Ben reports that this should not have been backported to the older
kernels as the rest of the macro is not empty. It was a clean-up patch
in 6.4-rc1 only, it did not add new device ids.
which may fail for example (or other errors could happen). When the
failure occurs, btrfs_relocate_block_group() detects the error and frees
rc and doesn't set fs_info->reloc_ctl to NULL. After that, in
btrfs_init_reloc_root(), rc is retrieved from fs_info->reloc_ctl and
then used, which may cause a use-after-free bug.
This possible bug can be triggered by calling btrfs_ioctl_balance()
before calling btrfs_ioctl_defrag().
To fix this possible bug, in prepare_to_relocate(), check if
btrfs_commit_transaction() fails. If the failure occurs,
unset_reloc_control() is called to set fs_info->reloc_ctl to NULL.
The error log in our fault-injection testing is shown as follows:
There are a few places where we don't check the return value of
btrfs_commit_transaction in relocation.c. Thankfully all these places
have straightforward error handling, so simply change all of the sites
at once.
Reviewed-by: Qu Wenruo <wqu@suse.com> Signed-off-by: Josef Bacik <josef@toxicpanda.com> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Stefan Ghinea <stefan.ghinea@windriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
ext4_xattr_block_set() relies on its caller to call dquot_initialize()
on the inode. To assure that this has happened there are WARN_ON
checks. Unfortunately, this is subject to false positives if there is
an antagonist thread which is flipping the file system at high rates
between r/o and rw. So only do the check if EXT4_XATTR_DEBUG is
enabled.
If pm runtime resume fails the .remove callback used to exit early. This
resulted in an error message by the driver core but the device gets
removed anyhow. This lets the registered i2c adapter stay around with an
unbound parent device.
So only skip clk disabling if resume failed, but do delete the adapter.
Similar to commit 0f7d9b31ce7a ("netfilter: nf_tables: fix use-after-free
in nft_set_catchall_destroy()"). We can not access k after kfree_rcu()
call.
Cc: stable@vger.kernel.org Signed-off-by: Min Li <lm0963hack@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is a race between capsnaps flush and removing the inode from
'mdsc->snap_flush_list' list:
== Thread A == == Thread B ==
ceph_queue_cap_snap()
-> allocate 'capsnapA'
->ihold('&ci->vfs_inode')
->add 'capsnapA' to 'ci->i_cap_snaps'
->add 'ci' to 'mdsc->snap_flush_list'
...
== Thread C ==
ceph_flush_snaps()
->__ceph_flush_snaps()
->__send_flush_snap()
handle_cap_flushsnap_ack()
->iput('&ci->vfs_inode')
this also will release 'ci'
...
== Thread D ==
ceph_handle_snap()
->flush_snaps()
->iterate 'mdsc->snap_flush_list'
->get the stale 'ci'
->remove 'ci' from ->ihold(&ci->vfs_inode) this
'mdsc->snap_flush_list' will WARNING
To fix this we will increase the inode's i_count ref when adding 'ci'
to the 'mdsc->snap_flush_list' list.
According to Alex, most APUs from that time seem to have the same issue
(vbios says 48Mhz, actual is 100Mhz). I only have a CHIP_STONEY so I
limit the fixup to CHIP_STONEY
While doing my research to improve the xpad device names I noticed
that the 1532:0037 VID/PID seems to be used by the DeathAdder 2013,
so that Razer Sabertooth instance looked wrong and very suspect to
me. I didn't see any mention in the official drivers, either.
After doing more research, it turns out that the xpad list
is used by many other projects (like Steam) as-is [1], this
issue was reported [2] and Valve/Sam Lantinga fixed it [3]:
(With multiple Internet users reporting similar issues, not linked here)
After not being able to find the correct VID/PID combination anywhere
on the Internet and not receiving any reply from Razer support I did
some additional detective work, it seems like it presents itself as
"Razer Sabertooth Gaming Controller (XBOX360)", code 1689:FE00.
Leaving us with this:
* Razer Sabertooth (1689:fe00)
* Razer Sabertooth Elite (24c6:5d04)
* Razer DeathAdder 2013 (1532:0037) [note: not a gamepad]
So, to sum things up; remove this conflicting/duplicate entry:
As the real/correct one is already present there, even if
the Internet as a whole insists on presenting it as the
Razer Sabertooth Elite, which (by all accounts) is not:
Syzkaller got a lot of crashes like:
KASAN: use-after-free Write in *_timers*
All of these crashes point to the same memory area:
The buggy address belongs to the object at ffff88801f870000
which belongs to the cache kmalloc-8k of size 8192
The buggy address is located 5320 bytes inside of
8192-byte region [ffff88801f870000, ffff88801f872000)
This area belongs to :
batadv_priv->batadv_priv_dat->delayed_work->timer_list
The reason for these issues is the lack of synchronization. Delayed
work (batadv_dat_purge) schedules new timer/work while the device
is being deleted. As the result new timer/delayed work is set after
cancel_delayed_work_sync() was called. So after the device is freed
the timer list contains pointer to already freed memory.
Found by Linux Verification Center (linuxtesting.org) with syzkaller.
Cc: stable@kernel.org Fixes: 2f1dfbe18507 ("batman-adv: Distributed ARP Table - implement local storage") Signed-off-by: Vladislav Efanov <VEfanov@ispras.ru> Acked-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
irq_cpu_rmap_release() calls cpu_rmap_put(), which may free the rmap.
So we need to clear the pointer to our glue structure in rmap before
doing that, not after.
Fixes: 4e0473f1060a ("lib: cpu_rmap: Avoid use after free on rmap->obj array entries") Signed-off-by: Ben Hutchings <ben@decadent.org.uk> Reviewed-by: Simon Horman <simon.horman@corigine.com> Link: https://lore.kernel.org/r/ZHo0vwquhOy3FaXc@decadent.org.uk Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
try_module_get will be called in tcf_proto_lookup_ops. So module_put needs
to be called to drop the refcount if ops don't implement the required
function.
Fixes: 9f407f1768d3 ("net: sched: introduce chain templates") Signed-off-by: Hangyu Hua <hbh25y@gmail.com> Reviewed-by: Larysa Zaremba <larysa.zaremba@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
rtm_tca_policy is used from net/sched/sch_api.c and net/sched/cls_api.c,
thus should be declared in an include file.
This fixes the following sparse warning:
net/sched/sch_api.c:1434:25: warning: symbol 'rtm_tca_policy' was not declared. Should it be static?
