[ Upstream commit ecdf985d76 ]
For aligned stack writes using BPF_ST instruction track stored values
in a same way BPF_STX is handled, e.g. make sure that the following
commands produce similar verifier knowledge:
fp[-8] = 42; r1 = 42;
fp[-8] = r1;
This covers two cases:
- non-null values written to stack are stored as spill of fake
registers;
- null values written to stack are stored as STACK_ZERO marks.
Previously both cases above used STACK_MISC marks instead.
Some verifier test cases relied on the old logic to obtain STACK_MISC
marks for some stack values. These test cases are updated in the same
commit to avoid failures during bisect.
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20230214232030.1502829-2-eddyz87@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Stable-dep-of: 713274f1f2 ("bpf: Fix verifier id tracking of scalars on spill")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 8681f71759 ]
Restore the host's PMUSERENR_EL0 value instead of clearing it,
before returning back to userspace, as the host's EL0 might have
a direct access to PMU registers (some bits of PMUSERENR_EL0 for
might not be zero for the host EL0).
Fixes: 83a7a4d643 ("arm64: perf: Enable PMU counter userspace access for perf event")
Signed-off-by: Reiji Watanabe <reijiw@google.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20230603025035.3781797-2-reijiw@google.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit a287f5b0cf ]
This change adds methods in the XFRM-I input path that ensures that
policies are checked prior to processing of the subsequent decapsulated
packet, after which the relevant policies may no longer be resolvable
(due to changing src/dst/proto/etc).
Notably, raw ESP/AH packets did not perform policy checks inherently,
whereas all other encapsulated packets (UDP, TCP encapsulated) do policy
checks after calling xfrm_input handling in the respective encapsulation
layer.
Fixes: b0355dbbf1 ("Fix XFRM-I support for nested ESP tunnels")
Test: Verified with additional Android Kernel Unit tests
Test: Verified against Android CTS
Signed-off-by: Benedict Wong <benedictwong@google.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 1f8b6df6a9 ]
This change allows inbound traffic through nested IPsec tunnels to
successfully match policies and templates, while retaining the secpath
stack trace as necessary for netfilter policies.
Specifically, this patch marks secpath entries that have already matched
against a relevant policy as having been verified, allowing it to be
treated as optional and skipped after a tunnel decapsulation (during
which the src/dst/proto/etc may have changed, and the correct policy
chain no long be resolvable).
This approach is taken as opposed to the iteration in b0355dbbf1,
where the secpath was cleared, since that breaks subsequent validations
that rely on the existence of the secpath entries (netfilter policies, or
transport-in-tunnel mode, where policies remain resolvable).
Fixes: b0355dbbf1 ("Fix XFRM-I support for nested ESP tunnels")
Test: Tested against Android Kernel Unit Tests
Test: Tested against Android CTS
Signed-off-by: Benedict Wong <benedictwong@google.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b8ada54fa1 ]
The driver overrides the error codes and IRQ0 returned by platform_get_irq()
to -EINVAL, so if it returns -EPROBE_DEFER, the driver will fail the probe
permanently instead of the deferred probing. Switch to propagating the error
codes upstream. Since commit ce753ad154 ("platform: finally disallow IRQ0
in platform_get_irq() and its ilk") IRQ0 is no longer returned by those APIs,
so we now can safely ignore it...
Fixes: cbcaac6d7d ("mmc: meson-gx-mmc: Fix platform_get_irq's error checking")
Cc: stable@vger.kernel.org # v5.19+
Signed-off-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20230617203622.6812-3-s.shtylyov@omp.ru
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit d082d48737 upstream.
KPTI keeps around two PGDs: one for userspace and another for the
kernel. Among other things, set_pgd() contains infrastructure to
ensure that updates to the kernel PGD are reflected in the user PGD
as well.
One side-effect of this is that set_pgd() expects to be passed whole
pages. Unfortunately, init_trampoline_kaslr() passes in a single entry:
'trampoline_pgd_entry'.
When KPTI is on, set_pgd() will update 'trampoline_pgd_entry' (an
8-Byte globally stored [.bss] variable) and will then proceed to
replicate that value into the non-existent neighboring user page
(located +4k away), leading to the corruption of other global [.bss]
stored variables.
Fix it by directly assigning 'trampoline_pgd_entry' and avoiding
set_pgd().
[ dhansen: tweak subject and changelog ]
Fixes: 0925dda596 ("x86/mm/KASLR: Use only one PUD entry for real mode trampoline")
Suggested-by: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Lee Jones <lee@kernel.org>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/all/20230614163859.924309-1-lee@kernel.org/g
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 782e53d0c1 upstream.
In a syzbot stress test that deliberately causes file system errors on
nilfs2 with a corrupted disk image, it has been reported that
nilfs_clear_dirty_page() called from nilfs_clear_dirty_pages() can cause a
general protection fault.
In nilfs_clear_dirty_pages(), when looking up dirty pages from the page
cache and calling nilfs_clear_dirty_page() for each dirty page/folio
retrieved, the back reference from the argument page to "mapping" may have
been changed to NULL (and possibly others). It is necessary to check this
after locking the page/folio.
