Certain device drivers allocate IO queues on a per-cpu basis.
On AMD EPYC platform, which can support up-to 256 cpu threads,
this can exceed the current MAX_IRQ_PER_TABLE limit of 256,
and result in the error message:
AMD-Vi: Failed to allocate IRTE
This has been observed with certain NVME devices.
AMD IOMMU hardware can actually support upto 512 interrupt
remapping table entries. Therefore, update the driver to
match the hardware limit.
Please note that this also increases the size of interrupt remapping
table to 8KB per device when using the 128-bit IRTE format.
The request may be executed asynchronously, and rq->state may be
changed to IDLE. To avoid repeated request completion, only
MQ_RQ_COMPLETE of rq->state is checked in nvme_tcp_complete_timed_out.
It is not safe, so need adding check IDLE for rq->state.
The request may be executed asynchronously, and rq->state may be
changed to IDLE. To avoid repeated request completion, only
MQ_RQ_COMPLETE of rq->state is checked in nvme_rdma_complete_timed_out.
It is not safe, so need adding check IDLE for rq->state.
Now use teardown_lock to serialize for time out and tear down. This may
cause abnormal: first cancel all request in tear down, then time out may
complete the request again, but the request may already be freed or
restarted.
To avoid race between time out and tear down, in tear down process,
first we quiesce the queue, and then delete the timer and cancel
the time out work for the queue. At the same time we need to delete
teardown_lock.
Now use teardown_lock to serialize for time out and tear down. This may
cause abnormal: first cancel all request in tear down, then time out may
complete the request again, but the request may already be freed or
restarted.
To avoid race between time out and tear down, in tear down process,
first we quiesce the queue, and then delete the timer and cancel
the time out work for the queue. At the same time we need to delete
teardown_lock.
While reenabling the IRQ after irq poll there may be small time window
where HBA firmware has posted some replies and raise the interrupts but
driver has not received the interrupts. So we may observe I/O timeouts as
the driver has not processed the replies as interrupts got missed while
reenabling the IRQ.
To fix this issue the driver has to go for one more round of processing the
reply descriptors from reply descriptor post queue after enabling the IRQ.
Link: https://lore.kernel.org/r/20201102072746.27410-1-sreekanth.reddy@broadcom.com Reported-by: Tomas Henzl <thenzl@redhat.com> Reviewed-by: Tomas Henzl <thenzl@redhat.com> Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
alua_bus_detach() might be running concurrently with alua_rtpg_work(), so
we might trip over h->sdev == NULL and call BUG_ON(). The correct way of
handling it is to not set h->sdev to NULL in alua_bus_detach(), and call
rcu_synchronize() before the final delete to ensure that all concurrent
threads have left the critical section. Then we can get rid of the
BUG_ON() and replace it with a simple if condition.
ARC HSDK platform stopped booting on released v5.10-rc1, getting stuck
in startup of non master SMP cores.
This was bisected to upstream commit 7fef431be9c9ac25
"(mm/page_alloc: place pages to tail in __free_pages_core())"
That commit itself is harmless, it just exposed a subtle assumption in
our platform code (hence CC'ing linux-mm just as FYI in case some other
arches / platforms trip on it).
The upstream commit is semantically disruptive as it reverses the order
of page allocations (actually it can be good test for hardware
verification to exercise different memory patterns altogether).
For ARC HSDK platform that meant a remapped memory region (pertaining to
unused Closely Coupled Memory) started getting used early for dynamice
allocations, while not effectively remapped on all the cores, triggering
memory error exception on those cores.
The fix is to move the CCM remapping from early platform code to to early core
boot code. And while it is undesirable to riddle common boot code with
platform quirks, there is no other way to do this since the faltering code
involves setting up stack itself so even function calls are not allowed at
that point.
If anyone is interested, all the gory details can be found at Link below.
Multiple CPUs may be mapped to the same hctx, allowing mulitple
submission contexts to attempt commit_rqs(). We need to verify we're
not writing the same doorbell value multiple times since that's a spec
violation.
If netfilter changes the packet mark, the packet is rerouted. The
ip_route_me_harder family of functions fails to use the right sk, opting
to instead use skb->sk, resulting in a routing loop when used with
tunnels. With the next change fixing this issue in netfilter, test for
the relevant condition inside our test suite, since wireguard was where
the bug was discovered.
Reported-by: Chen Minqiang <ptpt52@gmail.com> Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
There's a race condition in the netdev registration in that
NETDEV_REGISTER actually happens after the netdev is available,
and so if we initialize things only there, we might get called
with an uninitialized wdev through nl80211 - not using a wdev
but using a netdev interface index.
I found this while looking into a syzbot report, but it doesn't
really seem to be related, and unfortunately there's no repro
for it (yet). I can't (yet) explain how it managed to get into
cfg80211_release_pmsr() from nl80211_netlink_notify() without
the wdev having been initialized, as the latter only iterates
the wdevs that are linked into the rdev, which even without the
change here happened after init.
However, looking at this, it seems fairly clear that the init
needs to be done earlier, otherwise we might even re-init on a
netns move, when data might still be pending.
When ieee80211_skb_resize() is called from ieee80211_build_hdr()
the skb has no 802.11 header yet, in fact it consist only of the
payload as the ethernet frame is removed. As such, we're using
the payload data for ieee80211_is_mgmt(), which is of course
completely wrong. This didn't really hurt us because these are
always data frames, so we could only have added more tailroom
than we needed if we determined it was a management frame and
sdata->crypto_tx_tailroom_needed_cnt was false.
However, syzbot found that of course there need not be any payload,
so we're using at best uninitialized memory for the check.
Fix this to pass explicitly the kind of frame that we have instead
of checking there, by replacing the "bool may_encrypt" argument
with an argument that can carry the three possible states - it's
not going to be encrypted, it's a management frame, or it's a data
frame (and then we check sdata->crypto_tx_tailroom_needed_cnt).
This reverts commit f87812284172a9809820d10143b573d833cd3f75 ("drm/amdgpu:
Fix bug where DPM is not enabled after hibernate and resume").
It was intended to fix Hawaii S4(hibernation) issue but break S3. As
ixFEATURE_STATUS is filled with garbage data on resume which can be
only cleared by reloading smc firmware(but that will involve many
changes). So, we will revert this S4 fix and seek a new way.