Fixes: e331473fee3d ("net/sched: cls_api: add missing validation of netlink attributes") Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
When receiving a connect response we should make sure that the DCID is
within the valid range and that we don't already have another channel
allocated for the same DCID.
Missing checks may violate the specification (BLUETOOTH CORE SPECIFICATION
Version 5.4 | Vol 3, Part A, Page 1046).
Fixes: 40624183c202 ("Bluetooth: L2CAP: Add missing checks for invalid LE DCID") Signed-off-by: Sungwoo Kim <iam@sung-woo.kim> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
L2CAP assumes that the locks conn->chan_lock and chan->lock are
acquired in the order conn->chan_lock, chan->lock to avoid
potential deadlock.
For example, l2sock_shutdown acquires these locks in the order:
mutex_lock(&conn->chan_lock)
l2cap_chan_lock(chan)
However, l2cap_disconnect_req acquires chan->lock in
l2cap_get_chan_by_scid first and then acquires conn->chan_lock
before calling l2cap_chan_del. This means that these locks are
acquired in unexpected order, which leads to potential deadlock:
l2cap_chan_lock(c)
mutex_lock(&conn->chan_lock)
This patch releases chan->lock before acquiring the conn_chan_lock
to avoid the potential deadlock.
Fixes: a2a9339e1c9d ("Bluetooth: L2CAP: Fix use-after-free in l2cap_disconnect_{req,rsp}") Signed-off-by: Ying Hsu <yinghsu@chromium.org> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
LAN9303 doesn't associate FDB (ALR) entries with VLANs, it has just one
global Address Logic Resolution table [1].
Ignore VID in port_fdb_{add|del} methods, go on with the global table. This
is the same semantics as hellcreek or RZ/N1 implement.
Visible symptoms:
LAN9303_MDIO 5b050000.ethernet-1:00: port 2 failed to delete 00:xx:xx:xx:xx:cf vid 1 from fdb: -2
LAN9303_MDIO 5b050000.ethernet-1:00: port 2 failed to add 00:xx:xx:xx:xx:cf vid 1 to fdb: -95
It is usually better to request all necessary resources (clocks,
regulators, ...) before starting to make use of them. That way they do
not change state in case one of the resources is not available yet and
probe deferral (-EPROBE_DEFER) is necessary. This is particularly
important for DMA channels and IOMMUs which are not enforced by
fw_devlink yet (unless you use fw_devlink.strict=1).
spi-qup does this in the wrong order, the clocks are enabled and
disabled again when the DMA channels are not available yet.
This causes issues in some cases: On most SoCs one of the SPI QUP
clocks is shared with the UART controller. When using earlycon UART is
actively used during boot but might not have probed yet, usually for
the same reason (waiting for the DMA controller). In this case, the
brief enable/disable cycle ends up gating the clock and further UART
console output will halt the system completely.
Avoid this by requesting the DMA channels before changing the clock
state.
Not upstream as it was fixed in a much larger api change in newer
kernels.
gcc-13 rightfully complains that enum is not the same as an int, so fix
up the function prototypes in i40e_alloc.h to be correct, solving a
bunch of build warnings.
Not upstream as this function is no longer around anymore.
The function i40iw_manage_apbvt() has the wrong prototype match from the
.h file to the .c declaration, so fix it up, otherwise gcc-13 complains
(rightfully) that the type is incorrect.
Since IQK could spend time, we make a cache of IQK result matrix that looks
like iqk_matrix[channel_idx].val[x][y], and we can reload the matrix if we
have made a cache. To determine a cache is made, we check
iqk_matrix[channel_idx].val[0][0].
The initial commit 7274a8c22980 ("rtlwifi: rtl8192de: Merge phy routines")
make a mistake that checks incorrect iqk_matrix[channel_idx].val[0] that
is always true, and this mistake is found by commit ee3db469dd31
("wifi: rtlwifi: remove always-true condition pointed out by GCC 12"), so
I recall the vendor driver to find fix and apply the correctness.
Ben Hutchings [Sat, 27 May 2023 13:52:48 +0000 (15:52 +0200)]
scsi: dpt_i2o: Do not process completions with invalid addresses
adpt_isr() reads reply addresses from a hardware register, which
should always be within the DMA address range of the device's pool of
reply address buffers. In case the address is out of range, it tries
to muddle on, converting to a virtual address using bus_to_virt().
bus_to_virt() does not take DMA addresses, and it doesn't make sense
to try to handle the completion in this case. Ignore it and continue
looping to service the interrupt. If a completion has been lost then
the SCSI core should eventually time-out and trigger a reset.
There is no corresponding upstream commit, because this driver was
removed upstream.
Fixes: 67af2b060e02 ("[SCSI] dpt_i2o: move from virt_to_bus/bus_to_virt ...") Signed-off-by: Ben Hutchings <benh@debian.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
adpt_i2o_passthru() takes a user-provided message and passes it
through to the hardware with appropriate translation of addresses
and message IDs. It has a number of bugs:
- When a message requires scatter/gather, it doesn't verify that the
offset to the scatter/gather list is less than the message size.
- When a message requires scatter/gather, it overwrites the DMA
addresses with the user-space virtual addresses before unmapping the
DMA buffers.
- It reads the message from user memory multiple times. This allows
user-space to change the message and bypass validation.
- It assumes that the message is at least 4 words long, but doesn't
check that.
I tried fixing these, but even the maintainer of the corresponding
user-space in Debian doesn't have the hardware any more.
Instead, remove the pass-through ioctl (I2OUSRCMD) and supporting
code.
There is no corresponding upstream commit, because this driver was
removed upstream.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Fixes: 67af2b060e02 ("[SCSI] dpt_i2o: move from virt_to_bus/bus_to_virt ...") Signed-off-by: Ben Hutchings <benh@debian.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently, when regmap_raw_write() splits the data, it uses the
max_raw_write value defined for the bus. For any bus that includes
the target register address in the max_raw_write value, the chunked
transmission will always exceed the maximum transmission length.
To avoid this problem, subtract the length of the register and the
padding from the maximum transmission.
BUG: KASAN: null-ptr-deref in soft_cursor+0x384/0x6b4 drivers/video/fbdev/core/softcursor.c:70
Read of size 16 at addr 0000000000000200 by task kworker/u4:1/12
Treat i_data_sem for ea_inodes as being in their own lockdep class to
avoid lockdep complaints about ext4_setattr's use of inode_lock() on
normal inodes potentially causing lock ordering with i_data_sem on
ea_inodes in ext4_xattr_inode_write(). However, ea_inodes will be
operated on by ext4_setattr(), so this isn't a problem.
