From: Chris Metcalf
nohz_full is only useful with isolcpus also set, since otherwise the
scheduler has to run periodically to try to determine whether to steal
work from other cores.
Accordingly, when booting with nohz_full=xxx on the command line, we
should act as if isolcpus=xxx was also set,
From: Chris Metcalf
The "clear" API is useful, for example, to modify a cpumask to avoid
the nohz cores so that interrupts aren't sent to them.
Likewise the "set" API is useful to modify a cpumask indicating some
special nohz-type functionality so that the nohz cores are
automatically added to
From: Chris Metcalf
Change the default behavior of watchdog so it only runs on the
housekeeping cores when nohz_full is enabled at build and boot time.
Allow modifying the set of cores the watchdog is currently running
on with a new kernel.watchdog_cpumask sysctl.
Acked-by: Don Zickus
From: Chris Metcalf
This change allows some cores to be excluded from running the
smp_hotplug_thread tasks. The motivating example for this is
the watchdog threads, which by default we don't want to run
on any enabled nohz_full cores.
Signed-off-by: Chris Metcalf
---
v6: change from an
From: Chris Metcalf cmetc...@ezchip.com
Change the default behavior of watchdog so it only runs on the
housekeeping cores when nohz_full is enabled at build and boot time.
Allow modifying the set of cores the watchdog is currently running
on with a new kernel.watchdog_cpumask sysctl.
Acked-by:
From: Chris Metcalf cmetc...@ezchip.com
This change allows some cores to be excluded from running the
smp_hotplug_thread tasks. The motivating example for this is
the watchdog threads, which by default we don't want to run
on any enabled nohz_full cores.
Signed-off-by: Chris Metcalf
From: Chris Metcalf cmetc...@ezchip.com
The clear API is useful, for example, to modify a cpumask to avoid
the nohz cores so that interrupts aren't sent to them.
Likewise the set API is useful to modify a cpumask indicating some
special nohz-type functionality so that the nohz cores are
From: Chris Metcalf cmetc...@ezchip.com
nohz_full is only useful with isolcpus also set, since otherwise the
scheduler has to run periodically to try to determine whether to steal
work from other cores.
Accordingly, when booting with nohz_full=xxx on the command line, we
should act as if
From: Chris Metcalf
This change allows some cores to be excluded from running the
smp_hotplug_thread tasks. The motivating example for this is
the watchdog threads, which by default we don't want to run
on any enabled nohz_full cores.
Signed-off-by: Chris Metcalf
---
include/linux/smpboot.h
From: Chris Metcalf
Change the default behavior of watchdog so it only runs on the
housekeeping cores when nohz_full is enabled at build and boot time.
Allow modifying the set of cores the watchdog is currently running
on with a new kernel.watchdog_exclude sysctl.
Signed-off-by: Chris Metcalf
From: Chris Metcalf
These changes allow the watchdog to work cleanly with nohz_full
by default, and to be configurable if desired to enable the watchdog
on cores that would normally disable it due to being nohz_full.
Thomas, does the addition of an exclude_mask to smp_hotplug_thread
meet with
From: Chris Metcalf
The "clear" API is useful, for example, to modify a cpumask to avoid
the nohz cores so that interrupts aren't sent to them.
Likewise the "set" API is useful to modify a cpumask indicating some
special nohz-type functionality so that the nohz cores are
automatically added to
From: Chris Metcalf
It's not clear that nohz_full is useful without isolcpus also
set, since otherwise the scheduler has to run periodically to
try to determine whether to steal work from other cores.
Signed-off-by: Chris Metcalf
Acked-by: Mike Galbraith ["thumbs up!"]
Acked-by: Rik van Riel
From: Chris Metcalf cmetc...@ezchip.com
The clear API is useful, for example, to modify a cpumask to avoid
the nohz cores so that interrupts aren't sent to them.
Likewise the set API is useful to modify a cpumask indicating some
special nohz-type functionality so that the nohz cores are
From: Chris Metcalf cmetc...@ezchip.com
It's not clear that nohz_full is useful without isolcpus also
set, since otherwise the scheduler has to run periodically to
try to determine whether to steal work from other cores.
