From mboxrd@z Thu Jan 1 00:00:00 1970 From: Vitaly Kuznetsov Subject: Re: [PATCH] xen/events: Always allocate legacy interrupts on PV guests Date: Wed, 18 Nov 2015 15:28:51 +0100 Message-ID: <87d1v7v0t8.fsf__25444.1839679659$1447857032$gmane$org@vitty.brq.redhat.com> References: <1447774585-21857-1-git-send-email-boris.ostrovsky@oracle.com> <87lh9vv9pv.fsf@vitty.brq.redhat.com> <564C8531.2000608@oracle.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <564C8531.2000608@oracle.com> (Boris Ostrovsky's message of "Wed, 18 Nov 2015 09:03:29 -0500") List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Boris Ostrovsky Cc: jgross@suse.com, linux-kernel@vger.kernel.org, xen-devel@lists.xen.org, david.vrabel@citrix.com, tglx@linutronix.de List-Id: xen-devel@lists.xenproject.org Boris Ostrovsky writes: > On 11/18/2015 06:16 AM, Vitaly Kuznetsov wrote: >> Boris Ostrovsky writes: >> >>> After commit 8c058b0b9c34 ("x86/irq: Probe for PIC presence before >>> allocating descs for legacy IRQs") early_irq_init() will no longer >>> preallocate descriptors for legacy interrupts if PIT does not >>> exist. >> PIC? > > Right. David, can you fix this before committing? > >> >>> Therefore we need to allocate those descriptors for PV guests >>> ourselves. >>> >>> Signed-off-by: Boris Ostrovsky >>> Suggested-by: Thomas Gleixner >>> --- >>> drivers/xen/events/events_base.c | 8 ++++---- >>> 1 file changed, 4 insertions(+), 4 deletions(-) >>> >>> diff --git a/drivers/xen/events/events_base.c b/drivers/xen/events/events_base.c >>> index 849500e..a2bb333 100644 >>> --- a/drivers/xen/events/events_base.c >>> +++ b/drivers/xen/events/events_base.c >>> @@ -419,8 +419,8 @@ static int __must_check xen_allocate_irq_gsi(unsigned gsi) >>> if (xen_pv_domain() && !xen_initial_domain()) >>> return xen_allocate_irq_dynamic(); >>> >>> - /* Legacy IRQ descriptors are already allocated by the arch. */ >>> - if (gsi < NR_IRQS_LEGACY) >>> + /* On HVM legacy IRQ descriptors are already allocated by the arch. */ >>> + if (xen_hvm_domain() && gsi < NR_IRQS_LEGACY) >>> irq = gsi; >> Wouldn't it be better to write it as >> if (gsi < nr_legacy_irqs()) ? > > I don't think so: on PV we end up setting legacy_pic to > null_legacy_pic in probe_8259A() and that will make nr_legacy_irqs() > return 0. Yes, so the condition will always be false for PV and it equals to xen_hvm_domain() or am I missng something? > >> >> I think it's possible to have PIC-less HVM guests in future (btw, what >> about HVMlite?). I see nr_legacy_irqs() is x86-only but it can easily be >> defined to NR_IRQS_LEGACY on other arches. > > Yes, HVMlite (or however we will end up calling it) will allow these > sorts of thing. But HVMlite is not supported yet at all. And it's x86 > only. > > -boris > >> >>> else >>> irq = irq_alloc_desc_at(gsi, -1); >>> @@ -445,8 +445,8 @@ static void xen_free_irq(unsigned irq) >>> >>> kfree(info); >>> >>> - /* Legacy IRQ descriptors are managed by the arch. */ >>> - if (irq < NR_IRQS_LEGACY) >>> + /* On HVM legacy IRQ descriptors are managed by the arch. */ >>> + if (xen_hvm_domain() && irq < NR_IRQS_LEGACY) >>> return; >>> >>> irq_free_desc(irq); -- Vitaly