Patrice Chotard [Tue, 18 Jun 2024 13:29:50 +0000 (15:29 +0200)]
spi: stm32: qspi: Clamp stm32_qspi_get_mode() output to CCR_BUSWIDTH_4
In case usage of OCTAL mode, buswidth parameter can take the value 8.
As return value of stm32_qspi_get_mode() is used to configure fields
of CCR registers that are 2 bits only (fields IMODE, ADMODE, ADSIZE,
DMODE), clamp return value of stm32_qspi_get_mode() to 4.
Xin Long [Sat, 15 Jun 2024 21:47:30 +0000 (17:47 -0400)]
sched: act_ct: add netns into the key of tcf_ct_flow_table
zones_ht is a global hashtable for flow_table with zone as key. However,
it does not consider netns when getting a flow_table from zones_ht in
tcf_ct_init(), and it means an act_ct action in netns A may get a
flow_table that belongs to netns B if it has the same zone value.
In Shuang's test with the TOPO:
tcf2_c <---> tcf2_sw1 <---> tcf2_sw2 <---> tcf2_s
tcf2_sw1 and tcf2_sw2 saw the same flow and used the same flow table,
which caused their ct entries entering unexpected states and the
TCP connection not able to end normally.
This patch fixes the issue simply by adding netns into the key of
tcf_ct_flow_table so that an act_ct action gets a flow_table that
belongs to its own netns in tcf_ct_init().
Note that for easy coding we don't use tcf_ct_flow_table.nf_ft.net,
as the ct_ft is initialized after inserting it to the hashtable in
tcf_ct_flow_table_get() and also it requires to implement several
functions in rhashtable_params including hashfn, obj_hashfn and
obj_cmpfn.
Xin Long [Sat, 15 Jun 2024 18:27:20 +0000 (14:27 -0400)]
tipc: force a dst refcount before doing decryption
As it says in commit 3bc07321ccc2 ("xfrm: Force a dst refcount before
entering the xfrm type handlers"):
"Crypto requests might return asynchronous. In this case we leave the
rcu protected region, so force a refcount on the skb's destination
entry before we enter the xfrm type input/output handlers."
On TIPC decryption path it has the same problem, and skb_dst_force()
should be called before doing decryption to avoid a possible crash.
Shuang reported this issue when this warning is triggered:
David Ruth [Fri, 14 Jun 2024 19:03:26 +0000 (19:03 +0000)]
net/sched: act_api: fix possible infinite loop in tcf_idr_check_alloc()
syzbot found hanging tasks waiting on rtnl_lock [1]
A reproducer is available in the syzbot bug.
When a request to add multiple actions with the same index is sent, the
second request will block forever on the first request. This holds
rtnl_lock, and causes tasks to hang.
Return -EAGAIN to prevent infinite looping, while keeping documented
behavior.
====================
net: lan743x: Fixes for multiple WOL related issues
This patch series implement the following fixes:
1. Disable WOL upon resume in order to restore full data path operation
2. Support WOL at both the PHY and MAC appropriately
3. Remove interrupt mask clearing from config_init
Patch-3 was sent seperately earlier. Review comments in link:
https://lore.kernel.org/lkml/4a565d54-f468-4e32-8a2c-102c1203f72c@lunn.ch/T/
====================
Raju Lakkaraju [Fri, 14 Jun 2024 17:11:57 +0000 (22:41 +0530)]
net: phy: mxl-gpy: Remove interrupt mask clearing from config_init
When the system resumes from sleep, the phy_init_hw() function invokes
config_init(), which clears all interrupt masks and causes wake events to be
lost in subsequent wake sequences. Remove interrupt mask clearing from
config_init() and preserve relevant masks in config_intr().
Raju Lakkaraju [Fri, 14 Jun 2024 17:11:56 +0000 (22:41 +0530)]
net: lan743x: Support WOL at both the PHY and MAC appropriately
Prevent options not supported by the PHY from being requested to it by the MAC
Whenever a WOL option is supported by both, the PHY is given priority
since that usually leads to better power savings.
Fixes: e9e13b6adc33 ("lan743x: fix for potential NULL pointer dereference with bare card") Reviewed-by: Wojciech Drewek <[email protected]> Signed-off-by: Raju Lakkaraju <[email protected]> Signed-off-by: Paolo Abeni <[email protected]>
Raju Lakkaraju [Fri, 14 Jun 2024 17:11:55 +0000 (22:41 +0530)]
net: lan743x: disable WOL upon resume to restore full data path operation
When Wake-on-LAN (WoL) is active and the system is in suspend mode, triggering
a system event can wake the system from sleep, which may block the data path.
To restore normal data path functionality after waking, disable all wake-up
events. Furthermore, clear all Write 1 to Clear (W1C) status bits by writing
1's to them.
Fixes: 4d94282afd95 ("lan743x: Add power management support") Reviewed-by: Wojciech Drewek <[email protected]> Signed-off-by: Raju Lakkaraju <[email protected]> Signed-off-by: Paolo Abeni <[email protected]>
Stefan Wahren [Fri, 14 Jun 2024 14:50:30 +0000 (16:50 +0200)]
qca_spi: Make interrupt remembering atomic
The whole mechanism to remember occurred SPI interrupts is not atomic,
which could lead to unexpected behavior. So fix this by using atomic bit
operations instead.
Using wl183x devices in AP mode with various firmwares is not stable.
The driver currently adds a station to firmware with basic rates when it
is first known to the stack using the CMD_ADD_PEER command. Once the
station has finished authorising, another CMD_ADD_PEER command is issued
to update the firmware with the rates the station can use.
However, after a random amount of time, the firmware ignores the power
management nullfunc frames from the station, and tries to send packets
while the station is asleep, resulting in lots of retries dropping down
in rate due to no response. This restricts the available bandwidth.
With this happening with several stations, the user visible effect is
the latency of interactive connections increases significantly, packets
get dropped, and in general the WiFi connections become unreliable and
unstable.
Eventually, the firmware transmit queue appears to get stuck - with
packets and blocks allocated that never clear.
TI have a couple of patches that address this, but they touch the
mac80211 core to disable NL80211_FEATURE_FULL_AP_CLIENT_STATE for *all*
wireless drivers, which has the effect of not adding the station to the
stack until later when the rates are known. This is a sledge hammer
approach to solving the problem.
The solution implemented here has the same effect, but without
impacting all drivers.
We delay adding the station to firmware until it has been authorised
in the driver, and correspondingly remove the station when unwinding
from authorised state. Adding the station to firmware allocates a hlid,
which will now happen later than the driver expects. Therefore, we need
to track when this happens so that we transmit using the correct hlid.
This patch is an equivalent fix to these two patches in TI's
wilink8-wlan repository:
This is trigger as below:
ns0 ns1
tun_set_iff() //dev is tun0
tun->dev = dev
//ip link set tun0 netns ns1
put_net() //ref is 0
__tun_chr_ioctl() //TUNGETDEVNETNS
net = dev_net(tun->dev);
open_related_ns(&net->ns, get_net_ns); //ns1
get_net_ns()
get_net() //addition on 0
Use maybe_get_net() in get_net_ns in case net's ref is zero to fix this
It is important to have fixed (sub)test names in TAP, because these
names are used to identify them. If they are not fixed, tracking cannot
be done.
Some subtests from the userspace_pm selftest were using random numbers
in their names: the client and server address IDs from $RANDOM, and the
client port number randomly picked by the kernel when creating the
connection. These values have been replaced by 'client' and 'server'
words: that's even more helpful than showing random numbers. Note that
the addresses IDs are incremented and decremented in the test: +1 or -1
are then displayed in these cases.
Not to loose info that can be useful for debugging in case of issues,
these random numbers are now displayed at the beginning of the test.
Eric Dumazet [Fri, 14 Jun 2024 13:06:15 +0000 (13:06 +0000)]
tcp: clear tp->retrans_stamp in tcp_rcv_fastopen_synack()
Some applications were reporting ETIMEDOUT errors on apparently
good looking flows, according to packet dumps.
