Arseniy Krasnov says:
====================
fix header length on skb merging
this patchset fixes appending newly arrived skbuff to the last skbuff of
the socket's queue during rx path. Problem fires when we are trying to
append data to skbuff which was already processed in dequeue callback
at least once. Dequeue callback calls function 'skb_pull()' which changes
'skb->len'. In current implementation 'skb->len' is used to update length
in header of last skbuff after new data was copied to it. This is bug,
because value in header is used to calculate 'rx_bytes'/'fwd_cnt' and
thus must be constant during skbuff lifetime. Here is example, we have
two skbuffs: skb0 with length 10 and skb1 with length 4.
1) skb0 arrives, hdr->len == skb->len == 10, rx_bytes == 10
2) Read 3 bytes from skb0, skb->len == 7, hdr->len == 10, rx_bytes == 10
3) skb1 arrives, hdr->len == skb->len == 4, rx_bytes == 14
4) Append skb1 to skb0, skb0 now has skb->len == 11, hdr->len == 11.
But value of 11 in header is invalid.
5) Read whole skb0, update rx_bytes by 11 from skb0's header.
6) At this moment rx_bytes == 3, but socket's queue is empty.
This bug starts to fire since:
commit
0777061657 ("virtio/vsock: don't use skbuff state to account credit")
In fact, it presents before, but didn't triggered due to a little bit
buggy implementation of credit calculation logic. So i'll use Fixes tag
for it.
I really forgot about this branch in rx path when implemented patch
0777061657.
This patchset contains 3 patches:
1) Fix itself.
2) Patch with WARN_ONCE() to catch such problems in future.
3) Patch with test which triggers skb appending logic. It looks like
simple test with several 'send()' and 'recv()', but it checks, that
skbuff appending works ok.
====================
Link: https://lore.kernel.org/r/0683cc6e-5130-484c-1105-ef2eb792d355@sberdevices.ru
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
This adds test which checks case when data of newly received skbuff is
appended to the last skbuff in the socket's queue. It looks like simple
test with 'send()' and 'recv()', but internally it triggers logic which
appends one received skbuff to another. Test checks that this feature
works correctly.
This test is actual only for virtio transport.
Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
This adds WARN_ONCE() and return from stream dequeue callback when
socket's queue is empty, but 'rx_bytes' still non-zero. This allows
the detection of potential bugs due to packet merging (see previous
patch).
Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
This fixes appending newly arrived skbuff to the last skbuff of the
socket's queue. Problem fires when we are trying to append data to skbuff
which was already processed in dequeue callback at least once. Dequeue
callback calls function 'skb_pull()' which changes 'skb->len'. In current
implementation 'skb->len' is used to update length in header of the last
skbuff after new data was copied to it. This is bug, because value in
header is used to calculate 'rx_bytes'/'fwd_cnt' and thus must be not
be changed during skbuff's lifetime.
Bug starts to fire since:
commit 0777061657
("virtio/vsock: don't use skbuff state to account credit")
It presents before, but didn't triggered due to a little bit buggy
implementation of credit calculation logic. So use Fixes tag for it.
Fixes: 0777061657 ("virtio/vsock: don't use skbuff state to account credit")
Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Currently, with VHE, KVM enables the EL0 event counting for the
guest on vcpu_load() or KVM enables it as a part of the PMU
register emulation process, when needed. However, in the migration
case (with VHE), the same handling is lacking, as vPMU register
values that were restored by userspace haven't been propagated yet
(the PMU events haven't been created) at the vcpu load-time on the
first KVM_RUN (kvm_vcpu_pmu_restore_guest() called from vcpu_load()
on the first KVM_RUN won't do anything as events_{guest,host} of
kvm_pmu_events are still zero).
So, with VHE, enable the guest's EL0 event counting on the first
KVM_RUN (after the migration) when needed. More specifically,
have kvm_pmu_handle_pmcr() call kvm_vcpu_pmu_restore_guest()
so that kvm_pmu_handle_pmcr() on the first KVM_RUN can take
care of it.
