kernel.git/fs/xattr.c, branch linux-6.10.y

vfs: Fix potential circular locking through setxattr() and removexattr()

2024-09-12T09:13:03Z

[ Upstream commit c3a5e3e872f3688ae0dc57bb78ca633921d96a91 ] When using cachefiles, lockdep may emit something similar to the circular locking dependency notice below. The problem appears to stem from the following: (1) Cachefiles manipulates xattrs on the files in its cache when called from ->writepages(). (2) The setxattr() and removexattr() system call handlers get the name (and value) from userspace after taking the sb_writers lock, putting accesses of the vma->vm_lock and mm->mmap_lock inside of that. (3) The afs filesystem uses a per-inode lock to prevent multiple revalidation RPCs and in writeback vs truncate to prevent parallel operations from deadlocking against the server on one side and local page locks on the other. Fix this by moving the getting of the name and value in {get,remove}xattr() outside of the sb_writers lock. This also has the minor benefits that we don't need to reget these in the event of a retry and we never try to take the sb_writers lock in the event we can't pull the name and value into the kernel. Alternative approaches that might fix this include moving the dispatch of a write to the cache off to a workqueue or trying to do without the validation lock in afs. Note that this might also affect other filesystems that use netfslib and/or cachefiles. ====================================================== WARNING: possible circular locking dependency detected 6.10.0-build2+ #956 Not tainted ------------------------------------------------------ fsstress/6050 is trying to acquire lock: ffff888138fd82f0 (mapping.invalidate_lock#3){++++}-{3:3}, at: filemap_fault+0x26e/0x8b0 but task is already holding lock: ffff888113f26d18 (&vma->vm_lock->lock){++++}-{3:3}, at: lock_vma_under_rcu+0x165/0x250 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #4 (&vma->vm_lock->lock){++++}-{3:3}: __lock_acquire+0xaf0/0xd80 lock_acquire.part.0+0x103/0x280 down_write+0x3b/0x50 vma_start_write+0x6b/0xa0 vma_link+0xcc/0x140 insert_vm_struct+0xb7/0xf0 alloc_bprm+0x2c1/0x390 kernel_execve+0x65/0x1a0 call_usermodehelper_exec_async+0x14d/0x190 ret_from_fork+0x24/0x40 ret_from_fork_asm+0x1a/0x30 -> #3 (&mm->mmap_lock){++++}-{3:3}: __lock_acquire+0xaf0/0xd80 lock_acquire.part.0+0x103/0x280 __might_fault+0x7c/0xb0 strncpy_from_user+0x25/0x160 removexattr+0x7f/0x100 __do_sys_fremovexattr+0x7e/0xb0 do_syscall_64+0x9f/0x100 entry_SYSCALL_64_after_hwframe+0x76/0x7e -> #2 (sb_writers#14){.+.+}-{0:0}: __lock_acquire+0xaf0/0xd80 lock_acquire.part.0+0x103/0x280 percpu_down_read+0x3c/0x90 vfs_iocb_iter_write+0xe9/0x1d0 __cachefiles_write+0x367/0x430 cachefiles_issue_write+0x299/0x2f0 netfs_advance_write+0x117/0x140 netfs_write_folio.isra.0+0x5ca/0x6e0 netfs_writepages+0x230/0x2f0 afs_writepages+0x4d/0x70 do_writepages+0x1e8/0x3e0 filemap_fdatawrite_wbc+0x84/0xa0 __filemap_fdatawrite_range+0xa8/0xf0 file_write_and_wait_range+0x59/0x90 afs_release+0x10f/0x270 __fput+0x25f/0x3d0 __do_sys_close+0x43/0x70 do_syscall_64+0x9f/0x100 entry_SYSCALL_64_after_hwframe+0x76/0x7e -> #1 (&vnode->validate_lock){++++}-{3:3}: __lock_acquire+0xaf0/0xd80 lock_acquire.part.0+0x103/0x280 down_read+0x95/0x200 afs_writepages+0x37/0x70 do_writepages+0x1e8/0x3e0 filemap_fdatawrite_wbc+0x84/0xa0 filemap_invalidate_inode+0x167/0x1e0 netfs_unbuffered_write_iter+0x1bd/0x2d0 vfs_write+0x22e/0x320 ksys_write+0xbc/0x130 do_syscall_64+0x9f/0x100 entry_SYSCALL_64_after_hwframe+0x76/0x7e -> #0 (mapping.invalidate_lock#3){++++}-{3:3}: check_noncircular+0x119/0x160 check_prev_add+0x195/0x430 __lock_acquire+0xaf0/0xd80 lock_acquire.part.0+0x103/0x280 down_read+0x95/0x200 filemap_fault+0x26e/0x8b0 __do_fault+0x57/0xd0 do_pte_missing+0x23b/0x320 __handle_mm_fault+0x2d4/0x320 handle_mm_fault+0x14f/0x260 do_user_addr_fault+0x2a2/0x500 exc_page_fault+0x71/0x90 asm_exc_page_fault+0x22/0x30 other info that might help us debug this: Chain exists of: mapping.invalidate_lock#3 --> &mm->mmap_lock --> &vma->vm_lock->lock Possible unsafe locking scenario: CPU0 CPU1 ---- ---- rlock(&vma->vm_lock->lock); lock(&mm->mmap_lock); lock(&vma->vm_lock->lock); rlock(mapping.invalidate_lock#3); *** DEADLOCK *** 1 lock held by fsstress/6050: #0: ffff888113f26d18 (&vma->vm_lock->lock){++++}-{3:3}, at: lock_vma_under_rcu+0x165/0x250 stack backtrace: CPU: 0 PID: 6050 Comm: fsstress Not tainted 6.10.0-build2+ #956 Hardware name: ASUS All Series/H97-PLUS, BIOS 2306 10/09/2014 Call Trace: dump_stack_lvl+0x57/0x80 check_noncircular+0x119/0x160 ? queued_spin_lock_slowpath+0x4be/0x510 ? __pfx_check_noncircular+0x10/0x10 ? __pfx_queued_spin_lock_slowpath+0x10/0x10 ? mark_lock+0x47/0x160 ? init_chain_block+0x9c/0xc0 ? add_chain_block+0x84/0xf0 check_prev_add+0x195/0x430 __lock_acquire+0xaf0/0xd80 ? __pfx___lock_acquire+0x10/0x10 ? __lock_release.isra.0+0x13b/0x230 lock_acquire.part.0+0x103/0x280 ? filemap_fault+0x26e/0x8b0 ? __pfx_lock_acquire.part.0+0x10/0x10 ? rcu_is_watching+0x34/0x60 ? lock_acquire+0xd7/0x120 down_read+0x95/0x200 ? filemap_fault+0x26e/0x8b0 ? __pfx_down_read+0x10/0x10 ? __filemap_get_folio+0x25/0x1a0 filemap_fault+0x26e/0x8b0 ? __pfx_filemap_fault+0x10/0x10 ? find_held_lock+0x7c/0x90 ? __pfx___lock_release.isra.0+0x10/0x10 ? __pte_offset_map+0x99/0x110 __do_fault+0x57/0xd0 do_pte_missing+0x23b/0x320 __handle_mm_fault+0x2d4/0x320 ? __pfx___handle_mm_fault+0x10/0x10 handle_mm_fault+0x14f/0x260 do_user_addr_fault+0x2a2/0x500 exc_page_fault+0x71/0x90 asm_exc_page_fault+0x22/0x30 Signed-off-by: David Howells Link: https://lore.kernel.org/r/2136178.1721725194@warthog.procyon.org.uk cc: Alexander Viro cc: Christian Brauner cc: Jan Kara cc: Jeff Layton cc: Gao Xiang cc: Matthew Wilcox cc: netfs@lists.linux.dev cc: linux-erofs@lists.ozlabs.org cc: linux-fsdevel@vger.kernel.org [brauner: fix minor issues] Signed-off-by: Christian Brauner Signed-off-by: Sasha Levin

