hash_list);
-		if (entry->e_key == key && entry->e_reusable &&
+		if (entry->e_key == key &&
+		    test_bit(MBE_REUSABLE_B, &entry->e_flags) &&
 		    atomic_inc_not_zero(&entry->e_refcnt))
 			goto out;
 		node = node->next;

@@ -284,7 +286,7 @@ EXPORT_SYMBOL(mb_cache_entry_delete_or_g
 void mb_cache_entry_touch(struct mb_cache *cache,
 			  struct mb_cache_entry *entry)
 {
-	entry->e_referenced = 1;
+	set_bit(MBE_REFERENCED_B, &entry->e_flags);
 }
 EXPORT_SYMBOL(mb_cache_entry_touch);
 

@@ -309,9 +311,9 @@ static unsigned long mb_cache_shrink(str
 		entry = list_first_entry(&cache->c_list,
 					 struct mb_cache_entry, e_list);
 		/* Drop initial hash reference if there is no user */
-		if (entry->e_referenced ||
+		if (test_bit(MBE_REFERENCED_B, &entry->e_flags) ||
 		    atomic_cmpxchg(&entry->e_refcnt, 1, 0) != 1) {
-			entry->e_referenced = 0;
+			clear_bit(MBE_REFERENCED_B, &entry->e_flags);
 			list_move_tail(&entry->e_list, &cache->c_list);
 			continue;
 		}

@@ -10,6 +10,12 @@
 
 struct mb_cache;
 
+/* Cache entry flags */
+enum {
+	MBE_REFERENCED_B = 0,
+	MBE_REUSABLE_B
+};
+
 struct mb_cache_entry {
 	/* List of entries in cache - protected by cache->c_list_lock */
 	struct list_head	e_list;

@@ -26,8 +32,7 @@ struct mb_cache_entry {
 	atomic_t		e_refcnt;
 	/* Key in hash - stable during lifetime of the entry */
 	u32			e_key;
-	u32			e_referenced:1;
-	u32			e_reusable:1;
+	unsigned long		e_flags;
 	/* User provided value - stable during lifetime of the entry */
 	u64			e_value;
 };

@@ -1315,7 +1315,8 @@ static int ext4_write_end(struct file *f
 
 	trace_ext4_write_end(inode, pos, len, copied);
 
-	if (ext4_has_inline_data(inode))
+	if (ext4_has_inline_data(inode) &&
+	    ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA))
 		return ext4_write_inline_data_end(inode, pos, len, copied, page);
 
 	copied = block_write_end(file, mapping, pos, len, copied, page, fsdata);

@@ -1441,6 +1441,9 @@ static struct inode *ext4_xattr_inode_cr
 		if (!err)
 			err = ext4_inode_attach_jinode(ea_inode);
 		if (err) {
+			if (ext4_xattr_inode_dec_ref(handle, ea_inode))
+				ext4_warning_inode(ea_inode,
+					"cleanup dec ref error %d", err);
 			iput(ea_inode);
 			return ERR_PTR(err);
 		}

@@ -732,6 +732,10 @@ static int ext4_ioctl_setproject(struct
 	if (ext4_is_quota_file(inode))
 		return err;
 
+	err = dquot_initialize(inode);
+	if (err)
+		return err;
+
 	err = ext4_get_inode_loc(inode, &iloc);
 	if (err)
 		return err;

@@ -747,10 +751,6 @@ static int ext4_ioctl_setproject(struct
 		brelse(iloc.bh);
 	}
 
-	err = dquot_initialize(inode);
-	if (err)
-		return err;
-
 	handle = ext4_journal_start(inode, EXT4_HT_QUOTA,
 		EXT4_QUOTA_INIT_BLOCKS(sb) +
 		EXT4_QUOTA_DEL_BLOCKS(sb) + 3);

@@ -5875,6 +5875,14 @@ static int __ext4_expand_extra_isize(str
 		return 0;
 	}
 
+	/*
+	 * We may need to allocate external xattr block so we need quotas
+	 * initialized. Here we can be called with various locks held so we
+	 * cannot affort to initialize quotas ourselves. So just bail.
+	 */
+	if (dquot_initialize_needed(inode))
+		return -EAGAIN;
+
 	/* try to expand with EAs present */
 	error = ext4_expand_extra_isize_ea(inode, new_extra_isize,
 					   raw_inode, handle);