Fixes: d0c94c4979 ("KVM: arm64: Restore PMU configuration on first run")
Cc: stable@vger.kernel.org
Reviewed-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Reiji Watanabe <reijiw@google.com>
Link: https://lore.kernel.org/r/20230329023944.2488484-1-reijiw@google.com
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
Saeed Mahameed says:
====================
mlx5-updates-2023-03-28
Dragos Tatulea says:
====================
net/mlx5e: RX, Drop page_cache and fully use page_pool
For page allocation on the rx path, the mlx5e driver has been using an
internal page cache in tandem with the page pool. The internal page
cache uses a queue for page recycling which has the issue of head of
queue blocking.
This patch series drops the internal page_cache altogether and uses the
page_pool to implement everything that was done by the page_cache
before:
* Let the page_pool handle dma mapping and unmapping.
* Use fragmented pages with fragment counter instead of tracking via
page ref.
* Enable skb recycling.
The patch series has the following effects on the rx path:
* Improved performance for the cases when there was low page recycling
due to head of queue blocking in the internal page_cache. The test
for this was running a single iperf TCP stream to a rx queue
which is bound on the same cpu as the application.
|-------------+--------+--------+------+---------|
| rq type | before | after | unit | diff |
|-------------+--------+--------+------+---------|
| striding rq | 30.1 | 31.4 | Gbps | 4.14 % |
| legacy rq | 30.2 | 33.0 | Gbps | 8.48 % |
|-------------+--------+--------+------+---------|
* Small XDP performance degradation. The test was is XDP drop
program running on a single rx queue with small packets incoming
it looks like this:
|-------------+----------+----------+------+---------|
| rq type | before | after | unit | diff |
|-------------+----------+----------+------+---------|
| striding rq | 19725449 | 18544617 | pps | -6.37 % |
| legacy rq | 19879931 | 18631841 | pps | -6.70 % |
|-------------+----------+----------+------+---------|
This will be handled in a different patch series by adding support for
multi-packet per page.
* For other cases the performance is roughly the same.
The above numbers were obtained on the following system:
24 core Intel(R) Xeon(R) Platinum 8380 CPU @ 2.30GHz
32 GB RAM
ConnectX-7 single port
The breakdown on the patch series is the following:
* Preparations for introducing the mlx5e_frag_page struct.
* Delete the mlx5e_page_cache struct.
* Enable dma mapping from page_pool.
* Enable skb recycling and fragment counting.
* Do deferred release of pages (just before alloc) to ensure better
page_pool cache utilization.
====================
* tag 'mlx5-updates-2023-03-28' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux:
net/mlx5e: RX, Remove unnecessary recycle parameter and page_cache stats
net/mlx5e: RX, Break the wqe bulk refill in smaller chunks
net/mlx5e: RX, Increase WQE bulk size for legacy rq
net/mlx5e: RX, Split off release path for xsk buffers for legacy rq
net/mlx5e: RX, Defer page release in legacy rq for better recycling
net/mlx5e: RX, Change wqe last_in_page field from bool to bit flags
net/mlx5e: RX, Defer page release in striding rq for better recycling
net/mlx5e: RX, Rename xdp_xmit_bitmap to a more generic name
net/mlx5e: RX, Enable skb page recycling through the page_pool
net/mlx5e: RX, Enable dma map and sync from page_pool allocator
net/mlx5e: RX, Remove internal page_cache
net/mlx5e: RX, Store SHAMPO header pages in array
net/mlx5e: RX, Remove alloc unit layout constraint for striding rq
net/mlx5e: RX, Remove alloc unit layout constraint for legacy rq
net/mlx5e: RX, Remove mlx5e_alloc_unit argument in page allocation
====================
Link: https://lore.kernel.org/r/20230328205623.142075-1-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Michael Chan says:
====================
bnxt_en: 3 Bug fixes
This series contains 3 small bug fixes covering ethtool self test, PCI
ID string typos, and some missing 200G link speed ethtool reporting logic.
====================
Link: https://lore.kernel.org/r/20230329013021.5205-1-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
bnxt_fw_to_ethtool_speed() is missing the case statement for 200G
link speed reported by firmware. As a result, ethtool will report
unknown speed when the firmware reports 200G link speed.
Fixes: 532262ba3b ("bnxt_en: ethtool: support PAM4 link speeds up to 200G")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2023-03-28 (ice)
This series contains updates to ice driver only.
Jesse fixes mismatched header documentation reported when building with
W=1.
Brett restricts setting of VSI context to only applicable fields for the
given ICE_AQ_VSI_PROP_Q_OPT_VALID bit.