We were able to root cause the issue to an accidental setting
of tp->retrans_stamp in the following scenario:
- client sends TFO SYN with data.
- server has TFO disabled, ACKs only SYN but not payload.
- client receives SYNACK covering only SYN.
- tcp_ack() eats SYN and sets tp->retrans_stamp to 0.
- tcp_rcv_fastopen_synack() calls tcp_xmit_retransmit_queue()
to retransmit TFO payload w/o SYN, sets tp->retrans_stamp to "now",
but we are not in any loss recovery state.
- TFO payload is ACKed.
- we are not in any loss recovery state, and don't see any dupacks,
so we don't get to any code path that clears tp->retrans_stamp.
- tp->retrans_stamp stays non-zero for the lifetime of the connection.
- after first RTO, tcp_clamp_rto_to_user_timeout() clamps second RTO
to 1 jiffy due to bogus tp->retrans_stamp.
- on clamped RTO with non-zero icsk_retransmits, retransmits_timed_out()
sets start_ts from tp->retrans_stamp from TFO payload retransmit
hours/days ago, and computes bogus long elapsed time for loss recovery,
and suffers ETIMEDOUT early.
This was a wrongheaded workaround for an issue that had already been
fixed much better by commit 4ef9ad19e176 ("mm: huge_memory: don't force
huge page alignment on 32 bit").
Asking users questions at kernel compile time that they can't make sense
of is not a viable strategy. And the fact that even the kernel VM
maintainers apparently didn't catch that this "fix" is not a fix any
more pretty much proves the point that people can't be expected to
understand the implications of the question.
It may well be the case that we could improve things further, and that
__thp_get_unmapped_area() should take the mapping randomization into
account even for 64-bit kernels. Maybe we should not be so eager to use
THP mappings.
But in no case should this be a kernel config option.
Hans de Goede [Wed, 12 Jun 2024 10:42:20 +0000 (12:42 +0200)]
ACPI: scan: Ignore camera graph port nodes on all Dell Tiger, Alder and Raptor Lake models
Dell laptops with IPU6 camera (the Tiger Lake, Alder Lake and Raptor
Lake generations) have broken ACPI MIPI DISCO information (this results
from an OEM attempt to make Linux work by supplying it with custom data
in the ACPI tables which has never been supported in the mainline).
Instead of adding a lot of DMI quirks for this, check for Dell platforms
based on the processor generations in question and drop the ACPI graph
port nodes, likely to be created with the help of invalid data, on all
of them.
Fixes: bd721b934323 ("ACPI: scan: Extract CSI-2 connection graph from _CRS") Signed-off-by: Hans de Goede <[email protected]>
[ rjw: Changelog edits ] Signed-off-by: Rafael J. Wysocki <[email protected]>
Linus Torvalds [Mon, 17 Jun 2024 19:30:07 +0000 (12:30 -0700)]
Merge tag 'mm-hotfixes-stable-2024-06-17-11-43' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Pull misc fixes from Andrew Morton:
"Mainly MM singleton fixes. And a couple of ocfs2 regression fixes"
* tag 'mm-hotfixes-stable-2024-06-17-11-43' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm:
kcov: don't lose track of remote references during softirqs
mm: shmem: fix getting incorrect lruvec when replacing a shmem folio
mm/debug_vm_pgtable: drop RANDOM_ORVALUE trick
mm: fix possible OOB in numa_rebuild_large_mapping()
mm/migrate: fix kernel BUG at mm/compaction.c:2761!
selftests: mm: make map_fixed_noreplace test names stable
mm/memfd: add documentation for MFD_NOEXEC_SEAL MFD_EXEC
mm: mmap: allow for the maximum number of bits for randomizing mmap_base by default
gcov: add support for GCC 14
zap_pid_ns_processes: clear TIF_NOTIFY_SIGNAL along with TIF_SIGPENDING
mm: huge_memory: fix misused mapping_large_folio_support() for anon folios
lib/alloc_tag: fix RCU imbalance in pgalloc_tag_get()
lib/alloc_tag: do not register sysctl interface when CONFIG_SYSCTL=n
MAINTAINERS: remove Lorenzo as vmalloc reviewer
Revert "mm: init_mlocked_on_free_v3"
mm/page_table_check: fix crash on ZONE_DEVICE
gcc: disable '-Warray-bounds' for gcc-9
ocfs2: fix NULL pointer dereference in ocfs2_abort_trigger()
ocfs2: fix NULL pointer dereference in ocfs2_journal_dirty()
Raju Rangoju [Fri, 14 Jun 2024 14:01:49 +0000 (19:31 +0530)]
ACPICA: Revert "ACPICA: avoid Info: mapping multiple BARs. Your kernel is fine."
Undo the modifications made in commit d410ee5109a1 ("ACPICA: avoid
"Info: mapping multiple BARs. Your kernel is fine.""). The initial
purpose of this commit was to stop memory mappings for operation
regions from overlapping page boundaries, as it can trigger warnings
if different page attributes are present.
However, it was found that when this situation arises, mapping
continues until the boundary's end, but there is still an attempt to
read/write the entire length of the map, leading to a NULL pointer
deference. For example, if a four-byte mapping request is made but
only one byte is mapped because it hits the current page boundary's
end, a four-byte read/write attempt is still made, resulting in a NULL
pointer deference.
Instead, map the entire length, as the ACPI specification does not
mandate that it must be within the same page boundary. It is
permissible for it to be mapped across different regions.
- kunit/overflow: Adjust for __counted_by with DEFINE_RAW_FLEX()
- MAINTAINERS: Update entries for Kees Cook
* tag 'hardening-v6.10-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
MAINTAINERS: Update entries for Kees Cook
kunit/overflow: Adjust for __counted_by with DEFINE_RAW_FLEX()
yama: document function parameter
mm/util: Swap kmemdup_array() arguments
Kees Cook [Mon, 17 Jun 2024 18:13:01 +0000 (11:13 -0700)]
MAINTAINERS: Update entries for Kees Cook
Update current email address for Kees Cook in the MAINTAINER file to
match the change from commit 4e173c825b19 ("mailmap: update entry for
Kees Cook").
Linus Torvalds [Mon, 17 Jun 2024 18:05:56 +0000 (11:05 -0700)]
Merge tag 'hyperv-fixes-signed-20240616' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux
Pull Hyper-V fixes from Wei Liu:
- Some cosmetic changes for hv.c and balloon.c (Aditya Nagesh)
- Two documentation updates (Michael Kelley)
- Suppress the invalid warning for packed member alignment (Saurabh
Sengar)
- Two hv_balloon fixes (Michael Kelley)
* tag 'hyperv-fixes-signed-20240616' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux:
Drivers: hv: Cosmetic changes for hv.c and balloon.c
Documentation: hyperv: Improve synic and interrupt handling description
Documentation: hyperv: Update spelling and fix typo
tools: hv: suppress the invalid warning for packed member alignment
hv_balloon: Enable hot-add for memblock sizes > 128 MiB
hv_balloon: Use kernel macros to simplify open coded sequences
====================
bpf: Fix missed var_off related to movsx in verifier
Zac reported a verification issue ([1]) where verification unexpectedly succeeded.
This is due to missing proper var_off setting in verifier related to
movsx insn. I found another similar issue as well. This patch set fixed
both problems and added three inline asm tests to test these fixes.
Yonghong Song [Sat, 15 Jun 2024 17:46:32 +0000 (10:46 -0700)]
bpf: Add missed var_off setting in coerce_subreg_to_size_sx()
In coerce_subreg_to_size_sx(), for the case where upper
sign extension bits are the same for smax32 and smin32
values, we missed to setup properly. This is especially
problematic if both smax32 and smin32's sign extension
bits are 1.
The following is a simple example illustrating the inconsistent
verifier states due to missed var_off:
The var_off=(0x80, 0xf) is not correct, and the correct one should
be var_off=(0xffffff80; 0xf) since from insn 3, we know that at
insn 4, the sign extension bits will be 1. This patch fixed this
issue by setting var_off properly.