Signed-off-by: Evan Quan <evan.quan@amd.com> Tested-by: Sandeep Raghuraman <sandy.8925@gmail.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Before this patch, gfs2_fitrim was not properly checking for a "live" file
system. If the file system had something to trim and the file system
was read-only (or spectator) it would start the trim, but when it starts
the transaction, gfs2_trans_begin returns -EROFS (read-only file system)
and it errors out. However, if the file system was already trimmed so
there's no work to do, it never called gfs2_trans_begin. That code is
bypassed so it never returns the error. Instead, it returns a good
return code with 0 work. All this makes for inconsistent behavior:
The same fstrim command can return -EROFS in one case and 0 in another.
This tripped up xfstests generic/537 which reports the error as:
+fstrim with unrecovered metadata just ate your filesystem
This patch adds a check for a "live" (iow, active journal, iow, RW)
file system, and if not, returns the error properly.
Signed-off-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Gfs2 creates an address space for its rgrps called sd_aspace, but it never
called truncate_inode_pages_final on it. This confused vfs greatly which
tried to reference the address space after gfs2 had freed the superblock
that contained it.
This patch adds a call to truncate_inode_pages_final for sd_aspace, thus
avoiding the use-after-free.
Signed-off-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Function gfs2_clear_rgrpd calls kfree(rgd->rd_bits) before calling
return_all_reservations, but return_all_reservations still dereferences
rgd->rd_bits in __rs_deltree. Fix that by moving the call to kfree below the
call to return_all_reservations.
Signed-off-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The broken jack detection should be fixed by commit a6e7d0a4bdb0 ("ALSA:
hda: fix jack detection with Realtek codecs when in D3"), let's try
enabling runtime PM by default again.
Both pm_runtime_force_suspend() and pm_runtime_force_resume() have
some implicit checks, so it can make code flow more straightforward if
we separate runtime and system suspend callbacks.
High Definition Audio Specification, 4.5.9.3 Codec Wake From System S3
states that codec can wake the system up from S3 if WAKEEN is toggled.
Since HDA controller has different wakeup settings for runtime and
system susend, we also need to explicitly disable direct-complete which
can be enabled automatically by PCI core. In addition to that, avoid
waking up codec if runtime resume is for system suspend, to not break
direct-complete for codecs.
While at it, also remove AZX_DCAPS_SUSPEND_SPURIOUS_WAKEUP, as the
original bug commit a6630529aecb ("ALSA: hda: Workaround for spurious
wakeups on some Intel platforms") solves doesn't happen with this
patch.
Drop unneeded <linux/wait.h> header inclusion to fix pidfd compilation
errors seen in Fedora 32:
In file included from pidfd_open_test.c:9:
../../../../usr/include/linux/wait.h:17:16: error: expected identifier before numeric constant
17 | #define P_ALL 0
| ^
Signed-off-by: Tommi Rantala <tommi.t.rantala@nokia.com> Reviewed-by: Kees Cook <keescook@chromium.org> Acked-by: Christian Brauner <christian.brauner@ubuntu.com> Signed-off-by: Shuah Khan <skhan@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
More recent libc implementations are now using openat/openat2 system
calls so also add do_sys_openat2 to the tracing so that the test
passes on these systems because do_sys_open may not be called.
Thanks to Masami Hiramatsu for the help on getting this fix to work
correctly.
Signed-off-by: Colin Ian King <colin.king@canonical.com> Acked-by: Masami Hiramatsu <mhiramat@kernel.org> Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by: Shuah Khan <skhan@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When fetch 'event' from event queue, after copy its address
space content to user space, the 'event' the memory space
pointed to by the 'event' pointer need be freed.
goku_probe() goes to error label "err" and invokes goku_remove()
in case of failures of pci_enable_device(), pci_resource_start()
and ioremap(). goku_remove() gets a device from
pci_get_drvdata(pdev) and works with it without any checks, in
particular it dereferences a corresponding pointer. But
goku_probe() did not set this device yet. So, one can expect
various crashes. The patch moves setting the device just after
allocation of memory for it.
Found by Linux Driver Verification project (linuxtesting.org).
Reported-by: Pavel Andrianov <andrianov@ispras.ru> Signed-off-by: Evgeny Novikov <novikov@ispras.ru> Signed-off-by: Felipe Balbi <balbi@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
There is a lot of stuff here which can be done outside of the big
opp_table_lock, do that. This helps avoiding few circular dependency
lockdeps around debugfs and interconnects.
Reported-by: Rob Clark <robdclark@gmail.com> Reported-by: Dmitry Osipenko <digetx@gmail.com> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
A delay must be introduced before the shutdown down of the mclk,
as stated in CS42L51 datasheet. Otherwise the codec may
produce some noise after the end of DAPM power down sequence.
The delay between DAC and CLOCK_SUPPLY widgets is too short.
Add a delay in mclk shutdown request to manage the shutdown delay
explicitly. From experiments, at least 10ms delay is necessary.
Set delay to 20ms as recommended in Documentation/timers/timers-howto.rst
when using msleep().
This is not very UCM friendly both w.r.t to common up configs and
card identification, and UCM2 became totally not usefull with just
one ucm sdm845.conf for two machines which have different setups
w.r.t HDMI and other dais.
Reasons for such thing is partly because Qualcomm machine drivers never
cared to set driver_name.
This patch sets up driver name for the this driver to sort out the
UCM integration issues!
with this its possible to align with what UCM2 expects and we can have
sdm845/DB845.conf
sdm845/LENOVO-81JL-LenovoYOGAC630_13Q50-LNVNB161216.conf
... for board variants. This should scale much better!
According to the SMCCC spec[1](7.5.2 Discovery) the
ARM_SMCCC_ARCH_WORKAROUND_1 function id only returns 0, 1, and
SMCCC_RET_NOT_SUPPORTED.
0 is "workaround required and safe to call this function"
1 is "workaround not required but safe to call this function"
SMCCC_RET_NOT_SUPPORTED is "might be vulnerable or might not be, who knows, I give up!"
SMCCC_RET_NOT_SUPPORTED might as well mean "workaround required, except
calling this function may not work because it isn't implemented in some
cases". Wonderful. We map this SMC call to
0 is SPECTRE_MITIGATED
1 is SPECTRE_UNAFFECTED
SMCCC_RET_NOT_SUPPORTED is SPECTRE_VULNERABLE
For KVM hypercalls (hvc), we've implemented this function id to return
SMCCC_RET_NOT_SUPPORTED, 0, and SMCCC_RET_NOT_REQUIRED. One of those
isn't supposed to be there. Per the code we call
arm64_get_spectre_v2_state() to figure out what to return for this
feature discovery call.