Junfeng adds check when adding Flow Director filters that conflict with
existing filter rules.
Jakob Koschel adds interim variable for iterating to prevent possible
misuse after looping.
* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
ice: fix invalid check for empty list in ice_sched_assoc_vsi_to_agg()
ice: add profile conflict check for AVF FDIR
ice: Fix ice_cfg_rdma_fltr() to only update relevant fields
ice: fix W=1 headers mismatch
====================
Link: https://lore.kernel.org/r/20230328172035.3904953-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Stefan Schmidt says:
====================
ieee802154 for net 2023-03-29
Two small fixes this time.
Dongliang Mu removed an unnecessary null pointer check.
Harshit Mogalapalli fixed an int comparison unsigned against signed from a
recent other fix in the ca8210 driver.
* tag 'ieee802154-for-net-2023-03-29' of git://git.kernel.org/pub/scm/linux/kernel/git/wpan/wpan:
net: ieee802154: remove an unnecessary null pointer check
ca8210: Fix unsigned mac_len comparison with zero in ca8210_skb_tx()
====================
Link: https://lore.kernel.org/r/20230329064541.2147400-1-stefan@datenfreihafen.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In gsi_trans_pool_init_dma(), the total size of a pool of memory
used for DMA transactions is calculated. However the calculation is
done incorrectly.
For 4KB pages, this total size is currently always more than one
page, and as a result, the calculation produces a positive (though
incorrect) total size. The code still works in this case; we just
end up with fewer DMA pool entries than we intended.
Bjorn Andersson tested booting a kernel with 16KB pages, and hit a
null pointer derereference in sg_alloc_append_table_from_pages(),
descending from gsi_trans_pool_init_dma(). The cause of this was
that a 16KB total size was going to be allocated, and with 16KB
pages the order of that allocation is 0. The total_size calculation
yielded 0, which eventually led to the crash.
Correcting the total_size calculation fixes the problem.
Reported-by: Bjorn Andersson <quic_bjorande@quicinc.com>
Tested-by: Bjorn Andersson <quic_bjorande@quicinc.com>
Fixes: 9dd441e4ed ("soc: qcom: ipa: GSI transactions")
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Alex Elder <elder@linaro.org>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Link: https://lore.kernel.org/r/20230328162751.2861791-1-elder@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Adding a DFL (Device Feature List) device driver of ToD device for
Intel FPGA cards.
The Intel FPGA Time of Day(ToD) IP within the FPGA DFL bus is exposed
as PTP Hardware clock(PHC) device to the Linux PTP stack to synchronize
the system clock to its ToD information using phc2sys utility of the
Linux PTP stack. The DFL is a hardware List within FPGA, which defines
a linked list of feature headers within the device MMIO space to provide
an extensible way of adding subdevice features.
Signed-off-by: Raghavendra Khadatare <raghavendrax.anand.khadatare@intel.com>
Signed-off-by: Tianfei Zhang <tianfei.zhang@intel.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Link: https://lore.kernel.org/r/20230328142455.481146-1-tianfei.zhang@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When we allocate a nvme-tcp queue, we set the data_ready callback before
we actually need to use it. This creates the potential that if a stray
controller sends us data on the socket before we connect, we can trigger
the io_work and start consuming the socket.
In this case reported: we failed to allocate one of the io queues, and
as we start releasing the queues that we already allocated, we get
a UAF [1] from the io_work which is running before it should really.
Fix this by setting the socket ops callbacks only before we start the
queue, so that we can't accidentally schedule the io_work in the
initialization phase before the queue started. While we are at it,
rename nvme_tcp_restore_sock_calls to pair with nvme_tcp_setup_sock_ops.
[1]:
[16802.107284] nvme nvme4: starting error recovery
[16802.109166] nvme nvme4: Reconnecting in 10 seconds...
[16812.173535] nvme nvme4: failed to connect socket: -111
[16812.173745] nvme nvme4: Failed reconnect attempt 1
[16812.173747] nvme nvme4: Reconnecting in 10 seconds...
[16822.413555] nvme nvme4: failed to connect socket: -111
[16822.413762] nvme nvme4: Failed reconnect attempt 2
[16822.413765] nvme nvme4: Reconnecting in 10 seconds...
