We want to avoid leaking pointer info from xdp_frame (that is placed in
top of frame) like commit 6dfb970d3d ("xdp: avoid leaking info stored in
frame data on page reuse"), and followup commit 97e19cce05 ("bpf:
reserve xdp_frame size in xdp headroom") that reserve this headroom.
These changes also affected how cpumap constructed SKBs, as xdpf->headroom
size changed, the skb data starting point were in-effect shifted with 32
bytes (sizeof xdp_frame). This was still okay, as the cpumap frame_size
calculation also included xdpf->headroom which were reduced by same amount.
A bug was introduced in commit 77ea5f4cbe ("bpf/cpumap: make sure
frame_size for build_skb is aligned if headroom isn't"), where the
xdpf->headroom became part of the SKB_DATA_ALIGN rounding up. This
round-up to find the frame_size is in principle still correct as it does
not exceed the 2048 bytes frame_size (which is max for ixgbe and i40e),
but the 32 bytes offset of pkt_data_start puts this over the 2048 bytes
limit. This cause skb_shared_info to spill into next frame. It is a little
hard to trigger, as the SKB need to use above 15 skb_shinfo->frags[] as
far as I calculate. This does happen in practise for TCP streams when
skb_try_coalesce() kicks in.
KASAN can be used to detect these wrong memory accesses, I've seen:
BUG: KASAN: use-after-free in skb_try_coalesce+0x3cb/0x760
BUG: KASAN: wild-memory-access in skb_release_data+0xe2/0x250
Driver veth also construct a SKB from xdp_frame in this way, but is not
affected, as it doesn't reserve/deduct the room (used by xdp_frame) from
the SKB headroom. Instead is clears the pointers via xdp_scrub_frame(),
and allows SKB to use this area.
The fix in this patch is to do like veth and instead allow SKB to (re)use
the area occupied by xdp_frame, by clearing via xdp_scrub_frame(). (This
does kill the idea of the SKB being able to access (mem) info from this
area, but I guess it was a bad idea anyhow, and it was already killed by
the veth changes.)
Fixes: 77ea5f4cbe ("bpf/cpumap: make sure frame_size for build_skb is aligned if headroom isn't")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Currently there is a difference in how verifier checks memory access for
helper arguments for PTR_TO_MAP_VALUE and PTR_TO_STACK with regard to
variable part of offset.
check_map_access, that is used for PTR_TO_MAP_VALUE, can handle variable
offsets just fine, so that BPF program can call a helper like this:
some_helper(map_value_ptr + off, size);
, where offset is unknown at load time, but is checked by program to be
in a safe rage (off >= 0 && off + size < map_value_size).
But it's not the case for check_stack_boundary, that is used for
PTR_TO_STACK, and same code with pointer to stack is rejected by
verifier:
some_helper(stack_value_ptr + off, size);
For example:
0: (7a) *(u64 *)(r10 -16) = 0
1: (7a) *(u64 *)(r10 -8) = 0
2: (61) r2 = *(u32 *)(r1 +0)
3: (57) r2 &= 4
4: (17) r2 -= 16
5: (0f) r2 += r10
6: (18) r1 = 0xffff888111343a80
8: (85) call bpf_map_lookup_elem#1
invalid variable stack read R2 var_off=(0xfffffffffffffff0; 0x4)
Add support for variable offset access to check_stack_boundary so that
if offset is checked by program to be in a safe range it's accepted by
verifier.
Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
There are a few system calls (pselect, ppoll, etc) which replace a task
sigmask while they are running in a kernel-space
When a task calls one of these syscalls, the kernel saves a current
sigmask in task->saved_sigmask and sets a syscall sigmask.
On syscall-exit-stop, ptrace traps a task before restoring the
saved_sigmask, so PTRACE_GETSIGMASK returns the syscall sigmask and
PTRACE_SETSIGMASK does nothing, because its sigmask is replaced by
saved_sigmask, when the task returns to user-space.
This patch fixes this problem. PTRACE_GETSIGMASK returns saved_sigmask
if it's set. PTRACE_SETSIGMASK drops the TIF_RESTORE_SIGMASK flag.