Yonghong Song [Sat, 15 Jun 2024 17:46:26 +0000 (10:46 -0700)]
bpf: Add missed var_off setting in set_sext32_default_val()
Zac reported a verification failure and Alexei reproduced the issue
with a simple reproducer ([1]). The verification failure is due to missed
setting for var_off.
Note that after insn 1, the var_off for R7 is (0x0; 0x7f). This is not correct
since upper 24 bits of w7 could be 0 or 1. So correct var_off should be
(0x0; 0xffffffff). Missing var_off setting in set_sext32_default_val() caused later
incorrect analysis in zext_32_to_64(dst_reg) and reg_bounds_sync(dst_reg).
To fix the issue, set var_off correctly in set_sext32_default_val(). The correct
reg state after insn 1 becomes:
1: (bc) w7 = (s8)w3 ;
R3_w=scalar(smin=smin32=0,smax=umax=smax32=umax32=255,var_off=(0x0; 0xff))
R7_w=scalar(smin=0,smax=umax=0xffffffff,smin32=-128,smax32=127,var_off=(0x0; 0xffffffff))
and at insn 2, the verifier correctly determines either branch is possible.
Lorenzo Bianconi [Mon, 17 Jun 2024 14:26:26 +0000 (16:26 +0200)]
NFSD: grab nfsd_mutex in nfsd_nl_rpc_status_get_dumpit()
Grab nfsd_mutex lock in nfsd_nl_rpc_status_get_dumpit routine and remove
nfsd_nl_rpc_status_get_start() and nfsd_nl_rpc_status_get_done(). This
patch fix the syzbot log reported below:
Jeff Layton [Mon, 17 Jun 2024 11:54:08 +0000 (07:54 -0400)]
nfsd: fix oops when reading pool_stats before server is started
Sourbh reported an oops that is triggerable by trying to read the
pool_stats procfile before nfsd had been started. Move the check for a
NULL serv in svc_pool_stats_start above the mutex acquisition, and fix
the stop routine not to unlock the mutex if there is no serv yet.
Gavrilov Ilia [Thu, 13 Jun 2024 08:23:00 +0000 (08:23 +0000)]
netrom: Fix a memory leak in nr_heartbeat_expiry()
syzbot reported a memory leak in nr_create() [0].
Commit 409db27e3a2e ("netrom: Fix use-after-free of a listening socket.")
added sock_hold() to the nr_heartbeat_expiry() function, where
a) a socket has a SOCK_DESTROY flag or
b) a listening socket has a SOCK_DEAD flag.
But in the case "a," when the SOCK_DESTROY flag is set, the file descriptor
has already been closed and the nr_release() function has been called.
So it makes no sense to hold the reference count because no one will
call another nr_destroy_socket() and put it as in the case "b."
Reported-by: [email protected] Closes: https://syzkaller.appspot.com/bug?extid=d327a1f3b12e1e206c16 Fixes: 409db27e3a2e ("netrom: Fix use-after-free of a listening socket.") Signed-off-by: Gavrilov Ilia <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Merge tag 'thermal-v6.10-rc4' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/thermal/linux
Merge thermal driver fixes for 6.10-rc5 from Daniel Lezcano:
"- Remove the filtered mode for mt8188 as it is not supported on this
platform (Julien Panis)
- Fail in case the golden temperature is zero as that means the efuse
data is not correctly set (Julien Panis)"
* tag 'thermal-v6.10-rc4' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/thermal/linux:
thermal/drivers/mediatek/lvts_thermal: Return error in case of invalid efuse data
thermal/drivers/mediatek/lvts_thermal: Remove filtered mode for mt8188
Huang-Huang Bao [Thu, 6 Jun 2024 12:57:55 +0000 (20:57 +0800)]
pinctrl: rockchip: fix pinmux reset in rockchip_pmx_set
rockchip_pmx_set reset all pinmuxs in group to 0 in the case of error,
add missing bank data retrieval in that code to avoid setting mux on
unexpected pins.
Huang-Huang Bao [Thu, 6 Jun 2024 12:57:54 +0000 (20:57 +0800)]
pinctrl: rockchip: use dedicated pinctrl type for RK3328
rk3328_pin_ctrl uses type of RK3288 which has a hack in
rockchip_pinctrl_suspend and rockchip_pinctrl_resume to restore GPIO6-C6
at assume, the hack is not applicable to RK3328 as GPIO6 is not even
exist in it. So use a dedicated pinctrl type to skip this hack.
Huang-Huang Bao [Thu, 6 Jun 2024 12:57:53 +0000 (20:57 +0800)]
pinctrl: rockchip: fix pinmux bits for RK3328 GPIO3-B pins
The pinmux bits for GPIO3-B1 to GPIO3-B6 pins are not explicitly
specified in RK3328 TRM, however we can get hint from pad name and its
correspinding IOMUX setting for pins in interface descriptions. The
correspinding IOMIX settings for these pins can be found in the same
row next to occurrences of following pad names in RK3328 TRM.
Huang-Huang Bao [Thu, 6 Jun 2024 12:57:52 +0000 (20:57 +0800)]
pinctrl: rockchip: fix pinmux bits for RK3328 GPIO2-B pins
The pinmux bits for GPIO2-B0 to GPIO2-B6 actually have 2 bits width,
correct the bank flag for GPIO2-B. The pinmux bits for GPIO2-B7 is
recalculated so it remain unchanged.
The pinmux bits for those pins are not explicitly specified in RK3328
TRM, however we can get hint from pad name and its correspinding IOMUX
setting for pins in interface descriptions. The correspinding IOMIX
settings for GPIO2-B0 to GPIO2-B6 can be found in the same row next to
occurrences of following pad names in RK3328 TRM.
Hagar Hemdan [Tue, 4 Jun 2024 08:58:38 +0000 (08:58 +0000)]
pinctrl: fix deadlock in create_pinctrl() when handling -EPROBE_DEFER
In create_pinctrl(), pinctrl_maps_mutex is acquired before calling
add_setting(). If add_setting() returns -EPROBE_DEFER, create_pinctrl()
calls pinctrl_free(). However, pinctrl_free() attempts to acquire
pinctrl_maps_mutex, which is already held by create_pinctrl(), leading to
a potential deadlock.
This patch resolves the issue by releasing pinctrl_maps_mutex before
calling pinctrl_free(), preventing the deadlock.
This bug was discovered and resolved using Coverity Static Analysis
Security Testing (SAST) by Synopsys, Inc.
Stefan Wahren [Mon, 3 Jun 2024 18:19:37 +0000 (20:19 +0200)]
pinctrl: bcm2835: Fix permissions of persist_gpio_outputs
The commit 8ff05989b44e ("pinctrl: bcm2835: Make pin freeing behavior
configurable") unintentionally made the module parameter
persist_gpio_outputs changeable at runtime. So drop the write permission
in order to make the freeing behavior predictable for user applications.
Johan Hovold [Wed, 29 May 2024 16:29:53 +0000 (18:29 +0200)]
dt-bindings: pinctrl: qcom,pmic-gpio: drop pm8008
The binding for PM8008 is being reworked so that internal details like
interrupts and register offsets are no longer described. This
specifically also involves dropping the gpio child node and its
compatible string which is no longer needed.
Note that there are currently no users of the upstream binding and
driver.
Johan Hovold [Wed, 29 May 2024 16:29:52 +0000 (18:29 +0200)]
pinctrl: qcom: spmi-gpio: drop broken pm8008 support
The SPMI GPIO driver assumes that the parent device is an SPMI device
and accesses random data when backcasting the parent struct device
pointer for non-SPMI devices.
Fortunately this does not seem to cause any issues currently when the
parent device is an I2C client like the PM8008, but this could change if
the structures are reorganised (e.g. using structure randomisation).
Notably the interrupt implementation is also broken for non-SPMI devices.