Signed-off-by: Chris Metcalf cmetc...@ezchip.com
Acked-by: Mike Galbraith
From: Chris Metcalf cmetc...@ezchip.com
This change allows some cores to be excluded from running the
smp_hotplug_thread tasks. The motivating example for this is
the watchdog threads, which by default we don't want to run
on any enabled nohz_full cores.
Signed-off-by: Chris Metcalf
From: Chris Metcalf cmetc...@ezchip.com
Change the default behavior of watchdog so it only runs on the
housekeeping cores when nohz_full is enabled at build and boot time.
Allow modifying the set of cores the watchdog is currently running
on with a new kernel.watchdog_exclude sysctl.
From: Chris Metcalf cmetc...@ezchip.com
These changes allow the watchdog to work cleanly with nohz_full
by default, and to be configurable if desired to enable the watchdog
on cores that would normally disable it due to being nohz_full.
Thomas, does the addition of an exclude_mask to
From: Chris Metcalf
This is useful, for example, to modify a cpumask indicating some
special nohz-type functionality so that the nohz cores are
automatically added to that set.
Signed-off-by: Chris Metcalf
---
include/linux/tick.h | 7 +++
1 file changed, 7 insertions(+)
diff --git
From: Chris Metcalf
It's not clear that nohz_full is useful without isolcpus also
set, since otherwise the scheduler has to run periodically to
try to determine whether to steal work from other cores.
Signed-off-by: Chris Metcalf
---
I am puzzled why this has not been done before, so I suspect
From: Chris Metcalf
Change the default behavior of watchdog so it only runs on the
housekeeping cores when nohz_full is enabled at build and boot time.
Allow modifying the set of cores the watchdog is currently running
on with a new kernel.watchdog_exclude sysctl.
Signed-off-by: Chris Metcalf
From: Chris Metcalf
This change allows some cores to be excluded from running the
smp_hotplug_thread tasks. The motivating example for this is
the watchdog threads, which by default we don't want to run
on any enabled nohz_full cores.
Signed-off-by: Chris Metcalf
---
Relative to the quick
From: Chris Metcalf cmetc...@ezchip.com
This is useful, for example, to modify a cpumask indicating some
special nohz-type functionality so that the nohz cores are
automatically added to that set.
Signed-off-by: Chris Metcalf cmetc...@ezchip.com
---
include/linux/tick.h | 7 +++
1 file
From: Chris Metcalf cmetc...@ezchip.com
It's not clear that nohz_full is useful without isolcpus also
set, since otherwise the scheduler has to run periodically to
try to determine whether to steal work from other cores.
Signed-off-by: Chris Metcalf cmetc...@ezchip.com
---
I am puzzled why this
From: Chris Metcalf cmetc...@ezchip.com
Change the default behavior of watchdog so it only runs on the
housekeeping cores when nohz_full is enabled at build and boot time.
Allow modifying the set of cores the watchdog is currently running
on with a new kernel.watchdog_exclude sysctl.
From: Chris Metcalf cmetc...@ezchip.com
This change allows some cores to be excluded from running the
smp_hotplug_thread tasks. The motivating example for this is
the watchdog threads, which by default we don't want to run
on any enabled nohz_full cores.
Signed-off-by: Chris Metcalf
From: Chris Metcalf
Change the default behavior of watchdog so it only runs on the
housekeeping cores when nohz_full is enabled at build and boot time.
Allow modifying the set of cores the watchdog is currently running
on with a new kernel.watchdog_cpumask sysctl.
Signed-off-by: Chris Metcalf
From: Chris Metcalf cmetc...@ezchip.com
Change the default behavior of watchdog so it only runs on the
housekeeping cores when nohz_full is enabled at build and boot time.
Allow modifying the set of cores the watchdog is currently running
on with a new kernel.watchdog_cpumask sysctl.
From: Chris Metcalf
When queuing work, we should avoid queuing it on the local cpu if
we are using WORK_CPU_UNBOUND and the local cpu is nohz_full, since
the workqueue will mean a later interrupt of the nohz_full process
that presumably would prefer continuing to have 100% of the core
without
From: Chris Metcalf cmetc...@ezchip.com
When queuing work, we should avoid queuing it on the local cpu if
we are using WORK_CPU_UNBOUND and the local cpu is nohz_full, since
the workqueue will mean a later interrupt of the nohz_full process
that presumably would prefer continuing to have 100% of
From: Chris Metcalf
The "hypervisor shared" cores are ones that the Tilera hypervisor
uses to receive interrupts to manage hypervisor-owned devices.