[16832.661274] nvme nvme4: creating 32 I/O queues.
[16833.919887] BUG: kernel NULL pointer dereference, address: 0000000000000088
[16833.920068] nvme nvme4: Failed reconnect attempt 3
[16833.920094] #PF: supervisor write access in kernel mode
[16833.920261] nvme nvme4: Reconnecting in 10 seconds...
[16833.920368] #PF: error_code(0x0002) - not-present page
[16833.921086] Workqueue: nvme_tcp_wq nvme_tcp_io_work [nvme_tcp]
[16833.921191] RIP: 0010:_raw_spin_lock_bh+0x17/0x30
...
[16833.923138] Call Trace:
[16833.923271] <TASK>
[16833.923402] lock_sock_nested+0x1e/0x50
[16833.923545] nvme_tcp_try_recv+0x40/0xa0 [nvme_tcp]
[16833.923685] nvme_tcp_io_work+0x68/0xa0 [nvme_tcp]
[16833.923824] process_one_work+0x1e8/0x390
[16833.923969] worker_thread+0x53/0x3d0
[16833.924104] ? process_one_work+0x390/0x390
[16833.924240] kthread+0x124/0x150
[16833.924376] ? set_kthread_struct+0x50/0x50
[16833.924518] ret_from_fork+0x1f/0x30
[16833.924655] </TASK>
Reported-by: Yanjun Zhang <zhangyanjun@cestc.cn>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Tested-by: Yanjun Zhang <zhangyanjun@cestc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The two infinite loops in bound check cases added by commit
1a3148fc17 ("selftests/bpf: Check when bounds are not in the 32-bit range")
increased the execution time of test_verifier from about 6 seconds to
about 9 seconds. Rewrite these two infinite loops to finite loops to get
rid of this extra time cost.
Signed-off-by: Xu Kuohai <xukuohai@huawei.com>
Link: https://lore.kernel.org/r/20230329011048.1721937-1-xukuohai@huaweicloud.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Andrii Nakryiko says:
====================
Teach veristat how to deal with freplace BPF programs. As they can't be
directly loaded by veristat without custom user-space part that sets correct
target program FD, veristat always fails freplace programs. This patch set
teaches veristat to guess target program type that will be inherited by
freplace program itself, and subtitute it for BPF_PROG_TYPE_EXT (freplace) one
for the purposes of BPF verification.
Patch #1 fixes bug in libbpf preventing overriding freplace with specific
program type.
Patch #2 adds convenient -d flag to request veristat to emit libbpf debug
logs. It help debugging why a specific BPF program fails to load, if the
problem is not due to BPF verification itself.
v3->v4:
- fix optional kern_name check when guessing prog type (Alexei);
v2->v3:
- fix bpf_obj_id selftest that uses legacy bpf_prog_test_load() helper,
which always sets program type programmatically; teach the helper to do it
only if actually necessary (Stanislav);
v1->v2:
- fix compilation error reported by old GCC (my GCC v11 doesn't produce even
a warning) and Clang (see CI failure at [0]):
GCC version:
veristat.c: In function ‘fixup_obj’:
veristat.c:908:1: error: label at end of compound statement
908 | skip_freplace_fixup:
| ^~~~~~~~~~~~~~~~~~~
Clang version:
veristat.c:909:1: error: label at end of compound statement is a C2x extension [-Werror,-Wc2x-extensions]
}
^
1 error generated.
[0] https://github.com/kernel-patches/bpf/actions/runs/4515972059/jobs/7953845335
====================
Acked-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
SEC("freplace") (i.e., BPF_PROG_TYPE_EXT) programs are not loadable as
is through veristat, as kernel expects actual program's FD during
BPF_PROG_LOAD time, which veristat has no way of knowing.
Unfortunately, freplace programs are a pretty important class of
programs, especially when dealing with XDP chaining solutions, which
rely on EXT programs.
So let's do our best and teach veristat to try to guess the original
program type, based on program's context argument type. And if guessing
process succeeds, we manually override freplace/EXT with guessed program
type using bpf_program__set_type() setter to increase chances of proper
BPF verification.
We rely on BTF and maintain a simple lookup table. This process is
obviously not 100% bulletproof, as valid program might not use context
and thus wouldn't have to specify correct type. Also, __sk_buff is very
ambiguous and is the context type across many different program types.