evm: Move to LSM infrastructure

2024-02-16T04:43:47Z

As for IMA, move hardcoded EVM function calls from various places in the kernel to the LSM infrastructure, by introducing a new LSM named 'evm' (last and always enabled like 'ima'). The order in the Makefile ensures that 'evm' hooks are executed after 'ima' ones. Make EVM functions as static (except for evm_inode_init_security(), which is exported), and register them as hook implementations in init_evm_lsm(). Also move the inline functions evm_inode_remove_acl(), evm_inode_post_remove_acl(), and evm_inode_post_set_acl() from the public evm.h header to evm_main.c. Unlike before (see commit to move IMA to the LSM infrastructure), evm_inode_post_setattr(), evm_inode_post_set_acl(), evm_inode_post_remove_acl(), and evm_inode_post_removexattr() are not executed for private inodes. Finally, add the LSM_ID_EVM case in lsm_list_modules_test.c Signed-off-by: Roberto Sassu Reviewed-by: Casey Schaufler Acked-by: Christian Brauner Reviewed-by: Stefan Berger Reviewed-by: Mimi Zohar Acked-by: Mimi Zohar Signed-off-by: Paul Moore

security: Introduce inode_post_removexattr hook

2024-02-16T04:43:42Z

In preparation for moving IMA and EVM to the LSM infrastructure, introduce the inode_post_removexattr hook. At inode_removexattr hook, EVM verifies the file's existing HMAC value. At inode_post_removexattr, EVM re-calculates the file's HMAC with the passed xattr removed and other file metadata. Other LSMs could similarly take some action after successful xattr removal. The new hook cannot return an error and cannot cause the operation to be reverted. Signed-off-by: Roberto Sassu Reviewed-by: Stefan Berger Reviewed-by: Mimi Zohar Reviewed-by: Casey Schaufler Acked-by: Christian Brauner Signed-off-by: Paul Moore