Also note that the two GPIO pins on PM8008 are used for interrupts and
reset so their practical use should be limited.
And occurs when the the inode precommit handlers is attempt to look
up the inode cluster buffer to attach the inode for writeback.
The trail of logic that I can reconstruct is as follows.
1. the inode is clean when inodegc runs, so it is not
attached to a cluster buffer when precommit runs.
2. #1 implies the inode cluster buffer may be clean and not
pinned by dirty inodes when inodegc runs.
3. #2 implies that the inode cluster buffer can be reclaimed
by memory pressure at any time.
4. The assert failure implies that the cluster buffer was
attached to the transaction, but not marked done. It had
been accessed earlier in the transaction, but not marked
done.
5. #4 implies the cluster buffer has been invalidated (i.e.
marked stale).
6. #5 implies that the inode cluster buffer was instantiated
uninitialised in the transaction in xfs_ifree_cluster(),
which only instantiates the buffers to invalidate them
and never marks them as done.
Given factors 1-3, this issue is highly dependent on timing and
environmental factors. Hence the issue can be very difficult to
reproduce in some situations, but highly reliable in others. Luis
has an environment where it can be reproduced easily by g/531 but,
OTOH, I've reproduced it only once in ~2000 cycles of g/531.
I think the fix is to have xfs_ifree_cluster() set the XBF_DONE flag
on the cluster buffers, even though they may not be initialised. The
reasons why I think this is safe are:
1. A buffer cache lookup hit on a XBF_STALE buffer will
clear the XBF_DONE flag. Hence all future users of the
buffer know they have to re-initialise the contents
before use and mark it done themselves.
2. xfs_trans_binval() sets the XFS_BLI_STALE flag, which
means the buffer remains locked until the journal commit
completes and the buffer is unpinned. Hence once marked
XBF_STALE/XFS_BLI_STALE by xfs_ifree_cluster(), the only
context that can access the freed buffer is the currently
running transaction.
3. #2 implies that future buffer lookups in the currently
running transaction will hit the transaction match code
and not the buffer cache. Hence XBF_STALE and
XFS_BLI_STALE will not be cleared unless the transaction
initialises and logs the buffer with valid contents
again. At which point, the buffer will be marked marked
XBF_DONE again, so having XBF_DONE already set on the
stale buffer is a moot point.
4. #2 also implies that any concurrent access to that
cluster buffer will block waiting on the buffer lock
until the inode cluster has been fully freed and is no
longer an active inode cluster buffer.
5. #4 + #1 means that any future user of the disk range of
that buffer will always see the range of disk blocks
covered by the cluster buffer as not done, and hence must
initialise the contents themselves.
6. Setting XBF_DONE in xfs_ifree_cluster() then means the
unlinked inode precommit code will see a XBF_DONE buffer
from the transaction match as it expects. It can then
attach the stale but newly dirtied inode to the stale
but newly dirtied cluster buffer without unexpected
failures. The stale buffer will then sail through the
journal and do the right thing with the attached stale
inode during unpin.
Hence the fix is just one line of extra code. The explanation of
why we have to set XBF_DONE in xfs_ifree_cluster, OTOH, is long and
complex....
Linus Torvalds [Sun, 16 Jun 2024 18:50:16 +0000 (11:50 -0700)]
Merge tag 'parisc-for-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux
Pull parisc fix from Helge Deller:
"On parisc we have suffered since years from random segfaults which
seem to have been triggered due to cache inconsistencies. Those
segfaults happened more often on machines with PA8800 and PA8900 CPUs,
which have much bigger caches than the earlier machines.
Dave Anglin has worked over the last few weeks to fix this bug. His
patch has been successfully tested by various people on various
machines and with various kernels (6.6, 6.8 and 6.9), and the debian
buildd servers haven't shown a single random segfault with this patch.
Since the cache handling has been reworked, the patch is slightly
bigger than I would like in this stage, but the greatly improved
stability IMHO justifies the inclusion now"
* tag 'parisc-for-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
parisc: Try to fix random segmentation faults in package builds
Linus Torvalds [Sun, 16 Jun 2024 18:37:38 +0000 (11:37 -0700)]
Merge tag 'i2c-for-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux
Pull i2c fixes from Wolfram Sang:
"Two fixes to correctly report i2c functionality, ensuring that
I2C_FUNC_SLAVE is reported when a device operates solely as a slave
interface"
* tag 'i2c-for-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
i2c: designware: Fix the functionality flags of the slave-only interface
i2c: at91: Fix the functionality flags of the slave-only interface
Linus Torvalds [Sun, 16 Jun 2024 18:20:26 +0000 (11:20 -0700)]
Merge tag 'usb-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Pull USB / Thunderbolt fixes from Greg KH:
"Here are some small USB and Thunderbolt driver fixes for 6.10-rc4.
Included in here are:
- thunderbolt debugfs bugfix
- USB typec bugfixes
- kcov usb bugfix
- xhci bugfixes
- usb-storage bugfix
- dt-bindings bugfix
- cdc-wdm log message spam bugfix
All of these, except for the last cdc-wdm log level change, have been
in linux-next for a while with no reported problems. The cdc-wdm
bugfix has been tested by syzbot and proved to fix the reported cpu
lockup issues when the log is constantly spammed by a broken device"
* tag 'usb-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
USB: class: cdc-wdm: Fix CPU lockup caused by excessive log messages
xhci: Handle TD clearing for multiple streams case
xhci: Apply broken streams quirk to Etron EJ188 xHCI host
xhci: Apply reset resume quirk to Etron EJ188 xHCI host
xhci: Set correct transferred length for cancelled bulk transfers
usb-storage: alauda: Check whether the media is initialized
usb: typec: ucsi: Ack also failed Get Error commands
kcov, usb: disable interrupts in kcov_remote_start_usb_softirq
dt-bindings: usb: realtek,rts5411: Add missing "additionalProperties" on child nodes
usb: typec: tcpm: Ignore received Hard Reset in TOGGLING state
usb: typec: tcpm: fix use-after-free case in tcpm_register_source_caps
USB: xen-hcd: Traverse host/ when CONFIG_USB_XEN_HCD is selected
usb: typec: ucsi: glink: increase max ports for x1e80100
Revert "usb: chipidea: move ci_ulpi_init after the phy initialization"
thunderbolt: debugfs: Fix margin debugfs node creation condition
Linus Torvalds [Sun, 16 Jun 2024 18:05:47 +0000 (11:05 -0700)]
Merge tag 'tty-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Pull tty/serial driver fixes from Greg KH:
"Here are some small tty and serial driver fixes that resolve som
reported problems. Included in here are:
- n_tty lookahead buffer bugfix
- WARN_ON() removal where it was not needed
- 8250_dw driver bugfixes
- 8250_pxa bugfix
- sc16is7xx Kconfig fixes for reported build issues
All of these have been in linux-next for over a week with no reported
problems"
* tag 'tty-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
serial: drop debugging WARN_ON_ONCE() from uart_write()
serial: sc16is7xx: re-add Kconfig SPI or I2C dependency
serial: sc16is7xx: rename Kconfig CONFIG_SERIAL_SC16IS7XX_CORE
serial: port: Don't block system suspend even if bytes are left to xmit
serial: 8250_pxa: Configure tx_loadsz to match FIFO IRQ level
serial: 8250_dw: Revert "Move definitions to the shared header"
serial: 8250_dw: Don't use struct dw8250_data outside of 8250_dw
tty: n_tty: Fix buffer offsets when lookahead is used
Linus Torvalds [Sun, 16 Jun 2024 17:57:05 +0000 (10:57 -0700)]
Merge tag 'staging-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging
Pull staging driver fix from Greg KH:
"Here is a single staging driver fix, for the vc04 driver. It resolves
a reported problem that showed up in the merge window set of changes.
