Now, 's_err_report' timer is init after ext4_group_desc_init() when fill
super. Theoretically, ext4_group_desc_init() may access to error handle
as follows:
__ext4_fill_super
ext4_group_desc_init
ext4_check_descriptors
ext4_get_group_desc
ext4_error
ext4_handle_error
ext4_commit_super
ext4_update_super
if (!es->s_error_count)
mod_timer(&sbi->s_err_report, jiffies + 24*60*60*HZ);
--> Accessing Uninitialized Variables
timer_setup(&sbi->s_err_report, print_daily_error_info, 0);
Maybe above issue is just theoretical, as ext4_check_descriptors() didn't
judge 'gpd' which get from ext4_get_group_desc(), if access to error handle
ext4_get_group_desc() will return NULL, then will trigger null-ptr-deref in
ext4_check_descriptors().
However, from the perspective of pure code, it is better to initialize
resource that may need to be used first.
Signed-off-by: Ye Bin <yebin10@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20230119013711.86680-1-yebin@huaweicloud.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Syzbot reported a hung task problem:
==================================================================
INFO: task syz-executor232:5073 blocked for more than 143 seconds.
Not tainted 6.2.0-rc2-syzkaller-00024-g512dee0c00ad #0
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
task:syz-exec232 state:D stack:21024 pid:5073 ppid:5072 flags:0x00004004
Call Trace:
<TASK>
context_switch kernel/sched/core.c:5244 [inline]
__schedule+0x995/0xe20 kernel/sched/core.c:6555
schedule+0xcb/0x190 kernel/sched/core.c:6631
__wait_on_freeing_inode fs/inode.c:2196 [inline]
find_inode_fast+0x35a/0x4c0 fs/inode.c:950
iget_locked+0xb1/0x830 fs/inode.c:1273
__ext4_iget+0x22e/0x3ed0 fs/ext4/inode.c:4861
ext4_xattr_inode_iget+0x68/0x4e0 fs/ext4/xattr.c:389
ext4_xattr_inode_dec_ref_all+0x1a7/0xe50 fs/ext4/xattr.c:1148
ext4_xattr_delete_inode+0xb04/0xcd0 fs/ext4/xattr.c:2880
ext4_evict_inode+0xd7c/0x10b0 fs/ext4/inode.c:296
evict+0x2a4/0x620 fs/inode.c:664
ext4_orphan_cleanup+0xb60/0x1340 fs/ext4/orphan.c:474
__ext4_fill_super fs/ext4/super.c:5516 [inline]
ext4_fill_super+0x81cd/0x8700 fs/ext4/super.c:5644
get_tree_bdev+0x400/0x620 fs/super.c:1282
vfs_get_tree+0x88/0x270 fs/super.c:1489
do_new_mount+0x289/0xad0 fs/namespace.c:3145
do_mount fs/namespace.c:3488 [inline]
__do_sys_mount fs/namespace.c:3697 [inline]
__se_sys_mount+0x2d3/0x3c0 fs/namespace.c:3674
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x7fa5406fd5ea
RSP: 002b:00007ffc7232f968 EFLAGS: 00000202 ORIG_RAX: 00000000000000a5
RAX: ffffffffffffffda RBX: 0000000000000003 RCX: 00007fa5406fd5ea
RDX: 0000000020000440 RSI: 0000000020000000 RDI: 00007ffc7232f970
RBP: 00007ffc7232f970 R08: 00007ffc7232f9b0 R09: 0000000000000432
R10: 0000000000804a03 R11: 0000000000000202 R12: 0000000000000004
R13: 0000555556a7a2c0 R14: 00007ffc7232f9b0 R15: 0000000000000000
</TASK>
==================================================================
The problem is that the inode contains an xattr entry with ea_inum of 15
when cleaning up an orphan inode <15>. When evict inode <15>, the reference
counting of the corresponding EA inode is decreased. When EA inode <15> is
found by find_inode_fast() in __ext4_iget(), it is found that the EA inode
holds the I_FREEING flag and waits for the EA inode to complete deletion.
As a result, when inode <15> is being deleted, we wait for inode <15> to
complete the deletion, resulting in an infinite loop and triggering Hung
Task. To solve this problem, we only need to check whether the ino of EA
inode and parent is the same before getting EA inode.