Link: http://lkml.kernel.org/r/20181120060616.6043-1-avagin@gmail.com
Fixes: 29000caecb ("ptrace: add ability to get/set signal-blocked mask")
Signed-off-by: Andrei Vagin <avagin@gmail.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Randconfig builds with
CONFIG_TICK_ONESHOT=y
CONFIG_HOTPLUG_CPU=n
trigger
kernel/time/tick-broadcast.c:39:13: warning: ‘tick_broadcast_oneshot_offline’ \
declared ‘static’ but never defined [-Wunused-function]
due to that function's definition missing.
Move the CONFIG_HOTPLUG_CPU ifdeffery around its declaration too.
Fixes: 1b72d43237 ("tick: Remove outgoing CPU from broadcast masks")
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: x86@kernel.org
Link: https://lkml.kernel.org/r/20190329110508.6621-1-bp@alien8.de
When a module is loaded, its symbols' Elf_Sym information is stored
in a symtab. Further, type information is also captured. Since
Elf_Sym has no type field, historically the st_info field has been
hijacked for storing type: st_info was overwritten.
commit 5439c985c5 ("module: Overwrite
st_size instead of st_info") changes that practice, as its one-liner
indicates. Unfortunately, this change overwrites symbol size,
information that a tool like DTrace expects to find.
Allocate a typetab array to store type information so that no Elf_Sym
field needs to be overwritten.
Fixes: 5439c985c5 ("module: Overwrite st_size instead of st_info")
Signed-off-by: Eugene Loh <eugene.loh@oracle.com>
Reviewed-by: Nick Alcock <nick.alcock@oracle.com>
[jeyu: renamed typeoff -> typeoffs ]
Signed-off-by: Jessica Yu <jeyu@kernel.org>
Several people reported testing failures after setting CLOCK_REALTIME close
to the limits of the kernel internal representation in nanoseconds,
i.e. year 2262.
The failures are exposed in subsequent operations, i.e. when arming timers
or when the advancing CLOCK_MONOTONIC makes the calculation of
CLOCK_REALTIME overflow into negative space.
Now people start to paper over the underlying problem by clamping
calculations to the valid range, but that's just wrong because such
workarounds will prevent detection of real issues as well.
It is reasonable to force an upper bound for the various methods of setting
CLOCK_REALTIME. Year 2262 is the absolute upper bound. Assume a maximum
uptime of 30 years which is plenty enough even for esoteric embedded
systems. That results in an upper bound of year 2232 for setting the time.
Once that limit is reached in reality this limit is only a small part of
the problem space. But until then this stops people from trying to paper
over the problem at the wrong places.
Reported-by: Xiongfeng Wang <wangxiongfeng2@huawei.com>
Reported-by: Hongbo Yao <yaohongbo@huawei.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Stephen Boyd <sboyd@kernel.org>
Cc: Miroslav Lichvar <mlichvar@redhat.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lkml.kernel.org/r/alpine.DEB.2.21.1903231125480.2157@nanos.tec.linutronix.de
Tianyu reported a crash in a CPU hotplug teardown callback when booting a
kernel which has CONFIG_HOTPLUG_CPU disabled with the 'nosmt' boot
parameter.
It turns out that the SMP=y CONFIG_HOTPLUG_CPU=n case has been broken
forever in case that a bringup callback fails. Unfortunately this issue was
not recognized when the CPU hotplug code was reworked, so the shortcoming
just stayed in place.
When a bringup callback fails, the CPU hotplug code rolls back the
operation and takes the CPU offline.
The 'nosmt' command line argument uses a bringup failure to abort the
bringup of SMT sibling CPUs. This partial bringup is required due to the
MCE misdesign on Intel CPUs.
With CONFIG_HOTPLUG_CPU=y the rollback works perfectly fine, but
CONFIG_HOTPLUG_CPU=n lacks essential mechanisms to exercise the low level
teardown of a CPU including the synchronizations in various facilities like
RCU, NOHZ and others.
As a consequence the teardown callbacks which must be executed on the
outgoing CPU within stop machine with interrupts disabled are executed on
the control CPU in interrupt enabled and preemptible context causing the
kernel to crash and burn. The pre state machine code has a different
failure mode which is more subtle and resulting in a less obvious use after
free crash because the control side frees resources which are still in use
by the undead CPU.
