Make:
struct dl_rq::dl_nr_migratory
struct dl_rq::dl_nr_running
struct rt_rq::rt_nr_boosted
struct rt_rq::rt_nr_migratory
struct rt_rq::rt_nr_total
struct rq::nr_uninterruptible
32-bit.
If total number of tasks can't exceed 2**32 (and less due to futex pid
limits), then per-runqueue counters can't as well.
This patchset has been sponsored by REX Prefix Eradication Society.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/20210422200228.1423391-4-adobriyan@gmail.com
As pointed out by commit
de9b8f5dcb ("sched: Fix crash trying to dequeue/enqueue the idle thread")
init_idle() can and will be invoked more than once on the same idle
task. At boot time, it is invoked for the boot CPU thread by
sched_init(). Then smp_init() creates the threads for all the secondary
CPUs and invokes init_idle() on them.
As the hotplug machinery brings the secondaries to life, it will issue
calls to idle_thread_get(), which itself invokes init_idle() yet again.
In this case it's invoked twice more per secondary: at _cpu_up(), and at
bringup_cpu().
Given smp_init() already initializes the idle tasks for all *possible*
CPUs, no further initialization should be required. Now, removing
init_idle() from idle_thread_get() exposes some interesting expectations
with regards to the idle task's preempt_count: the secondary startup always
issues a preempt_disable(), requiring some reset of the preempt count to 0
between hot-unplug and hotplug, which is currently served by
idle_thread_get() -> idle_init().
Given the idle task is supposed to have preemption disabled once and never
see it re-enabled, it seems that what we actually want is to initialize its
preempt_count to PREEMPT_DISABLED and leave it there. Do that, and remove
init_idle() from idle_thread_get().
Secondary startups were patched via coccinelle:
@begone@
@@
-preempt_disable();
...
cpu_startup_entry(CPUHP_AP_ONLINE_IDLE);
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20210512094636.2958515-1-valentin.schneider@arm.com
This patch provides support for setting and copying core scheduling
'task cookies' between threads (PID), processes (TGID), and process
groups (PGID).
The value of core scheduling isn't that tasks don't share a core,
'nosmt' can do that. The value lies in exploiting all the sharing
opportunities that exist to recover possible lost performance and that
requires a degree of flexibility in the API.
From a security perspective (and there are others), the thread,
process and process group distinction is an existent hierarchal
categorization of tasks that reflects many of the security concerns
about 'data sharing'. For example, protecting against cache-snooping
by a thread that can just read the memory directly isn't all that
useful.
With this in mind, subcommands to CREATE/SHARE (TO/FROM) provide a
mechanism to create and share cookies. CREATE/SHARE_TO specify a
target pid with enum pidtype used to specify the scope of the targeted
tasks. For example, PIDTYPE_TGID will share the cookie with the
process and all of it's threads as typically desired in a security
scenario.
API:
prctl(PR_SCHED_CORE, PR_SCHED_CORE_GET, tgtpid, pidtype, &cookie)
prctl(PR_SCHED_CORE, PR_SCHED_CORE_CREATE, tgtpid, pidtype, NULL)
prctl(PR_SCHED_CORE, PR_SCHED_CORE_SHARE_TO, tgtpid, pidtype, NULL)
prctl(PR_SCHED_CORE, PR_SCHED_CORE_SHARE_FROM, srcpid, pidtype, NULL)
where 'tgtpid/srcpid == 0' implies the current process and pidtype is
kernel enum pid_type {PIDTYPE_PID, PIDTYPE_TGID, PIDTYPE_PGID, ...}.
For return values, EINVAL, ENOMEM are what they say. ESRCH means the
tgtpid/srcpid was not found. EPERM indicates lack of PTRACE permission
access to tgtpid/srcpid. ENODEV indicates your machines lacks SMT.