@@ -2550,7 +2550,7 @@ static int ext4_xattr_move_to_block(hand
 
 	is = kzalloc(sizeof(struct ext4_xattr_ibody_find), GFP_NOFS);
 	bs = kzalloc(sizeof(struct ext4_xattr_block_find), GFP_NOFS);
-	buffer = kmalloc(value_size, GFP_NOFS);
+	buffer = kvmalloc(value_size, GFP_NOFS);
 	b_entry_name = kmalloc(entry->e_name_len + 1, GFP_NOFS);
 	if (!is || !bs || !buffer || !b_entry_name) {
 		error = -ENOMEM;

@@ -2602,7 +2602,7 @@ static int ext4_xattr_move_to_block(hand
 	error = 0;
 out:
 	kfree(b_entry_name);
-	kfree(buffer);
+	kvfree(buffer);
 	if (is)
 		brelse(is->iloc.bh);
 	if (bs)

@@ -761,6 +761,9 @@ bool i915_gem_object_needs_ccs_pages(str
 	if (!HAS_FLAT_CCS(to_i915(obj->base.dev)))
 		return false;
 
+	if (obj->flags & I915_BO_ALLOC_CCS_AUX)
+		return true;
+
 	for (i = 0; i < obj->mm.n_placements; i++) {
 		/* Compression is not allowed for the objects with smem placement */
 		if (obj->mm.placements[i]->type == INTEL_MEMORY_SYSTEM)

@@ -327,16 +327,18 @@ struct drm_i915_gem_object {
  * dealing with userspace objects the CPU fault handler is free to ignore this.
  */
 #define I915_BO_ALLOC_GPU_ONLY	  BIT(6)
+#define I915_BO_ALLOC_CCS_AUX	  BIT(7)
 #define I915_BO_ALLOC_FLAGS (I915_BO_ALLOC_CONTIGUOUS | \
 			     I915_BO_ALLOC_VOLATILE | \
 			     I915_BO_ALLOC_CPU_CLEAR | \
 			     I915_BO_ALLOC_USER | \
 			     I915_BO_ALLOC_PM_VOLATILE | \
 			     I915_BO_ALLOC_PM_EARLY | \
-			     I915_BO_ALLOC_GPU_ONLY)
-#define I915_BO_READONLY          BIT(7)
-#define I915_TILING_QUIRK_BIT     8 /* unknown swizzling; do not release! */
-#define I915_BO_PROTECTED         BIT(9)
+			     I915_BO_ALLOC_GPU_ONLY | \
+			     I915_BO_ALLOC_CCS_AUX)
+#define I915_BO_READONLY          BIT(8)
+#define I915_TILING_QUIRK_BIT     9 /* unknown swizzling; do not release! */
+#define I915_BO_PROTECTED         BIT(10)
 	/**
 	 * @mem_flags - Mutable placement-related flags
 	 *

@@ -50,6 +50,7 @@ static int i915_ttm_backup(struct i915_g
 		container_of(bo->bdev, typeof(*i915), bdev);
 	struct drm_i915_gem_object *backup;
 	struct ttm_operation_ctx ctx = {};
+	unsigned int flags;
 	int err = 0;
 
 	if (bo->resource->mem_type == I915_PL_SYSTEM || obj->ttm.backup)

@@ -65,7 +66,22 @@ static int i915_ttm_backup(struct i915_g
 	if (obj->flags & I915_BO_ALLOC_PM_VOLATILE)
 		return 0;
 
