Vulnerabilities

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> SUNRPC: Trap RDMA segment overflows<br /> <br /> Prevent svc_rdma_build_writes() from walking off the end of a Write<br /> chunk&amp;#39;s segment array. Caught with KASAN.<br /> <br /> The test that this fix replaces is invalid, and might have been left<br /> over from an earlier prototype of the PCL work.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> NFSD: Fix potential use-after-free in nfsd_file_put()<br /> <br /> nfsd_file_put_noref() can free @nf, so don&amp;#39;t dereference @nf<br /> immediately upon return from nfsd_file_put_noref().

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> f2fs: fix to do sanity check on block address in f2fs_do_zero_range()<br /> <br /> As Yanming reported in bugzilla:<br /> <br /> https://bugzilla.kernel.org/show_bug.cgi?id=215894<br /> <br /> I have encountered a bug in F2FS file system in kernel v5.17.<br /> <br /> I have uploaded the system call sequence as case.c, and a fuzzed image can<br /> be found in google net disk<br /> <br /> The kernel should enable CONFIG_KASAN=y and CONFIG_KASAN_INLINE=y. You can<br /> reproduce the bug by running the following commands:<br /> <br /> kernel BUG at fs/f2fs/segment.c:2291!<br /> Call Trace:<br /> f2fs_invalidate_blocks+0x193/0x2d0<br /> f2fs_fallocate+0x2593/0x4a70<br /> vfs_fallocate+0x2a5/0xac0<br /> ksys_fallocate+0x35/0x70<br /> __x64_sys_fallocate+0x8e/0xf0<br /> do_syscall_64+0x3b/0x90<br /> entry_SYSCALL_64_after_hwframe+0x44/0xae<br /> <br /> The root cause is, after image was fuzzed, block mapping info in inode<br /> will be inconsistent with SIT table, so in f2fs_fallocate(), it will cause<br /> panic when updating SIT with invalid blkaddr.<br /> <br /> Let&amp;#39;s fix the issue by adding sanity check on block address before updating<br /> SIT table with it.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> af_unix: Fix a data-race in unix_dgram_peer_wake_me().<br /> <br /> unix_dgram_poll() calls unix_dgram_peer_wake_me() without `other`&amp;#39;s<br /> lock held and check if its receive queue is full. Here we need to<br /> use unix_recvq_full_lockless() instead of unix_recvq_full(), otherwise<br /> KCSAN will report a data-race.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: xfrm: unexport __init-annotated xfrm4_protocol_init()<br /> <br /> EXPORT_SYMBOL and __init is a bad combination because the .init.text<br /> section is freed up after the initialization. Hence, modules cannot<br /> use symbols annotated __init. The access to a freed symbol may end up<br /> with kernel panic.<br /> <br /> modpost used to detect it, but it has been broken for a decade.<br /> <br /> Recently, I fixed modpost so it started to warn it again, then this<br /> showed up in linux-next builds.<br /> <br /> There are two ways to fix it:<br /> <br /> - Remove __init<br /> - Remove EXPORT_SYMBOL<br /> <br /> I chose the latter for this case because the only in-tree call-site,<br /> net/ipv4/xfrm4_policy.c is never compiled as modular.<br /> (CONFIG_XFRM is boolean)