[peterz: complete rewrite]
Signed-off-by: Chris Hyser <chris.hyser@oracle.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Don Hiatt <dhiatt@digitalocean.com>
Tested-by: Hongyu Ning <hongyu.ning@linux.intel.com>
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lkml.kernel.org/r/20210422123309.039845339@infradead.org
- Don't migrate if there is a cookie mismatch
Load balance tries to move task from busiest CPU to the
destination CPU. When core scheduling is enabled, if the
task's cookie does not match with the destination CPU's
core cookie, this task may be skipped by this CPU. This
mitigates the forced idle time on the destination CPU.
- Select cookie matched idle CPU
In the fast path of task wakeup, select the first cookie matched
idle CPU instead of the first idle CPU.
- Find cookie matched idlest CPU
In the slow path of task wakeup, find the idlest CPU whose core
cookie matches with task's cookie
Signed-off-by: Aubrey Li <aubrey.li@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Don Hiatt <dhiatt@digitalocean.com>
Tested-by: Hongyu Ning <hongyu.ning@linux.intel.com>
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lkml.kernel.org/r/20210422123308.860083871@infradead.org
During force-idle, we end up doing cross-cpu comparison of vruntimes
during pick_next_task. If we simply compare (vruntime-min_vruntime)
across CPUs, and if the CPUs only have 1 task each, we will always
end up comparing 0 with 0 and pick just one of the tasks all the time.
This starves the task that was not picked. To fix this, take a snapshot
of the min_vruntime when entering force idle and use it for comparison.
This min_vruntime snapshot will only be used for cross-CPU vruntime
comparison, and nothing else.
A note about the min_vruntime snapshot and force idling:
During selection:
When we're not fi, we need to update snapshot.
when we're fi and we were not fi, we must update snapshot.
When we're fi and we were already fi, we must not update snapshot.
Which gives:
fib fi update
0 0 1
0 1 1
1 0 1
1 1 0
Where:
fi: force-idled now
fib: force-idled before
So the min_vruntime snapshot needs to be updated when: !(fib && fi).
Also, the cfs_prio_less() function needs to be aware of whether the
core is in force idle or not, since it will be use this information to
know whether to advance a cfs_rq's min_vruntime_fi in the hierarchy.
So pass this information along via pick_task() -> prio_less().
Suggested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Don Hiatt <dhiatt@digitalocean.com>
Tested-by: Hongyu Ning <hongyu.ning@linux.intel.com>
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lkml.kernel.org/r/20210422123308.738542617@infradead.org
The rationale is as follows. In the core-wide pick logic, even if
need_sync == false, we need to go look at other CPUs (non-local CPUs)
to see if they could be running RT.
Say the RQs in a particular core look like this:
Let CFS1 and CFS2 be 2 tagged CFS tags.
Let RT1 be an untagged RT task.
rq0 rq1
CFS1 (tagged) RT1 (no tag)
CFS2 (tagged)
Say schedule() runs on rq0. Now, it will enter the above loop and
pick_task(RT) will return NULL for 'p'. It will enter the above if()
block and see that need_sync == false and will skip RT entirely.
The end result of the selection will be (say prio(CFS1) > prio(CFS2)):
rq0 rq1
CFS1 IDLE
When it should have selected:
rq0 rq1
IDLE RT
Suggested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Don Hiatt <dhiatt@digitalocean.com>
Tested-by: Hongyu Ning <hongyu.ning@linux.intel.com>
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lkml.kernel.org/r/20210422123308.678425748@infradead.org
Introduce task_struct::core_cookie as an opaque identifier for core
scheduling. When enabled; core scheduling will only allow matching
task to be on the core; where idle matches everything.
When task_struct::core_cookie is set (and core scheduling is enabled)
these tasks are indexed in a second RB-tree, first on cookie value
then on scheduling function, such that matching task selection always
finds the most elegible match.
NOTE: *shudder* at the overhead...
NOTE: *sigh*, a 3rd copy of the scheduling function; the alternative
is per class tracking of cookies and that just duplicates a lot of
stuff for no raisin (the 2nd copy lives in the rt-mutex PI code).