It's a bad idea to try to use those cores with nohz_full, since
they will get interrupted unpredictably -- and invisibly to Linux
tracing tools,
From: Chris Metcalf
Running watchdog can be a helpful debugging feature on regular
cores, but it's incompatible with nohz_full, since it forces
regular scheduling events. Accordingly, just exit out immediately
from any nohz_full core.
An alternate approach would be to add a flags field or
From: Chris Metcalf cmetc...@ezchip.com
Running watchdog can be a helpful debugging feature on regular
cores, but it's incompatible with nohz_full, since it forces
regular scheduling events. Accordingly, just exit out immediately
from any nohz_full core.
An alternate approach would be to add a
From: Chris Metcalf cmetc...@ezchip.com
The hypervisor shared cores are ones that the Tilera hypervisor
uses to receive interrupts to manage hypervisor-owned devices.
It's a bad idea to try to use those cores with nohz_full, since
they will get interrupted unpredictably -- and invisibly to Linux
From: Chris Metcalf
Add the TIF_NOHZ flag appropriately.
Add call to user_exit() on entry to do_work_pending() and on entry
to syscalls via do_syscall_trace_enter(), and also the top of
do_syscall_trace_exit() just because it's done in x86.
Add call to user_enter() at the bottom of
From: Chris Metcalf
This is useful, for example, to modify a cpumask to avoid the
nohz cores so that interrupts aren't sent to them.
Signed-off-by: Chris Metcalf
---
Motivated by patch 4/4 in this series.
include/linux/tick.h | 7 +++
1 file changed, 7 insertions(+)
diff --git
From: Chris Metcalf
Signed-off-by: Chris Metcalf
---
drivers/net/ethernet/tile/tilegx.c | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/tile/tilegx.c
b/drivers/net/ethernet/tile/tilegx.c
index bea8cd2bb56c..cf9850c6a809 100644
---
From: Chris Metcalf
These changes are the necessary minimum to make the tile architecture
work relatively well with NOHZ_FULL. The tile architecture has had
out-of-tree support since 2008 for a DATAPLANE mode that provided
similar functionality (called Zero-Overhead Linux in our marketing
From: Chris Metcalf
Tile includes a hypervisor hook to deliver messages to arbitrary
tiles, so we can use that to raise an interrupt as soon as
possible on our own core. Unfortunately the Tilera hypervisor
disabled that support on principle in previous releases, but
it will be available in MDE
From: Chris Metcalf cmetc...@ezchip.com
Signed-off-by: Chris Metcalf cmetc...@ezchip.com
---
drivers/net/ethernet/tile/tilegx.c | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/tile/tilegx.c
b/drivers/net/ethernet/tile/tilegx.c
index
From: Chris Metcalf cmetc...@ezchip.com
These changes are the necessary minimum to make the tile architecture
work relatively well with NOHZ_FULL. The tile architecture has had
out-of-tree support since 2008 for a DATAPLANE mode that provided
similar functionality (called Zero-Overhead Linux in
From: Chris Metcalf cmetc...@ezchip.com
Tile includes a hypervisor hook to deliver messages to arbitrary
tiles, so we can use that to raise an interrupt as soon as
possible on our own core. Unfortunately the Tilera hypervisor
disabled that support on principle in previous releases, but
it will
From: Chris Metcalf cmetc...@ezchip.com
Add the TIF_NOHZ flag appropriately.
Add call to user_exit() on entry to do_work_pending() and on entry
to syscalls via do_syscall_trace_enter(), and also the top of
do_syscall_trace_exit() just because it's done in x86.
Add call to user_enter() at the
From: Chris Metcalf cmetc...@ezchip.com
This is useful, for example, to modify a cpumask to avoid the
nohz cores so that interrupts aren't sent to them.
Signed-off-by: Chris Metcalf cmetc...@ezchip.com
---
Motivated by patch 4/4 in this series.
include/linux/tick.h | 7 +++
1 file changed,
44 matches
Mail list logo