Link: https://syzkaller.appspot.com/bug?extid=77d6fcc37bbb92f26048
Reported-by: syzbot+77d6fcc37bbb92f26048@syzkaller.appspotmail.com
Signed-off-by: Baokun Li <libaokun1@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20230110133436.996350-1-libaokun1@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Following process will make data lost and could lead to a filesystem
corrupted problem:
1. jh(bh) is inserted into T1->t_checkpoint_list, bh is dirty, and
jh->b_transaction = NULL
2. T1 is added into journal->j_checkpoint_transactions.
3. Get bh prepare to write while doing checkpoing:
PA PB
do_get_write_access jbd2_log_do_checkpoint
spin_lock(&jh->b_state_lock)
if (buffer_dirty(bh))
clear_buffer_dirty(bh) // clear buffer dirty
set_buffer_jbddirty(bh)
transaction =
journal->j_checkpoint_transactions
jh = transaction->t_checkpoint_list
if (!buffer_dirty(bh))
__jbd2_journal_remove_checkpoint(jh)
// bh won't be flushed
jbd2_cleanup_journal_tail
__jbd2_journal_file_buffer(jh, transaction, BJ_Reserved)
4. Aborting journal/Power-cut before writing latest bh on journal area.
In this way we get a corrupted filesystem with bh's data lost.
Fix it by moving the clearing of buffer_dirty bit just before the call
to __jbd2_journal_file_buffer(), both bit clearing and jh->b_transaction
assignment are under journal->j_list_lock locked, so that
jbd2_log_do_checkpoint() will wait until jh's new transaction fininshed
even bh is currently not dirty. And journal_shrink_one_cp_list() won't
remove jh from checkpoint list if the buffer head is reused in
do_get_write_access().
Fetch a reproducer in [Link].
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216898
Cc: <stable@kernel.org>
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: zhanchengbin <zhanchengbin1@huawei.com>
Suggested-by: Jan Kara <jack@suse.cz>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20230110015327.1181863-1-chengzhihao1@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
When mounting a crafted ext4 image, s_journal_inum may change after journal
replay, which is obviously unreasonable because we have successfully loaded
and replayed the journal through the old s_journal_inum. And the new
s_journal_inum bypasses some of the checks in ext4_get_journal(), which
may trigger a null pointer dereference problem. So if s_journal_inum
changes after the journal replay, we ignore the change, and rewrite the
current journal_inum to the superblock.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216541
Reported-by: Luís Henriques <lhenriques@suse.de>
Signed-off-by: Baokun Li <libaokun1@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20230107032126.4165860-3-libaokun1@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
With clang's kernel control flow integrity (kCFI, CONFIG_CFI_CLANG),
indirect call targets are validated against the expected function
pointer prototype to make sure the call target is valid to help mitigate
ROP attacks. If they are not identical, there is a failure at run time,
which manifests as either a kernel panic or thread getting killed.
ext4_feat_ktype was setting the "release" handler to "kfree", which
doesn't have a matching function prototype. Add a simple wrapper
with the correct prototype.
This was found as a result of Clang's new -Wcast-function-type-strict
flag, which is more sensitive than the simpler -Wcast-function-type,
which only checks for type width mismatches.
Note that this code is only reached when ext4 is a loadable module and
it is being unloaded:
CFI failure at kobject_put+0xbb/0x1b0 (target: kfree+0x0/0x180; expected type: 0x7c4aa698)
...
RIP: 0010:kobject_put+0xbb/0x1b0
...
Call Trace:
<TASK>
ext4_exit_sysfs+0x14/0x60 [ext4]
cleanup_module+0x67/0xedb [ext4]
Fixes: b99fee58a2 ("ext4: create ext4_feat kobject dynamically")
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Eric Biggers <ebiggers@kernel.org>
Cc: stable@vger.kernel.org
Build-tested-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Link: https://lore.kernel.org/r/20230103234616.never.915-kees@kernel.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20230104210908.gonna.388-kees@kernel.org
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
If ENOMEM fails when the extent is splitting, we need to restore the length
of the split extent.
In the ext4_split_extent_at function, only in ext4_ext_create_new_leaf will
it alloc memory and change the shape of the extent tree,even if an ENOMEM
is returned at this time, the extent tree is still self-consistent, Just
restore the split extent lens in the function ext4_split_extent_at.
ext4_split_extent_at
ext4_ext_insert_extent
ext4_ext_create_new_leaf
1)ext4_ext_split
ext4_find_extent
2)ext4_ext_grow_indepth
ext4_find_extent
Signed-off-by: zhanchengbin <zhanchengbin1@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20230103022812.130603-1-zhanchengbin1@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
In the dio write path, we only take shared inode lock for the case of
aligned overwriting initialized blocks inside EOF. But for overwriting
preallocated blocks, it may only need to split unwritten extents, this
procedure has been protected under i_data_sem lock, it's safe to
release the exclusive inode lock and take shared inode lock.