But this is not a x86 only problem. Any architecture which supports the
SMP=y HOTPLUG_CPU=n combination suffers from the same issue. It's just less
likely to be triggered because in 99.99999% of the cases all bringup
callbacks succeed.
The easy solution of making HOTPLUG_CPU mandatory for SMP is not working on
all architectures as the following architectures have either no hotplug
support at all or not all subarchitectures support it:
alpha, arc, hexagon, openrisc, riscv, sparc (32bit), mips (partial).
Crashing the kernel in such a situation is not an acceptable state
either.
Implement a minimal rollback variant by limiting the teardown to the point
where all regular teardown callbacks have been invoked and leave the CPU in
the 'dead' idle state. This has the following consequences:
- the CPU is brought down to the point where the stop_machine takedown
would happen.
- the CPU stays there forever and is idle
- The CPU is cleared in the CPU active mask, but not in the CPU online
mask which is a legit state.
- Interrupts are not forced away from the CPU
- All facilities which only look at online mask would still see it, but
that is the case during normal hotplug/unplug operations as well. It's
just a (way) longer time frame.
This will expose issues, which haven't been exposed before or only seldom,
because now the normally transient state of being non active but online is
a permanent state. In testing this exposed already an issue vs. work queues
where the vmstat code schedules work on the almost dead CPU which ends up
in an unbound workqueue and triggers 'preemtible context' warnings. This is
not a problem of this change, it merily exposes an already existing issue.
Still this is better than crashing fully without a chance to debug it.
This is mainly thought as workaround for those architectures which do not
support HOTPLUG_CPU. All others should enforce HOTPLUG_CPU for SMP.
Fixes: 2e1a3483ce ("cpu/hotplug: Split out the state walk into functions")
Reported-by: Tianyu Lan <Tianyu.Lan@microsoft.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Tianyu Lan <Tianyu.Lan@microsoft.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Konrad Wilk <konrad.wilk@oracle.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Mukesh Ojha <mojha@codeaurora.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Jiri Kosina <jkosina@suse.cz>
Cc: Rik van Riel <riel@surriel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Micheal Kelley <michael.h.kelley@microsoft.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: K. Y. Srinivasan <kys@microsoft.com>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20190326163811.503390616@linutronix.de
Pull networking fixes from David Miller:
"Fixes here and there, a couple new device IDs, as usual:
1) Fix BQL race in dpaa2-eth driver, from Ioana Ciornei.
2) Fix 64-bit division in iwlwifi, from Arnd Bergmann.
3) Fix documentation for some eBPF helpers, from Quentin Monnet.
4) Some UAPI bpf header sync with tools, also from Quentin Monnet.
5) Set descriptor ownership bit at the right time for jumbo frames in
stmmac driver, from Aaro Koskinen.
6) Set IFF_UP properly in tun driver, from Eric Dumazet.
7) Fix load/store doubleword instruction generation in powerpc eBPF
JIT, from Naveen N. Rao.
8) nla_nest_start() return value checks all over, from Kangjie Lu.
9) Fix asoc_id handling in SCTP after the SCTP_*_ASSOC changes this
merge window. From Marcelo Ricardo Leitner and Xin Long.
10) Fix memory corruption with large MTUs in stmmac, from Aaro
Koskinen.
11) Do not use ipv4 header for ipv6 flows in TCP and DCCP, from Eric
Dumazet.
12) Fix topology subscription cancellation in tipc, from Erik Hugne.
13) Memory leak in genetlink error path, from Yue Haibing.
14) Valid control actions properly in packet scheduler, from Davide
Caratti.
15) Even if we get EEXIST, we still need to rehash if a shrink was
delayed. From Herbert Xu.
16) Fix interrupt mask handling in interrupt handler of r8169, from
Heiner Kallweit.
