Re: [RFC PATCH kernel] powerpc/xive: Drop deregistered irqs

2019-07-14 Thread Alexey Kardashevskiy
On 15/07/2019 05:14, Daniel Henrique Barboza wrote: This patch fixed an issue I was experiencing with virsh start/destroy of guests with mlx5 and GPU passthrough in a Power 9 server. I believe it's a similar situation which Alexey described in the post commit msg. Tested-by: Daniel Henrique

Re: [RFC PATCH kernel] powerpc/xive: Drop deregistered irqs

2019-07-14 Thread Benjamin Herrenschmidt
On Sun, 2019-07-14 at 21:44 +0200, Cédric Le Goater wrote: > > Well, best is probably to do just that though, but call it something > > like ppc_md.orphan_irq() or something like that instead. Another option > > as you mention is to try to scrub queues, but that's trickier to do due > > to the

Re: [RFC PATCH kernel] powerpc/xive: Drop deregistered irqs

2019-07-14 Thread Daniel Henrique Barboza
This patch fixed an issue I was experiencing with virsh start/destroy of guests with mlx5 and GPU passthrough in a Power 9 server. I believe it's a similar situation which Alexey described in the post commit msg. Tested-by: Daniel Henrique Barboza On 7/12/19 5:20 AM, Alexey Kardashevskiy

Re: [RFC PATCH kernel] powerpc/xive: Drop deregistered irqs

2019-07-14 Thread Cédric Le Goater
On 14/07/2019 03:31, Benjamin Herrenschmidt wrote: > On Sat, 2019-07-13 at 18:53 +1000, Alexey Kardashevskiy wrote: >> >> On 13/07/2019 09:47, Benjamin Herrenschmidt wrote: >>> On Fri, 2019-07-12 at 19:37 +1000, Alexey Kardashevskiy wrote: >>> >>> >

Re: [RFC PATCH kernel] powerpc/xive: Drop deregistered irqs

2019-07-13 Thread Benjamin Herrenschmidt
On Sat, 2019-07-13 at 18:53 +1000, Alexey Kardashevskiy wrote: > > On 13/07/2019 09:47, Benjamin Herrenschmidt wrote: > > On Fri, 2019-07-12 at 19:37 +1000, Alexey Kardashevskiy wrote: > > > > > > >

Re: [RFC PATCH kernel] powerpc/xive: Drop deregistered irqs

2019-07-13 Thread Alexey Kardashevskiy
On 13/07/2019 09:47, Benjamin Herrenschmidt wrote: > On Fri, 2019-07-12 at 19:37 +1000, Alexey Kardashevskiy wrote: >> > > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/powerpc/kernel/irq.c#n614 >> >> If so, then in order to do EOI, I'll need the desc which is

Re: [RFC PATCH kernel] powerpc/xive: Drop deregistered irqs

2019-07-12 Thread Benjamin Herrenschmidt
On Fri, 2019-07-12 at 19:37 +1000, Alexey Kardashevskiy wrote: > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/powerpc/kernel/irq.c#n614 > > If so, then in order to do EOI, I'll need the desc which is gone, or > I am missing the point? All you need is drop the

Re: [RFC PATCH kernel] powerpc/xive: Drop deregistered irqs

2019-07-12 Thread Alexey Kardashevskiy
On 12/07/2019 18:29, Benjamin Herrenschmidt wrote: On Fri, 2019-07-12 at 18:20 +1000, Alexey Kardashevskiy wrote: diff --git a/arch/powerpc/sysdev/xive/common.c b/arch/powerpc/sysdev/xive/common.c index 082c7e1c20f0..65742e280337 100644 --- a/arch/powerpc/sysdev/xive/common.c +++

Re: [RFC PATCH kernel] powerpc/xive: Drop deregistered irqs

2019-07-12 Thread Benjamin Herrenschmidt
On Fri, 2019-07-12 at 18:20 +1000, Alexey Kardashevskiy wrote: > > diff --git a/arch/powerpc/sysdev/xive/common.c > b/arch/powerpc/sysdev/xive/common.c > index 082c7e1c20f0..65742e280337 100644 > --- a/arch/powerpc/sysdev/xive/common.c > +++ b/arch/powerpc/sysdev/xive/common.c > @@ -148,8

[RFC PATCH kernel] powerpc/xive: Drop deregistered irqs

2019-07-12 Thread Alexey Kardashevskiy
There is a race between releasing an irq on one cpu and fetching it from XIVE on another cpu as there does not seem to be any locking between these, probably because xive_irq_chip::irq_shutdown() is supposed to remove the irq from all queues in the system which it does not do. As a result, when