0 is SPECTRE_MITIGATED
SMCCC_RET_NOT_REQUIRED is SPECTRE_UNAFFECTED
SMCCC_RET_NOT_SUPPORTED is SPECTRE_VULNERABLE
Let's clean this up so that KVM tells the guest this mapping:
0 is SPECTRE_MITIGATED
1 is SPECTRE_UNAFFECTED
SMCCC_RET_NOT_SUPPORTED is SPECTRE_VULNERABLE
Note: SMCCC_RET_NOT_AFFECTED is 1 but isn't part of the SMCCC spec
Fixes: c118bbb52743 ("arm64: KVM: Propagate full Spectre v2 workaround state to KVM guests") Signed-off-by: Stephen Boyd <swboyd@chromium.org> Acked-by: Marc Zyngier <maz@kernel.org> Acked-by: Will Deacon <will@kernel.org> Cc: Andre Przywara <andre.przywara@arm.com> Cc: Steven Price <steven.price@arm.com> Cc: Marc Zyngier <maz@kernel.org> Cc: stable@vger.kernel.org Link: https://developer.arm.com/documentation/den0028/latest Link: https://lore.kernel.org/r/20201023154751.1973872-1-swboyd@chromium.org Signed-off-by: Will Deacon <will@kernel.org> Signed-off-by: Stephen Boyd <swboyd@chromium.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Mimic the pre-existing ACPI and Device Tree event log behavior by not
creating the binary_bios_measurements file when the EFI TPM event log is
empty.
This fixes the following NULL pointer dereference that can occur when
reading /sys/kernel/security/tpm0/binary_bios_measurements after the
kernel received an empty event log from the firmware:
In this situation, the bios_event_log pointer in the tpm_bios_log struct
was not NULL but was equal to the ZERO_SIZE_PTR (0x10) value. This was
due to the following kmemdup() in tpm_read_log_efi():
int tpm_read_log_efi(struct tpm_chip *chip)
{
...
/* malloc EventLog space */
log->bios_event_log = kmemdup(log_tbl->log, log_size, GFP_KERNEL);
if (!log->bios_event_log) {
ret = -ENOMEM;
goto out;
}
...
}
When log_size is zero, due to an empty event log from firmware,
ZERO_SIZE_PTR is returned from kmemdup(). Upon a read of the
binary_bios_measurements file, the tpm2_bios_measurements_start()
function does not perform a ZERO_OR_NULL_PTR() check on the
bios_event_log pointer before dereferencing it.
Rather than add a ZERO_OR_NULL_PTR() check in functions that make use of
the bios_event_log pointer, simply avoid creating the
binary_bios_measurements_file as is done in other event log retrieval
backends.
Explicitly ignore all of the events in the final event log when the main
event log is empty. The list of events in the final event log cannot be
accurately parsed without referring to the first event in the main event
log (the event log header) so the final event log is useless in such a
situation.
Fixes: 58cc1e4faf10 ("tpm: parse TPM event logs based on EFI table") Link: https://lore.kernel.org/linux-integrity/E1FDCCCB-CA51-4AEE-AC83-9CDE995EAE52@canonical.com/ Reported-by: Kai-Heng Feng <kai.heng.feng@canonical.com> Reported-by: Kenneth R. Crudup <kenny@panix.com> Reported-by: Mimi Zohar <zohar@linux.ibm.com> Cc: Thiébaud Weksteen <tweek@google.com> Cc: Ard Biesheuvel <ardb@kernel.org> Signed-off-by: Tyler Hicks <tyhicks@linux.microsoft.com> Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
pm_runtime_get_sync() will increment pm usage at first and it
will resume the device later. If runtime of the device has
error or device is in inaccessible state(or other error state),
resume operation will fail. If we do not call put operation to
decrease the reference, the result is that this device cannot
enter the idle state and always stay busy or other non-idle
state.
Fixes: 249fa8217b846 ("USB: Add driver to control USB fast charge for iOS devices") Signed-off-by: Zhang Qilong <zhangqilong3@huawei.com> Link: https://lore.kernel.org/r/20201102022650.67115-1-zhangqilong3@huawei.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
We were relying on GNU ld's ability to re-link executable files in order
to extract our VDSO symbols. This behavior was deemed a bug as of
binutils-2.35 (specifically the binutils-gdb commit a87e1817a4 ("Have
the linker fail if any attempt to link in an executable is made."), but
as that has been backported to at least Debian's binutils-2.34 in may
manifest in other places.
The previous version of this was a bit of a mess: we were linking a
static executable version of the VDSO, containing only a subset of the
input symbols, which we then linked into the kernel. This worked, but
certainly wasn't a supported path through the toolchain. Instead this
new version parses the textual output of nm to produce a symbol table.
Both rely on near-zero addresses being linkable, but as we rely on weak
undefined symbols being linkable elsewhere I don't view this as a major
issue.
ld's --build-id defaults to "sha1" style, while lld defaults to "fast".
The build IDs are very different between the two, which may confuse
programs that reference them.
Signed-off-by: Bill Wendling <morbo@google.com> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
condition leads to a return before the task flag is set. Similarly,
ib_prctl_get() will return PR_SPEC_DISABLE even though IBPB is set to
conditional.
More generally, the following cases are possible:
1. STIBP = conditional && IBPB = on for spectre_v2_user=seccomp,ibpb
2. STIBP = on && IBPB = conditional for AMD CPUs with
X86_FEATURE_AMD_STIBP_ALWAYS_ON
The first case functions correctly today, but only because
spectre_v2_user_ibpb isn't updated to reflect the IBPB mode.
At a high level, this change does one thing. If either STIBP or IBPB
is set to conditional, allow the prctl to change the task flag.
Also, reflect that capability when querying the state. This isn't
perfect since it doesn't take into account if only STIBP or IBPB is
unconditionally on. But it allows the conditional feature to work as
expected, without affecting the unconditional one.