The Linux Kernel currently only requires make v3.82 while the grouped
target functionality requires make v4.3. Removed the grouped target
introduced in 4ce1f694eb5d ("selinux: ensure av_permissions.h is
built when needed") as well as the multiple header file targets in
the make rule. This effectively reverts the problem commit.
We will revisit this change when make >= 4.3 is required by the rest
of the kernel.
Cc: stable@vger.kernel.org Fixes: 4ce1f694eb5d ("selinux: ensure av_permissions.h is built when needed") Reported-by: Erwan Velu <e.velu@criteo.com> Reported-by: Luiz Capitulino <luizcap@amazon.com> Tested-by: Luiz Capitulino <luizcap@amazon.com> Signed-off-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
LPUART IP now has two known bugs, one is that CTS has higher priority
than the break signal, which causes the break signal sending through
UARTCTRL_SBK may impacted by the CTS input if the HW flow control is
enabled. It exists on all platforms we support in this driver.
So we add a workaround patch for this issue: commit c4c81db5cf8b
("tty: serial: fsl_lpuart: disable the CTS when send break signal").
Another IP bug is i.MX8QM LPUART may have an additional break character
being sent after SBK was cleared. It may need to add some delay between
clearing SBK and re-enabling CTS to ensure that the SBK latch are
completely cleared.
But we found that during the delay period before CTS is enabled, there
is still a risk that Bluetooth data in TX FIFO may be sent out during
this period because of break off and CTS disabled(even if BT sets CTS
line deasserted, data is still sent to BT).
Due to this risk, we have to drop the CTS-disabling workaround for SBK
bugs, use TXINV seems to be a better way to replace SBK feature and
avoid above risk. Also need to disable the transmitter to prevent any
data from being sent out during break, then invert the TX line to send
break. Then disable the TXINV when turn off break and re-enable
transmitter.
Fixes: c4c81db5cf8b ("tty: serial: fsl_lpuart: disable the CTS when send break signal") Cc: stable <stable@kernel.org> Signed-off-by: Sherry Sun <sherry.sun@nxp.com> Link: https://lore.kernel.org/r/20230519094751.28948-1-sherry.sun@nxp.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We may get an empty response with zero length at the beginning of
the driver start and get following UBSAN error. Since there is no
content(SDRT_NONE) for the response, just return and skip the response
handling to avoid this problem.
Test pass : SDIO wifi throughput test with this patch
Clang warns that the address of a pointer will always evaluated as true
in a boolean context:
drivers/regulator/da9052-regulator.c:423:22: warning: address of array
'pdata->regulators' will always evaluate to 'true'
[-Wpointer-bool-conversion]
if (pdata && pdata->regulators) {
~~ ~~~~~~~^~~~~~~~~~
drivers/regulator/da9055-regulator.c:615:22: warning: address of array
'pdata->regulators' will always evaluate to 'true'
[-Wpointer-bool-conversion]
if (pdata && pdata->regulators) {
~~ ~~~~~~~^~~~~~~~~~
Link: https://github.com/ClangBuiltLinux/linux/issues/142 Signed-off-by: Nathan Chancellor <natechancellor@gmail.com> Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Clang warns when the address of a pointer is used in a boolean context
as it will always return true.
drivers/hwmon/scmi-hwmon.c:59:24: warning: address of array
'sensor->name' will always evaluate to 'true'
[-Wpointer-bool-conversion]
if (sensor && sensor->name)
~~ ~~~~~~~~^~~~
1 warning generated.
Remove the check as it isn't doing anything currently; if validation
of the contents of the data structure was intended by the original
author (since this line has been present from the first version of
this driver), it can be added in a follow-up patch.
../lib/dynamic_debug.c:1034:24: warning: array comparison always
evaluates to false [-Wtautological-compare]
if (__start___verbose == __stop___verbose) {
^
1 warning generated.
These are not true arrays, they are linker defined symbols, which are just
addresses. Using the address of operator silences the warning and does
not change the resulting assembly with either clang/ld.lld or gcc/ld
(tested with diff + objdump -Dr).
../kernel/extable.c:37:52: warning: array comparison always evaluates to
a constant [-Wtautological-compare]
if (main_extable_sort_needed && __stop___ex_table > __start___ex_table) {
^
1 warning generated.
These are not true arrays, they are linker defined symbols, which are just
addresses. Using the address of operator silences the warning and does
not change the resulting assembly with either clang/ld.lld or gcc/ld
(tested with diff + objdump -Dr).
drivers/net/ethernet/sun/cassini.c:1316:29: error: comparison between two arrays [-Werror=array-compare]
drivers/net/ethernet/sun/cassini.c:3783:34: error: comparison between two arrays [-Werror=array-compare]
Note that 2 arrays should be compared by comparing of their addresses:
note: use ‘&cas_prog_workaroundtab[0] == &cas_prog_null[0]’ to compare the addresses
Signed-off-by: Martin Liska <mliska@suse.cz> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
While the concept of checking for dangling pointers to local variables
at function exit is really interesting, the gcc-12 implementation is not
compatible with reality, and results in false positives.
For example, gcc sees us putting things on a local list head allocated
on the stack, which involves exactly those kinds of pointers to the
local stack entry:
In function ‘__list_add’,
inlined from ‘list_add_tail’ at include/linux/list.h:102:2,
inlined from ‘rebuild_snap_realms’ at fs/ceph/snap.c:434:2:
include/linux/list.h:74:19: warning: storing the address of local variable ‘realm_queue’ in ‘*&realm_27(D)->rebuild_item.prev’ [-Wdangling-pointer=]
74 | new->prev = prev;
| ~~~~~~~~~~^~~~~~
But then gcc - understandably - doesn't really understand the big
picture how the doubly linked list works, so doesn't see how we then end
up emptying said list head in a loop and the pointer we added has been
removed.
Gcc also complains about us (intentionally) using this as a way to store
a kind of fake stack trace, eg
drivers/acpi/acpica/utdebug.c:40:38: warning: storing the address of local variable ‘current_sp’ in ‘acpi_gbl_entry_stack_pointer’ [-Wdangling-pointer=]
40 | acpi_gbl_entry_stack_pointer = ¤t_sp;
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~
which is entirely reasonable from a compiler standpoint, and we may want
to change those kinds of patterns, but not not.