It's been in linux-next for over a week with no reported problems"
* tag 'staging-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
staging: vchiq_debugfs: Fix NPD in vchiq_dump_state
Linus Torvalds [Sun, 16 Jun 2024 17:43:04 +0000 (10:43 -0700)]
Merge tag 'driver-core-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core
Pull driver core and sysfs fixes from Greg KH:
"Here are three small changes for 6.10-rc4 that resolve reported
problems, and finally drop an unused api call. These are:
- removal of devm_device_add_groups(), all the callers of this are
finally gone after the 6.10-rc1 merge (changes came in through
different trees), so it's safe to remove.
- much reported sysfs build error fixed up for systems that did not
have sysfs enabled
- driver core sync issue fix for a many reported issue over the years
that no one really paid much attention to, until Dirk finally
tracked down the real issue and made the "obviously correct and
simple" fix for it.
All of these have been in linux-next for over a week with no reported
problems"
* tag 'driver-core-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core:
drivers: core: synchronize really_probe() and dev_uevent()
sysfs: Unbreak the build around sysfs_bin_attr_simple_read()
driver core: remove devm_device_add_groups()
Linus Torvalds [Sun, 16 Jun 2024 17:29:37 +0000 (10:29 -0700)]
Merge tag 'char-misc-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc
Pull char/misc driver fixes from Greg KH:
"Here are a number of small char/misc and iio driver fixes for
6.10-rc4. Included in here are the following:
- iio driver fixes for a bunch of reported problems.
- mei driver fixes for a number of reported issues.
- amiga parport driver build fix.
- .editorconfig fix that was causing lots of unintended whitespace
changes to happen to files when they were being edited. Unless we
want to sweep the whole tree and remove all trailing whitespace at
once, this is needed for the .editorconfig file to be able to be
used at all. This change is required because the original
submitters never touched older files in the tree.
- jfs bugfix for a buffer overflow
The jfs bugfix is in here as I didn't know where else to put it, and
it's been ignored for a while as the filesystem seems to be abandoned
and I'm tired of seeing the same issue reported in multiple places.
All of these have been in linux-next with no reported issues"
* tag 'char-misc-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc: (25 commits)
.editorconfig: remove trim_trailing_whitespace option
jfs: xattr: fix buffer overflow for invalid xattr
misc: microchip: pci1xxxx: Fix a memory leak in the error handling of gp_aux_bus_probe()
misc: microchip: pci1xxxx: fix double free in the error handling of gp_aux_bus_probe()
parport: amiga: Mark driver struct with __refdata to prevent section mismatch
mei: vsc: Fix wrong invocation of ACPI SID method
mei: vsc: Don't stop/restart mei device during system suspend/resume
mei: me: release irq in mei_me_pci_resume error path
mei: demote client disconnect warning on suspend to debug
iio: inkern: fix channel read regression
iio: imu: inv_mpu6050: stabilized timestamping in interrupt
iio: adc: ad7173: Fix sampling frequency setting
iio: adc: ad7173: Clear append status bit
iio: imu: inv_icm42600: delete unneeded update watermark call
iio: imu: inv_icm42600: stabilized timestamp in interrupt
iio: invensense: fix odr switching to same value
iio: adc: ad7173: Remove index from temp channel
iio: adc: ad7173: Add ad7173_device_info names
iio: adc: ad7173: fix buffers enablement for ad7176-2
iio: temperature: mlx90635: Fix ERR_PTR dereference in mlx90635_probe()
...
Linus Torvalds [Sun, 16 Jun 2024 17:20:18 +0000 (10:20 -0700)]
Merge tag 'ata-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux
Pull ata fix from Niklas Cassel:
"Fix a bug where the SCSI Removable Media Bit (RMB) was incorrectly set
for hot-plug capable (and eSATA) ports.
The RMB bit means that the media is removable (e.g. floppy or CD-ROM),
not that the device server is removable. If the RMB bit is set, SCSI
will set the removable media sysfs attribute.
If the removable media sysfs attribute is set on a device,
GNOME/udisks will automatically mount the device on boot.
We only want to set the SCSI RMB bit (and thus the removable media
sysfs attribute) for devices where the ATA removable media device bit
is set"
* tag 'ata-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux:
ata: libata-scsi: Set the RMB bit only for removable media devices
Linus Torvalds [Sun, 16 Jun 2024 17:11:11 +0000 (10:11 -0700)]
Merge tag 'edac_urgent_for_v6.10_rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras
Pull EDAC fixes from Borislav Petkov:
- Fix two issues with MI300 address translation logic
* tag 'edac_urgent_for_v6.10_rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
RAS/AMD/ATL: Use system settings for MI300 DRAM to normalized address translation
RAS/AMD/ATL: Fix MI300 bank hash
Linus Torvalds [Sun, 16 Jun 2024 16:58:02 +0000 (09:58 -0700)]
Merge tag 'firewire-fixes-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394
Pull firewire fixes from Takashi Sakamoto:
- Update tracepoints events introduced in v6.10-rc1 so that it includes
the numeric identifier of host card in which the event happens
- replace wiki URL with the current website URL in Kconfig
* tag 'firewire-fixes-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394:
firewire: core: record card index in bus_reset_handle tracepoints event
firewire: core: record card index in tracepoinrts events derived from bus_reset_arrange_template
firewire: core: record card index in async_phy_inbound tracepoints event
firewire: core: record card index in async_phy_outbound_complete tracepoints event
firewire: core: record card index in async_phy_outbound_initiate tracepoints event
firewire: core: record card index in tracepoinrts events derived from async_inbound_template
firewire: core: record card index in tracepoinrts events derived from async_outbound_initiate_template
firewire: core: record card index in tracepoinrts events derived from async_outbound_complete_template
firewire: fix website URL in Kconfig
Hans de Goede [Fri, 7 Jun 2024 10:18:47 +0000 (12:18 +0200)]
leds: class: Revert: "If no default trigger is given, make hw_control trigger the default trigger"
Commit 66601a29bb23 ("leds: class: If no default trigger is given, make
hw_control trigger the default trigger") causes ledtrig-netdev to get
set as default trigger on various network LEDs.
This causes users to hit a pre-existing AB-BA deadlock issue in
ledtrig-netdev between the LED-trigger locks and the rtnl mutex,
resulting in hung tasks in kernels >= 6.9.
Solving the deadlock is non trivial, so for now revert the change to
set the hw_control trigger as default trigger, so that ledtrig-netdev
no longer gets activated automatically for various network LEDs.
The netdev trigger is not needed because the network LEDs are usually under
hw-control and the netdev trigger tries to leave things that way so setting
it as the active trigger for the LED class device is a no-op.
Gergely Meszaros [Sun, 16 Jun 2024 08:52:33 +0000 (10:52 +0200)]
ALSA: hda/realtek: Add quirk for Lenovo Yoga Pro 7 14ARP8
Similarly to other Lenovo laptops these also have a dual speaker
setup with a shared amplifier.
The model also seems to have a conflicting PCI SSID with the codec SSID for
the Legion Y9000X 2022 IAH7. Only tested on the Yoga Pro 7, as I don't have
access to the other laptop.
Wolfram Sang [Sun, 16 Jun 2024 10:48:30 +0000 (12:48 +0200)]
Merge tag 'i2c-host-fixes-6.10-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/andi.shyti/linux into i2c/for-current
Two fixes from Jean aim to correctly report i2c functionality,
specifically ensuring that I2C_FUNC_SLAVE is reported when a
device operates solely as a slave interface.
memblock: use numa_valid_node() helper to check for invalid node ID
Introduce numa_valid_node(nid) that verifies that nid is a valid node ID
and use that instead of comparing nid parameter with either NUMA_NO_NODE
or MAX_NUMNODES.
This makes the checks for valid node IDs consistent and more robust and
allows to get rid of multiple WARNings.
Fabio Estevam [Wed, 29 May 2024 03:48:54 +0000 (00:48 -0300)]
arm64: dts: imx93-11x11-evk: Remove the 'no-sdio' property
The usdhc2 port is connected to the microSD slot. The presence of the
'no-sdio' property prevents Wifi SDIO cards, such as CMP9010-X-EVB [1]
to be detected.