[ bp: Massage commit message and comment; space out statements for
better readability. ]
Fixes: 21998a351512 ("x86/speculation: Avoid force-disabling IBPB based on STIBP and enhanced IBRS.") Signed-off-by: Anand K Mistry <amistry@google.com> Signed-off-by: Borislav Petkov <bp@suse.de> Acked-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Tom Lendacky <thomas.lendacky@amd.com> Link: https://lkml.kernel.org/r/20201105163246.v2.1.Ifd7243cd3e2c2206a893ad0a5b9a4f19549e22c6@changeid Signed-off-by: Sasha Levin <sashal@kernel.org>
Some messages sent by the MDS entail a session sequence number
increment, and the MDS will drop certain types of requests on the floor
when the sequence numbers don't match.
In particular, a REQUEST_CLOSE message can cross with one of the
sequence morphing messages from the MDS which can cause the client to
stall, waiting for a response that will never come.
Originally, this meant an up to 5s delay before the recurring workqueue
job kicked in and resent the request, but a recent change made it so
that the client would never resend, causing a 60s stall unmounting and
sometimes a blockisting event.
Add a new helper for incrementing the session sequence and then testing
to see whether a REQUEST_CLOSE needs to be resent, and move the handling
of CEPH_MDS_SESSION_CLOSING into that function. Change all of the
bare sequence counter increments to use the new helper.
Reorganize check_session_state with a switch statement. It should no
longer be called when the session is CLOSING, so throw a warning if it
ever is (but still handle that case sanely).
With commit 669cbc708122 ("PCI: Move DT resource setup into
devm_pci_alloc_host_bridge()"), the DT 'ranges' is parsed and populated
into resources when the host bridge is allocated. The resources are
requested as well, but that happens a second time for the mvebu driver in
mvebu_pcie_parse_request_resources(). We should only be requesting the
additional resources added in mvebu_pcie_parse_request_resources(). These
are not added by default because they use custom properties rather than
standard DT address translation.
Also, the bus ranges was also populated by default, so we can remove it
from mvebu_pci_host_probe().
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=209729 Fixes: 669cbc708122 ("PCI: Move DT resource setup into devm_pci_alloc_host_bridge()") Link: https://lore.kernel.org/r/20201023145252.2691779-1-robh@kernel.org Reported-by: vtolkm@googlemail.com Tested-by: Jan Kundrát <jan.kundrat@cesnet.cz> Signed-off-by: Rob Herring <robh@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Russell King <linux@armlinux.org.uk> Signed-off-by: Sasha Levin <sashal@kernel.org>
Since commit 530b5affc675 ("spi: fsl-dspi: fix use-after-free in
remove path"), this driver causes a "NULL pointer dereference"
in dspi_suspend/resume.
This is because since this commit, the drivers private data point to
"dspi" instead of "ctlr", the codes in suspend and resume func were
not modified correspondly.
Fixes: 530b5affc675 ("spi: fsl-dspi: fix use-after-free in remove path") Signed-off-by: Zhao Qiang <qiang.zhao@nxp.com> Reviewed-by: Vladimir Oltean <olteanv@gmail.com> Link: https://lore.kernel.org/r/20201103020546.1822-1-qiang.zhao@nxp.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
In commit 7588cbeec6df, we tried to fix a race stemming from the lack of
coordination between higher level code that wants to allocate and remap
CoW fork extents into the data fork. Christoph cites as examples the
always_cow mode, and a directio write completion racing with writeback.
According to the comments before the goto retry, we want to restart the
lookup to catch the extent in the data fork, but we don't actually reset
whichfork or cow_fsb, which means the second try executes using stale
information. Up until now I think we've gotten lucky that either
there's something left in the CoW fork to cause cow_fsb to be reset, or
either data/cow fork sequence numbers have advanced enough to force a
fresh lookup from the data fork. However, if we reach the retry with an
empty stable CoW fork and a stable data fork, neither of those things
happens. The retry foolishly re-calls xfs_convert_blocks on the CoW
fork which fails again. This time, we toss the write.
I've recently been working on extending reflink to the realtime device.
When the realtime extent size is larger than a single block, we have to
force the page cache to CoW the entire rt extent if a write (or
fallocate) are not aligned with the rt extent size. The strategy I've
chosen to deal with this is derived from Dave's blocksize > pagesize
series: dirtying around the write range, and ensuring that writeback
always starts mapping on an rt extent boundary. This has brought this
race front and center, since generic/522 blows up immediately.
However, I'm pretty sure this is a bug outright, independent of that.
Fixes: 7588cbeec6df ("xfs: retry COW fork delalloc conversion when no extent was found") Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
It is possible to expose non-zeroed post-EOF data in XFS if the new
EOF page is dirty, backed by an unwritten block and the truncate
happens to race with writeback. iomap_truncate_page() will not zero
the post-EOF portion of the page if the underlying block is
unwritten. The subsequent call to truncate_setsize() will, but
doesn't dirty the page. Therefore, if writeback happens to complete
after iomap_truncate_page() (so it still sees the unwritten block)
but before truncate_setsize(), the cached page becomes inconsistent
with the on-disk block. A mapped read after the associated page is
reclaimed or invalidated exposes non-zero post-EOF data.
For example, consider the following sequence when run on a kernel
modified to explicitly flush the new EOF page within the race
window:
$ xfs_io -fc "falloc 0 4k" -c fsync /mnt/file
$ xfs_io -c "pwrite 0 4k" -c "truncate 1k" /mnt/file
...
$ xfs_io -c "mmap 0 4k" -c "mread -v 1k 8" /mnt/file 00000400: 00 00 00 00 00 00 00 00 ........
$ umount /mnt/; mount <dev> /mnt/
$ xfs_io -c "mmap 0 4k" -c "mread -v 1k 8" /mnt/file 00000400: cd cd cd cd cd cd cd cd ........
Update xfs_setattr_size() to explicitly flush the new EOF page prior
to the page truncate to ensure iomap has the latest state of the
underlying block.
Fixes: 68a9f5e7007c ("xfs: implement iomap based buffered write path") Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
With below sequence, we can see wakeup default is enabled after re-load module,
if it was enabled before, so we need disable wakeup in flexcan_remove().
After double check with Layerscape CAN owner (Pankaj Bansal), confirm that
LS1021A doesn't support ECC feature, so remove FLEXCAN_QUIRK_DISABLE_MECR
quirk.