This could give a significant speed up for multi-threaded writes. Test
on Intel Xeon Gold 6140 and nvme SSD with below fio parameters.
direct=1
ioengine=libaio
iodepth=10
numjobs=10
runtime=60
rw=randwrite
size=100G
And the test result are:
Before:
bs=4k IOPS=11.1k, BW=43.2MiB/s
bs=16k IOPS=11.1k, BW=173MiB/s
bs=64k IOPS=11.2k, BW=697MiB/s
After:
bs=4k IOPS=41.4k, BW=162MiB/s
bs=16k IOPS=41.3k, BW=646MiB/s
bs=64k IOPS=13.5k, BW=843MiB/s
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20221226062015.3479416-1-yi.zhang@huaweicloud.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Pull another io_uring fix from Jens Axboe:
"Just a single fix for a regression that happened in this release due
to a poll change. Normally I would've just deferred it to next week,
but since the original fix got picked up by stable, I think it's
better to just send this one off separately.
The issue is around the poll race fix, and how it mistakenly also got
applied to multishot polling. Those don't need the race fix, and we
should not be doing any reissues for that case. Exhaustive test cases
were written and committed to the liburing regression suite for the
reported issue, and additions for similar issues"
* tag 'io_uring-6.2-2023-01-21' of git://git.kernel.dk/linux:
io_uring/poll: don't reissue in case of poll race on multishot request
Pull char/misc driver fixes from Greg KH:
"Here are some small char/misc and other subsystem driver fixes for
6.2-rc5 to resolve a few reported issues. They include:
- long time pending fastrpc fixes (should have gone into 6.1, my
fault)
- mei driver/bus fixes and new device ids
- interconnect driver fixes for reported problems
- vmci bugfix
- w1 driver bugfixes for reported problems
Almost all of these have been in linux-next with no reported problems,
the rest have all passed 0-day bot testing in my tree and on the
mailing lists where they have sat too long due to me taking a long
time to catch up on my pending patch queue"
* tag 'char-misc-6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
VMCI: Use threaded irqs instead of tasklets
misc: fastrpc: Pass bitfield into qcom_scm_assign_mem
gsmi: fix null-deref in gsmi_get_variable
misc: fastrpc: Fix use-after-free race condition for maps
misc: fastrpc: Don't remove map on creater_process and device_release
misc: fastrpc: Fix use-after-free and race in fastrpc_map_find
misc: fastrpc: fix error code in fastrpc_req_mmap()
mei: me: add meteor lake point M DID
mei: bus: fix unlink on bus in error path
w1: fix WARNING after calling w1_process()
w1: fix deadloop in __w1_remove_master_device()
comedi: adv_pci1760: Fix PWM instruction handling
interconnect: qcom: rpm: Use _optional func for provider clocks
interconnect: qcom: msm8996: Fix regmap max_register values
interconnect: qcom: msm8996: Provide UFS clocks to A2NoC
dt-bindings: interconnect: Add UFS clocks to MSM8996 A2NoC
Pull driver core fixes from Greg KH:
"Here are three small driver and kernel core fixes for 6.2-rc5. They
include:
- potential gadget fixup in do_prlimit
- device property refcount leak fix
- test_async_probe bugfix for reported problem"
* tag 'driver-core-6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core:
prlimit: do_prlimit needs to have a speculation check
driver core: Fix test_async_probe_init saves device in wrong array
device property: fix of node refcount leak in fwnode_graph_get_next_endpoint()
Pull staging driver fix from Greg KH:
"Here is a single staging driver fix for 6.2-rc5. It resolves a build
issue reported and Fixed by Arnd in the vc04_services driver. It's
been in linux-next this week with no reported problems"
* tag 'staging-6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
staging: vchiq_arm: fix enum vchiq_status return types
Pull tty/serial driver fixes from Greg KH:
"Here are some small tty and serial driver fixes for 6.2-rc5 that
resolve a number of tiny reported issues and some new device ids. They
include:
- new device id for the exar serial driver
- speakup tty driver bugfix
- atmel serial driver baudrate fixup
- stm32 serial driver bugfix and then revert as the bugfix broke the
build. That will come back in a later pull request once it is all
worked out properly.