Remove the 'no-sdio' property so that SDIO cards could also work.
Michael Ellerman [Fri, 14 Jun 2024 12:29:10 +0000 (22:29 +1000)]
KVM: PPC: Book3S HV: Prevent UAF in kvm_spapr_tce_attach_iommu_group()
Al reported a possible use-after-free (UAF) in kvm_spapr_tce_attach_iommu_group().
It looks up `stt` from tablefd, but then continues to use it after doing
fdput() on the returned fd. After the fdput() the tablefd is free to be
closed by another thread. The close calls kvm_spapr_tce_release() and
then release_spapr_tce_table() (via call_rcu()) which frees `stt`.
Although there are calls to rcu_read_lock() in
kvm_spapr_tce_attach_iommu_group() they are not sufficient to prevent
the UAF, because `stt` is used outside the locked regions.
With an artifcial delay after the fdput() and a userspace program which
triggers the race, KASAN detects the UAF:
Fix it by delaying the fdput() until `stt` is no longer in use, which
is effectively the entire function. To keep the patch minimal add a call
to fdput() at each of the existing return paths. Future work can convert
the function to goto or __cleanup style cleanup.
With the fix in place the test case no longer triggers the UAF.
Linus Torvalds [Sat, 15 Jun 2024 19:03:32 +0000 (12:03 -0700)]
Merge tag 'xfs-6.10-fixes-3' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Pull xfs fix from Chandan Babu:
"Ensure xfs incore superblock's allocated inode counter, free inode
counter, and free data block counter are all zero or positive when
they are copied over from xfs_mount->m_[icount,ifree,fdblocks]
respectively"
* tag 'xfs-6.10-fixes-3' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
xfs: make sure sb_fdblocks is non-negative
Linus Torvalds [Sat, 15 Jun 2024 19:00:25 +0000 (12:00 -0700)]
Merge tag '6.10-rc3-smb3-server-fixes' of git://git.samba.org/ksmbd
Pull smb server fixes from Steve French:
"Two small smb3 server fixes:
- set xatttr fix
- pathname parsing check fix"
* tag '6.10-rc3-smb3-server-fixes' of git://git.samba.org/ksmbd:
ksmbd: fix missing use of get_write in in smb2_set_ea()
ksmbd: move leading slash check to smb2_get_name()
Linus Torvalds [Sat, 15 Jun 2024 18:03:05 +0000 (11:03 -0700)]
Merge tag 'x86-urgent-2024-06-15' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Ingo Molnar:
- Fix the 8 bytes get_user() logic on x86-32
- Fix build bug that creates weird & mistaken target directory under
arch/x86/
* tag 'x86-urgent-2024-06-15' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/boot: Don't add the EFI stub to targets, again
x86/uaccess: Fix missed zeroing of ia32 u64 get_user() range checking
Linus Torvalds [Sat, 15 Jun 2024 17:54:24 +0000 (10:54 -0700)]
Merge tag 'timers-urgent-2024-06-15' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull timer fix from Ingo Molnar:
"Fix boot-time warning in tick_setup_device()"
* tag 'timers-urgent-2024-06-15' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
tick/nohz_full: Don't abuse smp_call_function_single() in tick_setup_device()
Aleksandr Nogikh [Tue, 11 Jun 2024 13:32:29 +0000 (15:32 +0200)]
kcov: don't lose track of remote references during softirqs
In kcov_remote_start()/kcov_remote_stop(), we swap the previous KCOV
metadata of the current task into a per-CPU variable. However, the
kcov_mode_enabled(mode) check is not sufficient in the case of remote KCOV
coverage: current->kcov_mode always remains KCOV_MODE_DISABLED for remote
KCOV objects.
If the original task that has invoked the KCOV_REMOTE_ENABLE ioctl happens
to get interrupted and kcov_remote_start() is called, it ultimately leads
to kcov_remote_stop() NOT restoring the original KCOV reference. So when
the task exits, all registered remote KCOV handles remain active forever.
The most uncomfortable effect (at least for syzkaller) is that the bug
prevents the reuse of the same /sys/kernel/debug/kcov descriptor. If
we obtain it in the parent process and then e.g. drop some
capabilities and continuously fork to execute individual programs, at
some point current->kcov of the forked process is lost,
kcov_task_exit() takes no action, and all KCOV_REMOTE_ENABLE ioctls
calls from subsequent forks fail.
And, yes, the efficiency is also affected if we keep on losing remote
kcov objects.
a) kcov_remote_map keeps on growing forever.
b) (If I'm not mistaken), we're also not freeing the memory referenced
by kcov->area.
Fix it by introducing a special kcov_mode that is assigned to the task
that owns a KCOV remote object. It makes kcov_mode_enabled() return true
and yet does not trigger coverage collection in __sanitizer_cov_trace_pc()
and write_comp_data().
Baolin Wang [Thu, 13 Jun 2024 08:21:19 +0000 (16:21 +0800)]
mm: shmem: fix getting incorrect lruvec when replacing a shmem folio
When testing shmem swapin, I encountered the warning below on my machine.
The reason is that replacing an old shmem folio with a new one causes
mem_cgroup_migrate() to clear the old folio's memcg data. As a result,
the old folio cannot get the correct memcg's lruvec needed to remove
itself from the LRU list when it is being freed. This could lead to
possible serious problems, such as LRU list crashes due to holding the
wrong LRU lock, and incorrect LRU statistics.
To fix this issue, we can fallback to use the mem_cgroup_replace_folio()
to replace the old shmem folio.
Peter Xu [Thu, 23 May 2024 13:21:39 +0000 (09:21 -0400)]
mm/debug_vm_pgtable: drop RANDOM_ORVALUE trick
Macro RANDOM_ORVALUE was used to make sure the pgtable entry will be
populated with !none data in clear tests.
The RANDOM_ORVALUE tried to cover mostly all the bits in a pgtable entry,
even if there's no discussion on whether all the bits will be vaild. Both
S390 and PPC64 have their own masks to avoid touching some bits. Now it's
the turn for x86_64.
The issue is there's a recent report from Mikhail Gavrilov showing that
this can cause a warning with the newly added pte set check in commit 8430557fc5 on writable v.s. userfaultfd-wp bit, even though the check
itself was valid, the random pte is not. We can choose to mask more bits
out.
However the need to have such random bits setup is questionable, as now
it's already guaranteed to be true on below:
- For pte level, the pgtable entry will be installed with value from
pfn_pte(), where pfn points to a valid page. Hence the pte will be
!none already if populated with pfn_pte().
- For upper-than-pte level, the pgtable entry should contain a directory
entry always, which is also !none.
All the cases look like good enough to test a pxx_clear() helper. Instead
of extending the bitmask, drop the "set random bits" trick completely. Add
some warning guards to make sure the entries will be !none before clear().
Kefeng Wang [Wed, 12 Jun 2024 12:28:22 +0000 (20:28 +0800)]
mm: fix possible OOB in numa_rebuild_large_mapping()
The large folio is mapped with folio size(not greater PMD_SIZE) aligned
virtual address during the pagefault, ie, 'addr = ALIGN_DOWN(vmf->address,
nr_pages * PAGE_SIZE)' in do_anonymous_page(). But after the mremap(),
the virtual address only requires PAGE_SIZE alignment. Also pte is moved
to new in move_page_tables(), then traversal of the new pte in the
numa_rebuild_large_mapping() could hit the following issue,
Hugh Dickins [Wed, 12 Jun 2024 05:06:20 +0000 (22:06 -0700)]
mm/migrate: fix kernel BUG at mm/compaction.c:2761!
I hit the VM_BUG_ON(!list_empty(&cc->migratepages)) in compact_zone(); and
if DEBUG_VM were off, then pages would be lost on a local list.
Our convention is that if migrate_pages() reports complete success (0),
then the migratepages list will be empty; but if it reports an error or
some pages remaining, then its caller must putback_movable_pages().