So, fix this issue by not calling nilfs_clear_dirty_page() on a page/folio
after locking it in nilfs_clear_dirty_pages() if the back reference
"mapping" from the page/folio is different from the "mapping" that held
the page/folio just before.
Link: https://lkml.kernel.org/r/20230612021456.3682-1-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Reported-by: syzbot+53369d11851d8f26735c@syzkaller.appspotmail.com
Closes: https://lkml.kernel.org/r/000000000000da4f6b05eb9bf593@google.com
Tested-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit ef7dfac51d upstream.
We selectively grab the ctx->uring_lock for poll update/removal, but
we really should grab it from the start to fully synchronize with
linked timeouts. Normally this is indeed the case, but if requests
are forced async by the application, we don't fully cover removal
and timer disarm within the uring_lock.
Make this simpler by having consistent locking state for poll removal.
Cc: stable@vger.kernel.org # 6.1+
Reported-by: Querijn Voet <querijnqyn@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit c8e796895e ]
The max_raw_write member of the regmap_spi_avmm_bus structure is defined
as:
.max_raw_write = SPI_AVMM_VAL_SIZE * MAX_WRITE_CNT
SPI_AVMM_VAL_SIZE == 4 and MAX_WRITE_CNT == 1 so this results in a
maximum write transfer size of 4 bytes which provides only enough space to
transfer the address of the target register. It provides no space for the
value to be transferred. This bug became an issue (divide-by-zero in
_regmap_raw_write()) after the following was accepted into mainline:
commit 3981514180 ("regmap: Account for register length when chunking")
Change max_raw_write to include space (4 additional bytes) for both the
register address and value:
.max_raw_write = SPI_AVMM_REG_SIZE + SPI_AVMM_VAL_SIZE * MAX_WRITE_CNT
Fixes: 7f9fb67358 ("regmap: add Intel SPI Slave to AVMM Bus Bridge support")
Reviewed-by: Matthew Gerlach <matthew.gerlach@linux.intel.com>
Signed-off-by: Russ Weight <russell.h.weight@intel.com>
Link: https://lore.kernel.org/r/20230620202824.380313-1-russell.h.weight@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 0108a4e9f3 upstream.
When subprograms are in use, the main program is not jit'd after the
subprograms because jit_subprogs sets a value for prog->bpf_func upon
success. Subsequent calls to the JIT are bypassed when this value is
non-NULL. This leads to a situation where the main program and its
func[0] counterpart are both in the bpf kallsyms tree, but only func[0]
has an extable. Extables are only created during JIT. Now there are
two nearly identical program ksym entries in the tree, but only one has
an extable. Depending upon how the entries are placed, there's a chance
that a fault will call search_extable on the aux with the NULL entry.
Since jit_subprogs already copies state from func[0] to the main
program, include the extable pointer in this state duplication.
Additionally, ensure that the copy of the main program in func[0] is not
added to the bpf_prog_kallsyms table. Instead, let the main program get
added later in bpf_prog_load(). This ensures there is only a single
copy of the main program in the kallsyms table, and that its tag matches
the tag observed by tooling like bpftool.
Cc: stable@vger.kernel.org
Fixes: 1c2a088a66 ("bpf: x64: add JIT support for multi-function programs")
Signed-off-by: Krister Johansen <kjlx@templeofstupid.com>
Acked-by: Yonghong Song <yhs@fb.com>
Acked-by: Ilya Leoshkevich <iii@linux.ibm.com>
Tested-by: Ilya Leoshkevich <iii@linux.ibm.com>
Link: https://lore.kernel.org/r/6de9b2f4b4724ef56efbb0339daaa66c8b68b1e7.1686616663.git.kjlx@templeofstupid.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit c2df53c580 upstream.
The driver overrides the error codes and IRQ0 returned by platform_get_irq()
to -EINVAL, so if it returns -EPROBE_DEFER, the driver will fail the probe
permanently instead of the deferred probing. Switch to propagating the error
codes upstream. Since commit ce753ad154 ("platform: finally disallow IRQ0
in platform_get_irq() and its ilk") IRQ0 is no longer returned by those APIs,
so we now can safely ignore it...
Fixes: 2408a08583 ("mmc: sunxi-mmc: Handle return value of platform_get_irq")
Cc: stable@vger.kernel.org # v5.19+
Signed-off-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Reviewed-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Link: https://lore.kernel.org/r/20230617203622.6812-12-s.shtylyov@omp.ru
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 71150ac125 upstream.
The driver overrides the error codes and IRQ0 returned by platform_get_irq()
to -EINVAL, so if it returns -EPROBE_DEFER, the driver will fail the probe
permanently instead of the deferred probing. Switch to propagating the error
codes upstream. Since commit ce753ad154 ("platform: finally disallow IRQ0
in platform_get_irq() and its ilk") IRQ0 is no longer returned by those APIs,
so we now can safely ignore it...
Fixes: 660fc733bd ("mmc: bcm2835: Add new driver for the sdhost controller.")
Cc: stable@vger.kernel.org # v5.19+
Signed-off-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Link: https://lore.kernel.org/r/20230617203622.6812-2-s.shtylyov@omp.ru
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit e6f9e590b7 upstream.