Fixes: 99b7668c04b27 ("can: flexcan: adding platform specific details for LS1021A") Cc: Pankaj Bansal <pankaj.bansal@nxp.com> Signed-off-by: Joakim Zhang <qiangqing.zhang@nxp.com> Link: https://lore.kernel.org/r/20201020155402.30318-4-qiangqing.zhang@nxp.com Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Echo management is driven by PUCAN_MSG_LOOPED_BACK bit, while loopback
frames are identified with PUCAN_MSG_SELF_RECEIVE bit. Those bits are set
for each outgoing frame written to the IP core so that a copy of each one
will be placed into the rx path. Thus,
- when PUCAN_MSG_LOOPED_BACK is set then the rx frame is an echo of a
previously sent frame,
- when PUCAN_MSG_LOOPED_BACK+PUCAN_MSG_SELF_RECEIVE are set, then the rx
frame is an echo AND a loopback frame. Therefore, this frame must be
put into the socket rx path too.
This patch fixes how CAN frames are handled when these are sent while the
can interface is configured in "loopback on" mode.
Signed-off-by: Stephane Grosjean <s.grosjean@peak-system.com> Link: https://lore.kernel.org/r/20201013153947.28012-1-s.grosjean@peak-system.com Fixes: 8ac8321e4a79 ("can: peak: add support for PEAK PCAN-PCIe FD CAN-FD boards") Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Fabian Inostroza <fabianinostrozap@gmail.com> has discovered a potential
problem in the hardware timestamp reporting from the PCAN-USB USB CAN interface
(only), related to the fact that a timestamp of an event may precede the
timestamp used for synchronization when both records are part of the same USB
packet. However, this case was used to detect the wrapping of the time counter.
This patch details and fixes the two identified cases where this problem can
occur.
These values come from skb->data so Smatch considers them untrusted. I
believe Smatch is correct but I don't have a way to test this.
The usb_if->dev[] array has 2 elements but the index is in the 0-15
range without checks. The cfd->len can be up to 255 but the maximum
valid size is CANFD_MAX_DLEN (64) so that could lead to memory
corruption.
Fixes: 0a25e1f4f185 ("can: peak_usb: add support for PEAK new CANFD USB adapters") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Link: https://lore.kernel.org/r/20200813140604.GA456946@mwanda Acked-by: Stephane Grosjean <s.grosjean@peak-system.com> Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Calling pm_runtime_get_sync increments the counter even in case of
failure, causing incorrect ref count. Call pm_runtime_put if
pm_runtime_get_sync fails.
When a netdev down event occurs after a successful call to
j1939_sk_bind(), j1939_netdev_notify() can handle it correctly.
But if the netdev already in down state before calling j1939_sk_bind(),
j1939_sk_release() will stay in wait_event_interruptible() blocked
forever. Because in this case, j1939_netdev_notify() won't be called and
j1939_tp_txtimer() won't call j1939_session_cancel() or other function
to clear session for ENETDOWN error, this lead to mismatch of
j1939_session_get/put() and jsk->skb_pending will never decrease to
zero.
To reproduce it use following commands:
1. ip link add dev vcan0 type vcan
2. j1939acd -r 100,80-120 1122334455667788 vcan0
3. presses ctrl-c and thread will be blocked forever
This patch adds check for ndev->flags in j1939_sk_bind() to avoid this
kind of situation and return with -ENETDOWN.
Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol") Signed-off-by: Zhang Changzhong <zhangchangzhong@huawei.com> Link: https://lore.kernel.org/r/1599460308-18770-1-git-send-email-zhangchangzhong@huawei.com Acked-by: Oleksij Rempel <o.rempel@pengutronix.de> Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
All user space generated SKBs are owned by a socket (unless injected into the
key via AF_PACKET). If a socket is closed, all associated skbs will be cleaned
up.
This leads to a problem when a CAN driver calls can_put_echo_skb() on a
unshared SKB. If the socket is closed prior to the TX complete handler,
can_get_echo_skb() and the subsequent delivering of the echo SKB to all
registered callbacks, a SKB with a refcount of 0 is delivered.
To avoid the problem, in can_get_echo_skb() the original SKB is now always
cloned, regardless of shared SKB or not. If the process exists it can now
safely discard its SKBs, without disturbing the delivery of the echo SKB.
The problem shows up in the j1939 stack, when it clones the incoming skb, which
detects the already 0 refcount.
We can easily reproduce this with following example:
The can_get_echo_skb() function returns the number of received bytes to
be used for netdev statistics. In the case of RTR frames we get a valid
(potential non-zero) data length value which has to be passed for further
operations. But on the wire RTR frames have no payload length. Therefore
the value to be used in the statistics has to be zero for RTR frames.
Reported-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr> Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Link: https://lore.kernel.org/r/20201020064443.80164-1-socketcan@hartkopp.net Fixes: cf5046b309b3 ("can: dev: let can_get_echo_skb() return dlc of CAN frame") Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
If a driver calls can_get_echo_skb() during a hardware IRQ (which is often, but
not always, the case), the 'WARN_ON(in_irq)' in
net/core/skbuff.c#skb_release_head_state() might be triggered, under network
congestion circumstances, together with the potential risk of a NULL pointer
dereference.
The root cause of this issue is the call to kfree_skb() instead of
dev_kfree_skb_irq() in net/core/dev.c#enqueue_to_backlog().
This patch prevents the skb to be freed within the call to netif_rx() by
incrementing its reference count with skb_get(). The skb is finally freed by
one of the in-irq-context safe functions: dev_consume_skb_any() or
dev_kfree_skb_any(). The "any" version is used because some drivers might call
can_get_echo_skb() in a normal context.
The reason for this issue to occur is that initially, in the core network
stack, loopback skb were not supposed to be received in hardware IRQ context.
The CAN stack is an exeption.
This bug was previously reported back in 2017 in [1] but the proposed patch
never got accepted.
While [1] directly modifies net/core/dev.c, we try to propose here a
smoother modification local to CAN network stack (the assumption
behind is that only CAN devices are affected by this issue).
A CAN driver, using the rx-offload infrastructure, is reading CAN frames
(usually in IRQ context) from the hardware and placing it into the rx-offload
queue to be delivered to the networking stack via NAPI.