There's a new case in which migrate_pages() has been reporting complete
success, but returning with pages left on the migratepages list: when
migrate_pages_batch() successfully split a folio on the deferred list, but
then the "Failure isn't counted" call does not dispose of them all.
Since that block is expecting the large folio to have been counted as 1
failure already, and since the return code is later adjusted to success
whenever the returned list is found empty, the simple way to fix this
safely is to count splitting the deferred folio as "a failure".
Mark Brown [Wed, 5 Jun 2024 22:36:12 +0000 (23:36 +0100)]
selftests: mm: make map_fixed_noreplace test names stable
KTAP parsers interpret the output of ksft_test_result_*() as being the
name of the test. The map_fixed_noreplace test uses a dynamically
allocated base address for the mmap()s that it tests and currently
includes this in the test names that it logs so the test names that are
logged are not stable between runs. It also uses multiples of PAGE_SIZE
which mean that runs for kernels with different PAGE_SIZE configurations
can't be directly compared. Both these factors cause issues for CI
systems when interpreting and displaying results.
Fix this by replacing the current test names with fixed strings describing
the intent of the mappings that are logged, the existing messages with the
actual addresses and sizes are retained as diagnostic prints to aid in
debugging.
Jeff Xu [Fri, 7 Jun 2024 20:35:41 +0000 (20:35 +0000)]
mm/memfd: add documentation for MFD_NOEXEC_SEAL MFD_EXEC
When MFD_NOEXEC_SEAL was introduced, there was one big mistake: it didn't
have proper documentation. This led to a lot of confusion, especially
about whether or not memfd created with the MFD_NOEXEC_SEAL flag is
sealable. Before MFD_NOEXEC_SEAL, memfd had to explicitly set
MFD_ALLOW_SEALING to be sealable, so it's a fair question.
As one might have noticed, unlike other flags in memfd_create,
MFD_NOEXEC_SEAL is actually a combination of multiple flags. The idea is
to make it easier to use memfd in the most common way, which is NOEXEC +
F_SEAL_EXEC + MFD_ALLOW_SEALING. This works with sysctl vm.noexec to help
existing applications move to a more secure way of using memfd.
Proposals have been made to put MFD_NOEXEC_SEAL non-sealable, unless
MFD_ALLOW_SEALING is set, to be consistent with other flags [1], Those
are based on the viewpoint that each flag is an atomic unit, which is a
reasonable assumption. However, MFD_NOEXEC_SEAL was designed with the
intent of promoting the most secure method of using memfd, therefore a
combination of multiple functionalities into one bit.
Furthermore, the MFD_NOEXEC_SEAL has been added for more than one year,
and multiple applications and distributions have backported and utilized
it. Altering ABI now presents a degree of risk and may lead to
disruption.
MFD_NOEXEC_SEAL is a new flag, and applications must change their code to
use it. There is no backward compatibility problem.
When sysctl vm.noexec == 1 or 2, applications that don't set
MFD_NOEXEC_SEAL or MFD_EXEC will get MFD_NOEXEC_SEAL memfd. And
old-application might break, that is by-design, in such a system vm.noexec
= 0 shall be used. Also no backward compatibility problem.
I propose to include this documentation patch to assist in clarifying the
semantics of MFD_NOEXEC_SEAL, thereby preventing any potential future
confusion.
Finally, I would like to express my gratitude to David Rheinsberg and
Barnabás Pőcze for initiating the discussion on the topic of sealability.
Rafael Aquini [Thu, 6 Jun 2024 18:06:22 +0000 (14:06 -0400)]
mm: mmap: allow for the maximum number of bits for randomizing mmap_base by default
An ASLR regression was noticed [1] and tracked down to file-mapped areas
being backed by THP in recent kernels. The 21-bit alignment constraint
for such mappings reduces the entropy for randomizing the placement of
64-bit library mappings and breaks ASLR completely for 32-bit libraries.
The reported issue is easily addressed by increasing vm.mmap_rnd_bits and
vm.mmap_rnd_compat_bits. This patch just provides a simple way to set
ARCH_MMAP_RND_BITS and ARCH_MMAP_RND_COMPAT_BITS to their maximum values
allowed by the architecture at build time.
Using gcov on kernels compiled with GCC 14 results in truncated 16-byte
long .gcda files with no usable data. To fix this, update GCOV_COUNTERS
to match the value defined by GCC 14.
Oleg Nesterov [Sat, 8 Jun 2024 12:06:16 +0000 (14:06 +0200)]
zap_pid_ns_processes: clear TIF_NOTIFY_SIGNAL along with TIF_SIGPENDING
kernel_wait4() doesn't sleep and returns -EINTR if there is no
eligible child and signal_pending() is true.
That is why zap_pid_ns_processes() clears TIF_SIGPENDING but this is not
enough, it should also clear TIF_NOTIFY_SIGNAL to make signal_pending()
return false and avoid a busy-wait loop.
Ran Xiaokai [Fri, 7 Jun 2024 09:40:48 +0000 (17:40 +0800)]
mm: huge_memory: fix misused mapping_large_folio_support() for anon folios
When I did a large folios split test, a WARNING "[ 5059.122759][ T166]
Cannot split file folio to non-0 order" was triggered. But the test cases
are only for anonmous folios. while mapping_large_folio_support() is only
reasonable for page cache folios.
In split_huge_page_to_list_to_order(), the folio passed to
mapping_large_folio_support() maybe anonmous folio. The folio_test_anon()
check is missing. So the split of the anonmous THP is failed. This is
also the same for shmem_mapping(). We'd better add a check for both. But
the shmem_mapping() in __split_huge_page() is not involved, as for
anonmous folios, the end parameter is set to -1, so (head[i].index >= end)
is always false. shmem_mapping() is not called.
Also add a VM_WARN_ON_ONCE() in mapping_large_folio_support() for anon
mapping, So we can detect the wrong use more easily.
THP folios maybe exist in the pagecache even the file system doesn't
support large folio, it is because when CONFIG_TRANSPARENT_HUGEPAGE is
enabled, khugepaged will try to collapse read-only file-backed pages to
THP. But the mapping does not actually support multi order large folios
properly.
Using /sys/kernel/debug/split_huge_pages to verify this, with this patch,
large anon THP is successfully split and the warning is ceased.
lib/alloc_tag: fix RCU imbalance in pgalloc_tag_get()
put_page_tag_ref() should be called only when get_page_tag_ref() returns a
valid reference because only in that case get_page_tag_ref() enters RCU
read section while put_page_tag_ref() will call rcu_read_unlock() even if
the provided reference is NULL. Fix pgalloc_tag_get() which does not
follow this rule causing RCU imbalance. Add a warning in
put_page_tag_ref() to catch any future mistakes.
lib/alloc_tag: do not register sysctl interface when CONFIG_SYSCTL=n
Memory allocation profiling is trying to register sysctl interface even
when CONFIG_SYSCTL=n, resulting in proc_do_static_key() being undefined.
Prevent that by skipping sysctl registration for such configurations.
Lorenzo Stoakes [Sun, 2 Jun 2024 20:55:10 +0000 (21:55 +0100)]
MAINTAINERS: remove Lorenzo as vmalloc reviewer
I haven't had the bandwidth to review vmalloc patches recently and I
suspect I won't be able to do so consistently moving forwards, so I think
it's best if I remove myself as reviewer for the time being.
There was insufficient review and no agreement that this is the right
approach.
There are serious flaws with the implementation that make processes using
mlock() not even work with simple fork() [1] and we get reliable crashes
when rebooting.
Further, simply because we might be unmapping a single PTE of a large
mlocked folio, we shouldn't zero out the whole folio.
... especially because the code can also *corrupt* urelated memory because
kernel_init_pages(page, folio_nr_pages(folio));
Could end up writing outside of the actual folio if we work with a tail
page.
Let's revert it. Once there is agreement that this is the right approach,
the issues were fixed and there was reasonable review and proper testing,
we can consider it again.