While SDHCI claims to support 64-bit DMA on MSM8916 it does not seem to
be properly functional. It is not immediately obvious because SDHCI is
usually used with IOMMU bypassed on this SoC, and all physical memory
has 32-bit addresses. But when trying to enable the IOMMU it quickly
fails with an error such as the following:
arm-smmu 1e00000.iommu: Unhandled context fault:
fsr=0x402, iova=0xfffff200, fsynr=0xe0000, cbfrsynra=0x140, cb=3
mmc1: ADMA error: 0x02000000
mmc1: sdhci: ============ SDHCI REGISTER DUMP ===========
mmc1: sdhci: Sys addr: 0x00000000 | Version: 0x00002e02
mmc1: sdhci: Blk size: 0x00000008 | Blk cnt: 0x00000000
mmc1: sdhci: Argument: 0x00000000 | Trn mode: 0x00000013
mmc1: sdhci: Present: 0x03f80206 | Host ctl: 0x00000019
mmc1: sdhci: Power: 0x0000000f | Blk gap: 0x00000000
mmc1: sdhci: Wake-up: 0x00000000 | Clock: 0x00000007
mmc1: sdhci: Timeout: 0x0000000a | Int stat: 0x00000001
mmc1: sdhci: Int enab: 0x03ff900b | Sig enab: 0x03ff100b
mmc1: sdhci: ACmd stat: 0x00000000 | Slot int: 0x00000000
mmc1: sdhci: Caps: 0x322dc8b2 | Caps_1: 0x00008007
mmc1: sdhci: Cmd: 0x0000333a | Max curr: 0x00000000
mmc1: sdhci: Resp[0]: 0x00000920 | Resp[1]: 0x5b590000
mmc1: sdhci: Resp[2]: 0xe6487f80 | Resp[3]: 0x0a404094
mmc1: sdhci: Host ctl2: 0x00000008
mmc1: sdhci: ADMA Err: 0x00000001 | ADMA Ptr: 0x0000000ffffff224
mmc1: sdhci_msm: ----------- VENDOR REGISTER DUMP -----------
mmc1: sdhci_msm: DLL sts: 0x00000000 | DLL cfg: 0x60006400 | DLL cfg2: 0x00000000
mmc1: sdhci_msm: DLL cfg3: 0x00000000 | DLL usr ctl: 0x00000000 | DDR cfg: 0x00000000
mmc1: sdhci_msm: Vndr func: 0x00018a9c | Vndr func2 : 0xf88018a8 Vndr func3: 0x00000000
mmc1: sdhci: ============================================
mmc1: sdhci: fffffffff200: DMA 0x0000ffffffffe100, LEN 0x0008, Attr=0x21
mmc1: sdhci: fffffffff20c: DMA 0x0000000000000000, LEN 0x0000, Attr=0x03
Looking closely it's obvious that only the 32-bit part of the address
(0xfffff200) arrives at the SMMU, the higher 16-bit (0xffff...) get
lost somewhere. This might not be a limitation of the SDHCI itself but
perhaps the bus/interconnect it is connected to, or even the connection
to the SMMU.
Work around this by setting SDHCI_QUIRK2_BROKEN_64_BIT_DMA to avoid
using 64-bit addresses.
Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230518-msm8916-64bit-v1-1-5694b0f35211@gerhold.net
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 6f363f5aa8 upstream.
We found a refcount UAF bug as follows:
refcount_t: addition on 0; use-after-free.
WARNING: CPU: 1 PID: 342 at lib/refcount.c:25 refcount_warn_saturate+0xa0/0x148
Workqueue: events cpuset_hotplug_workfn
Call trace:
refcount_warn_saturate+0xa0/0x148
__refcount_add.constprop.0+0x5c/0x80
css_task_iter_advance_css_set+0xd8/0x210
css_task_iter_advance+0xa8/0x120
css_task_iter_next+0x94/0x158
update_tasks_root_domain+0x58/0x98
rebuild_root_domains+0xa0/0x1b0
rebuild_sched_domains_locked+0x144/0x188
cpuset_hotplug_workfn+0x138/0x5a0
process_one_work+0x1e8/0x448
worker_thread+0x228/0x3e0
kthread+0xe0/0xf0
ret_from_fork+0x10/0x20
then a kernel panic will be triggered as below:
Unable to handle kernel paging request at virtual address 00000000c0000010
Call trace:
cgroup_apply_control_disable+0xa4/0x16c
rebind_subsystems+0x224/0x590
cgroup_destroy_root+0x64/0x2e0
css_free_rwork_fn+0x198/0x2a0
process_one_work+0x1d4/0x4bc
worker_thread+0x158/0x410
kthread+0x108/0x13c
ret_from_fork+0x10/0x18
The race that cause this bug can be shown as below:
(hotplug cpu) | (umount cpuset)
mutex_lock(&cpuset_mutex) | mutex_lock(&cgroup_mutex)
cpuset_hotplug_workfn |
rebuild_root_domains | rebind_subsystems
update_tasks_root_domain | spin_lock_irq(&css_set_lock)
css_task_iter_start | list_move_tail(&cset->e_cset_node[ss->id]
while(css_task_iter_next) | &dcgrp->e_csets[ss->id]);
css_task_iter_end | spin_unlock_irq(&css_set_lock)
mutex_unlock(&cpuset_mutex) | mutex_unlock(&cgroup_mutex)
Inside css_task_iter_start/next/end, css_set_lock is hold and then
released, so when iterating task(left side), the css_set may be moved to
another list(right side), then it->cset_head points to the old list head
and it->cset_pos->next points to the head node of new list, which can't
be used as struct css_set.