We pick BPF_PROG_TYPE_CGROUP_SKB for now, which seems to work fine in
practice so far. Similarly, some program types require specifying attach
type, and so we pick one out of possible few variants.
Best effort at its best. But this makes veristat even more widely
applicable.
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Tested-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20230327185202.1929145-4-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
If user explicitly overrides programs's type with
bpf_program__set_type() API call, we need to disassociate whatever
SEC_DEF handler libbpf determined initially based on program's SEC()
definition, as it's not goind to be valid anymore and could lead to
crashes and/or confusing failures.
Also, fix up bpf_prog_test_load() helper in selftests/bpf, which is
force-setting program type (even if that's completely unnecessary; this
is quite a legacy piece of code), and thus should expect auto-attach to
not work, yet one of the tests explicitly relies on auto-attach for
testing.
Instead, force-set program type only if it differs from the desired one.
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20230327185202.1929145-2-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
8b/10b encoding needs to add 3% fec overhead into the pbn.
In the Synapcis Cascaded MST hub, the first stage MST branch device
needs the information to determine the timeslot count for the
second stage MST branch device. Missing this overhead will leads to
insufficient timeslot allocation.
Cc: stable@vger.kernel.org
Cc: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Fangzhi Zuo <Jerry.Zuo@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Yixin Shen says:
====================
This series allow BPF TCP CCs to write app_limited of struct
tcp_sock. A built-in CC or one from a kernel module is already
able to write to app_limited of struct tcp_sock. Until now,
a BPF CC doesn't have write access to this member of struct
tcp_sock.
v2:
- Merge the test of writing app_limited into the test of
writing sk_pacing. (Martin KaFai Lau)
====================
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Conor Dooley <conor.dooley@microchip.com> says:
Here's my attempt at fixing both the use of an FPU on XIP kernels and
the issue that Jason ran into where CONFIG_FPU, which needs the
alternatives frame work for has_fpu() checks, could be enabled without
the alternatives actually being present.
For the former, a "slow" fallback that does not use alternatives is
added to riscv_has_extension_[un]likely() that can be used with XIP.
Obviously, we want to make use of Jisheng's alternatives based approach
where possible, so any users of riscv_has_extension_[un]likely() will
want to make sure that they select RISCV_ALTERNATIVE.
If they don't however, they'll hit the fallback path which (should,
sparing a silly mistake from me!) behave in the same way, thus
succeeding silently. Sounds like a
To prevent "depends on !XIP_KERNEL; select RISCV_ALTERNATIVE" spreading
like the plague through the various places that want to check for the
presence of extensions, and sidestep the potential silent "success"
mentioned above, all users RISCV_ALTERNATIVE are converted from selects
to dependencies, with the option being selected for all !XIP_KERNEL
builds.
I know that the VDSO was a key place that Jisheng wanted to use the new
helper rather than static branches, and I think the fallback path
should not cause issues there.
See the thread at [1] for the prior discussion.
1 - https://lore.kernel.org/linux-riscv/20230128172856.3814-1-jszhang@kernel.org/T/#m21390d570997145d31dd8bb95002fd61f99c6573
[Palmer: merging in the fixes as a branch as there's some features that
depend on it.]
* b4-shazam-merge:
RISC-V: always select RISCV_ALTERNATIVE for non-xip kernels
RISC-V: add non-alternative fallback for riscv_has_extension_[un]likely()
Link: https://lore.kernel.org/r/20230324100538.3514663-1-conor.dooley@microchip.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
When moving switch_to's has_fpu() over to using
riscv_has_extension_likely() rather than static branches, the FPU code
gained a dependency on the alternatives framework.
That dependency has now been removed, as riscv_has_extension_ikely() now
contains a fallback path, using __riscv_isa_extension_available(), but
if CONFIG_RISCV_ALTERNATIVE isn't selected when CONFIG_FPU is, has_fpu()
checks will not benefit from the "fast path" that the alternatives
framework provides.
We want to ensure that alternatives are available whenever
riscv_has_extension_[un]likely() is used, rather than silently falling
back to the slow path, but rather than rely on selecting
RISCV_ALTERNATIVE in the myriad of locations that may use
riscv_has_extension_[un]likely(), select it (almost) always instead by
adding it to the main RISCV config entry.
xip kernels cannot make use of the alternatives framework, so it is not
enabled for those configurations, although this is the status quo.