17) Fix leak in ehea driver, from Wen Yang"
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (168 commits)
dpaa2-eth: fix race condition with bql frame accounting
chelsio: use BUG() instead of BUG_ON(1)
net: devlink: skip info_get op call if it is not defined in dumpit
net: phy: bcm54xx: Encode link speed and activity into LEDs
tipc: change to check tipc_own_id to return in tipc_net_stop
net: usb: aqc111: Extend HWID table by QNAP device
net: sched: Kconfig: update reference link for PIE
net: dsa: qca8k: extend slave-bus implementations
net: dsa: qca8k: remove leftover phy accessors
dt-bindings: net: dsa: qca8k: support internal mdio-bus
dt-bindings: net: dsa: qca8k: fix example
net: phy: don't clear BMCR in genphy_soft_reset
bpf, libbpf: clarify bump in libbpf version info
bpf, libbpf: fix version info and add it to shared object
rxrpc: avoid clang -Wuninitialized warning
tipc: tipc clang warning
net: sched: fix cleanup NULL pointer exception in act_mirr
r8169: fix cable re-plugging issue
net: ethernet: ti: fix possible object reference leak
net: ibm: fix possible object reference leak
...
Have the IMA architecture specific policy require signed kernel modules
on systems with secure boot mode enabled; and coordinate the different
signature verification methods, so only one signature is required.
Requiring appended kernel module signatures may be configured, enabled
on the boot command line, or with this patch enabled in secure boot
mode. This patch defines set_module_sig_enforced().
To coordinate between appended kernel module signatures and IMA
signatures, only define an IMA MODULE_CHECK policy rule if
CONFIG_MODULE_SIG is not enabled. A custom IMA policy may still define
and require an IMA signature.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Reviewed-by: Luis Chamberlain <mcgrof@kernel.org>
Acked-by: Jessica Yu <jeyu@kernel.org>
Alexei Starovoitov says:
====================
pull-request: bpf-next 2019-03-26
The following pull-request contains BPF updates for your *net-next* tree.
The main changes are:
1) introduce bpf_tcp_check_syncookie() helper for XDP and tc, from Lorenz.
2) allow bpf_skb_ecn_set_ce() in tc, from Peter.
3) numerous bpf tc tunneling improvements, from Willem.
4) and other miscellaneous improvements from Adrian, Alan, Daniel, Ivan, Stanislav.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, lock_torture_cleanup() uses the values of cxt.lwsa and cxt.lrsa
to detect bad parameters that prevented locktorture from initializing,
let alone running. In this case, lock_torture_cleanup() does no cleanup
aside from invoking torture_cleanup_begin() and torture_cleanup_end(),
as required to permit future torture tests to run. However, this
heuristic fails if the run with bad parameters was preceded by a previous
run that actually ran: In this case, both cxt.lwsa and cxt.lrsa will
remain non-zero, which means that the current lock_torture_cleanup()
invocation will be unable to detect the fact that it should skip cleanup,
which can result in charming outcomes such as double frees.
This commit therefore NULLs out both cxt.lwsa and cxt.lrsa at the end
of any run that actually ran.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Josh Triplett <josh@joshtriplett.org>
If the specified rcuperf.perf_type is not in the rcu_perf_init()
function's perf_ops[] array, rcuperf prints some console messages and
then invokes rcu_perf_cleanup() to set state so that a future torture
test can run. However, rcu_perf_cleanup() also attempts to end the
test that didn't actually start, and in doing so relies on the value
of cur_ops, a value that is not particularly relevant in this case.
This can result in confusing output or even follow-on failures due to
attempts to use facilities that have not been properly initialized.
This commit therefore sets the value of cur_ops to NULL in this case and
inserts a check near the beginning of rcu_perf_cleanup(), thus avoiding
relying on an irrelevant cur_ops value.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
If the specified rcutorture.torture_type is not in the rcu_torture_init()
function's torture_ops[] array, rcutorture prints some console messages
and then invokes rcu_torture_cleanup() to set state so that a future
torture test can run. However, rcu_torture_cleanup() also attempts to
end the test that didn't actually start, and in doing so relies on the
value of cur_ops, a value that is not particularly relevant in this case.
This can result in confusing output or even follow-on failures due to
attempts to use facilities that have not been properly initialized.
This commit therefore sets the value of cur_ops to NULL in this case
and inserts a check near the beginning of rcu_torture_cleanup(),
thus avoiding relying on an irrelevant cur_ops value.
Reported-by: kernel test robot <rong.a.chen@intel.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The rcutorture_oom_notify() function has a misplaced close parenthesis
that results in increasingly long delays in rcu_fwd_progress_check()'s
checking for various RCU forward-progress problems. This commit therefore
puts the parenthesis in the right place.