In case the rx-offload queue is full, trying to add more skbs results in the
skbs being dropped using kfree_skb(). If done from hard-IRQ context this
results in the following warning:
[ 682.552693] ------------[ cut here ]------------
[ 682.557360] WARNING: CPU: 0 PID: 3057 at net/core/skbuff.c:650 skb_release_head_state+0x74/0x84
[ 682.566075] Modules linked in: can_raw can coda_vpu flexcan dw_hdmi_ahb_audio v4l2_jpeg imx_vdoa can_dev
[ 682.575597] CPU: 0 PID: 3057 Comm: cansend Tainted: G W 5.7.0+ #18
[ 682.583098] Hardware name: Freescale i.MX6 Quad/DualLite (Device Tree)
[ 682.589657] [<c0112628>] (unwind_backtrace) from [<c010c1c4>] (show_stack+0x10/0x14)
[ 682.597423] [<c010c1c4>] (show_stack) from [<c06c481c>] (dump_stack+0xe0/0x114)
[ 682.604759] [<c06c481c>] (dump_stack) from [<c0128f10>] (__warn+0xc0/0x10c)
[ 682.611742] [<c0128f10>] (__warn) from [<c0129314>] (warn_slowpath_fmt+0x5c/0xc0)
[ 682.619248] [<c0129314>] (warn_slowpath_fmt) from [<c0b95dec>] (skb_release_head_state+0x74/0x84)
[ 682.628143] [<c0b95dec>] (skb_release_head_state) from [<c0b95e08>] (skb_release_all+0xc/0x24)
[ 682.636774] [<c0b95e08>] (skb_release_all) from [<c0b95eac>] (kfree_skb+0x74/0x1c8)
[ 682.644479] [<c0b95eac>] (kfree_skb) from [<bf001d1c>] (can_rx_offload_queue_sorted+0xe0/0xe8 [can_dev])
[ 682.654051] [<bf001d1c>] (can_rx_offload_queue_sorted [can_dev]) from [<bf001d6c>] (can_rx_offload_get_echo_skb+0x48/0x94 [can_dev])
[ 682.666007] [<bf001d6c>] (can_rx_offload_get_echo_skb [can_dev]) from [<bf01efe4>] (flexcan_irq+0x194/0x5dc [flexcan])
[ 682.676734] [<bf01efe4>] (flexcan_irq [flexcan]) from [<c019c1ec>] (__handle_irq_event_percpu+0x4c/0x3ec)
[ 682.686322] [<c019c1ec>] (__handle_irq_event_percpu) from [<c019c5b8>] (handle_irq_event_percpu+0x2c/0x88)
[ 682.695993] [<c019c5b8>] (handle_irq_event_percpu) from [<c019c64c>] (handle_irq_event+0x38/0x5c)
[ 682.704887] [<c019c64c>] (handle_irq_event) from [<c01a1058>] (handle_fasteoi_irq+0xc8/0x180)
[ 682.713432] [<c01a1058>] (handle_fasteoi_irq) from [<c019b2c0>] (generic_handle_irq+0x30/0x44)
[ 682.722063] [<c019b2c0>] (generic_handle_irq) from [<c019b8f8>] (__handle_domain_irq+0x64/0xdc)
[ 682.730783] [<c019b8f8>] (__handle_domain_irq) from [<c06df4a4>] (gic_handle_irq+0x48/0x9c)
[ 682.739158] [<c06df4a4>] (gic_handle_irq) from [<c0100b30>] (__irq_svc+0x70/0x98)
[ 682.746656] Exception stack(0xe80e9dd8 to 0xe80e9e20)
[ 682.751725] 9dc0: 00000001e80e8000
[ 682.759922] 9de0: e820cf8000000000ffffe00000000000eaf08fe400000000600d001300000000
[ 682.768117] 9e00: c1732e3cc16093a8e820d4c0e80e9e28c018a57cc018b870600d0013ffffffff
[ 682.776315] [<c0100b30>] (__irq_svc) from [<c018b870>] (lock_acquire+0x108/0x4e8)
[ 682.783821] [<c018b870>] (lock_acquire) from [<c0e938e4>] (down_write+0x48/0xa8)
[ 682.791242] [<c0e938e4>] (down_write) from [<c02818dc>] (unlink_file_vma+0x24/0x40)
[ 682.798922] [<c02818dc>] (unlink_file_vma) from [<c027a258>] (free_pgtables+0x34/0xb8)
[ 682.806858] [<c027a258>] (free_pgtables) from [<c02835a4>] (exit_mmap+0xe4/0x170)
[ 682.814361] [<c02835a4>] (exit_mmap) from [<c01248e0>] (mmput+0x5c/0x110)
[ 682.821171] [<c01248e0>] (mmput) from [<c012e910>] (do_exit+0x374/0xbe4)
[ 682.827892] [<c012e910>] (do_exit) from [<c0130888>] (do_group_exit+0x38/0xb4)
[ 682.835132] [<c0130888>] (do_group_exit) from [<c0130914>] (__wake_up_parent+0x0/0x14)
[ 682.843063] irq event stamp: 1936
[ 682.846399] hardirqs last enabled at (1935): [<c02938b0>] rmqueue+0xf4/0xc64
[ 682.853553] hardirqs last disabled at (1936): [<c0100b20>] __irq_svc+0x60/0x98
[ 682.860799] softirqs last enabled at (1878): [<bf04cdcc>] raw_release+0x108/0x1f0 [can_raw]
[ 682.869256] softirqs last disabled at (1876): [<c0b8f478>] release_sock+0x18/0x98
[ 682.876753] ---[ end trace 7bca4751ce44c444 ]---
This patch fixes the problem by replacing the kfree_skb() by
dev_kfree_skb_any(), as rx-offload might be called from threaded IRQ handlers
as well.
The ioeventfd is called under spinlock with interrupts disabled,
therefore if the memory lock is contended defer code that might
sleep to a thread context.
Fixes: bc93b9ae0151 ("vfio-pci: Avoid recursive read-lock usage") Link: https://bugzilla.kernel.org/show_bug.cgi?id=209253#c1 Reported-by: Ian Pilcher <arequipeno@gmail.com> Tested-by: Ian Pilcher <arequipeno@gmail.com> Tested-by: Justin Gatzen <justin.gatzen@gmail.com> Signed-off-by: Alex Williamson <alex.williamson@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The cleanup for the yfs_store_opaque_acl2_operation calls the wrong
function to destroy the ACL content buffer. It's an afs_acl struct, not
a yfs_acl struct - and the free function for latter may pass invalid
pointers to kfree().
Fix this by using the afs_acl_put() function. The yfs_acl_put()
function is then no longer used and can be removed.
