Re: [GIT PULL] workqueue fixes for v4.13-rc3

On Mon, Aug 28, 2017 at 11:07:28AM +0200, Geert Uytterhoeven wrote: > Hi Tejun, > > On Thu, Aug 24, 2017 at 4:33 PM, Tejun Heo wrote: > > On Thu, Aug 24, 2017 at 03:32:04PM +0200, Geert Uytterhoeven wrote: > >> > Ah, okay, so it has multiple nodes but not NUMA. The generic numa

Re: [GIT PULL] workqueue fixes for v4.13-rc3

On Mon, Aug 28, 2017 at 11:07:28AM +0200, Geert Uytterhoeven wrote: > Hi Tejun, > > On Thu, Aug 24, 2017 at 4:33 PM, Tejun Heo wrote: > > On Thu, Aug 24, 2017 at 03:32:04PM +0200, Geert Uytterhoeven wrote: > >> > Ah, okay, so it has multiple nodes but not NUMA. The generic numa > >> > topology

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hi Tejun, On Thu, Aug 24, 2017 at 4:33 PM, Tejun Heo wrote: > On Thu, Aug 24, 2017 at 03:32:04PM +0200, Geert Uytterhoeven wrote: >> > Ah, okay, so it has multiple nodes but not NUMA. The generic numa >> > topology code assumes that there's only one node if !NUMA and reports >>

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hi Tejun, On Thu, Aug 24, 2017 at 4:33 PM, Tejun Heo wrote: > On Thu, Aug 24, 2017 at 03:32:04PM +0200, Geert Uytterhoeven wrote: >> > Ah, okay, so it has multiple nodes but not NUMA. The generic numa >> > topology code assumes that there's only one node if !NUMA and reports >> > all online

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hello, On Thu, Aug 24, 2017 at 03:32:04PM +0200, Geert Uytterhoeven wrote: > > Ah, okay, so it has multiple nodes but not NUMA. The generic numa > > topology code assumes that there's only one node if !NUMA and reports > > all online cpus regardless of the node number, which makes the same > >

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hello, On Thu, Aug 24, 2017 at 03:32:04PM +0200, Geert Uytterhoeven wrote: > > Ah, okay, so it has multiple nodes but not NUMA. The generic numa > > topology code assumes that there's only one node if !NUMA and reports > > all online cpus regardless of the node number, which makes the same > >

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hi Tejun, On Wed, Aug 23, 2017 at 7:08 PM, Tejun Heo wrote: > On Wed, Aug 23, 2017 at 04:47:41PM +0200, Geert Uytterhoeven wrote: >> Some code is mixing up multiple memory nodes with multiple cpu nodes. >> M68k uses DISCONTIGMEM, but not NUMA (no SMP): >> >> config

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hi Tejun, On Wed, Aug 23, 2017 at 7:08 PM, Tejun Heo wrote: > On Wed, Aug 23, 2017 at 04:47:41PM +0200, Geert Uytterhoeven wrote: >> Some code is mixing up multiple memory nodes with multiple cpu nodes. >> M68k uses DISCONTIGMEM, but not NUMA (no SMP): >> >> config NEED_MULTIPLE_NODES >>

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hello, Geert. On Wed, Aug 23, 2017 at 04:47:41PM +0200, Geert Uytterhoeven wrote: > Some code is mixing up multiple memory nodes with multiple cpu nodes. > M68k uses DISCONTIGMEM, but not NUMA (no SMP): > > config NEED_MULTIPLE_NODES > def_bool y > depends on DISCONTIGMEM ||

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hello, Geert. On Wed, Aug 23, 2017 at 04:47:41PM +0200, Geert Uytterhoeven wrote: > Some code is mixing up multiple memory nodes with multiple cpu nodes. > M68k uses DISCONTIGMEM, but not NUMA (no SMP): > > config NEED_MULTIPLE_NODES > def_bool y > depends on DISCONTIGMEM ||

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hi Tejun, On Wed, Aug 23, 2017 at 4:24 PM, Tejun Heo wrote: > On Wed, Aug 23, 2017 at 10:10:54AM +0200, Geert Uytterhoeven wrote: >> > + pr_warn_once("WARNING: workqueue empty cpumask: node=%d >> > cpu_going_down=%d cpumask=%*pb online=%*pb possible=%*pb\n", >> >

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hi Tejun, On Wed, Aug 23, 2017 at 4:24 PM, Tejun Heo wrote: > On Wed, Aug 23, 2017 at 10:10:54AM +0200, Geert Uytterhoeven wrote: >> > + pr_warn_once("WARNING: workqueue empty cpumask: node=%d >> > cpu_going_down=%d cpumask=%*pb online=%*pb possible=%*pb\n", >> > +

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hello, Geert. Something is really fishy. On Wed, Aug 23, 2017 at 10:10:54AM +0200, Geert Uytterhoeven wrote: > > + pr_warn_once("WARNING: workqueue empty cpumask: node=%d > > cpu_going_down=%d cpumask=%*pb online=%*pb possible=%*pb\n", > > +node,

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hello, Geert. Something is really fishy. On Wed, Aug 23, 2017 at 10:10:54AM +0200, Geert Uytterhoeven wrote: > > + pr_warn_once("WARNING: workqueue empty cpumask: node=%d > > cpu_going_down=%d cpumask=%*pb online=%*pb possible=%*pb\n", > > +node,

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hi Tejun, On Mon, Aug 7, 2017 at 7:06 PM, Tejun Heo wrote: > On Mon, Aug 07, 2017 at 02:18:51PM +0200, Geert Uytterhoeven wrote: >> This triggers on m68k, which doesn't have SMP. >> Haven't tried it yet on any other system due to holidays. > > That's weird. Can you please apply

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hi Tejun, On Mon, Aug 7, 2017 at 7:06 PM, Tejun Heo wrote: > On Mon, Aug 07, 2017 at 02:18:51PM +0200, Geert Uytterhoeven wrote: >> This triggers on m68k, which doesn't have SMP. >> Haven't tried it yet on any other system due to holidays. > > That's weird. Can you please apply the following

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hello, On Mon, Aug 07, 2017 at 02:18:51PM +0200, Geert Uytterhoeven wrote: > This triggers on m68k, which doesn't have SMP. > Haven't tried it yet on any other system due to holidays. That's weird. Can you please apply the following patch and report the messages? Thanks. diff --git

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hello, On Mon, Aug 07, 2017 at 02:18:51PM +0200, Geert Uytterhoeven wrote: > This triggers on m68k, which doesn't have SMP. > Haven't tried it yet on any other system due to holidays. That's weird. Can you please apply the following patch and report the messages? Thanks. diff --git

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hi Tejun, On Mon, Jul 31, 2017 at 5:38 PM, Tejun Heo wrote: > Two notable fixes. > * Workqueue assumes that CPU <-> NUMA node mapping remains static. > This is a general assumption - we don't have any synchronization > mechanism around CPU <-> node mapping. Unfortunately,

Re: [GIT PULL] workqueue fixes for v4.13-rc3

Hi Tejun, On Mon, Jul 31, 2017 at 5:38 PM, Tejun Heo wrote: > Two notable fixes. > * Workqueue assumes that CPU <-> NUMA node mapping remains static. > This is a general assumption - we don't have any synchronization > mechanism around CPU <-> node mapping. Unfortunately, powerpc may >