Signed-off-by: Neeraj Upadhyay <neeraju@codeaurora.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Back when there was a separate RCU-bh flavor, the ->ext_irq_conflict
field was used to prevent executing local_bh_enable() while interrupts
were disabled. However, there is no longer an RCU-bh flavor, so this
commit removes the no-longer-needed ->ext_irq_conflict field.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The code actually rarely uses more than one type of RCU read-side
protection, as is actually desired given that we need some reasonable
probability of preempting RCU read-side critical sections, which cannot
happen with multiple types of protection. This comment therefore adjusts
the comment.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
If there is only one online CPU, it doesn't make sense to try to offline
it, as any such attempt is guaranteed to fail. This commit therefore
check for this condition and refuses to attempt the nonsensical.
Reported-by: Su Yue <suy.fnst@cn.fujitsu.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Tested-By: Su Yue <suy.fnst@cn.fujitsu.com>
The Documentation/RCU/stallwarn.txt file says that stall warnings
print "D" if dyntick-idle processing is enabled, but the code in
print_cpu_stall_fast_no_hz() prints "." instead. This commit therefore
reverses the sense of the test to make the code match the documentation.
Signed-off-by: Neeraj Upadhyay <neeraju@codeaurora.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
This commit further consolidates stall-warning functionality by moving
forward-progress checkers into kernel/rcu/tree_stall.h, updating a
comment or two while in the area. More specifically, this commit moves
show_rcu_gp_kthreads(), rcu_check_gp_start_stall(), rcu_fwd_progress_check(),
sysrq_rcu, sysrq_show_rcu(), sysrq_rcudump_op, and rcu_sysrq_init() from
kernel/rcu/tree.c to kernel/rcu/tree_stall.h.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The rcu_iw_handler() function's sole purpose in life is to indicate
whether a stalled CPU had interrupts disabled, so it belongs in
kernel/rcu/tree_stall.h. This commit therefore makes that move,
clarifying its header comment while in the area.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
This commit does only code movement, removal of now-unneeded forward
declarations, and addition of comments. It organizes the functions
that implement RCU CPU stall warnings for normal grace periods into
three categories:
1. Control of RCU CPU stall warnings, including computing timeouts.
2. Interaction of stall warnings with grace periods.
3. Actual printing of the RCU CPU stall-warning messages.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
This commit further consolidates the stall-warning code by moving
print_cpu_stall_info() and its helper functions along with
zero_cpu_stall_ticks() to kernel/rcu/tree_stall.h.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The print_cpu_stall_info_begin() and print_cpu_stall_info_end() print a
single character each onto the console, and are a holdover from a time
when RCU CPU stall warning messages could be abbreviated using a long-gone
Kconfig option. This commit therefore adds these single characters to
already-printed strings in the calling functions, and then eliminates
both print_cpu_stall_info_begin() and print_cpu_stall_info_end().
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Because expedited CPU stall warnings are contained within the
kernel/rcu/tree_exp.h file, rcu_print_task_exp_stall() should live
there too. This commit carries out the required code motion.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The rcu_print_detail_task_stall(), rcu_print_task_stall_begin(), and
rcu_print_task_stall_end() functions were defined to allow long-gone
Kconfig options to provide an abbreviated RCU CPU stall warning printout.
This commit saves a few lines of code by inlining them into their sole
callers.
While in the area, a useless call of rcu_print_detail_task_stall_rnp()
on the root rcu_node structure was eliminated. If there is only one
rcu_node structure, its tasks get printed twice, but if there are more,
the root rcu_node structure is guaranteed to have an empty list of blocked
tasks, hence the uselessness. (Long ago, root rcu_node structures with
non-empty ->blkd_tasks lists could happen, but no longer.)
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
This commit completes the process of consolidating the code for RCU CPU
stall warnings for normal grace periods by moving the remaining such
code from kernel/rcu/tree.c to kernel/rcu/tree_stall.h.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The RCU CPU stall-warning code for normal grace periods is currently
scattered across two files, due to earlier Tiny RCU support for RCU
CPU stall warnings and for old Kconfig options that have long since
been retired. Given that it is hard for the lead RCU maintainer to
find relevant stall-warning code, it would be good to consolidate it.