[Joel: folded fixes]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Don Hiatt <dhiatt@digitalocean.com>
Tested-by: Hongyu Ning <hongyu.ning@linux.intel.com>
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lkml.kernel.org/r/20210422123308.496975854@infradead.org
When switching on core-sched, CPUs need to agree which lock to use for
their RQ.
The new rule will be that rq->core_enabled will be toggled while
holding all rq->__locks that belong to a core. This means we need to
double check the rq->core_enabled value after each lock acquire and
retry if it changed.
This also has implications for those sites that take multiple RQ
locks, they need to be careful that the second lock doesn't end up
being the first lock.
Verify the lock pointer after acquiring the first lock, because if
they're on the same core, holding any of the rq->__lock instances will
pin the core state.
While there, change the rq->__lock order to CPU number, instead of rq
address, this greatly simplifies the next patch.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Don Hiatt <dhiatt@digitalocean.com>
Tested-by: Hongyu Ning <hongyu.ning@linux.intel.com>
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lkml.kernel.org/r/YJUNY0dmrJMD/BIm@hirez.programming.kicks-ass.net
Assuming this stuff isn't actually used much; disable it by default
and avoid allocating and tracking the task_delay_info structure.
taskstats is changed to still report the regular sched and sched_info
and only skip the missing task_delay_info fields instead of not
reporting anything.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Link: https://lkml.kernel.org/r/20210505111525.308018373@infradead.org
The situation around sched_info is somewhat complicated, it is used by
sched_stats and delayacct and, indirectly, kvm.
If SCHEDSTATS=Y (but disabled by default) sched_info_on() is
unconditionally true -- this is the case for all distro kernel configs
I checked.
If for some reason SCHEDSTATS=N, but TASK_DELAY_ACCT=Y, then
sched_info_on() can return false when delayacct is disabled,
presumably because there would be no other users left; except kvm is.
Instead of complicating matters further by accurately accounting
sched_stat and kvm state, simply unconditionally enable when
SCHED_INFO=Y, matching the common distro case.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Link: https://lkml.kernel.org/r/20210505111525.121458839@infradead.org
find_energy_efficient_cpu() (feec()) searches the best energy CPU
to place a task on. To do so, compute_energy() estimates the energy
impact of placing the task on a CPU, based on CPU and task utilization
signals.
Utilization signals can be concurrently updated while evaluating a
performance domain (pd). In some cases, this leads to having a
'negative delta', i.e. placing the task in the pd is seen as an
energy gain. Thus, any further energy comparison is biased.
In case of a 'negative delta', return prev_cpu since:
1. a 'negative delta' happens in less than 0.5% of feec() calls,
on a Juno with 6 CPUs (4 little, 2 big)
2. it is unlikely to have two consecutive 'negative delta' for
a task, so if the first call fails, feec() will correctly
place the task in the next feec() call
3. EAS current behavior tends to select prev_cpu if the task
doesn't raise the OPP of its current pd. prev_cpu is EAS's
generic decision
4. prev_cpu should be preferred to returning an error code.
In the latter case, select_idle_sibling() would do the placement,
selecting a big (and not energy efficient) CPU. As 3., the task
would potentially reside on the big CPU for a long time
Reported-by: Xuewen Yan <xuewen.yan@unisoc.com>
Suggested-by: Xuewen Yan <xuewen.yan@unisoc.com>
Signed-off-by: Pierre Gondois <Pierre.Gondois@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Reviewed-by: Vincent Donnefort <vincent.donnefort@arm.com>
Link: https://lkml.kernel.org/r/20210504090743.9688-3-Pierre.Gondois@arm.com
find_energy_efficient_cpu() searches the best energy CPU
to place a task on. To do so, the energy of each performance domain
(pd) is computed w/ and w/o the task placed on it.
The energy of a pd w/o the task (base_energy_pd) is computed prior
knowing whether a CPU is available in the pd.
Move the base_energy_pd computation after looping through the CPUs
of a pd and only compute it if at least one CPU is available.