- amba-pl011 serial driver rs486 mode bugfix
- qcom_geni serial driver bugfix
Most of these have been in linux-next with no reported problems (well,
other than the build breakage which generated the revert), the new
device id passed 0-day testing"
* tag 'tty-6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
serial: exar: Add support for Sealevel 7xxxC serial cards
Revert "serial: stm32: Merge hard IRQ and threaded IRQ handling into single IRQ handler"
tty: serial: qcom_geni: avoid duplicate struct member init
serial: atmel: fix incorrect baudrate setup
tty: fix possible null-ptr-defer in spk_ttyio_release
serial: stm32: Merge hard IRQ and threaded IRQ handling into single IRQ handler
serial: amba-pl011: fix high priority character transmission in rs486 mode
serial: pch_uart: Pass correct sg to dma_unmap_sg()
tty: serial: qcom-geni-serial: fix slab-out-of-bounds on RX FIFO buffer
Pull USB / Thunderbolt fixes from Greg KH:
"Here are a number of small USB and Thunderbolt driver fixes and new
device id changes for 6.2-rc5. Included in here are:
- thunderbolt bugfixes for reported problems
- new usb-serial driver ids added
- onboard_hub usb driver fixes for much-reported problems
- xhci bugfixes
- typec bugfixes
- ehci-fsl driver module alias fix
- iowarrior header size fix
- usb gadget driver fixes
All of these, except for the iowarrior fix, have been in linux-next
with no reported issues. The iowarrior fix passed the 0-day testing
and is a one digit change based on a reported problem in the driver
(which was written to a spec, not the real device that is now
available)"
* tag 'usb-6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (40 commits)
USB: misc: iowarrior: fix up header size for USB_DEVICE_ID_CODEMERCS_IOW100
usb: host: ehci-fsl: Fix module alias
usb: dwc3: fix extcon dependency
usb: core: hub: disable autosuspend for TI TUSB8041
USB: fix misleading usb_set_intfdata() kernel doc
usb: gadget: f_ncm: fix potential NULL ptr deref in ncm_bitrate()
USB: gadget: Add ID numbers to configfs-gadget driver names
usb: typec: tcpm: Fix altmode re-registration causes sysfs create fail
usb: gadget: g_webcam: Send color matching descriptor per frame
usb: typec: altmodes/displayport: Use proper macro for pin assignment check
usb: typec: altmodes/displayport: Fix pin assignment calculation
usb: typec: altmodes/displayport: Add pin assignment helper
usb: gadget: f_fs: Ensure ep0req is dequeued before free_request
usb: gadget: f_fs: Prevent race during ffs_ep0_queue_wait
usb: misc: onboard_hub: Move 'attach' work to the driver
usb: misc: onboard_hub: Invert driver registration order
usb: ucsi: Ensure connector delayed work items are flushed
usb: musb: fix error return code in omap2430_probe()
usb: chipidea: core: fix possible constant 0 if use IS_ERR(ci->role_switch)
xhci: Detect lpm incapable xHC USB3 roothub ports from ACPI tables
...
Pull Kbuild fixes from Masahiro Yamada:
- Hide LDFLAGS_vmlinux from decompressor Makefiles to fix error
messages when GNU Make 4.4 is used.
- Fix 'make modules' build error when CONFIG_DEBUG_INFO_BTF_MODULES=y.
- Fix warnings emitted by GNU Make 4.4 in scripts/kconfig/Makefile.
- Support GNU Make 4.4 for scripts/jobserver-exec.
- Show clearer error message when kernel/gen_kheaders.sh fails due to
missing cpio.
* tag 'kbuild-fixes-v6.2-3' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild:
kheaders: explicitly validate existence of cpio command
scripts: support GNU make 4.4 in jobserver-exec
kconfig: Update all declared targets
scripts: rpm: make clear that mkspec script contains 4.13 feature
init/Kconfig: fix LOCALVERSION_AUTO help text
kbuild: fix 'make modules' error when CONFIG_DEBUG_INFO_BTF_MODULES=y
kbuild: export top-level LDFLAGS_vmlinux only to scripts/Makefile.vmlinux
init/version-timestamp.c: remove unneeded #include <linux/version.h>
docs: kbuild: remove mention to dropped $(objtree) feature
do_prlimit() adds the user-controlled resource value to a pointer that
will subsequently be dereferenced. In order to help prevent this
codepath from being used as a spectre "gadget" a barrier needs to be
added after checking the range.
Reported-by: Jordy Zomer <jordyzomer@google.com>
Tested-by: Jordy Zomer <jordyzomer@google.com>
Suggested-by: Linus Torvalds <torvalds@linuxfoundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Ye Bin
Baokun Li
Zhihao Cheng
Kees Cook
XU pengfei
zhanchengbin
Jun Nie
Tanmay Bhushan
Zhang Yi
Wang Jianjian
Eric Biggers
Theodore Ts'o
Linus Torvalds
Greg Kroah-Hartman