All current sites that select RISCV_ALTERNATIVE are converted to
dependencies on the option instead. The explicit dependencies on
!XIP_KERNEL can be dropped, as RISCV_ALTERNATIVE is not user selectable.
Fixes: 702e64550b ("riscv: fpu: switch has_fpu() to riscv_has_extension_likely()")
Link: https://lore.kernel.org/all/ZBruFRwt3rUVngPu@zx2c4.com/
Reported-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Jason A. Donenfeld <Jason@zx2c4.com>
Link: https://lore.kernel.org/r/20230324100538.3514663-3-conor.dooley@microchip.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
The has_fpu() check, which in turn calls riscv_has_extension_likely(),
relies on alternatives to figure out whether the system has an FPU.
As a result, it will malfunction on XIP kernels, as they do not support
the alternatives mechanism.
When alternatives support is not present, fall back to using
__riscv_isa_extension_available() in riscv_has_extension_[un]likely()
instead stead, which handily takes the same argument, so that kernels
that do not support alternatives can accurately report the presence of
FPU support.
Fixes: 702e64550b ("riscv: fpu: switch has_fpu() to riscv_has_extension_likely()")
Link: https://lore.kernel.org/all/ad445951-3d13-4644-94d9-e0989cda39c3@spud/
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Jason A. Donenfeld <Jason@zx2c4.com>
Link: https://lore.kernel.org/r/20230324100538.3514663-2-conor.dooley@microchip.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Commit 52f04f10b9 ("thermal: intel: int340x: processor_thermal: Fix
deadlock") addressed deadlock issue during user space trip update. But it
missed a case when thermal zone device is disabled when user writes 0.
Call to thermal_zone_device_disable() also causes deadlock as it also
tries to lock tz->lock, which is already claimed by trip_point_temp_store()
in the thermal core code.
Remove call to thermal_zone_device_disable() in the function
sys_set_trip_temp(), which is called from trip_point_temp_store().
Fixes: 52f04f10b9 ("thermal: intel: int340x: processor_thermal: Fix deadlock")
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: 6.2+ <stable@vger.kernel.org> # 6.2+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Commit 3e45352259 ("md: Free resources in __md_stop") tried to fix
null-ptr-deference for 'active_io' by moving percpu_ref_exit() to
__md_stop(), however, the commit also moving 'writes_pending' to
__md_stop(), and this will cause mdadm tests broken:
BUG: kernel NULL pointer dereference, address: 0000000000000038
Oops: 0000 [#1] PREEMPT SMP
CPU: 15 PID: 17830 Comm: mdadm Not tainted 6.3.0-rc3-next-20230324-00009-g520d37
RIP: 0010:free_percpu+0x465/0x670
Call Trace:
<TASK>
__percpu_ref_exit+0x48/0x70
percpu_ref_exit+0x1a/0x90
__md_stop+0xe9/0x170
do_md_stop+0x1e1/0x7b0
md_ioctl+0x90c/0x1aa0
blkdev_ioctl+0x19b/0x400
vfs_ioctl+0x20/0x50
__x64_sys_ioctl+0xba/0xe0
do_syscall_64+0x6c/0xe0
entry_SYSCALL_64_after_hwframe+0x63/0xcd
And the problem can be reporduced 100% by following test:
mdadm -CR /dev/md0 -l1 -n1 /dev/sda --force
echo inactive > /sys/block/md0/md/array_state
echo read-auto > /sys/block/md0/md/array_state
echo inactive > /sys/block/md0/md/array_state
Root cause:
// start raid
raid1_run
mddev_init_writes_pending
percpu_ref_init
// inactive raid
array_state_store
do_md_stop
__md_stop
percpu_ref_exit
// start raid again
array_state_store
do_md_run
raid1_run
mddev_init_writes_pending
if (mddev->writes_pending.percpu_count_ptr)
// won't reinit
// inactive raid again
...
percpu_ref_exit
-> null-ptr-deference
Before the commit, 'writes_pending' is exited when mddev is freed, and
it's safe to restart raid because mddev_init_writes_pending() already make
sure that 'writes_pending' will only be initialized once.