Suggested-by: Xuewen Yan <xuewen.yan@unisoc.com>
Signed-off-by: Pierre Gondois <Pierre.Gondois@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Reviewed-by: Vincent Donnefort <vincent.donnefort@arm.com>
Link: https://lkml.kernel.org/r/20210504090743.9688-2-Pierre.Gondois@arm.com
The try_to_wake_up function has an optimization where it can queue
a task for wakeup on its previous CPU, if the task is still in the
middle of going to sleep inside schedule().
Once schedule() re-enables IRQs, the task will be woken up with an
IPI, and placed back on the runqueue.
If we have such a wakeup pending, there is no need to search other
CPUs for runnable tasks. Just skip (or bail out early from) newidle
balancing, and run the just woken up task.
For a memcache like workload test, this reduces total CPU use by
about 2%, proportionally split between user and system time,
and p99 and p95 application response time by 10% on average.
The schedstats run_delay number shows a similar improvement.
Signed-off-by: Rik van Riel <riel@surriel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Link: https://lkml.kernel.org/r/20210422130236.0bb353df@imladris.surriel.com
In commit:
9fe1f127b9 ("sched/fair: Merge select_idle_core/cpu()")
in select_idle_cpu(), we check if an idle core is present in the LLC
of the target CPU via the flag "has_idle_cores". We look for the idle
core in select_idle_cores(). If select_idle_cores() isn't able to find
an idle core/CPU, we need to unset the has_idle_cores flag in the LLC
of the target to prevent other CPUs from going down this route.
However, the current code is unsetting it in the LLC of the current
CPU instead of the target CPU. This patch fixes this issue.
Fixes: 9fe1f127b9 ("sched/fair: Merge select_idle_core/cpu()")
Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Reviewed-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Link: https://lore.kernel.org/r/1620746169-13996-1-git-send-email-ego@linux.vnet.ibm.com
Pull scheduler fixes from Thomas Gleixner:
"A set of scheduler updates:
- Prevent PSI state corruption when schedule() races with cgroup
move.
A recent commit combined two PSI callbacks to reduce the number of
cgroup tree updates, but missed that schedule() can drop rq::lock
for load balancing, which opens the race window for
cgroup_move_task() which then observes half updated state.
The fix is to solely use task::ps_flags instead of looking at the
potentially mismatching scheduler state
- Prevent an out-of-bounds access in uclamp caused bu a rounding
division which can lead to an off-by-one error exceeding the
buckets array size.
- Prevent unfairness caused by missing load decay when a task is
attached to a cfs runqueue.
The old load of the task was attached to the runqueue and never
removed. Fix it by enforcing the load update through the hierarchy
for unthrottled run queue instances.
- A documentation fix fot the 'sched_verbose' command line option"
* tag 'sched-urgent-2021-05-09' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
sched/fair: Fix unfairness caused by missing load decay
sched: Fix out-of-bound access in uclamp
psi: Fix psi state corruption when schedule() races with cgroup move
sched,doc: sched_debug_verbose cmdline should be sched_verbose
Pull locking fixes from Thomas Gleixner:
"A set of locking related fixes and updates:
- Two fixes for the futex syscall related to the timeout handling.
FUTEX_LOCK_PI does not support the FUTEX_CLOCK_REALTIME bit and
because it's not set the time namespace adjustment for clock
MONOTONIC is applied wrongly.
FUTEX_WAIT cannot support the FUTEX_CLOCK_REALTIME bit because its
always a relative timeout.
- Cleanups in the futex syscall entry points which became obvious
when the two timeout handling bugs were fixed.