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: dsa: lantiq_gswip: Fix refcount leak in gswip_gphy_fw_list<br /> <br /> Every iteration of for_each_available_child_of_node() decrements<br /> the reference count of the previous node.<br /> when breaking early from a for_each_available_child_of_node() loop,<br /> we need to explicitly call of_node_put() on the gphy_fw_np.<br /> Add missing of_node_put() to avoid refcount leak.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ext4: fix bug_on in ext4_writepages<br /> <br /> we got issue as follows:<br /> EXT4-fs error (device loop0): ext4_mb_generate_buddy:1141: group 0, block bitmap and bg descriptor inconsistent: 25 vs 31513 free cls<br /> ------------[ cut here ]------------<br /> kernel BUG at fs/ext4/inode.c:2708!<br /> invalid opcode: 0000 [#1] PREEMPT SMP KASAN PTI<br /> CPU: 2 PID: 2147 Comm: rep Not tainted 5.18.0-rc2-next-20220413+ #155<br /> RIP: 0010:ext4_writepages+0x1977/0x1c10<br /> RSP: 0018:ffff88811d3e7880 EFLAGS: 00010246<br /> RAX: 0000000000000000 RBX: 0000000000000001 RCX: ffff88811c098000<br /> RDX: 0000000000000000 RSI: ffff88811c098000 RDI: 0000000000000002<br /> RBP: ffff888128140f50 R08: ffffffffb1ff6387 R09: 0000000000000000<br /> R10: 0000000000000007 R11: ffffed10250281ea R12: 0000000000000001<br /> R13: 00000000000000a4 R14: ffff88811d3e7bb8 R15: ffff888128141028<br /> FS: 00007f443aed9740(0000) GS:ffff8883aef00000(0000) knlGS:0000000000000000<br /> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033<br /> CR2: 0000000020007200 CR3: 000000011c2a4000 CR4: 00000000000006e0<br /> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000<br /> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400<br /> Call Trace:<br /> <br /> do_writepages+0x130/0x3a0<br /> filemap_fdatawrite_wbc+0x83/0xa0<br /> filemap_flush+0xab/0xe0<br /> ext4_alloc_da_blocks+0x51/0x120<br /> __ext4_ioctl+0x1534/0x3210<br /> __x64_sys_ioctl+0x12c/0x170<br /> do_syscall_64+0x3b/0x90<br /> <br /> It may happen as follows:<br /> 1. write inline_data inode<br /> vfs_write<br /> new_sync_write<br /> ext4_file_write_iter<br /> ext4_buffered_write_iter<br /> generic_perform_write<br /> ext4_da_write_begin<br /> ext4_da_write_inline_data_begin -&gt; If inline data size too<br /> small will allocate block to write, then mapping will has<br /> dirty page<br /> ext4_da_convert_inline_data_to_extent -&gt;clear EXT4_STATE_MAY_INLINE_DATA<br /> 2. fallocate<br /> do_vfs_ioctl<br /> ioctl_preallocate<br /> vfs_fallocate<br /> ext4_fallocate<br /> ext4_convert_inline_data<br /> ext4_convert_inline_data_nolock<br /> ext4_map_blocks -&gt; fail will goto restore data<br /> ext4_restore_inline_data<br /> ext4_create_inline_data<br /> ext4_write_inline_data<br /> ext4_set_inode_state -&gt; set inode EXT4_STATE_MAY_INLINE_DATA<br /> 3. writepages<br /> __ext4_ioctl<br /> ext4_alloc_da_blocks<br /> filemap_flush<br /> filemap_fdatawrite_wbc<br /> do_writepages<br /> ext4_writepages<br /> if (ext4_has_inline_data(inode))<br /> BUG_ON(ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA))<br /> <br /> The root cause of this issue is we destory inline data until call<br /> ext4_writepages under delay allocation mode. But there maybe already<br /> convert from inline to extent. To solve this issue, we call<br /> filemap_flush first..

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ext4: filter out EXT4_FC_REPLAY from on-disk superblock field s_state<br /> <br /> The EXT4_FC_REPLAY bit in sbi-&gt;s_mount_state is used to indicate that<br /> we are in the middle of replay the fast commit journal. This was<br /> actually a mistake, since the sbi-&gt;s_mount_info is initialized from<br /> es-&gt;s_state. Arguably s_mount_state is misleadingly named, but the<br /> name is historical --- s_mount_state and s_state dates back to ext2.<br /> <br /> What should have been used is the ext4_{set,clear,test}_mount_flag()<br /> inline functions, which sets EXT4_MF_* bits in sbi-&gt;s_mount_flags.<br /> <br /> The problem with using EXT4_FC_REPLAY is that a maliciously corrupted<br /> superblock could result in EXT4_FC_REPLAY getting set in<br /> s_mount_state. This bypasses some sanity checks, and this can trigger<br /> a BUG() in ext4_es_cache_extent(). As a easy-to-backport-fix, filter<br /> out the EXT4_FC_REPLAY bit for now. We should eventually transition<br /> away from EXT4_FC_REPLAY to something like EXT4_MF_REPLAY.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: mdio: unexport __init-annotated mdio_bus_init()<br /> <br /> EXPORT_SYMBOL and __init is a bad combination because the .init.text<br /> section is freed up after the initialization. Hence, modules cannot<br /> use symbols annotated __init. The access to a freed symbol may end up<br /> with kernel panic.<br /> <br /> modpost used to detect it, but it has been broken for a decade.<br /> <br /> Recently, I fixed modpost so it started to warn it again, then this<br /> showed up in linux-next builds.<br /> <br /> There are two ways to fix it:<br /> <br /> - Remove __init<br /> - Remove EXPORT_SYMBOL<br /> <br /> I chose the latter for this case because the only in-tree call-site,<br /> drivers/net/phy/phy_device.c is never compiled as modular.<br /> (CONFIG_PHYLIB is boolean)