This commit continues this process by moving stall-warning code from
kernel/rcu/tree_plugin.c to a new kernel/rcu/tree_stall.h file.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The RCU CPU stall-warning code for normal grace periods is currently
scattered across three files, due to earlier Tiny RCU support for RCU
CPU stall warnings and for old Kconfig options that have long since
been retired. Given that it is hard for the lead RCU maintainer to
find relevant stall-warning code, it would be good to consolidate it.
This commit starts this process by moving stall-warning code from
kernel/rcu/update.c to a new kernel/rcu/tree_stall.h file.
Note that the definitions of rcu_cpu_stall_suppress and
rcu_cpu_stall_timeout must remain in kernel/rcu/update.h to provide
compatibility for kernel boot parameter lists.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The cleanup_srcu_struct_quiesced() function was added because NVME
used WQ_MEM_RECLAIM workqueues and SRCU did not, which meant that
NVME workqueues waiting on SRCU workqueues could result in deadlocks
during low-memory conditions. However, SRCU now also has WQ_MEM_RECLAIM
workqueues, so there is no longer a potential for deadlock. Furthermore,
it turns out to be extremely hard to use cleanup_srcu_struct_quiesced()
correctly due to the fact that SRCU callback invocation accesses the
srcu_struct structure's per-CPU data area just after callbacks are
invoked. Therefore, the usual practice of using srcu_barrier() to wait
for callbacks to be invoked before invoking cleanup_srcu_struct_quiesced()
fails because SRCU's callback-invocation workqueue handler might be
delayed, which can result in cleanup_srcu_struct_quiesced() being invoked
(and thus freeing the per-CPU data) before the SRCU's callback-invocation
workqueue handler is finished using that per-CPU data. Nor is this a
theoretical problem: KASAN emitted use-after-free warnings because of
this problem on actual runs.
In short, NVME can now safely invoke cleanup_srcu_struct(), which
avoids the use-after-free scenario. And cleanup_srcu_struct_quiesced()
is quite difficult to use safely. This commit therefore removes
cleanup_srcu_struct_quiesced(), switching its sole user back to
cleanup_srcu_struct(). This effectively reverts the following pair
of commits:
f7194ac32c ("srcu: Add cleanup_srcu_struct_quiesced()")
4317228ad9 ("nvme: Avoid flush dependency in delete controller flow")
Reported-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Bart Van Assche <bvanassche@acm.org>
If someone fails to drain the corresponding SRCU callbacks (for
example, by failing to invoke srcu_barrier()) before invoking either
cleanup_srcu_struct() or cleanup_srcu_struct_quiesced(), the resulting
diagnostic is an ambiguous use-after-free diagnostic, and even then
only if you are running something like KASAN. This commit therefore
improves SRCU diagnostics by adding checks for in-flight callbacks at
_cleanup_srcu_struct() time.
Note that these diagnostics can still be defeated, for example, by
invoking call_srcu() concurrently with cleanup_srcu_struct(). Which is
a really bad idea, but sometimes all too easy to do. But even then,
these diagnostics have at least some probability of catching the problem.
Reported-by: Sagi Grimberg <sagi@grimberg.me>
Reported-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Tested-by: Bart Van Assche <bvanassche@acm.org>
The task_struct structure's ->rcu_read_unlock_special field is only ever
read or written by the owning task, but it is accessed both at process
and interrupt levels. It may therefore be accessed using plain reads
and writes while interrupts are disabled, but must be accessed using
READ_ONCE() and WRITE_ONCE() or better otherwise. This commit makes a
few adjustments to align with this discipline.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
This commit changes a rcu_exp_handler() comment from rcu_preempt_defer_qs()
to rcu_preempt_deferred_qs() in order to better match reality.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Because rcu_wake_cond() checks for a null task_struct pointer, there is
no need for its callers to do so. This commit eliminates the redundant
check.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Previously, threads blocked on offlining CPUS were migrated to the
root rcu_node structure, thus requiring RCU priority boosting on this
structure. However, since commit d19fb8d1f3 ("rcu: Don't migrate
blocked tasks even if all corresponding CPUs offline"), RCU does not
migrate blocked tasks. Consequently, RCU no longer does RCU priority
boosting on the root rcu_node structure as of commit 1be0085b51 ("rcu:
Don't initiate RCU priority boosting on root rcu_node").
This commit therefore brings comments for the force_qs_rnp() function's
header comment in line with this new no-root-boosting reality.