- Cleanup of queued_write_lock_slowpath() as suggested by Linus
- Fixup of the smp_call_function_single_async() prototype"
* tag 'locking-urgent-2021-05-09' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
futex: Make syscall entry points less convoluted
futex: Get rid of the val2 conditional dance
futex: Do not apply time namespace adjustment on FUTEX_LOCK_PI
Revert 337f13046f ("futex: Allow FUTEX_CLOCK_REALTIME with FUTEX_WAIT op")
locking/qrwlock: Cleanup queued_write_lock_slowpath()
smp: Fix smp_call_function_single_async prototype
Pull more Kbuild updates from Masahiro Yamada:
- Convert sh and sparc to use generic shell scripts to generate the
syscall headers
- refactor .gitignore files
- Update kernel/config_data.gz only when the content of the .config
is really changed, which avoids the unneeded re-link of vmlinux
- move "remove stale files" workarounds to scripts/remove-stale-files
- suppress unused-but-set-variable warnings by default for Clang
as well
- fix locale setting LANG=C to LC_ALL=C
- improve 'make distclean'
- always keep intermediate objects from scripts/link-vmlinux.sh
- move IF_ENABLED out of <linux/kconfig.h> to make it self-contained
- misc cleanups
* tag 'kbuild-v5.13-2' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild: (25 commits)
linux/kconfig.h: replace IF_ENABLED() with PTR_IF() in <linux/kernel.h>
kbuild: Don't remove link-vmlinux temporary files on exit/signal
kbuild: remove the unneeded comments for external module builds
kbuild: make distclean remove tag files in sub-directories
kbuild: make distclean work against $(objtree) instead of $(srctree)
kbuild: refactor modname-multi by using suffix-search
kbuild: refactor fdtoverlay rule
kbuild: parameterize the .o part of suffix-search
arch: use cross_compiling to check whether it is a cross build or not
kbuild: remove ARCH=sh64 support from top Makefile
.gitignore: prefix local generated files with a slash
kbuild: replace LANG=C with LC_ALL=C
Makefile: Move -Wno-unused-but-set-variable out of GCC only block
kbuild: add a script to remove stale generated files
kbuild: update config_data.gz only when the content of .config is changed
.gitignore: ignore only top-level modules.builtin
.gitignore: move tags and TAGS close to other tag files
kernel/.gitgnore: remove stale timeconst.h and hz.bc
usr/include: refactor .gitignore
genksyms: fix stale comment
...
Pull networking fixes from Jakub Kicinski:
"Networking fixes for 5.13-rc1, including fixes from bpf, can and
netfilter trees. Self-contained fixes, nothing risky.
Current release - new code bugs:
- dsa: ksz: fix a few bugs found by static-checker in the new driver
- stmmac: fix frame preemption handshake not triggering after
interface restart
Previous releases - regressions:
- make nla_strcmp handle more then one trailing null character
- fix stack OOB reads while fragmenting IPv4 packets in openvswitch
and net/sched
- sctp: do asoc update earlier in sctp_sf_do_dupcook_a
- sctp: delay auto_asconf init until binding the first addr
- stmmac: clear receive all(RA) bit when promiscuous mode is off
- can: mcp251x: fix resume from sleep before interface was brought up
Previous releases - always broken:
- bpf: fix leakage of uninitialized bpf stack under speculation
- bpf: fix masking negation logic upon negative dst register
- netfilter: don't assume that skb_header_pointer() will never fail
- only allow init netns to set default tcp cong to a restricted algo
- xsk: fix xp_aligned_validate_desc() when len == chunk_size to avoid
false positive errors
- ethtool: fix missing NLM_F_MULTI flag when dumping
- can: m_can: m_can_tx_work_queue(): fix tx_skb race condition
- sctp: fix a SCTP_MIB_CURRESTAB leak in sctp_sf_do_dupcook_b
- bridge: fix NULL-deref caused by a races between assigning
rx_handler_data and setting the IFF_BRIDGE_PORT bit
Latecomer:
- seg6: add counters support for SRv6 Behaviors"
* tag 'net-5.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (73 commits)
atm: firestream: Use fallthrough pseudo-keyword
net: stmmac: Do not enable RX FIFO overflow interrupts
mptcp: fix splat when closing unaccepted socket
i40e: Remove LLDP frame filters
i40e: Fix PHY type identifiers for 2.5G and 5G adapters
i40e: fix the restart auto-negotiation after FEC modified
i40e: Fix use-after-free in i40e_client_subtask()
i40e: fix broken XDP support
netfilter: nftables: avoid potential overflows on 32bit arches
netfilter: nftables: avoid overflows in nft_hash_buckets()
tcp: Specify cmsgbuf is user pointer for receive zerocopy.