Peter Xu [Wed, 5 Jun 2024 21:21:46 +0000 (17:21 -0400)]
mm/page_table_check: fix crash on ZONE_DEVICE
Not all pages may apply to pgtable check. One example is ZONE_DEVICE
pages: they map PFNs directly, and they don't allocate page_ext at all
even if there's struct page around. One may reference
devm_memremap_pages().
When both ZONE_DEVICE and page-table-check enabled, then try to map some
dax memories, one can trigger kernel bug constantly now when the kernel
was trying to inject some pfn maps on the dax device:
kernel BUG at mm/page_table_check.c:55!
While it's pretty legal to use set_pxx_at() for ZONE_DEVICE pages for page
fault resolutions, skip all the checks if page_ext doesn't even exist in
pgtable checker, which applies to ZONE_DEVICE but maybe more.
Yury Norov [Wed, 22 May 2024 22:58:30 +0000 (15:58 -0700)]
gcc: disable '-Warray-bounds' for gcc-9
'-Warray-bounds' is already disabled for gcc-10+. Now that we've merged
bitmap_{read,write), I see the following error when building the kernel
with gcc-9.4 (Ubuntu 20.04.4 LTS) for x86_64 allmodconfig:
drivers/pinctrl/pinctrl-cy8c95x0.c: In function `cy8c95x0_read_regs_mask.isra.0':
include/linux/bitmap.h:756:18: error: array subscript [1, 288230376151711744] is outside array bounds of `long unsigned int[1]' [-Werror=array-bounds]
756 | value_high = map[index + 1] & BITMAP_LAST_WORD_MASK(start + nbits);
| ~~~^~~~~~~~~~~
The immediate reason is that the commit b44759705f7d ("bitmap: make
bitmap_{get,set}_value8() use bitmap_{read,write}()") switched the
bitmap_get_value8() to an alias of bitmap_read(); the same for 'set'.
Now; the code that triggers Warray-bounds, calls the function like this:
In case of bitmap_get_value8(), it's impossible to violate the boundary
because 'space >= nbits' is never the true for byte-aligned 8-bit access.
So, this is clearly a false-positive.
The same type of false-positives break my allmodconfig build in many
places. gcc-8, is clear, however.
Joseph Qi [Thu, 30 May 2024 11:06:30 +0000 (19:06 +0800)]
ocfs2: fix NULL pointer dereference in ocfs2_abort_trigger()
bdev->bd_super has been removed and commit 8887b94d9322 change the usage
from bdev->bd_super to b_assoc_map->host->i_sb. Since ocfs2 hasn't set
bh->b_assoc_map, it will trigger NULL pointer dereference when calling
into ocfs2_abort_trigger().
Actually this was pointed out in history, see commit 74e364ad1b13. But
I've made a mistake when reviewing commit 8887b94d9322 and then
re-introduce this regression.
Since we cannot revive bdev in buffer head, so fix this issue by
initializing all types of ocfs2 triggers when fill super, and then get the
specific ocfs2 trigger from ocfs2_caching_info when access journal.
Joseph Qi [Thu, 30 May 2024 11:06:29 +0000 (19:06 +0800)]
ocfs2: fix NULL pointer dereference in ocfs2_journal_dirty()
bdev->bd_super has been removed and commit 8887b94d9322 change the usage
from bdev->bd_super to b_assoc_map->host->i_sb. This introduces the
following NULL pointer dereference in ocfs2_journal_dirty() since
b_assoc_map is still not initialized. This can be easily reproduced by
running xfstests generic/186, which simulate no more credits.
Fix it by only aborting transaction and journal in ocfs2_journal_dirty()
now, and leave ocfs2_abort() later when detecting an aborted handle,
e.g. start next transaction. Also log the handle details in this case.
The memory object in this case is for efi_rt_stack_top and is allocated
in an initcall. So this is certainly a false positive. Mark the object
as not a leak to quash it.
Ard Biesheuvel [Mon, 10 Jun 2024 14:02:13 +0000 (16:02 +0200)]
efi/x86: Free EFI memory map only when installing a new one.
The logic in __efi_memmap_init() is shared between two different
execution flows:
- mapping the EFI memory map early or late into the kernel VA space, so
that its entries can be accessed;
- the x86 specific cloning of the EFI memory map in order to insert new
entries that are created as a result of making a memory reservation
via a call to efi_mem_reserve().
In the former case, the underlying memory containing the kernel's view
of the EFI memory map (which may be heavily modified by the kernel
itself on x86) is not modified at all, and the only thing that changes
is the virtual mapping of this memory, which is different between early
and late boot.
In the latter case, an entirely new allocation is created that carries a
new, updated version of the kernel's view of the EFI memory map. When
installing this new version, the old version will no longer be
referenced, and if the memory was allocated by the kernel, it will leak
unless it gets freed.
The logic that implements this freeing currently lives on the code path
that is shared between these two use cases, but it should only apply to
the latter. So move it to the correct spot.
While at it, drop the dummy definition for non-x86 architectures, as
that is no longer needed.
Ard Biesheuvel [Mon, 10 Jun 2024 12:12:45 +0000 (14:12 +0200)]
efi/arm: Disable LPAE PAN when calling EFI runtime services
EFI runtime services are remapped into the lower 1 GiB of virtual
address space at boot, so they are guaranteed to be able to co-exist
with the kernel virtual mappings without the need to allocate space for
them in the kernel's vmalloc region, which is rather small.
This means those mappings are covered by TTBR0 when LPAE PAN is enabled,
and so 'user' access must be enabled while such calls are in progress.
Liu Ying [Tue, 14 May 2024 03:07:18 +0000 (11:07 +0800)]
arm: dts: imx53-qsb-hdmi: Disable panel instead of deleting node
We cannot use /delete-node/ directive to delete a node in a DT
overlay. The node won't be deleted effectively. Instead, set
the node's status property to "disabled" to achieve something
similar.
Fixes: eeb403df953f ("ARM: dts: imx53-qsb: add support for the HDMI expander") Signed-off-by: Liu Ying <[email protected]> Reviewed-by: Dmitry Baryshkov <[email protected]> Signed-off-by: Shawn Guo <[email protected]>
Marek Vasut [Sat, 15 Jun 2024 08:00:43 +0000 (16:00 +0800)]
arm64: dts: imx8mp: Fix TC9595 input clock on DH i.MX8M Plus DHCOM SoM
The IMX8MP_CLK_CLKOUT2 supplies the TC9595 bridge with 13 MHz reference
clock. The IMX8MP_CLK_CLKOUT2 is supplied from IMX8MP_AUDIO_PLL2_OUT.
The IMX8MP_CLK_CLKOUT2 operates only as a power-of-two divider, and the
current 156 MHz is not power-of-two divisible to achieve 13 MHz.
To achieve 13 MHz output from IMX8MP_CLK_CLKOUT2, set IMX8MP_AUDIO_PLL2_OUT
to 208 MHz, because 208 MHz / 16 = 13 MHz.
Fixes: 20d0b83e712b ("arm64: dts: imx8mp: Add TC9595 bridge on DH electronics i.MX8M Plus DHCOM") Signed-off-by: Marek Vasut <[email protected]> Signed-off-by: Shawn Guo <[email protected]>
Takashi Sakamoto [Thu, 13 Jun 2024 13:14:40 +0000 (22:14 +0900)]
firewire: core: record card index in bus_reset_handle tracepoints event
The bus reset event occurs in the bus managed by one of 1394 OHCI
controller in Linux system, however the existing tracepoints events has
the lack of data about it to distinguish the issued hardware from the
others.
This commit adds card_index member into event structure to store the index
of host controller in use, and prints it.
Takashi Sakamoto [Thu, 13 Jun 2024 13:14:39 +0000 (22:14 +0900)]
firewire: core: record card index in tracepoinrts events derived from bus_reset_arrange_template
The asynchronous transmission of phy packet is initiated on one of 1394
OHCI controller, however the existing tracepoints events has the lack of
data about it.
This commit adds card_index member into event structure to store the index
of host controller in use, and prints it.
projects / linux.git / log