Fix the prblem by moving 'writes_pending' back, it's a litter hard to find
the relationship between alloc memory and free memory, however, code
changes is much less and we lived with this for a long time already.
Fixes: 3e45352259 ("md: Free resources in __md_stop")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230328094400.1448955-1-yukuai1@huaweicloud.com
Pull xtensa fixes from Max Filippov:
- fix KASAN report in show_stack
- drop linux-xtensa mailing list from the MAINTAINERS file
* tag 'xtensa-20230327' of https://github.com/jcmvbkbc/linux-xtensa:
MAINTAINERS: xtensa: drop linux-xtensa@linux-xtensa.org mailing list
xtensa: fix KASAN report for show_stack
Pull f2fs fix from Jaegeuk Kim:
"This fixes a tracepoint field size in f2fs in preparation for stricter
rules for tracing fields"
* tag 'f2fs-fix-6.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs:
f2fs: Fix f2fs_truncate_partial_nodes ftrace event
The drm_buddy_test KUnit tests verify that returned blocks have sizes
which are powers of two using is_power_of_2(). However, is_power_of_2()
operations on a 'long', but the block size is a u64. So on systems where
long is 32-bit, this can sometimes fail even on correctly sized blocks.
This only reproduces randomly, as the parameters passed to the buddy
allocator in this test are random. The seed 0xb2e06022 reproduced it
fine here.
For now, just hardcode an is_power_of_2() implementation using
x & (x - 1).
Signed-off-by: David Gow <davidgow@google.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Reviewed-by: Arunpravin Paneer Selvam <arunpravin.paneerselvam@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230329065532.2122295-2-davidgow@google.com
Signed-off-by: Christian König <christian.koenig@amd.com>
The hypervisor supports user-mode NX from Power10.
pseries_vas_dlpar_cpu() is called from lparcfg_write() to update VAS
windows for DLPAR event in shared processor mode and the kernel gets
-ENOTSUPP for HCALLs if the user-mode NX is not supported. The current
VAS implementation also supports only with Radix page tables. Whereas in
dedicated processor mode, pseries_vas_notifier() is registered only if
the copy/paste feature is enabled. So instead of displaying HCALL error
messages, update VAS capabilities if the copy/paste feature is
available.
This patch ignores updating VAS capabilities in pseries_vas_dlpar_cpu()
and returns success if the copy/paste feature is not enabled. Then
lparcfg_write() completes the processor DLPAR operations without any
failures.
Fixes: 2147783d6b ("powerpc/pseries: Use lparcfg to reconfig VAS windows for DLPAR CPU")
Cc: stable@vger.kernel.org # v6.1+
Signed-off-by: Haren Myneni <haren@linux.ibm.com>
Reviewed-by: Nathan Lynch <nathanl@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/1d0e727e7dbd9a28627ef08ca9df9c86a50175e2.camel@linux.ibm.com
After entering 6.3-rc1 the LLC cacheinfo is not exported on our ACPI
based arm64 server. This is because the LLC cacheinfo is partly reset
when secondary CPUs boot up. On arm64 the primary cpu will allocate
and setup cacheinfo:
init_cpu_topology()
for_each_possible_cpu()
fetch_cache_info() // Allocate cacheinfo and init levels
detect_cache_attributes()
cache_shared_cpu_map_setup()
if (!last_level_cache_is_valid()) // not valid, setup LLC
cache_setup_properties() // setup LLC
On secondary CPU boot up:
detect_cache_attributes()
populate_cache_leaves()
get_cache_type() // Get cache type from clidr_el1,
// for LLC type=CACHE_TYPE_NOCACHE
cache_shared_cpu_map_setup()
if (!last_level_cache_is_valid()) // Valid and won't go to this branch,
// leave LLC's type=CACHE_TYPE_NOCACHE
The last_level_cache_is_valid() use cacheinfo->{attributes, fw_token} to
test it's valid or not, but populate_cache_leaves() will only reset
LLC's type, so we won't try to re-setup LLC's type and leave it
CACHE_TYPE_NOCACHE and won't export it through sysfs.
This patch tries to fix this by not re-populating the cache leaves if
the LLC is valid.