So this is one of those "it would be lovely if the compiler were to
complain about us leaving dangling pointers to the stack", but not this
way.
Address of a field inside a struct can't possibly be null; gcc-12 warns
about this.
Signed-off-by: Adam Borowski <kilobyte@angband.pl> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
GCC 11 (incorrectly[1]) assumes that literal values cast to (void *)
should be treated like a NULL pointer with an offset, and raises
diagnostics when doing bounds checking under -Warray-bounds. GCC 12
got "smarter" about finding these:
In function 'rdfs8',
inlined from 'vga_recalc_vertical' at /srv/code/arch/x86/boot/video-mode.c:124:29,
inlined from 'set_mode' at /srv/code/arch/x86/boot/video-mode.c:163:3:
/srv/code/arch/x86/boot/boot.h:114:9: warning: array subscript 0 is outside array bounds of 'u8[0]' {aka 'unsigned char[]'} [-Warray-bounds]
114 | asm volatile("movb %%fs:%1,%0" : "=q" (v) : "m" (*(u8 *)addr));
| ^~~
This has been solved in other places[2] already by using the recently
added absolute_pointer() macro. Do the same here.
For devices not attached to a port multiplier and managed directly by
libata, the device number passed to ata_find_dev() must always be lower
than the maximum number of devices returned by ata_link_max_devices().
That is 1 for SATA devices or 2 for an IDE link with master+slave
devices. This device number is the SCSI device ID which matches these
constraints as the IDs are generated per port and so never exceed the
maximum number of devices for the link being used.
However, for libsas managed devices, SCSI device IDs are assigned per
struct scsi_host, leading to device IDs for SATA devices that can be
well in excess of libata per-link maximum number of devices. This
results in ata_find_dev() to always return NULL for libsas managed
devices except for the first device of the target scsi_host with ID
(device number) equal to 0. This issue is visible by executing the
hdparm utility, which fails. E.g.:
hdparm -i /dev/sdX
/dev/sdX:
HDIO_GET_IDENTITY failed: No message of desired type
Fix this by rewriting ata_find_dev() to ignore the device number for
non-PMP attached devices with a link with at most 1 device, that is SATA
devices. For these, the device number 0 is always used to
return the correct pointer to the struct ata_device of the port link.
This change excludes IDE master/slave setups (maximum number of devices
per link is 2) and port-multiplier attached devices. Also, to be
consistant with the fact that SCSI device IDs and channel numbers used
as device numbers are both unsigned int, change the devno argument of
ata_find_dev() to unsigned int.
Reported-by: Xingui Yang <yangxingui@huawei.com> Fixes: 41bda9c98035 ("libata-link: update hotplug to handle PMP links") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <dlemoal@kernel.org> Reviewed-by: Jason Yan <yanaijie@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
gcc 13 may assign another type to enumeration constants than gcc 12. Split
the large enum at the top of source file stex.c such that the type of the
constants used in time expressions is changed back to the same type chosen
by gcc 12. This patch suppresses compiler warnings like this one:
In file included from ./include/linux/bitops.h:7,
from ./include/linux/kernel.h:22,
from drivers/scsi/stex.c:13:
drivers/scsi/stex.c: In function ‘stex_common_handshake’:
./include/linux/typecheck.h:12:25: error: comparison of distinct pointer types lacks a cast [-Werror]
12 | (void)(&__dummy == &__dummy2); \
| ^~
./include/linux/jiffies.h:106:10: note: in expansion of macro ‘typecheck’
106 | typecheck(unsigned long, b) && \
| ^~~~~~~~~
drivers/scsi/stex.c:1035:29: note: in expansion of macro ‘time_after’
1035 | if (time_after(jiffies, before + MU_MAX_DELAY * HZ)) {
| ^~~~~~~~~~
See also https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107405.
Cc: stable@vger.kernel.org Acked-by: Randy Dunlap <rdunlap@infradead.org> Tested-by: Randy Dunlap <rdunlap@infradead.org> # build-tested Signed-off-by: Bart Van Assche <bvanassche@acm.org> Link: https://lore.kernel.org/r/20230529195034.3077-1-bvanassche@acm.org Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
While exercising the unbind path, with the current implementation
the functionfs_unbind would be calling which waits for the ffs->mutex
to be available, however within the same time ffs_ep0_read is invoked
& if no setup packets are pending, it will invoke function
wait_event_interruptible_exclusive_locked_irq which by definition waits
for the ev.count to be increased inside the same mutex for which
functionfs_unbind is waiting.
This creates deadlock situation because the functionfs_unbind won't
get the lock until ev.count is increased which can only happen if
the caller ffs_func_unbind can proceed further.
Commit 28d1a7ac2a0d ("iio: dac: Add AD5758 support") adds the config AD5758
and the corresponding driver ad5758.c. In the Makefile, the ad5758 driver
is however included when AD5755 is selected, not when AD5758 is selected.
Probably, this was simply a mistake that happened by copy-and-paste and
forgetting to adjust the actual line. Surprisingly, no one has ever noticed
that this driver is actually only included when AD5755 is selected and that
the config AD5758 has actually no effect on the build.
The i2c_master_send() returns number of sent bytes on success,
or negative on error. The suspend/resume callbacks expect zero
on success and non-zero on error. Adapt the return value of the
i2c_master_send() to the expectation of the suspend and resume
callbacks, including proper validation of the return value.
If high bit is set to 1 in ((data[3] & 0x0f << 28), after all arithmetic
operations and integer promotions are done, high bits in
wacom->serial[idx] will be filled with 1s as well.
Avoid this, albeit unlikely, issue by specifying left operand's __u64
type for the right operand.
Found by Linux Verification Center (linuxtesting.org) with static
analysis tool SVACE.
the order of three init operation:
1.mxs_lradc_adc_trigger_init
2.iio_triggered_buffer_setup
3.mxs_lradc_adc_hw_init
thus, the order of three cleanup operation should be:
1.mxs_lradc_adc_hw_stop
2.iio_triggered_buffer_cleanup
3.mxs_lradc_adc_trigger_remove
we exchange the order of two cleanup operations,
introducing the following differences:
1.if mxs_lradc_adc_trigger_init fails, returns directly;
2.if trigger_init succeeds but iio_triggered_buffer_setup fails,
goto err_trig and remove the trigger.
In addition, we also reorder the unwind that goes on in the
remove() callback to match the new ordering.