-	backup = i915_gem_object_create_shmem(i915, obj->base.size);
+	/*
+	 * It seems that we might have some framebuffers still pinned at this
+	 * stage, but for such objects we might also need to deal with the CCS
+	 * aux state. Make sure we force the save/restore of the CCS state,
+	 * otherwise we might observe display corruption, when returning from
+	 * suspend.
+	 */
+	flags = 0;
+	if (i915_gem_object_needs_ccs_pages(obj)) {
+		WARN_ON_ONCE(!i915_gem_object_is_framebuffer(obj));
+		WARN_ON_ONCE(!pm_apply->allow_gpu);
+
+		flags = I915_BO_ALLOC_CCS_AUX;
+	}
+	backup = i915_gem_object_create_region(i915->mm.regions[INTEL_REGION_SMEM],
+					       obj->base.size, 0, flags);
 	if (IS_ERR(backup))
 		return PTR_ERR(backup);
 

@@ -1512,6 +1512,7 @@ static int amdgpu_dm_init(struct amdgpu_
 		case IP_VERSION(3, 0, 1):
 		case IP_VERSION(3, 1, 2):
 		case IP_VERSION(3, 1, 3):
+		case IP_VERSION(3, 1, 4):
 		case IP_VERSION(3, 1, 5):
 		case IP_VERSION(3, 1, 6):
 			init_data.flags.gpu_vm_support = true;

@@ -2040,6 +2040,15 @@ static int amdgpu_pci_probe(struct pci_d
 			 "See modparam exp_hw_support\n");
 		return -ENODEV;
 	}
+	/* differentiate between P10 and P11 asics with the same DID */
+	if (pdev->device == 0x67FF &&
+	    (pdev->revision == 0xE3 ||
+	     pdev->revision == 0xE7 ||
+	     pdev->revision == 0xF3 ||
+	     pdev->revision == 0xF7)) {
+		flags &= ~AMD_ASIC_MASK;
+		flags |= CHIP_POLARIS10;
+	}
 
 	/* Due to hardware bugs, S/G Display on raven requires a 1:1 IOMMU mapping,
 	 * however, SME requires an indirect IOMMU mapping because the encryption

@@ -2109,12 +2118,12 @@ static int amdgpu_pci_probe(struct pci_d
 
 	pci_set_drvdata(pdev, ddev);
 
-	ret = amdgpu_driver_load_kms(adev, ent->driver_data);
+	ret = amdgpu_driver_load_kms(adev, flags);
 	if (ret)
 		goto err_pci;
 
 retry_init:
-	ret = drm_dev_register(ddev, ent->driver_data);
+	ret = drm_dev_register(ddev, flags);
 	if (ret == -EAGAIN && ++retry <= 3) {
 		DRM_INFO("retry init %d\n", retry);
 		/* Don't request EX mode too frequently which is attacking */

@@ -1509,7 +1509,8 @@ u64 amdgpu_bo_gpu_offset_no_check(struct
 uint32_t amdgpu_bo_get_preferred_domain(struct amdgpu_device *adev,
 					    uint32_t domain)
 {
-	if (domain == (AMDGPU_GEM_DOMAIN_VRAM | AMDGPU_GEM_DOMAIN_GTT)) {
+	if ((domain == (AMDGPU_GEM_DOMAIN_VRAM | AMDGPU_GEM_DOMAIN_GTT)) &&
+	    ((adev->asic_type == CHIP_CARRIZO) || (adev->asic_type == CHIP_STONEY))) {
 		domain = AMDGPU_GEM_DOMAIN_VRAM;
 		if (adev->gmc.real_vram_size <= AMDGPU_SG_THRESHOLD)
 			domain = AMDGPU_GEM_DOMAIN_GTT;

@@ -729,32 +729,69 @@ static int eb_reserve(struct i915_execbu
 	bool unpinned;
 