xattr: make the xattr array itself const

2023-10-09T14:24:16Z

As it is currently declared, the xattr_handler structs are const but the array containing their pointers is not. This patch makes it so that fs modules can place them in .rodata, which makes it harder for accidental/malicious modifications at runtime. Signed-off-by: Wedson Almeida Filho Link: https://lore.kernel.org/r/20230930050033.41174-2-wedsonaf@gmail.com Signed-off-by: Christian Brauner

tmpfs,xattr: GFP_KERNEL_ACCOUNT for simple xattrs

2023-08-22T08:57:46Z

It is particularly important for the userns mount case (when a sensible nr_inodes maximum may not be enforced) that tmpfs user xattrs be subject to memory cgroup limiting. Leave temporary buffer allocations as is, but change the persistent simple xattr allocations from GFP_KERNEL to GFP_KERNEL_ACCOUNT. This limits kernfs's cgroupfs too, but that's good. (I had intended to send this change earlier, but had been confused by shmem_alloc_inode() using GFP_KERNEL, and thought a discussion would be needed to change that too: no, I was forgetting the SLAB_ACCOUNT on that kmem_cache, which implicitly adds __GFP_ACCOUNT to all its allocations.) Signed-off-by: Hugh Dickins Reviewed-by: Jan Kara Message-Id: Signed-off-by: Christian Brauner

tmpfs,xattr: enable limited user extended attributes

2023-08-10T10:06:04Z

Enable "user." extended attributes on tmpfs, limiting them by tracking the space they occupy, and deducting that space from the limited ispace (unless tmpfs mounted with nr_inodes=0 to leave that ispace unlimited). tmpfs inodes and simple xattrs are both unswappable, and have to be in lowmem on a 32-bit highmem kernel: so the ispace limit is appropriate for xattrs, without any need for a further mount option. Add simple_xattr_space() to give approximate but deterministic estimate of the space taken up by each xattr: with simple_xattrs_free() outputting the space freed if required (but kernfs and even some tmpfs usages do not require that, so don't waste time on strlen'ing if not needed). Security and trusted xattrs were already supported: for consistency and simplicity, account them from the same pool; though there's a small risk that a tmpfs with enough space before would now be considered too small. When extended attributes are used, "df -i" does show more IUsed and less IFree than can be explained by the inodes: document that (manpage later). xfstests tests/generic which were not run on tmpfs before but now pass: 020 037 062 070 077 097 103 117 337 377 454 486 523 533 611 618 728 with no new failures. Signed-off-by: Hugh Dickins Reviewed-by: Jan Kara Reviewed-by: Carlos Maiolino Message-Id: <2e63b26e-df46-5baa-c7d6-f9a8dd3282c5@google.com> Signed-off-by: Christian Brauner

xattr: simple_xattr_set() return old_xattr to be freed

2023-08-09T07:15:51Z

tmpfs wants to support limited user extended attributes, but kernfs (or cgroupfs, the only kernfs with KERNFS_ROOT_SUPPORT_USER_XATTR) already supports user extended attributes through simple xattrs: but limited by a policy (128KiB per inode) too liberal to be used on tmpfs. To allow a different limiting policy for tmpfs, without affecting the policy for kernfs, change simple_xattr_set() to return the replaced or removed xattr (if any), leaving the caller to update their accounting then free the xattr (by simple_xattr_free(), renamed from the static free_simple_xattr()). Signed-off-by: Hugh Dickins Reviewed-by: Jan Kara Reviewed-by: Christian Brauner Reviewed-by: Carlos Maiolino Message-Id: <158c6585-2aa7-d4aa-90ff-f7c3f8fe407c@google.com> Signed-off-by: Christian Brauner

fs: don't call posix_acl_listxattr in generic_listxattr

2023-05-17T13:25:20Z

Commit f2620f166e2a caused the kernel to start emitting POSIX ACL xattrs for NFSv4 inodes, which it doesn't support. The only other user of generic_listxattr is HFS (classic) and it doesn't support POSIX ACLs either. Fixes: f2620f166e2a xattr: simplify listxattr helpers Reported-by: Ondrej Valousek Signed-off-by: Jeff Layton Message-Id: <20230516124655.82283-1-jlayton@kernel.org> Signed-off-by: Christian Brauner

acl: don't depend on IOP_XATTR

2023-03-06T08:59:20Z

All codepaths that don't want to implement POSIX ACLs should simply not implement the associated inode operations instead of relying on IOP_XATTR. That's the case for all filesystems today. For vfs_listxattr() all filesystems that explicitly turn of xattrs for a given inode all set inode->i_op to a dedicated set of inode operations that doesn't implement ->listxattr(). We can remove the dependency of vfs_listxattr() on IOP_XATTR. Removing this dependency will allow us to decouple POSIX ACLs from IOP_XATTR and they can still be listed even if no other xattr handlers are implemented. Otherwise we would have to implement elaborate schemes to raise IOP_XATTR even if sb->s_xattr is set to NULL. Signed-off-by: Christian Brauner (Microsoft)

xattr: remove unused argument

2023-03-06T08:57:11Z

his helpers is really just used to check for user.* xattr support so don't make it pointlessly generic. Reviewed-by: Christoph Hellwig Signed-off-by: Christian Brauner (Microsoft)