Fixes: 6dd112b9f85e ("iio: adc: mxs-lradc: Add support for ADC driver") Signed-off-by: Jiakai Luo <jkluo@hust.edu.cn> Reviewed-by: Dongliang Mu <dzm91@hust.edu.cn> Link: https://lore.kernel.org/r/20230422133407.72908-1-jkluo@hust.edu.cn Cc: <Stable@vger.kernel.org> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When CONFIG_PROC_FS is disabled, the function declarations for some
procfs functions are hidden, but the definitions are still build,
as shown by this compiler warning:
net/atm/resources.c:403:7: error: no previous prototype for 'atm_dev_seq_start' [-Werror=missing-prototypes]
net/atm/resources.c:409:6: error: no previous prototype for 'atm_dev_seq_stop' [-Werror=missing-prototypes]
net/atm/resources.c:414:7: error: no previous prototype for 'atm_dev_seq_next' [-Werror=missing-prototypes]
Add another #ifdef to leave these out of the build.
Two functions are defined and used in pcm_oss.c but also optionally
used from io.c, with an optional prototype. If CONFIG_SND_PCM_OSS_PLUGINS
is disabled, this causes a warning as the functions are not static
and have no prototype:
sound/core/oss/pcm_oss.c:1235:19: error: no previous prototype for 'snd_pcm_oss_write3' [-Werror=missing-prototypes]
sound/core/oss/pcm_oss.c:1266:19: error: no previous prototype for 'snd_pcm_oss_read3' [-Werror=missing-prototypes]
Avoid this by making the prototypes unconditional.
gcc with W=1 and ! CONFIG_NF_NAT
net/netfilter/nf_conntrack_netlink.c:3463:32: error:
‘exp_nat_nla_policy’ defined but not used [-Werror=unused-const-variable=]
3463 | static const struct nla_policy exp_nat_nla_policy[CTA_EXPECT_NAT_MAX+1] = {
| ^~~~~~~~~~~~~~~~~~
net/netfilter/nf_conntrack_netlink.c:2979:33: error:
‘any_addr’ defined but not used [-Werror=unused-const-variable=]
2979 | static const union nf_inet_addr any_addr;
| ^~~~~~~~
These variables use is controlled by CONFIG_NF_NAT, so should their definitions.
Signed-off-by: Tom Rix <trix@redhat.com> Reviewed-by: Simon Horman <simon.horman@corigine.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
clang warns about an unpacked structure inside of a packed one:
drivers/net/wireless/broadcom/b43/b43.h:654:4: error: field data within 'struct b43_iv' is less aligned than 'union (unnamed union at /home/arnd/arm-soc/drivers/net/wireless/broadcom/b43/b43.h:651:2)' and is usually due to 'struct b43_iv' being packed, which can lead to unaligned accesses [-Werror,-Wunaligned-access]
The problem here is that the anonymous union has the default alignment
from its members, apparently because the original author mixed up the
placement of the __packed attribute by placing it next to the struct
member rather than the union definition. As the struct itself is
also marked as __packed, there is no need to mark its members, so just
move the annotation to the inner type instead.
As Michael noted, the same problem is present in b43legacy, so
change both at the same time.
If scsi_dispatch_cmd() failed, the SCSI command was not sent to the target,
scsi_queue_rq() would return BLK_STS_RESOURCE and the related request would
be requeued. The timeout of this request would not fire, no one would
increase iodone_cnt.
The above flow would result the iodone_cnt smaller than iorequest_cnt. So
decrease the iorequest_cnt if dispatch failed to workaround the issue.
Signed-off-by: Wenchao Hao <haowenchao2@huawei.com> Reported-by: Ming Lei <ming.lei@redhat.com> Closes: https://lore.kernel.org/r/ZF+zB+bB7iqe0wGd@ovpn-8-17.pek2.redhat.com Link: https://lore.kernel.org/r/20230515070156.1790181-3-haowenchao2@huawei.com Reviewed-by: Ming Lei <ming.lei@redhat.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When using rtl8192cu with rtl8xxxu driver to connect wifi, there is a
probability of failure, which shows "authentication with ... timed out".
Through debugging, it was found that the RCR register has been inexplicably
modified to an incorrect value, resulting in the nic not being able to
receive authenticated frames.
To fix this problem, add regrcr in rtl8xxxu_priv struct, and store
the RCR value every time the register is written, and use it the next
time the register need to be modified.
Using a semaphore in the wait_event*() condition is no good idea.
It hits a kernel WARN_ON() at prepare_to_wait_event() like:
do not call blocking ops when !TASK_RUNNING; state=1 set at
prepare_to_wait_event+0x6d/0x690
For avoiding the potential deadlock, rewrite to an open-coded loop
instead. Unlike the loop in wait_event*(), this uses wait_woken()
after the condition check, hence the task state stays consistent.
A race condition may occur between the .disconnect function, which
is called when the device is disconnected, and the dvb_device_open()
function, which is called when the device node is open()ed.
This results in several types of UAFs.
The root cause of this is that you use the dvb_device_open() function,
which does not implement a conditional statement
that checks 'dvbnet->exit'.
So, add 'remove_mutex` to protect 'dvbnet->exit' and use
locked_dvb_net_open() function to check 'dvbnet->exit'.
The driver will match mostly by DT table (even thought there is regular
ID table) so there is little benefit in of_match_ptr (this also allows
ACPI matching via PRP0001, even though it might not be relevant here).
This also fixes !CONFIG_OF error:
drivers/media/dvb-frontends/mn88443x.c:782:34: error: ‘mn88443x_of_match’ defined but not used [-Werror=unused-const-variable=]
Since dvb_frontend_detach() is not called in ttusb_dec_exit_dvb(),
which is called when the device is disconnected, dvb_frontend_free()
is not finally called.
This causes a memory leak just by repeatedly plugging and
unplugging the device.
Fix this issue by adding dvb_frontend_detach() to ttusb_dec_exit_dvb().
The function of "dvb_ca_en50221_write_data" at source/drivers/media
/dvb-core/dvb_ca_en50221.c is used for two cases.
The first case is for writing APDU data in the function of
"dvb_ca_en50221_io_write" at source/drivers/media/dvb-core/
dvb_ca_en50221.c.
The second case is for writing the host link buf size on the
Command Register in the function of "dvb_ca_en50221_link_init"
at source/drivers/media/dvb-core/dvb_ca_en50221.c.
In the second case, there exists a bug like following.
In the function of the "dvb_ca_en50221_link_init",
after a TV host calculates the host link buf_size,
the TV host writes the calculated host link buf_size on the
Size Register.