To fix this issue, switch from all css_sets to only scgrp's css_sets to
patch in-flight iterators to preserve correct iteration, and then
update it->cset_head as well.
Reported-by: Gaosheng Cui <cuigaosheng1@huawei.com>
Link: https://www.spinics.net/lists/cgroups/msg37935.html
Suggested-by: Michal Koutný <mkoutny@suse.com>
Link: https://lore.kernel.org/all/20230526114139.70274-1-xiujianfeng@huaweicloud.com/
Signed-off-by: Xiu Jianfeng <xiujianfeng@huawei.com>
Fixes: 2d8f243a5e ("cgroup: implement cgroup->e_csets[]")
Cc: stable@vger.kernel.org # v3.16+
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 81c1d02901 upstream.
An orphaned msk releases the used resources via the worker,
when the latter first see the msk in CLOSED status.
If the msk status transitions to TCP_CLOSE in the release callback
invoked by the worker's final release_sock(), such instance of the
workqueue will not take any action.
Additionally the MPTCP code prevents scheduling the worker once the
socket reaches the CLOSE status: such msk resources will be leaked.
The only code path that can trigger the above scenario is the
__mptcp_check_send_data_fin() in fallback mode.
Address the issue removing the special handling of fallback socket
in __mptcp_check_send_data_fin(), consolidating the state machine
for fallback and non fallback socket.
Since non-fallback sockets do not send and do not receive data_fin,
the mptcp code can update the msk internal status to match the next
step in the SM every time data fin (ack) should be generated or
received.
As a consequence we can remove a bunch of checks for fallback from
the fastpath.
Fixes: 6e628cd3a8 ("mptcp: use mptcp release_cb for delayed tasks")
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Mat Martineau <martineau@kernel.org>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 56a666c48b upstream.
At passive MPJ time, if the msk socket lock is held by the user,
the new subflow is appended to the msk->join_list under the msk
data lock.
In mptcp_release_cb()/__mptcp_flush_join_list(), the subflows in
that list are moved from the join_list into the conn_list under the
msk socket lock.
Append and removal could race, possibly corrupting such list.
Address the issue splicing the join list into a temporary one while
still under the msk data lock.
Found by code inspection, the race itself should be almost impossible
to trigger in practice.
Fixes: 3e5014909b ("mptcp: cleanup MPJ subflow list handling")
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit c2b2ae3925 upstream.
Currently the mptcp code has assumes that disconnect() can fail only
at mptcp_sendmsg_fastopen() time - to avoid a deadlock scenario - and
don't even bother returning an error code.
Soon mptcp_disconnect() will handle more error conditions: let's track
them explicitly.
As a bonus, explicitly annotate TCP-level disconnect as not failing:
the mptcp code never blocks for event on the subflows.
Fixes: 7d803344fd ("mptcp: fix deadlock in fastopen error path")
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Tested-by: Christoph Paasch <cpaasch@apple.com>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 78d0d2063b upstream.
We cannot sanely handle partial retries for recvmsg if we have cmsg
attached. If we don't, then we'd just be overwriting the initial cmsg
header on retries. Alternatively we could increment and handle this
appropriately, but it doesn't seem worth the complication.
Move the MSG_WAITALL check into the non-multishot case while at it,
since MSG_WAITALL is explicitly disabled for multishot anyway.
Link: https://lore.kernel.org/io-uring/0b0d4411-c8fd-4272-770b-e030af6919a0@kernel.dk/
Cc: stable@vger.kernel.org # 5.10+
Reported-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit b1dc492087 upstream.
If we have cmsg attached AND we transferred partial data at least, clear
msg_controllen on retry so we don't attempt to send that again.
Cc: stable@vger.kernel.org # 5.10+
Fixes: cac9e4418f ("io_uring/net: save msghdr->msg_control for retries")
Reported-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 067d6ec7ed upstream.
In the case of fast device addition/removal, it's possible that
hv_eject_device_work() can start to run before create_root_hv_pci_bus()
starts to run; as a result, the pci_get_domain_bus_and_slot() in
hv_eject_device_work() can return a 'pdev' of NULL, and
hv_eject_device_work() can remove the 'hpdev', and immediately send a
message PCI_EJECTION_COMPLETE to the host, and the host immediately
unassigns the PCI device from the guest; meanwhile,
create_root_hv_pci_bus() and the PCI device driver can be probing the
dead PCI device and reporting timeout errors.