mlxsw: spectrum_mr: Update egress RIF list before route's action
net: ipa: fix inter-EE IRQ register definitions
can: m_can: m_can_tx_work_queue(): fix tx_skb race condition
can: mcp251x: fix resume from sleep before interface was brought up
can: mcp251xfd: mcp251xfd_probe(): add missing can_rx_offload_del() in error path
can: mcp251xfd: mcp251xfd_probe(): fix an error pointer dereference in probe
netfilter: nftables: Fix a memleak from userdata error path in new objects
netfilter: remove BUG_ON() after skb_header_pointer()
netfilter: nfnetlink_osf: Fix a missing skb_header_pointer() NULL check
...
Allow the developer to specifiy the initial value of the modprobe_path[]
string. This can be used to set it to the empty string initially, thus
effectively disabling request_module() during early boot until userspace
writes a new value via the /proc/sys/kernel/modprobe interface. [1]
When building a custom kernel (often for an embedded target), it's normal
to build everything into the kernel that is needed for booting, and indeed
the initramfs often contains no modules at all, so every such
request_module() done before userspace init has mounted the real rootfs is
a waste of time.
This is particularly useful when combined with the previous patch, which
made the initramfs unpacking asynchronous - for that to work, it had to
make any usermodehelper call wait for the unpacking to finish before
attempting to invoke the userspace helper. By eliminating all such
(known-to-be-futile) calls of usermodehelper, the initramfs unpacking and
the {device,late}_initcalls can proceed in parallel for much longer.
For a relatively slow ppc board I'm working on, the two patches combined
lead to 0.2s faster boot - but more importantly, the fact that the
initramfs unpacking proceeds completely in the background while devices
get probed means I get to handle the gpio watchdog in time without getting
reset.
[1] __request_module() already has an early -ENOENT return when
modprobe_path is the empty string.
Link: https://lkml.kernel.org/r/20210313212528.2956377-3-linux@rasmusvillemoes.dk
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Jessica Yu <jeyu@kernel.org>
Acked-by: Luis Chamberlain <mcgrof@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "background initramfs unpacking, and CONFIG_MODPROBE_PATH", v3.
These two patches are independent, but better-together.
The second is a rather trivial patch that simply allows the developer to
change "/sbin/modprobe" to something else - e.g. the empty string, so
that all request_module() during early boot return -ENOENT early, without
even spawning a usermode helper, needlessly synchronizing with the
initramfs unpacking.
The first patch delegates decompressing the initramfs to a worker thread,
allowing do_initcalls() in main.c to proceed to the device_ and late_
initcalls without waiting for that decompression (and populating of
rootfs) to finish. Obviously, some of those later calls may rely on the
initramfs being available, so I've added synchronization points in the
firmware loader and usermodehelper paths - there might be other places
that would need this, but so far no one has been able to think of any
places I have missed.
There's not much to win if most of the functionality needed during boot is
only available as modules. But systems with a custom-made .config and
initramfs can boot faster, partly due to utilizing more than one cpu
earlier, partly by avoiding known-futile modprobe calls (which would still
trigger synchronization with the initramfs unpacking, thus eliminating
most of the first benefit).
This patch (of 2):
Most of the boot process doesn't actually need anything from the
initramfs, until of course PID1 is to be executed. So instead of doing
the decompressing and populating of the initramfs synchronously in
populate_rootfs() itself, push that off to a worker thread.