Accroding to the en50221 Spec (the page 60 of
https://dvb.org/wp-content/uploads/2020/02/En50221.V1.pdf),
before this writing operation, the "SW(CMDREG_SW)" flag in the
Command Register should be set. We can see this setting operation
in the function of the "dvb_ca_en50221_link_init" like below.
...
if ((ret = ca->pub->write_cam_control(ca->pub, slot,
CTRLIF_COMMAND, IRQEN | CMDREG_SW)) != 0)
return ret;
...
But, after that, the real writing operation is implemented using
the function of the "dvb_ca_en50221_write_data" in the function of
"dvb_ca_en50221_link_init", and the "dvb_ca_en50221_write_data"
includes the function of "ca->pub->write_cam_control",
and the function of the "ca->pub->write_cam_control" in the
function of the "dvb_ca_en50221_wrte_data" does not include
"CMDREG_SW" flag like below.
...
if ((status = ca->pub->write_cam_control(ca->pub, slot,
CTRLIF_COMMAND, IRQEN | CMDREG_HC)) != 0)
...
In the above source code, we can see only the "IRQEN | CMDREG_HC",
but we cannot see the "CMDREG_SW".
The "CMDREG_SW" flag which was set in the function of the
"dvb_ca_en50221_link_init" was rollbacked by the follwoing function
of the "dvb_ca_en50221_write_data".
This is a bug. and this bug causes that the calculated host link buf_size
is not properly written in the CI module.
Through this patch, we fix this bug.
IRQ handler netup_spi_interrupt() takes spinlock spi->lock. The lock
is initialized in netup_spi_init(). However, irq handler is registered
before initializing the lock.
Spinlock dma->lock and i2c->lock suffer from the same problem.
Fix this by registering the irq at the end of probe.
In su3000_read_mac_address, if i2c_transfer fails to execute two
messages, array mac address will not be initialized. Without handling
such error, later in function dvb_usb_adapter_dvb_init, proposed_mac
is accessed before initialization.
Fix this error by returning a negative value if message execution fails.
In digitv_i2c_xfer, msg is controlled by user. When msg[i].buf
is null and msg[i].len is zero, former checks on msg[i].buf would be
passed. Malicious data finally reach digitv_i2c_xfer. If accessing
msg[i].buf[0] without sanity check, null ptr deref would happen. We add
check on msg[i].len to prevent crash.
Similar commit:
commit 0ed554fd769a ("media: dvb-usb: az6027: fix null-ptr-deref in az6027_i2c_xfer()")
In rtl28xxu_i2c_xfer, msg is controlled by user. When msg[i].buf
is null and msg[i].len is zero, former checks on msg[i].buf would be
passed. Malicious data finally reach rtl28xxu_i2c_xfer. If accessing
msg[i].buf[0] without sanity check, null ptr deref would happen.
We add check on msg[i].len to prevent crash.
Similar commit:
commit 0ed554fd769a
("media: dvb-usb: az6027: fix null-ptr-deref in az6027_i2c_xfer()")
In ce6230_i2c_master_xfer, msg is controlled by user. When msg[i].buf
is null and msg[i].len is zero, former checks on msg[i].buf would be
passed. Malicious data finally reach ce6230_i2c_master_xfer. If accessing
msg[i].buf[0] without sanity check, null ptr deref would happen. We add
check on msg[i].len to prevent crash.
Similar commit:
commit 0ed554fd769a ("media: dvb-usb: az6027: fix null-ptr-deref in az6027_i2c_xfer()")
In ec168_i2c_xfer, msg is controlled by user. When msg[i].buf is null
and msg[i].len is zero, former checks on msg[i].buf would be passed.
If accessing msg[i].buf[0] without sanity check, null pointer deref
would happen. We add check on msg[i].len to prevent crash.
Similar commit:
commit 0ed554fd769a ("media: dvb-usb: az6027: fix null-ptr-deref in az6027_i2c_xfer()")
In az6027_i2c_xfer, msg is controlled by user. When msg[i].buf is null,
commit 0ed554fd769a ("media: dvb-usb: az6027: fix null-ptr-deref in
az6027_i2c_xfer()") fix the null-ptr-deref bug when msg[i].addr is 0x99.
However, null-ptr-deref also happens when msg[i].addr is 0xd0 and 0xc0.
We add check on msg[i].len to prevent null-ptr-deref.
In dvb_demux.c, some logics exist which compare the expected
continuity counter and the real continuity counter. If they
are not matched each other, both of the expected continuity
counter and the real continuity counter should be printed.
But there exists a bug that the expected continuity counter
is not correctly printed. The expected continuity counter is
replaced with the real countinuity counter + 1 so that
the epected continuity counter is not correclty printed.
This is wrong. This bug is fixed.
Apply a workaround for what appears to be a hardware quirk.
The problem seems to happen when enabling "whole chip power" (bit D7
register R6) for the very first time after the chip receives power. If
either "output" (D4) or "DAC" (D3) aren't powered on at that time,
playback becomes very distorted later on.
This happens on the Google Chameleon v3, as well as on a ZYBO Z7-10:
https://ez.analog.com/audio/f/q-a/543726/solved-ssm2603-right-output-offset-issue/480229
I suspect this happens only when using an external MCLK signal (which
is the case for both of these boards).
Here are some experiments run on a Google Chameleon v3. These were run
in userspace using a wrapper around the i2cset utility:
ssmset() {
i2cset -y 0 0x1a $(($1*2)) $2
}
For each of the following sequences, we apply power to the ssm2603
chip, set the configuration registers R0-R5 and R7-R8, run the selected
sequence, and check for distortions on playback.
ssmset 0x09 0x01 # core
ssmset 0x06 0x1f # chip
ssmset 0x06 0x07 # out, dac
NOT OK
ssmset 0x06 0x1f # chip
ssmset 0x09 0x01 # core
ssmset 0x06 0x07 # out, dac
NOT OK
ssmset 0x09 0x01 # core
ssmset 0x06 0x0f # chip, out
ssmset 0x06 0x07 # dac
NOT OK
ssmset 0x09 0x01 # core
ssmset 0x06 0x17 # chip, dac
ssmset 0x06 0x07 # out
NOT OK
For each of the following sequences, we apply power to the ssm2603
chip, run the selected sequence, issue a reset with R15, configure
R0-R5 and R7-R8, run one of the NOT OK sequences from above, and check
for distortions.