Fix the issue by adding a per-bus mutex 'state_lock' and grabbing the
mutex before powering on the PCI bus in hv_pci_enter_d0(): when
hv_eject_device_work() starts to run, it's able to find the 'pdev' and call
pci_stop_and_remove_bus_device(pdev): if the PCI device driver has
loaded, the PCI device driver's probe() function is already called in
create_root_hv_pci_bus() -> pci_bus_add_devices(), and now
hv_eject_device_work() -> pci_stop_and_remove_bus_device() is able
to call the PCI device driver's remove() function and remove the device
reliably; if the PCI device driver hasn't loaded yet, the function call
hv_eject_device_work() -> pci_stop_and_remove_bus_device() is able to
remove the PCI device reliably and the PCI device driver's probe()
function won't be called; if the PCI device driver's probe() is already
running (e.g., systemd-udev is loading the PCI device driver), it must
be holding the per-device lock, and after the probe() finishes and releases
the lock, hv_eject_device_work() -> pci_stop_and_remove_bus_device() is
able to proceed to remove the device reliably.
Fixes: 4daace0d8c ("PCI: hv: Add paravirtual PCI front-end for Microsoft Hyper-V VMs")
Signed-off-by: Dexuan Cui <decui@microsoft.com>
Reviewed-by: Michael Kelley <mikelley@microsoft.com>
Acked-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230615044451.5580-6-decui@microsoft.com
Signed-off-by: Wei Liu <wei.liu@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 2738d5ab79 upstream.
When the host tries to remove a PCI device, the host first sends a
PCI_EJECT message to the guest, and the guest is supposed to gracefully
remove the PCI device and send a PCI_EJECTION_COMPLETE message to the host;
the host then sends a VMBus message CHANNELMSG_RESCIND_CHANNELOFFER to
the guest (when the guest receives this message, the device is already
unassigned from the guest) and the guest can do some final cleanup work;
if the guest fails to respond to the PCI_EJECT message within one minute,
the host sends the VMBus message CHANNELMSG_RESCIND_CHANNELOFFER and
removes the PCI device forcibly.
In the case of fast device addition/removal, it's possible that the PCI
device driver is still configuring MSI-X interrupts when the guest receives
the PCI_EJECT message; the channel callback calls hv_pci_eject_device(),
which sets hpdev->state to hv_pcichild_ejecting, and schedules a work
hv_eject_device_work(); if the PCI device driver is calling
pci_alloc_irq_vectors() -> ... -> hv_compose_msi_msg(), we can break the
while loop in hv_compose_msi_msg() due to the updated hpdev->state, and
leave data->chip_data with its default value of NULL; later, when the PCI
device driver calls request_irq() -> ... -> hv_irq_unmask(), the guest
crashes in hv_arch_irq_unmask() due to data->chip_data being NULL.
Fix the issue by not testing hpdev->state in the while loop: when the
guest receives PCI_EJECT, the device is still assigned to the guest, and
the guest has one minute to finish the device removal gracefully. We don't
really need to (and we should not) test hpdev->state in the loop.
Fixes: de0aa7b2f9 ("PCI: hv: Fix 2 hang issues in hv_compose_msi_msg()")
Signed-off-by: Dexuan Cui <decui@microsoft.com>
Reviewed-by: Michael Kelley <mikelley@microsoft.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230615044451.5580-3-decui@microsoft.com
Signed-off-by: Wei Liu <wei.liu@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit a847234e24 upstream.
This reverts commit d6af2ed29c.
The statement "the hv_pci_bus_exit() call releases structures of all its
child devices" in commit d6af2ed29c is not true: in the path
hv_pci_probe() -> hv_pci_enter_d0() -> hv_pci_bus_exit(hdev, true): the
parameter "keep_devs" is true, so hv_pci_bus_exit() does *not* release the
child "struct hv_pci_dev *hpdev" that is created earlier in
pci_devices_present_work() -> new_pcichild_device().
The commit d6af2ed29c was originally made in July 2020 for RHEL 7.7,
where the old version of hv_pci_bus_exit() was used; when the commit was
rebased and merged into the upstream, people didn't notice that it's
not really necessary. The commit itself doesn't cause any issue, but it
makes hv_pci_probe() more complicated. Revert it to facilitate some
upcoming changes to hv_pci_probe().
Signed-off-by: Dexuan Cui <decui@microsoft.com>
Reviewed-by: Michael Kelley <mikelley@microsoft.com>
Acked-by: Wei Hu <weh@microsoft.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230615044451.5580-5-decui@microsoft.com
Signed-off-by: Wei Liu <wei.liu@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 440b5e3663 upstream.
Since day 1 of the driver, there has been a race between
hv_pci_query_relations() and survey_child_resources(): during fast
device hotplug, hv_pci_query_relations() may error out due to
device-remove and the stack variable 'comp' is no longer valid;
however, pci_devices_present_work() -> survey_child_resources() ->
complete() may be running on another CPU and accessing the no-longer-valid
'comp'. Fix the race by flushing the workqueue before we exit from
hv_pci_query_relations().