 	/*
-	 * Attempt to pin all of the buffers into the GTT.
-	 * This is done in 2 phases:
+	 * We have one more buffers that we couldn't bind, which could be due to
+	 * various reasons. To resolve this we have 4 passes, with every next
+	 * level turning the screws tighter:
 	 *
-	 * 1. Unbind all objects that do not match the GTT constraints for
-	 *    the execbuffer (fenceable, mappable, alignment etc).
-	 * 2. Bind new objects.
+	 * 0. Unbind all objects that do not match the GTT constraints for the
+	 * execbuffer (fenceable, mappable, alignment etc). Bind all new
+	 * objects.  This avoids unnecessary unbinding of later objects in order
+	 * to make room for the earlier objects *unless* we need to defragment.
 	 *
-	 * This avoid unnecessary unbinding of later objects in order to make
-	 * room for the earlier objects *unless* we need to defragment.
+	 * 1. Reorder the buffers, where objects with the most restrictive
+	 * placement requirements go first (ignoring fixed location buffers for
+	 * now).  For example, objects needing the mappable aperture (the first
+	 * 256M of GTT), should go first vs objects that can be placed just
+	 * about anywhere. Repeat the previous pass.
 	 *
-	 * Defragmenting is skipped if all objects are pinned at a fixed location.
+	 * 2. Consider buffers that are pinned at a fixed location. Also try to
+	 * evict the entire VM this time, leaving only objects that we were
+	 * unable to lock. Try again to bind the buffers. (still using the new
+	 * buffer order).
+	 *
+	 * 3. We likely have object lock contention for one or more stubborn
+	 * objects in the VM, for which we need to evict to make forward
+	 * progress (perhaps we are fighting the shrinker?). When evicting the
+	 * VM this time around, anything that we can't lock we now track using
+	 * the busy_bo, using the full lock (after dropping the vm->mutex to
+	 * prevent deadlocks), instead of trylock. We then continue to evict the
+	 * VM, this time with the stubborn object locked, which we can now
+	 * hopefully unbind (if still bound in the VM). Repeat until the VM is
+	 * evicted. Finally we should be able bind everything.
 	 */
-	for (pass = 0; pass <= 2; pass++) {
+	for (pass = 0; pass <= 3; pass++) {
 		int pin_flags = PIN_USER | PIN_VALIDATE;
 
 		if (pass == 0)
 			pin_flags |= PIN_NONBLOCK;
 
 		if (pass >= 1)
-			unpinned = eb_unbind(eb, pass == 2);
+			unpinned = eb_unbind(eb, pass >= 2);
 
 		if (pass == 2) {
 			err = mutex_lock_interruptible(&eb->context->vm->mutex);
 			if (!err) {
-				err = i915_gem_evict_vm(eb->context->vm, &eb->ww);
+				err = i915_gem_evict_vm(eb->context->vm, &eb->ww, NULL);
+				mutex_unlock(&eb->context->vm->mutex);
+			}
+			if (err)
+				return err;
+		}
+
+		if (pass == 3) {
+retry:
+			err = mutex_lock_interruptible(&eb->context->vm->mutex);
+			if (!err) {
+				struct drm_i915_gem_object *busy_bo = NULL;
+
+				err = i915_gem_evict_vm(eb->context->vm, &eb->ww, &busy_bo);
 				mutex_unlock(&eb->context->vm->mutex);
+				if (err && busy_bo) {
+					err = i915_gem_object_lock(busy_bo, &eb->ww);
+					i915_gem_object_put(busy_bo);
+					if (!err)
+						goto retry;
+				}
 			}
 			if (err)
 				return err;

@@ -369,7 +369,7 @@ retry:
 		if (vma == ERR_PTR(-ENOSPC)) {
 			ret = mutex_lock_interruptible(&ggtt->vm.mutex);
 			if (!ret) {
-				ret = i915_gem_evict_vm(&ggtt->vm, &ww);
+				ret = i915_gem_evict_vm(&ggtt->vm, &ww, NULL);
 				mutex_unlock(&ggtt->vm.mutex);
 			}
 			if (ret)

@@ -416,6 +416,11 @@ int i915_gem_evict_for_node(struct i915_
  * @vm: Address space to cleanse
  * @ww: An optional struct i915_gem_ww_ctx. If not NULL, i915_gem_evict_vm
  * will be able to evict vma's locked by the ww as well.
+ * @busy_bo: Optional pointer to struct drm_i915_gem_object. If not NULL, then
+ * in the event i915_gem_evict_vm() is unable to trylock an object for eviction,
+ * then @busy_bo will point to it. -EBUSY is also returned. The caller must drop
+ * the vm->mutex, before trying again to acquire the contended lock. The caller
+ * also owns a reference to the object.
  *
  * This function evicts all vmas from a vm.
  *