This change ensures that if configured in the policy, the if_id set in
the policy and secpath states match during the inbound policy check.
Without this, there is potential for ambiguity where entries in the
secpath differing by only the if_id could be mismatched.
Notably, this is checked in the outbound direction when resolving
templates to SAs, but not on the inbound path when matching SAs and
policies.
Test: Tested against Android kernel unit tests & CTS Signed-off-by: Benedict Wong <benedictwong@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
On slow CPU (FPGA/QEMU emulated) printing overrun messages from
interrupt handler to uart console may leads to more overrun errors.
So use dev_err_ratelimited to limit the number of error messages.
The debugfs_create_dir function returns ERR_PTR in case of error, and the
only correct way to check if an error occurred is 'IS_ERR' inline function.
This patch will replace the null-comparison with IS_ERR.
When adding proper support for V4L2_FIELD_ALTERNATE it was missed that
this field format should trigger an interrupt for each field, not just
for the whole frame. Fix this by marking it as progressive in the
capture setup, which will then select the correct interrupt mode.
Tested on both Gen2 and Gen3 with the result of a doubling of the frame
rate for V4L2_FIELD_ALTERNATE. From a PAL video source the frame rate is
now 50, which is expected for alternate field capture.
When unwind instruction is 0xb2,the subsequent instructions
are uleb128 bytes.
For now,it uses only the first uleb128 byte in code.
For vsp increments of 0x204~0x400,use one uleb128 byte like below:
0xc06a00e4 <unwind_test_work>: 0x80b27fac
Compact model index: 0
0xb2 0x7f vsp = vsp + 1024
0xac pop {r4, r5, r6, r7, r8, r14}
For vsp increments larger than 0x400,use two uleb128 bytes like below:
0xc06a00e4 <unwind_test_work>: @0xc0cc9e0c
Compact model index: 1
0xb2 0x81 0x01 vsp = vsp + 1032
0xac pop {r4, r5, r6, r7, r8, r14}
The unwind works well since the decoded uleb128 byte is also 0x81.
For vsp increments larger than 0x600,use two uleb128 bytes like below:
0xc06a00e4 <unwind_test_work>: @0xc0cc9e0c
Compact model index: 1
0xb2 0x81 0x02 vsp = vsp + 1544
0xac pop {r4, r5, r6, r7, r8, r14}
In this case,the decoded uleb128 result is 0x101(vsp=0x204+(0x101<<2)).
While the uleb128 used in code is 0x81(vsp=0x204+(0x81<<2)).
The unwind aborts at this frame since it gets incorrect vsp.
To fix this,add uleb128 decode to cover all the above case.
Signed-off-by: Haibo Li <haibo.li@mediatek.com> Reviewed-by: Linus Walleij <linus.walleij@linaro.org> Reviewed-by: Alexandre Mergnat <amergnat@baylibre.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> Signed-off-by: Sasha Levin <sashal@kernel.org>
If a user can make copy_from_user() fail, there is a potential for
UAF/DF due to a lack of locking around the allocation, use and freeing
of the data buffers.
This issue is not theoretical. I managed to author a POC for it:
Doing a 'cat /dev/watchdog0' with menz069_wdt as watchdog0 will result in
a NULL pointer dereference.
This happens because we're passing the wrong pointer to
watchdog_register_device(). Fix this by getting rid of the static
watchdog_device structure and use the one embedded into the driver's
per-instance private data.
A switch held in reset by default needs to wait longer until we can
reliably detect it.
An issue was observed when testing on the Marvell 88E6393X (Link Street).
The driver failed to detect the switch on some upstarts. Increasing the
wait time after reset deactivation solves this issue.
The updated wait time is now also the same as the wait time in the
mv88e6xxx_hardware_reset function.
Fixes: 7b75e49de424 ("net: dsa: mv88e6xxx: wait after reset deactivation") Signed-off-by: Andreas Svensson <andreas.svensson@axis.com> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Link: https://lore.kernel.org/r/20230530145223.1223993-1-andreas.svensson@axis.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
If we send two TCA_FLOWER_KEY_ENC_OPTS_GENEVE packets and their total
size is 252 bytes(key->enc_opts.len = 252) then
key->enc_opts.len = opt->length = data_len / 4 = 0 when the third
TCA_FLOWER_KEY_ENC_OPTS_GENEVE packet enters fl_set_geneve_opt. This
bypasses the next bounds check and results in an out-of-bounds.
Fixes: 0a6e77784f49 ("net/sched: allow flower to match tunnel options") Signed-off-by: Hangyu Hua <hbh25y@gmail.com> Reviewed-by: Simon Horman <simon.horman@corigine.com> Reviewed-by: Pieter Jansen van Vuuren <pieter.jansen-van-vuuren@amd.com> Link: https://lore.kernel.org/r/20230531102805.27090-1-hbh25y@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Syzkaller got the following report:
BUG: KASAN: use-after-free in sk_setup_caps+0x621/0x690 net/core/sock.c:2018
Read of size 8 at addr ffff888027f82780 by task syz-executor276/3255
The function sk_setup_caps (called by ip6_sk_dst_store_flow->
ip6_dst_store) referenced already freed memory as this memory was
freed by parallel task in udpv6_sendmsg->ip6_sk_dst_lookup_flow->
sk_dst_check.
The reason for this race condition is: sk_setup_caps() keeps using
the dst after transferring the ownership to the dst cache.
Found by Linux Verification Center (linuxtesting.org) with syzkaller.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Vladislav Efanov <VEfanov@ispras.ru> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The current code for the length calculation wrongly truncates the reported
length of the groups array, causing an under report of the subscribed
groups. To fix this, use 'BITS_TO_BYTES()' which rounds up the
division by 8.
Fixes: b42be38b2778 ("netlink: add API to retrieve all group memberships") Signed-off-by: Pedro Tammela <pctammela@mojatatu.com> Reviewed-by: Simon Horman <simon.horman@corigine.com> Link: https://lore.kernel.org/r/20230529153335.389815-1-pctammela@mojatatu.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
There are users already and will be more of BITS_TO_BYTES() macro. Move
it to bitops.h for wider use.
In the case of ocfs2 the replacement is identical.