Fixes: 4daace0d8c ("PCI: hv: Add paravirtual PCI front-end for Microsoft Hyper-V VMs")
Signed-off-by: Dexuan Cui <decui@microsoft.com>
Reviewed-by: Michael Kelley <mikelley@microsoft.com>
Acked-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230615044451.5580-2-decui@microsoft.com
Signed-off-by: Wei Liu <wei.liu@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 320805ab61 upstream.
vmbus_wait_for_unload() may be called in the panic path after other
CPUs are stopped. vmbus_wait_for_unload() currently loops through
online CPUs looking for the UNLOAD response message. But the values of
CONFIG_KEXEC_CORE and crash_kexec_post_notifiers affect the path used
to stop the other CPUs, and in one of the paths the stopped CPUs
are removed from cpu_online_mask. This removal happens in both
x86/x64 and arm64 architectures. In such a case, vmbus_wait_for_unload()
only checks the panic'ing CPU, and misses the UNLOAD response message
except when the panic'ing CPU is CPU 0. vmbus_wait_for_unload()
eventually times out, but only after waiting 100 seconds.
Fix this by looping through *present* CPUs in vmbus_wait_for_unload().
The cpu_present_mask is not modified by stopping the other CPUs in the
panic path, nor should it be.
Also, in a CoCo VM the synic_message_page is not allocated in
hv_synic_alloc(), but is set and cleared in hv_synic_enable_regs()
and hv_synic_disable_regs() such that it is set only when the CPU is
online. If not all present CPUs are online when vmbus_wait_for_unload()
is called, the synic_message_page might be NULL. Add a check for this.
Fixes: cd95aad557 ("Drivers: hv: vmbus: handle various crash scenarios")
Cc: stable@vger.kernel.org
Reported-by: John Starks <jostarks@microsoft.com>
Signed-off-by: Michael Kelley <mikelley@microsoft.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Link: https://lore.kernel.org/r/1684422832-38476-1-git-send-email-mikelley@microsoft.com
Signed-off-by: Wei Liu <wei.liu@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit ec97e11298 upstream.
Commit 572086325c ("Drivers: hv: vmbus: Cleanup synic memory free path")
says "Any memory allocations that succeeded will be freed when the caller
cleans up by calling hv_synic_free()", but if the get_zeroed_page() in
hv_synic_alloc() fails, currently hv_synic_free() is not really called
in vmbus_bus_init(), consequently there will be a memory leak, e.g.
hv_context.hv_numa_map is not freed in the error path. Fix this by
updating the goto labels.
Cc: stable@kernel.org
Signed-off-by: Dexuan Cui <decui@microsoft.com>
Fixes: 4df4cb9e99 ("x86/hyperv: Initialize clockevents earlier in CPU onlining")
Reviewed-by: Michael Kelley <mikelley@microsoft.com>
Link: https://lore.kernel.org/r/20230504224155.10484-1-decui@microsoft.com
Signed-off-by: Wei Liu <wei.liu@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 2230f9e117 upstream.
We run into guest hang in edk2 firmware when KSM is kept as running on
the host. The edk2 firmware is waiting for status 0x80 from QEMU's pflash
device (TYPE_PFLASH_CFI01) during the operation of sector erasing or
buffered write. The status is returned by reading the memory region of
the pflash device and the read request should have been forwarded to QEMU
and emulated by it. Unfortunately, the read request is covered by an
illegal stage2 mapping when the guest hang issue occurs. The read request
is completed with QEMU bypassed and wrong status is fetched. The edk2
firmware runs into an infinite loop with the wrong status.
The illegal stage2 mapping is populated due to same page sharing by KSM
at (C) even the associated memory slot has been marked as invalid at (B)
when the memory slot is requested to be deleted. It's notable that the
active and inactive memory slots can't be swapped when we're in the middle
of kvm_mmu_notifier_change_pte() because kvm->mn_active_invalidate_count
is elevated, and kvm_swap_active_memslots() will busy loop until it reaches
to zero again. Besides, the swapping from the active to the inactive memory
slots is also avoided by holding &kvm->srcu in __kvm_handle_hva_range(),
corresponding to synchronize_srcu_expedited() in kvm_swap_active_memslots().
CPU-A CPU-B
----- -----
ioctl(kvm_fd, KVM_SET_USER_MEMORY_REGION)
kvm_vm_ioctl_set_memory_region
kvm_set_memory_region
__kvm_set_memory_region
kvm_set_memslot(kvm, old, NULL, KVM_MR_DELETE)
kvm_invalidate_memslot
kvm_copy_memslot
kvm_replace_memslot
kvm_swap_active_memslots (A)
kvm_arch_flush_shadow_memslot (B)
same page sharing by KSM
kvm_mmu_notifier_invalidate_range_start
:
kvm_mmu_notifier_change_pte
kvm_handle_hva_range
__kvm_handle_hva_range
kvm_set_spte_gfn (C)
:
kvm_mmu_notifier_invalidate_range_end
Fix the issue by skipping the invalid memory slot at (C) to avoid the
illegal stage2 mapping so that the read request for the pflash's status
is forwarded to QEMU and emulated by it. In this way, the correct pflash's
status can be returned from QEMU to break the infinite loop in the edk2
firmware.
We tried a git-bisect and the first problematic commit is cd4c718352 ("
KVM: arm64: Convert to the gfn-based MMU notifier callbacks"). With this,
clean_dcache_guest_page() is called after the memory slots are iterated
in kvm_mmu_notifier_change_pte(). clean_dcache_guest_page() is called
before the iteration on the memory slots before this commit. This change
literally enlarges the racy window between kvm_mmu_notifier_change_pte()
and memory slot removal so that we're able to reproduce the issue in a
practical test case. However, the issue exists since commit d5d8184d35
("KVM: ARM: Memory virtualization setup").