Signed-off-by: Zhouyi Zhou <zhouzhouyi@gmail.com>
[ paulmck: Also remove obsolete comment on suppressing new grace periods. ]
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
This commit better documents the jiffies_to_sched_qs default-value
strategy used by adjust_jiffies_till_sched_qs()
Reported-by: Joel Fernandes <joel@joelfernandes.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The current code only calls adjust_jiffies_till_sched_qs() if
jiffies_till_sched_qs is left at its default value, so when the
jiffies_till_sched_qs kernel-boot parameter actually is specified,
jiffies_to_sched_qs will be left with the value zero, which
will result in useless slowdowns of cond_resched(). This commit
therefore changes rcu_init_geometry() to unconditionally invoke
adjust_jiffies_till_sched_qs(), which ensures that jiffies_to_sched_qs
will be initialized in all cases, thus maintaining good cond_resched()
performance.
Signed-off-by: Neeraj Upadhyay <neeraju@codeaurora.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The current rcu_gp_kthread_wake() function uses in_interrupt()
and thus does a self-wakeup from all interrupt contexts, including
the pointless case where the GP kthread happens to be running with
bottom halves disabled, along with the impossible case where the GP
kthread is running within an NMI handler (you are not supposed to invoke
rcu_gp_kthread_wake() from within an NMI handler. This commit therefore
replaces the in_interrupt() with in_irq(), so that the self-wakeups
happen only from handlers for hardware interrupts and softirqs.
This also makes the code match the comment.
Signed-off-by: Neeraj Upadhyay <neeraju@codeaurora.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
This commit prints a console message when cpulist_parse() reports a
bad list of CPUs, and sets all CPUs' bits in that case. The reason for
setting all CPUs' bits is that this is the safe(r) choice for real-time
workloads, which would normally be the ones using the rcu_nocbs= kernel
boot parameter. Either way, later RCU console log messages list the
actual set of CPUs whose RCU callbacks will be offloaded.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Currently, the rcu_nocbs= kernel boot parameter requires that a specific
list of CPUs be specified, and has no way to say "all of them".
As noted by user RavFX in a comment to Phoronix topic 1002538, this
is an inconvenient side effect of the removal of the RCU_NOCB_CPU_ALL
Kconfig option. This commit therefore enables the rcu_nocbs= kernel boot
parameter to be given the string "all", as in "rcu_nocbs=all" to specify
that all CPUs on the system are to have their RCU callbacks offloaded.
Another approach would be to make cpulist_parse() check for "all", but
there are uses of cpulist_parse() that do other checking, which could
conflict with an "all". This commit therefore focuses on the specific
use of cpulist_parse() in rcu_nocb_setup().
Just a note to other people who would like changes to Linux-kernel RCU:
If you send your requests to me directly, they might get fixed somewhat
faster. RavFX's comment was posted on January 22, 2018 and I first saw
it on March 5, 2019. And the only reason that I found it -at- -all- was
that I was looking for projects using RCU, and my search engine showed
me that Phoronix comment quite by accident. Your choice, though! ;-)
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
As the result of recent addition of "rdp->core_needs_qs = false;" in
the "if" block, now both branches of the if-else have the same
assignment.
Factor it out and reduce line count.
Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
Cc: Joel Fernandes <joel@joelfernandes.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Acked-by: Joel Fernandes (Google) <joel@joelfernandes.org>
The rcutree.kthread_prio kernel-boot parameter is used to set the
priority for boost (rcub), per-CPU (rcuc), and grace-period (rcu_preempt
or rcu_sched) kthreads. It is also used by rcutorture to check whether
it is possible to meaningfully test RCU priority boosting. However,
all of these cases will either ignore or be confused by any post-boot
changes to rcutree.kthread_prio.
Note that the user really can change the priorities of all of these
kthreads using chrt, given sufficient privileges. Therefore, the
read-write nature of sysfs access to rcutree.kthread_prio is thus at
best an attractive nuisance.
This commit therefore changes sysfs access to rcutree.kthread_prio to
be read-only.