This is primarily motivated by an embedded ppc target, where unpacking
even the rather modest sized initramfs takes 0.6 seconds, which is long
enough that the external watchdog becomes unhappy that it doesn't get
attention soon enough. By doing the initramfs decompression in a worker
thread, we get to do the device_initcalls and hence start petting the
watchdog much sooner.
Normal desktops might benefit as well. On my mostly stock Ubuntu kernel,
my initramfs is a 26M xz-compressed blob, decompressing to around 126M.
That takes almost two seconds:
[ 0.201454] Trying to unpack rootfs image as initramfs...
[ 1.976633] Freeing initrd memory: 29416K
Before this patch, these lines occur consecutively in dmesg. With this
patch, the timestamps on these two lines is roughly the same as above, but
with 172 lines inbetween - so more than one cpu has been kept busy doing
work that would otherwise only happen after the populate_rootfs()
finished.
Should one of the initcalls done after rootfs_initcall time (i.e., device_
and late_ initcalls) need something from the initramfs (say, a kernel
module or a firmware blob), it will simply wait for the initramfs
unpacking to be done before proceeding, which should in theory make this
completely safe.
But if some driver pokes around in the filesystem directly and not via one
of the official kernel interfaces (i.e. request_firmware*(),
call_usermodehelper*) that theory may not hold - also, I certainly might
have missed a spot when sprinkling wait_for_initramfs(). So there is an
escape hatch in the form of an initramfs_async= command line parameter.
Link: https://lkml.kernel.org/r/20210313212528.2956377-1-linux@rasmusvillemoes.dk
Link: https://lkml.kernel.org/r/20210313212528.2956377-2-linux@rasmusvillemoes.dk
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Reviewed-by: Luis Chamberlain <mcgrof@kernel.org>
Cc: Jessica Yu <jeyu@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
It's currently nigh impossible to get these pr_debug()s to print
something. Being guarded by initcall_debug means one has to enable tons
of other debug output during boot, and the system_state condition further
means it's impossible to get them when loading modules later.
Also, the compiler can't know that these global conditions do not change,
so there are W=2 warnings
kernel/async.c:125:9: warning: `calltime' may be used uninitialized in this function [-Wmaybe-uninitialized]
kernel/async.c:300:9: warning: `starttime' may be used uninitialized in this function [-Wmaybe-uninitialized]
Make it possible, for a DYNAMIC_DEBUG kernel, to get these to print their
messages by booting with appropriate 'dyndbg="file async.c +p"' command
line argument. For a non-DYNAMIC_DEBUG kernel, pr_debug() compiles to
nothing.
This does cost doing an unconditional ktime_get() for the starttime value,
but the corresponding ktime_get for the end time can be elided by
factoring it into a function which only gets called if the printk()
arguments end up being evaluated.
Link: https://lkml.kernel.org/r/20210309151723.1907838-1-linux@rasmusvillemoes.dk
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
request_free_mem_region() is used to find an empty range of physical
addresses for hotplugging ZONE_DEVICE memory. It does this by iterating
over the range of possible addresses using region_intersects() to see if
the range is free before calling request_mem_region() to allocate the
region.
However the resource_lock is dropped between these two calls meaning by
the time request_mem_region() is called in request_free_mem_region()
another thread may have already reserved the requested region. This
results in unexpected failures and a message in the kernel log from
hitting this condition:
/*
* mm/hmm.c reserves physical addresses which then
* become unavailable to other users. Conflicts are
* not expected. Warn to aid debugging if encountered.
*/
if (conflict->desc == IORES_DESC_DEVICE_PRIVATE_MEMORY) {
pr_warn("Unaddressable device %s %pR conflicts with %pR",
conflict->name, conflict, res);
These unexpected failures can be corrected by holding resource_lock across
the two calls. This also requires memory allocation to be performed prior
to taking the lock.
Link: https://lkml.kernel.org/r/20210419070109.4780-3-apopple@nvidia.com
Signed-off-by: Alistair Popple <apopple@nvidia.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Balbir Singh <bsingharora@gmail.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Muchun Song <smuchun@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