Cc: stable@vger.kernel.org # v3.9+
Fixes: d5d8184d35 ("KVM: ARM: Memory virtualization setup")
Reported-by: Shuai Hu <hshuai@redhat.com>
Reported-by: Zhenyu Zhang <zhenyzha@redhat.com>
Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Oliver Upton <oliver.upton@linux.dev>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Shaoqin Huang <shahuang@redhat.com>
Message-Id: <20230615054259.14911-1-gshan@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 22db06337f upstream.
The addition of might_sleep() to down_timeout() caused the latter to
enable interrupts unconditionally in some cases, which in turn broke
the ACPI S3 wakeup path in acpi_suspend_enter(), where down_timeout()
is called by acpi_disable_all_gpes() via acpi_ut_acquire_mutex().
Namely, if CONFIG_DEBUG_ATOMIC_SLEEP is set, might_sleep() causes
might_resched() to be used and if CONFIG_PREEMPT_VOLUNTARY is set,
this triggers __cond_resched() which may call preempt_schedule_common(),
so __schedule() gets invoked and it ends up with enabled interrupts (in
the prev == next case).
Now, enabling interrupts early in the S3 wakeup path causes the kernel
to crash.
Address this by modifying acpi_suspend_enter() to disable GPEs without
attempting to acquire the sleeping lock which is not needed in that code
path anyway.
Fixes: 99409b935c ("locking/semaphore: Add might_sleep() to down_*() family")
Reported-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: 5.15+ <stable@vger.kernel.org> # 5.15+
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 679bd7ebdd upstream.
As a result of analysis of a syzbot report, it turned out that in three
cases where nilfs2 allocates block device buffers directly via sb_getblk,
concurrent reads to the device can corrupt the allocated buffers.
Nilfs2 uses sb_getblk for segment summary blocks, that make up a log
header, and the super root block, that is the trailer, and when moving and
writing the second super block after fs resize.
In any of these, since the uptodate flag is not set when storing metadata
to be written in the allocated buffers, the stored metadata will be
overwritten if a device read of the same block occurs concurrently before
the write. This causes metadata corruption and misbehavior in the log
write itself, causing warnings in nilfs_btree_assign() as reported.
Fix these issues by setting an uptodate flag on the buffer head on the
first or before modifying each buffer obtained with sb_getblk, and
clearing the flag on failure.
When setting the uptodate flag, the lock_buffer/unlock_buffer pair is used
to perform necessary exclusive control, and the buffer is filled to ensure
that uninitialized bytes are not mixed into the data read from others. As
for buffers for segment summary blocks, they are filled incrementally, so
if the uptodate flag was unset on their allocation, set the flag and zero
fill the buffer once at that point.
Also, regarding the superblock move routine, the starting point of the
memset call to zerofill the block is incorrectly specified, which can
cause a buffer overflow on file systems with block sizes greater than
4KiB. In addition, if the superblock is moved within a large block, it is
necessary to assume the possibility that the data in the superblock will
be destroyed by zero-filling before copying. So fix these potential
issues as well.
Link: https://lkml.kernel.org/r/20230609035732.20426-1-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Reported-by: syzbot+31837fe952932efc8fb9@syzkaller.appspotmail.com
Closes: https://lkml.kernel.org/r/00000000000030000a05e981f475@google.com
Tested-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 54abe19e00 upstream.
When commit 19343b5bdd ("mm/page-writeback: introduce tracepoint for
wait_on_page_writeback()") repurposed the writeback_dirty_page trace event
as a template to create its new wait_on_page_writeback trace event, it
ended up opening a window to NULL pointer dereference crashes due to the
(infrequent) occurrence of a race where an access to a page in the
swap-cache happens concurrently with the moment this page is being written
to disk and the tracepoint is enabled:
BUG: kernel NULL pointer dereference, address: 0000000000000040
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 800000010ec0a067 P4D 800000010ec0a067 PUD 102353067 PMD 0
Oops: 0000 [#1] PREEMPT SMP PTI
CPU: 1 PID: 1320 Comm: shmem-worker Kdump: loaded Not tainted 6.4.0-rc5+ #13
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS edk2-20230301gitf80f052277c8-1.fc37 03/01/2023
RIP: 0010:trace_event_raw_event_writeback_folio_template+0x76/0xf0
Code: 4d 85 e4 74 5c 49 8b 3c 24 e8 06 98 ee ff 48 89 c7 e8 9e 8b ee ff ba 20 00 00 00 48 89 ef 48 89 c6 e8 fe d4 1a 00 49 8b 04 24 <48> 8b 40 40 48 89 43 28 49 8b 45 20 48 89 e7 48 89 43 30 e8 a2 4d
RSP: 0000:ffffaad580b6fb60 EFLAGS: 00010246
RAX: 0000000000000000 RBX: ffff90e38035c01c RCX: 0000000000000000
RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff90e38035c044
RBP: ffff90e38035c024 R08: 0000000000000002 R09: 0000000000000006