@@ -425,7 +430,8 @@ int i915_gem_evict_for_node(struct i915_
  * To clarify: This is for freeing up virtual address space, not for freeing
  * memory in e.g. the shrinker.
  */
-int i915_gem_evict_vm(struct i915_address_space *vm, struct i915_gem_ww_ctx *ww)
+int i915_gem_evict_vm(struct i915_address_space *vm, struct i915_gem_ww_ctx *ww,
+		      struct drm_i915_gem_object **busy_bo)
 {
 	int ret = 0;
 

@@ -457,15 +463,22 @@ int i915_gem_evict_vm(struct i915_addres
 			 * the resv is shared among multiple objects, we still
 			 * need the object ref.
 			 */
-			if (dying_vma(vma) ||
+			if (!i915_gem_object_get_rcu(vma->obj) ||
 			    (ww && (dma_resv_locking_ctx(vma->obj->base.resv) == &ww->ctx))) {
 				__i915_vma_pin(vma);
 				list_add(&vma->evict_link, &locked_eviction_list);
 				continue;
 			}
 
-			if (!i915_gem_object_trylock(vma->obj, ww))
+			if (!i915_gem_object_trylock(vma->obj, ww)) {
+				if (busy_bo) {
+					*busy_bo = vma->obj; /* holds ref */
+					ret = -EBUSY;
+					break;
+				}
+				i915_gem_object_put(vma->obj);
 				continue;
+			}
 
 			__i915_vma_pin(vma);
 			list_add(&vma->evict_link, &eviction_list);

@@ -473,25 +486,29 @@ int i915_gem_evict_vm(struct i915_addres
 		if (list_empty(&eviction_list) && list_empty(&locked_eviction_list))
 			break;
 
-		ret = 0;
 		/* Unbind locked objects first, before unlocking the eviction_list */
 		list_for_each_entry_safe(vma, vn, &locked_eviction_list, evict_link) {
 			__i915_vma_unpin(vma);
 
-			if (ret == 0)
+			if (ret == 0) {
 				ret = __i915_vma_unbind(vma);
-			if (ret != -EINTR) /* "Get me out of here!" */
-				ret = 0;
+				if (ret != -EINTR) /* "Get me out of here!" */
+					ret = 0;
+			}
+			if (!dying_vma(vma))
+				i915_gem_object_put(vma->obj);
 		}
 
 		list_for_each_entry_safe(vma, vn, &eviction_list, evict_link) {
 			__i915_vma_unpin(vma);
-			if (ret == 0)
+			if (ret == 0) {
 				ret = __i915_vma_unbind(vma);
-			if (ret != -EINTR) /* "Get me out of here!" */
-				ret = 0;
+				if (ret != -EINTR) /* "Get me out of here!" */
+					ret = 0;
+			}
 
 			i915_gem_object_unlock(vma->obj);
+			i915_gem_object_put(vma->obj);
 		}
 	} while (ret == 0);
 

@@ -11,6 +11,7 @@
 struct drm_mm_node;
 struct i915_address_space;
 struct i915_gem_ww_ctx;
+struct drm_i915_gem_object;
 
 int __must_check i915_gem_evict_something(struct i915_address_space *vm,
 					  struct i915_gem_ww_ctx *ww,

@@ -23,6 +24,7 @@ int __must_check i915_gem_evict_for_node
 					 struct drm_mm_node *node,
 					 unsigned int flags);
 int i915_gem_evict_vm(struct i915_address_space *vm,
-		      struct i915_gem_ww_ctx *ww);
+		      struct i915_gem_ww_ctx *ww,
+		      struct drm_i915_gem_object **busy_bo);
 
 #endif /* __I915_GEM_EVICT_H__ */

@@ -1569,7 +1569,7 @@ static int __i915_ggtt_pin(struct i915_v
 			 * locked objects when called from execbuf when pinning
 			 * is removed. This would probably regress badly.
 			 */
-			i915_gem_evict_vm(vm, NULL);
+			i915_gem_evict_vm(vm, NULL, NULL);
 			mutex_unlock(&vm->mutex);
 		}
 	} while (1);