As for bnx2x, there are two places where floor version is used. In the
first case to calculate the amount of structures that can fit one memory
page. In this case obviously the ceiling variant is correct and
original code might have a potential bug, if amount of bits % 8 is not
0. In the second case the macro is used to calculate bytes transmitted
in one microsecond. This will work for all speeds which is multiply of
1Gbps without any change, for the rest new code will give ceiling value,
for instance 100Mbps will give 13 bytes, while old code gives 12 bytes
and the arithmetically correct one is 12.5 bytes. Further the value is
used to setup timer threshold which in any case has its own margins due
to certain resolution. I don't see here an issue with slightly shifting
thresholds for low speed connections, the card is supposed to utilize
highest available rate, which is usually 10Gbps.
Link: http://lkml.kernel.org/r/20200108121316.22411-1-andriy.shevchenko@linux.intel.com Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com> Acked-by: Sudarsana Reddy Kalluru <skalluru@marvell.com> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: Gang He <ghe@suse.com> Cc: Jun Piao <piaojun@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Stable-dep-of: f4e4534850a9 ("net/netlink: fix NETLINK_LIST_MEMBERSHIPS length report") Signed-off-by: Sasha Levin <sashal@kernel.org>
When use the following command to test:
1)ip link add bond0 type bond
2)ip link set bond0 up
3)tc qdisc add dev bond0 root handle ffff: mq
4)tc qdisc replace dev bond0 parent ffff:fff1 handle ffff: mq
This is because when mq is added for the first time, qdiscs in mq is set
to NULL in mq_attach(). Therefore, when replacing mq after adding mq, we
need to initialize qdiscs in the mq before continuing to graft. Otherwise,
it will couse NULL pointer dereference issue in mq_attach(). And the same
issue will occur in the attach functions of mqprio, taprio and htb.
ffff:fff1 means that the repalce qdisc is ingress. Ingress does not allow
any qdisc to be attached. Therefore, ffff:fff1 is incorrectly used, and
the command should be dropped.
Currently, after creating an ingress (or clsact) Qdisc and grafting it
under TC_H_INGRESS (TC_H_CLSACT), it is possible to graft it again under
e.g. a TBF Qdisc:
$ ip link add ifb0 type ifb
$ tc qdisc add dev ifb0 handle 1: root tbf rate 20kbit buffer 1600 limit 3000
$ tc qdisc add dev ifb0 clsact
$ tc qdisc link dev ifb0 handle ffff: parent 1:1
$ tc qdisc show dev ifb0
qdisc tbf 1: root refcnt 2 rate 20Kbit burst 1600b lat 560.0ms
qdisc clsact ffff: parent ffff:fff1 refcnt 2
^^^^^^^^
clsact's refcount has increased: it is now grafted under both
TC_H_CLSACT and 1:1.
ingress and clsact Qdiscs should only be used under TC_H_INGRESS
(TC_H_CLSACT). Prohibit regrafting them.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Fixes: 1f211a1b929c ("net, sched: add clsact qdisc") Tested-by: Pedro Tammela <pctammela@mojatatu.com> Acked-by: Jamal Hadi Salim <jhs@mojatatu.com> Reviewed-by: Jamal Hadi Salim <jhs@mojatatu.com> Reviewed-by: Vlad Buslov <vladbu@nvidia.com> Signed-off-by: Peilin Ye <peilin.ye@bytedance.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently it is possible to add e.g. an HTB Qdisc under ffff:fff1
(TC_H_INGRESS, TC_H_CLSACT):
$ ip link add name ifb0 type ifb
$ tc qdisc add dev ifb0 parent ffff:fff1 htb
$ tc qdisc add dev ifb0 clsact
Error: Exclusivity flag on, cannot modify.
$ drgn
...
>>> ifb0 = netdev_get_by_name(prog, "ifb0")
>>> qdisc = ifb0.ingress_queue.qdisc_sleeping
>>> print(qdisc.ops.id.string_().decode())
htb
>>> qdisc.flags.value_() # TCQ_F_INGRESS
2
Only allow ingress and clsact Qdiscs under ffff:fff1. Return -EINVAL
for everything else. Make TCQ_F_INGRESS a static flag of ingress and
clsact Qdiscs.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Fixes: 1f211a1b929c ("net, sched: add clsact qdisc") Tested-by: Pedro Tammela <pctammela@mojatatu.com> Acked-by: Jamal Hadi Salim <jhs@mojatatu.com> Reviewed-by: Jamal Hadi Salim <jhs@mojatatu.com> Reviewed-by: Vlad Buslov <vladbu@nvidia.com> Signed-off-by: Peilin Ye <peilin.ye@bytedance.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
This patch replaces the tp->mss_cache check in getting TCP_MAXSEG
with tp->rx_opt.user_mss check for CLOSE/LISTEN sock. Since
tp->mss_cache is initialized with TCP_MSS_DEFAULT, checking if
it's zero is probably a bug.
With this change, getting TCP_MAXSEG before connecting will return
default MSS normally, and return user_mss if user_mss is set.
Historically connect(AF_UNSPEC) has been abused by syzkaller
and other fuzzers to trigger various bugs.
A recent one triggers a divide-by-zero [1], and Paolo Abeni
was able to diagnose the issue.
tcp_recvmsg_locked() has tests about sk_state being not TCP_LISTEN
and TCP REPAIR mode being not used.
Then later if socket lock is released in sk_wait_data(),
another thread can call connect(AF_UNSPEC), then make this
socket a TCP listener.
When recvmsg() is resumed, it can eventually call tcp_cleanup_rbuf()
and attempt a divide by 0 in tcp_rcv_space_adjust() [1]
This patch adds a new socket field, counting number of threads
blocked in sk_wait_event() and inet_wait_for_connect().
If this counter is not zero, tcp_disconnect() returns an error.
This patch adds code in blocking socket system calls, thus should
not hurt performance of non blocking ones.
Note that we probably could revert commit 499350a5a6e7 ("tcp:
initialize rcv_mss to TCP_MIN_MSS instead of 0") to restore
original tcpi_rcv_mss meaning (was 0 if no payload was ever
received on a socket)
In the event of a change in XGBE mode, the current auto-negotiation
needs to be reset and the AN cycle needs to be re-triggerred. However,
the current code ignores the return value of xgbe_set_mode(), leading to
false information as the link is declared without checking the status
register.
Fix this by propagating the mode switch status information to
xgbe_phy_status().
Fixes: e57f7a3feaef ("amd-xgbe: Prepare for working with more than one type of phy") Co-developed-by: Sudheesh Mavila <sudheesh.mavila@amd.com> Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com> Reviewed-by: Simon Horman <simon.horman@corigine.com> Acked-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com> Signed-off-by: Raju Rangoju <Raju.Rangoju@amd.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>