R10: ffff90e38035c02e R11: 0000000000000020 R12: ffff90e380bac000
R13: ffffe3a7456d9200 R14: 0000000000001b81 R15: ffffe3a7456d9200
FS: 00007f2e4e8a15c0(0000) GS:ffff90e3fbc80000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000040 CR3: 00000001150c6003 CR4: 0000000000170ee0
Call Trace:
<TASK>
? __die+0x20/0x70
? page_fault_oops+0x76/0x170
? kernelmode_fixup_or_oops+0x84/0x110
? exc_page_fault+0x65/0x150
? asm_exc_page_fault+0x22/0x30
? trace_event_raw_event_writeback_folio_template+0x76/0xf0
folio_wait_writeback+0x6b/0x80
shmem_swapin_folio+0x24a/0x500
? filemap_get_entry+0xe3/0x140
shmem_get_folio_gfp+0x36e/0x7c0
? find_busiest_group+0x43/0x1a0
shmem_fault+0x76/0x2a0
? __update_load_avg_cfs_rq+0x281/0x2f0
__do_fault+0x33/0x130
do_read_fault+0x118/0x160
do_pte_missing+0x1ed/0x2a0
__handle_mm_fault+0x566/0x630
handle_mm_fault+0x91/0x210
do_user_addr_fault+0x22c/0x740
exc_page_fault+0x65/0x150
asm_exc_page_fault+0x22/0x30
This problem arises from the fact that the repurposed writeback_dirty_page
trace event code was written assuming that every pointer to mapping
(struct address_space) would come from a file-mapped page-cache object,
thus mapping->host would always be populated, and that was a valid case
before commit 19343b5bdd. The swap-cache address space
(swapper_spaces), however, doesn't populate its ->host (struct inode)
pointer, thus leading to the crashes in the corner-case aforementioned.
commit 19343b5bdd ended up breaking the assignment of __entry->name and
__entry->ino for the wait_on_page_writeback tracepoint -- both dependent
on mapping->host carrying a pointer to a valid inode. The assignment of
__entry->name was fixed by commit 68f23b8906 ("memcg: fix a crash in
wb_workfn when a device disappears"), and this commit fixes the remaining
case, for __entry->ino.
Link: https://lkml.kernel.org/r/20230606233613.1290819-1-aquini@redhat.com
Fixes: 19343b5bdd ("mm/page-writeback: introduce tracepoint for wait_on_page_writeback()")
Signed-off-by: Rafael Aquini <aquini@redhat.com>
Reviewed-by: Yafang Shao <laoar.shao@gmail.com>
Cc: Aristeu Rozanski <aris@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It looks like this test was broken in v6.1 after the backport of commit
48d73f609d ("selftests: mptcp: update userspace pm addr tests").
It was not working because the commit ad3493746e ("selftests: mptcp:
add test-cases for mixed v4/v6 subflows") is not in v6.1. This commit
changes how the connections are being created in mptcp_join.sh selftest:
with IPv6 support always enabled. But then in v6.1, the server still
create IPv4 only connections, so without the v4-mapped-v6 format with
the "::ffff:" prefix like we have in v6.3.
The modification here adds a support for connections created in v4 as
well so it fixes the issue in v6.1. This patch is not needed for the
selftests in v6.3 because only IPv6 listening sockets are being created.
Fixes: 8f0ba8ec18 ("selftests: mptcp: update userspace pm addr tests")
Cc: stable@vger.kernel.org
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 4a0b866a3f upstream
Selftests are supposed to run on any kernels, including the old ones not
supporting all MPTCP features.
Some tests are using IPTables and/or TC commands to force some
behaviours. If one of these commands fails -- likely because some
features are not available due to missing kernel config -- we should
intercept the error and skip the tests requiring these features.
Note that if we expect to have these features available and if
SELFTESTS_MPTCP_LIB_EXPECT_ALL_FEATURES env var is set to 1, the tests
will be marked as failed instead of skipped.
This patch also replaces the 'exit 1' by 'return 1' not to stop the
selftest in the middle without the conclusion if there is an issue with
NF or TC.
Link: https://github.com/multipath-tcp/mptcp_net-next/issues/368
Fixes: 8d014eaa92 ("selftests: mptcp: add ADD_ADDR timeout test case")
Cc: stable@vger.kernel.org
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit dc97251bf0 upstream.
Selftests are supposed to run on any kernels, including the old ones not
supporting all MPTCP features.
One of them is the listen diag dump support introduced by
commit 4fa39b701c ("mptcp: listen diag dump support").
It looks like there is no good pre-check to do here, i.e. dedicated
function available in kallsyms. Instead, we try to get info if nothing
is returned, the test is marked as skipped.
That's not ideal because something could be wrong with the feature and
instead of reporting an error, the test could be marked as skipped. If
we know in advanced that the feature is supposed to be supported, the
tester can set SELFTESTS_MPTCP_LIB_EXPECT_ALL_FEATURES env var to 1: in
this case the test will report an error instead of marking the test as
skipped if nothing is returned.
Link: https://github.com/multipath-tcp/mptcp_net-next/issues/368
Fixes: f2ae0fa68e ("selftests/mptcp: add diag listen tests")
Cc: stable@vger.kernel.org
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