Signed-off-by: Liu Song <liu.song11@zte.com.cn>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The purpose of exit_rcu() is to handle cases where buggy code causes a
task to exit within an RCU read-side critical section. It currently
does that in the case where said RCU read-side critical section was
preempted at least once, but fails to handle cases where preemption did
not occur. This case needs to be handled because otherwise the final
context switch away from the exiting task will incorrectly behave as if
task exit were instead a preemption of an RCU read-side critical section,
and will therefore queue the exiting task. The exiting task will have
exited, and thus won't ever execute rcu_read_unlock(), which means that
it will remain queued forever, blocking all subsequent grace periods,
and eventually resulting in OOM.
Although this is arguably better than letting grace periods proceed
and having a later rcu_read_unlock() access the now-freed task
structure that once belonged to the exiting tasks, it would obviously
be better to correctly handle this case. This commit therefore sets
->rcu_read_lock_nesting to 1 in that case, so that the subsequence call
to __rcu_read_unlock() causes the exiting task to exit its dangling RCU
read-side critical section.
Note that deferred quiescent states need not be considered. The reason
is that removing the task from the ->blkd_tasks[] list in the call to
rcu_preempt_deferred_qs() handles the per-task component of any deferred
quiescent state, and all other components of any deferred quiescent state
are associated with the CPU, which isn't going anywhere until some later
CPU-hotplug operation, which will report any remaining deferred quiescent
states from within the rcu_report_dead() function.
Note also that negative values of ->rcu_read_lock_nesting need not be
considered. First, these won't show up in exit_rcu() unless there is
a serious bug in RCU, and second, setting ->rcu_read_lock_nesting sets
the state so that the RCU read-side critical section will be exited
normally.
Again, this code has no effect unless there has been some prior bug
that prevents a task from leaving an RCU read-side critical section
before exiting. Furthermore, there have been no reports of the bug
fixed by this commit appearing in production. This commit is therefore
absolutely -not- recommended for backporting to -stable.
Reported-by: ABHISHEK DUBEY <dabhishek@iisc.ac.in>
Reported-by: BHARATH Y MOURYA <bharathm@iisc.ac.in>
Reported-by: Aravinda Prasad <aravinda@iisc.ac.in>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Tested-by: ABHISHEK DUBEY <dabhishek@iisc.ac.in>
The rcu_qs is disabling IRQs by self so no need to do the same in raise_softirq
but instead we can save some cycles using raise_softirq_irqoff directly.
CC: Paul E. McKenney <paulmck@linux.ibm.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
When there are no callbacks pending on an idle system, I noticed that
RCU softirq is continuously firing. During this the cpu_no_qs is set to
false, and core_needs_qs is set to true indefinitely. This causes
rcu_process_callbacks to be repeatedly called, even though the node
corresponding to the CPU has that CPU's mask bit cleared and the system
is idle. I believe the race is when such mask clearing is done during
idle CPU scan of the quiescent state forcing stage in the kthread
instead of the softirq. Since the rnp mask is cleared, but the flags on
the CPU's rdp are not cleared, the CPU thinks it still needs to report
to core RCU.
Cure this by clearing the core_needs_qs flag when the CPU detects that
its node is already updated which will avoid the unwanted softirq raises
to the benefit of real-time systems.
Test: Ran rcutorture for various tree RCU configs.
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The rcu_pm_notify() function refuses to switch to/from expedited grace
periods on systems with more than 256 CPUs due to the serialized
initialization of expedited grace periods. However, expedited grace
periods are now initialized in parallel, removing this concern.
This commit therefore removes the checks from rcu_pm_notify(), so that
expedited grace periods are used unconditionally during suspend/resume
and hibernate/wake operations.
As always, real-time workloads wishing to completely avoid expedited
grace periods can use the rcupdate.rcu_normal= kernel parameter.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
The BPF verifier checks the maximum number of call stack frames twice,
first in the main CFG traversal (do_check) and then in a subsequent
traversal (check_max_stack_depth). If the second check fails, it logs a
'verifier bug' warning and errors out, as the number of call stack frames
should have been verified already.
However, the second check may fail without indicating a verifier bug: if
the excessive function calls reside in dead code, the main CFG traversal
may not visit them; the subsequent traversal visits all instructions,
including dead code.
This case raises the question of how invalid dead code should be treated.
This patch implements the conservative option and rejects such code.
Signed-off-by: Paul Chaignon <paul.chaignon@orange.com>
Tested-by: Xiao Han <xiao.han@orange.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