When using the afs.yfs.acl xattr to change an AuriStor ACL, a warning
can be generated when the request is marshalled because the buffer
pointer isn't increased after adding the last element, thereby
triggering the check at the end if the ACL wasn't empty. This just
causes something like the following warning, but doesn't stop the call
from happening successfully:
In prq_event_thread(), the QI_PGRP_PDP is wrongly set by
'req->pasid_present' which should be replaced to
'req->priv_data_present'.
Fixes: 5b438f4ba315 ("iommu/vt-d: Support page request in scalable mode") Signed-off-by: Liu, Yi L <yi.l.liu@intel.com> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com> Acked-by: Lu Baolu <baolu.lu@linux.intel.com> Link: https://lore.kernel.org/r/1604025444-6954-3-git-send-email-yi.y.sun@linux.intel.com Signed-off-by: Joerg Roedel <jroedel@suse.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Should get correct sid and set it into sdev. Because we execute
'sdev->sid != req->rid' in the loop of prq_event_thread().
Fixes: eb8d93ea3c1d ("iommu/vt-d: Report page request faults for guest SVA") Signed-off-by: Liu Yi L <yi.l.liu@intel.com> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com> Acked-by: Lu Baolu <baolu.lu@linux.intel.com> Link: https://lore.kernel.org/r/1604025444-6954-2-git-send-email-yi.y.sun@linux.intel.com Signed-off-by: Joerg Roedel <jroedel@suse.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
When unloading the call to pm_runtime_put_sync_suspend() will attempt to
turn the GPU cores off, however panfrost_device_fini() will have turned
the clocks off. This leads to the hardware locking up.
Instead don't call pm_runtime_put_sync_suspend() and instead simply mark
the device as suspended using pm_runtime_set_suspended(). And also
include this on the error path in panfrost_probe().
Fixes: aebe8c22a912 ("drm/panfrost: Fix possible suspend in panfrost_remove") Signed-off-by: Steven Price <steven.price@arm.com> Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com> Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com> Link: https://patchwork.freedesktop.org/patch/msgid/20201030145833.29006-1-steven.price@arm.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Later we will introduce devfreq probing regulator if they
are present. As regulator should be probe only one time we
need to get this logic in the device_init().
panfrost_device is already taking care of devfreq_resume()
and devfreq_suspend(), so it's not totally illogic to move
the devfreq_init() and devfreq_fini() here.
memcg_page_state will get the specified number in hierarchical memcg, It
should multiply by HPAGE_PMD_NR rather than an page if the item is
NR_ANON_THPS.
[akpm@linux-foundation.org: fix printk warning]
[akpm@linux-foundation.org: use u64 cast, per Michal]
Fixes: 468c398233da ("mm: memcontrol: switch to native NR_ANON_THPS counter") Signed-off-by: zhongjiang-ali <zhongjiang-ali@linux.alibaba.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Michal Hocko <mhocko@suse.com> Link: https://lkml.kernel.org/r/1603722395-72443-1-git-send-email-zhongjiang-ali@linux.alibaba.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Failure in srpt_refresh_port() for the second port will leave MAD
registered for the first one, however, the srpt_add_one() will be marked
as "failed" and SRPT will leak resources for that registered but not used
and released first port.
Unregister the MAD agent for all ports in case of failure.
Fixes: a42d985bd5b2 ("ib_srpt: Initial SRP Target merge for v3.3-rc1") Link: https://lore.kernel.org/r/20201028065051.112430-1-leon@kernel.org Signed-off-by: Maor Gottlieb <maorg@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The BO cache needs to be cleaned up using vc4_bo_cache_destroy, but it's
not used consistently (vc4_drv's bind calls it in its error path, but
doesn't in unbind), and we can make that automatic through a managed
action. Let's remove the requirement to call vc4_bo_cache_destroy.
Lockdep complains that a possible deadlock below in
eeh_addr_cache_show() because it is acquiring a lock with IRQ enabled,
but eeh_addr_cache_insert_dev() needs to acquire the same lock with IRQ
disabled. Let's just make eeh_addr_cache_show() acquire the lock with
IRQ disabled as well.
Fixes: 5ca85ae6318d ("powerpc/eeh_cache: Add a way to dump the EEH address cache") Signed-off-by: Qian Cai <cai@redhat.com> Reviewed-by: Oliver O'Halloran <oohall@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20201028152717.8967-1-cai@redhat.com Signed-off-by: Sasha Levin <sashal@kernel.org>
In ip_set_match_extensions(), for sets with counters, we take care of
updating counters themselves by calling ip_set_update_counter(), and of
checking if the given comparison and values match, by calling
ip_set_match_counter() if needed.
However, if a given comparison on counters doesn't match the configured
values, that doesn't mean the set entry itself isn't matching.
This fix restores the behaviour we had before commit 4750005a85f7
("netfilter: ipset: Fix "don't update counters" mode when counters used
at the matching"), without reintroducing the issue fixed there: back
then, mtype_data_match() first updated counters in any case, and then
took care of matching on counters.
Now, if the IPSET_FLAG_SKIP_COUNTER_UPDATE flag is set,
ip_set_update_counter() will anyway skip counter updates if desired.
The issue observed is illustrated by this reproducer:
ipset create c hash:ip counters
ipset add c 192.0.2.1
iptables -I INPUT -m set --match-set c src --bytes-gt 800 -j DROP
if we now send packets from 192.0.2.1, bytes and packets counters
for the entry as shown by 'ipset list' are always zero, and, no
matter how many bytes we send, the rule will never match, because
counters themselves are not updated.
Reported-by: Mithil Mhatre <mmhatre@redhat.com> Fixes: 4750005a85f7 ("netfilter: ipset: Fix "don't update counters" mode when counters used at the matching") Signed-off-by: Stefano Brivio <sbrivio@redhat.com> Signed-off-by: Jozsef Kadlecsik <kadlec@netfilter.org> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Some devices support ACS functionality even though they don't have a
spec-compliant ACS Capability; pci_enable_acs() has a quirk mechanism to
handle them.
We want to enable ACS whenever possible, but 52fbf5bdeeef ("PCI: Cache ACS
capability offset in device") inadvertently broke this by calling
pci_enable_acs() only if we find an ACS Capability.