Fixes: 5944ce092b ("arch_topology: Build cacheinfo from primary CPU")
Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Reviewed-by: Pierre Gondois <pierre.gondois@arm.com>
Reviewed-by: Sudeep Holla <sudeep.holla@arm.com>
Link: https://lore.kernel.org/r/20230328114915.33340-1-yangyicong@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The nouveau code used to call drm_fb_helper_initial_config() from
nouveau_fbcon_init() before calling drm_dev_register(). This would
probe all connectors so that drm_connector->status could be used during
backlight registration which runs from nouveau_connector_late_register().
After commit 4a16dd9d18 ("drm/nouveau/kms: switch to drm fbdev helpers")
the fbdev emulation code, which now is a drm-client, can only run after
drm_dev_register(). So during backlight registration the connectors are
not probed yet and the drm_connector->status == connected check in
nv50_backlight_init() would now always fail.
Replace the drm_connector->status == connected check with
a drm_helper_probe_detect() == connected check to fix nv_backlight
no longer getting registered because of this.
Fixes: 4a16dd9d18 ("drm/nouveau/kms: switch to drm fbdev helpers")
Link: https://gitlab.freedesktop.org/drm/nouveau/-/issues/202
Link: https://bugzilla.redhat.com/show_bug.cgi?id=2181941
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230326205433.36485-1-hdegoede@redhat.com
According to LPUART RM, Transmission Complete Flag becomes 0 if queuing
a break character by writing 1 to CTRL[SBK], so here need to avoid
checking for transmission complete when UARTCTRL_SBK is asserted,
otherwise the lpuart32_tx_empty may never get TIOCSER_TEMT.
Commit 2411fd94ceaa("tty: serial: fsl_lpuart: skip waiting for
transmission complete when UARTCTRL_SBK is asserted") only fix it in
lpuart32_set_termios(), here also fix it in lpuart32_tx_empty().
Fixes: 380c966c09 ("tty: serial: fsl_lpuart: add 32-bit register interface support")
Cc: stable <stable@kernel.org>
Signed-off-by: Sherry Sun <sherry.sun@nxp.com>
Link: https://lore.kernel.org/r/20230323054415.20363-1-sherry.sun@nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Hans de Goede reported Bluetooth adapters (HCIs) connected over an UART
connection failed due corrupted Rx payload. The problem was narrowed
down to DMA Rx starting on UART_IIR_THRI interrupt. The problem occurs
despite LSR having DR bit set, which is precondition for attempting to
start DMA Rx in the first place.
From a debug patch:
[x.807834] 8250irq: iir=cc lsr+saved=60 received=0/15 ier=0f dma_t/rx/err=0/0/0
[x.808676] 8250irq: iir=c2 lsr+saved=61 received=0/0 ier=0f dma_t/rx/err=0/0/0
[x.808776] 8250irq: iir=cc lsr+saved=60 received=1/12 ier=0d dma_t/rx/err=0/1/0
[x.808870] Bluetooth: hci0: Frame reassembly failed (-84)
In the debug snippet, received field indicates 1 byte was transferred
over DMA and 12 bytes after that with the non-DMA Rx. The sole byte DMA
handled was corrupted (gets zeroed) which leads to the HCI failure.
This problem became apparent after commit e8ffbb71f7 ("serial: 8250:
use THRE & __stop_tx also with DMA") changed Tx stop behavior. Tx stop
is now triggered from a THRI interrupt.
Despite that this problem looks like a HW bug, this fix is not adding
UART_BUG_xx flag to the driver beucase it seems useful in general to
avoid starting DMA when there are only a few bytes to transfer.
Skipping DMA for small transfers avoids the extra overhead DMA incurs.
Thus, don't setup DMA Rx on UART_IIR_THRI but leave it to a subsequent
interrupt which has Rx a related IIR value.
By returning false from handle_rx_dma(), the DMA vs non-DMA decision is
postponed until either UART_IIR_RDI (FIFO threshold worth of bytes
awaiting) or UART_IIR_TIMEOUT (inter-character timeout) triggers at a
later time which allows better to discern whether the number of bytes
warrants starting DMA or not.
Reported-by: Hans de Goede <hdegoede@redhat.com>
Tested-by: Hans de Goede <hdegoede@redhat.com>
Fixes: e8ffbb71f7 ("serial: 8250: use THRE & __stop_tx also with DMA")
Cc: stable@vger.kernel.org
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20230317103034.12881-1-ilpo.jarvinen@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