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: altera: Fix refcount leak in altera_tse_mdio_create<br /> <br /> Every iteration of for_each_child_of_node() decrements<br /> the reference count of the previous node.<br /> When break from a for_each_child_of_node() loop,<br /> we need to explicitly call of_node_put() on the child node when<br /> not need anymore.<br /> Add missing of_node_put() to avoid refcount leak.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ext4: fix warning in ext4_handle_inode_extension<br /> <br /> We got issue as follows:<br /> EXT4-fs error (device loop0) in ext4_reserve_inode_write:5741: Out of memory<br /> EXT4-fs error (device loop0): ext4_setattr:5462: inode #13: comm syz-executor.0: mark_inode_dirty error<br /> EXT4-fs error (device loop0) in ext4_setattr:5519: Out of memory<br /> EXT4-fs error (device loop0): ext4_ind_map_blocks:595: inode #13: comm syz-executor.0: Can&amp;#39;t allocate blocks for non-extent mapped inodes with bigalloc<br /> ------------[ cut here ]------------<br /> WARNING: CPU: 1 PID: 4361 at fs/ext4/file.c:301 ext4_file_write_iter+0x11c9/0x1220<br /> Modules linked in:<br /> CPU: 1 PID: 4361 Comm: syz-executor.0 Not tainted 5.10.0+ #1<br /> RIP: 0010:ext4_file_write_iter+0x11c9/0x1220<br /> RSP: 0018:ffff924d80b27c00 EFLAGS: 00010282<br /> RAX: ffffffff815a3379 RBX: 0000000000000000 RCX: 000000003b000000<br /> RDX: ffff924d81601000 RSI: 00000000000009cc RDI: 00000000000009cd<br /> RBP: 000000000000000d R08: ffffffffbc5a2c6b R09: 0000902e0e52a96f<br /> R10: ffff902e2b7c1b40 R11: ffff902e2b7c1b40 R12: 000000000000000a<br /> R13: 0000000000000001 R14: ffff902e0e52aa10 R15: ffffffffffffff8b<br /> FS: 00007f81a7f65700(0000) GS:ffff902e3bc80000(0000) knlGS:0000000000000000<br /> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033<br /> CR2: ffffffffff600400 CR3: 000000012db88001 CR4: 00000000003706e0<br /> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000<br /> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400<br /> Call Trace:<br /> do_iter_readv_writev+0x2e5/0x360<br /> do_iter_write+0x112/0x4c0<br /> do_pwritev+0x1e5/0x390<br /> __x64_sys_pwritev2+0x7e/0xa0<br /> do_syscall_64+0x37/0x50<br /> entry_SYSCALL_64_after_hwframe+0x44/0xa9<br /> <br /> Above issue may happen as follows:<br /> Assume<br /> inode.i_size=4096<br /> EXT4_I(inode)-&gt;i_disksize=4096<br /> <br /> step 1: set inode-&gt;i_isize = 8192<br /> ext4_setattr<br /> if (attr-&gt;ia_size != inode-&gt;i_size)<br /> EXT4_I(inode)-&gt;i_disksize = attr-&gt;ia_size;<br /> rc = ext4_mark_inode_dirty<br /> ext4_reserve_inode_write<br /> ext4_get_inode_loc<br /> __ext4_get_inode_loc<br /> sb_getblk --&gt; return -ENOMEM<br /> ...<br /> if (!error) -&gt;will not update i_size<br /> i_size_write(inode, attr-&gt;ia_size);<br /> Now:<br /> inode.i_size=4096<br /> EXT4_I(inode)-&gt;i_disksize=8192<br /> <br /> step 2: Direct write 4096 bytes<br /> ext4_file_write_iter<br /> ext4_dio_write_iter<br /> iomap_dio_rw -&gt;return error<br /> if (extend)<br /> ext4_handle_inode_extension<br /> WARN_ON_ONCE(i_size_read(inode) i_disksize);<br /> -&gt;Then trigger warning.<br /> <br /> To solve above issue, if mark inode dirty failed in ext4_setattr just<br /> set &amp;#39;EXT4_I(inode)-&gt;i_disksize&amp;#39; with old value.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ext4: avoid cycles in directory h-tree<br /> <br /> A maliciously corrupted filesystem can contain cycles in the h-tree<br /> stored inside a directory. That can easily lead to the kernel corrupting<br /> tree nodes that were already verified under its hands while doing a node<br /> split and consequently accessing unallocated memory. Fix the problem by<br /> verifying traversed block numbers are unique.