This resulted in ACS not being enabled for these non-compliant devices,
which means devices can't be separated into different IOMMU groups, which
in turn means we may not be able to pass those devices through to VMs, as
reported by Boris V:
If userspace does not include the trailing end of batch message, then
nfnetlink aborts the transaction. This allows to check that ruleset
updates trigger no errors.
After this patch, invoking this command from the prerouting chain:
# nft -c add rule x y fib saddr . oif type local
fails since oif is not supported there.
This patch fixes the lack of rule validation from the abort/check path
to catch configuration errors such as the one above.
If netfilter changes the packet mark when mangling, the packet is
rerouted using the route_me_harder set of functions. Prior to this
commit, there's one big difference between route_me_harder and the
ordinary initial routing functions, described in the comment above
__ip_queue_xmit():
/* Note: skb->sk can be different from sk, in case of tunnels */
int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
That function goes on to correctly make use of sk->sk_bound_dev_if,
rather than skb->sk->sk_bound_dev_if. And indeed the comment is true: a
tunnel will receive a packet in ndo_start_xmit with an initial skb->sk.
It will make some transformations to that packet, and then it will send
the encapsulated packet out of a *new* socket. That new socket will
basically always have a different sk_bound_dev_if (otherwise there'd be
a routing loop). So for the purposes of routing the encapsulated packet,
the routing information as it pertains to the socket should come from
that socket's sk, rather than the packet's original skb->sk. For that
reason __ip_queue_xmit() and related functions all do the right thing.
One might argue that all tunnels should just call skb_orphan(skb) before
transmitting the encapsulated packet into the new socket. But tunnels do
*not* do this -- and this is wisely avoided in skb_scrub_packet() too --
because features like TSQ rely on skb->destructor() being called when
that buffer space is truely available again. Calling skb_orphan(skb) too
early would result in buffers filling up unnecessarily and accounting
info being all wrong. Instead, additional routing must take into account
the new sk, just as __ip_queue_xmit() notes.
So, this commit addresses the problem by fishing the correct sk out of
state->sk -- it's already set properly in the call to nf_hook() in
__ip_local_out(), which receives the sk as part of its normal
functionality. So we make sure to plumb state->sk through the various
route_me_harder functions, and then make correct use of it following the
example of __ip_queue_xmit().
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Reviewed-by: Florian Westphal <fw@strlen.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
After the previous similar bugfix there was another bug here,
if no VHT elements were found we also disabled HE. Fix this to
disable HE only on the 5 GHz band; on 6 GHz it was already not
disabled, and on 2.4 GHz there need (should) not be any VHT.
Make sure that we actually initialize xefi_discard when we're scheduling
a deferred free of an AGFL block. This was (eventually) found by the
UBSAN while I was banging on realtime rmap problems, but it exists in
the upstream codebase. While we're at it, rearrange the structure to
reduce the struct size from 64 to 56 bytes.
Fixes: fcb762f5de2e ("xfs: add bmapi nodiscard flag") Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit 1056d3d2c97e ("selftests: enforce local header dependency in
lib.mk") added header dependency to the rule, but as the rule uses $^,
the headers are added to the compiler command line.
This can cause unexpected precompiled header files being generated when
compilation fails:
Due to the raw_output() function on kunit_parser.py actually being a
generator, it only runs if something reads the lines it returns. Since
we no-longer do that (parsing doesn't actually happen if raw_output is
enabled), it was not printing anything.
Fixes: 45ba7a893ad8 ("kunit: kunit_tool: Separate out config/build/exec/parse") Signed-off-by: David Gow <davidgow@google.com> Reviewed-by: Brendan Higgins <brendanhiggins@google.com> Tested-by: Brendan Higgins <brendanhiggins@google.com> Signed-off-by: Shuah Khan <skhan@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
An interrupt submitted to an affinity change will always be left enabled
after plic_set_affinity() has been called, while the expectation is that
it should stay in whatever state it was before the call.
Preserving the configuration fixes a PWM hang issue on the Unleashed
board.
[ 919.015783] rcu: INFO: rcu_sched detected stalls on CPUs/tasks:
[ 919.020922] rcu: 0-...0: (0 ticks this GP)
idle=7d2/1/0x4000000000000002 softirq=1424/1424 fqs=105807
[ 919.030295] (detected by 1, t=225825 jiffies, g=1561, q=3496)
[ 919.036109] Task dump for CPU 0:
[ 919.039321] kworker/0:1 R running task 0 30 2 0x00000008
[ 919.046359] Workqueue: events set_brightness_delayed
[ 919.051302] Call Trace:
[ 919.053738] [<ffffffe000930d92>] __schedule+0x194/0x4de
[ 982.035783] rcu: INFO: rcu_sched detected stalls on CPUs/tasks:
[ 982.040923] rcu: 0-...0: (0 ticks this GP)
idle=7d2/1/0x4000000000000002 softirq=1424/1424 fqs=113325
[ 982.050294] (detected by 1, t=241580 jiffies, g=1561, q=3509)
[ 982.056108] Task dump for CPU 0:
[ 982.059321] kworker/0:1 R running task 0 30 2 0x00000008
[ 982.066359] Workqueue: events set_brightness_delayed
[ 982.071302] Call Trace:
[ 982.073739] [<ffffffe000930d92>] __schedule+0x194/0x4de
[..]
Fixes: bb0fed1c60cc ("irqchip/sifive-plic: Switch to fasteoi flow") Signed-off-by: Greentime Hu <greentime.hu@sifive.com>
[maz: tidy-up commit message] Signed-off-by: Marc Zyngier <maz@kernel.org> Reviewed-by: Anup Patel <anup@brainfault.org> Link: https://lore.kernel.org/r/20201020081532.2377-1-greentime.hu@sifive.com Signed-off-by: Sasha Levin <sashal@kernel.org>
525c9e5a32bd introduced pm_runtime support for the i.MX SPI driver. With
this pm_runtime is used to bring up the clocks initially. When CONFIG_PM
is disabled the clocks are no longer enabled and the driver doesn't work
anymore. Fix this by enabling the clocks in the probe function and
telling pm_runtime that the device is active using
pm_runtime_set_active().
Fixes: 525c9e5a32bd spi: imx: enable runtime pm support Tested-by: Christian Eggers <ceggers@arri.de> [tested for !CONFIG_PM only] Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de> Link: https://lore.kernel.org/r/20201021104513.21560-1-s.hauer@pengutronix.de Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
projects / thirdparty / kernel / stable.git / log
