[v3,02/11] x86/hvm: drop domain parameter from vioapic/vpic EOI callbacks

Commit Message

Roger Pau Monne March 31, 2021, 10:32 a.m. UTC

EOIs are always executed in guest vCPU context, so there's no reason to
pass a domain parameter around as can be fetched from current->domain.

No functional change intended.

Signed-off-by: Roger Pau Monné <[e-mail]>
Reviewed-by: Paul Durrant <[e-mail]>
---
Changes since v1:
 - New in this version.
---
 xen/arch/x86/hvm/vioapic.c        | 4 ++--
 xen/arch/x86/hvm/vpic.c           | 6 ++----
 xen/drivers/passthrough/x86/hvm.c | 3 ++-
 xen/include/asm-x86/hvm/io.h      | 2 +-
 4 files changed, 7 insertions(+), 8 deletions(-)

Comments

Jan Beulich March 31, 2021, 4:04 p.m. UTC | #1

On 31.03.2021 12:32, Roger Pau Monne wrote:
> EOIs are always executed in guest vCPU context, so there's no reason to
> pass a domain parameter around as can be fetched from current->domain.
> 
> No functional change intended.
> 
> Signed-off-by: Roger Pau Monné <[e-mail]>
> Reviewed-by: Paul Durrant <[e-mail]>
> ---
> Changes since v1:
>  - New in this version.

Just to make it explicit - possibly same thing as with patch 1,
depending on how exactly the issue there gets taken care of.

Jan

Roger Pau Monne April 1, 2021, 9:15 a.m. UTC | #2

On Wed, Mar 31, 2021 at 06:04:43PM +0200, Jan Beulich wrote:
> On 31.03.2021 12:32, Roger Pau Monne wrote:
> > EOIs are always executed in guest vCPU context, so there's no reason to
> > pass a domain parameter around as can be fetched from current->domain.
> > 
> > No functional change intended.
> > 
> > Signed-off-by: Roger Pau Monné <[e-mail]>
> > Reviewed-by: Paul Durrant <[e-mail]>
> > ---
> > Changes since v1:
> >  - New in this version.
> 
> Just to make it explicit - possibly same thing as with patch 1,
> depending on how exactly the issue there gets taken care of.

I don't think we have any scenario ATM where we allow EOI'ing of
interrupts from a different vCPU context, neither I can see us
allowing such in the future, but I don't want this discussion to block
the series, hence I'm going to drop this patch together with patch 1.

Roger.

Jan Beulich April 1, 2021, 9:28 a.m. UTC | #3

On 01.04.2021 11:15, Roger Pau Monné wrote:
> On Wed, Mar 31, 2021 at 06:04:43PM +0200, Jan Beulich wrote:
>> On 31.03.2021 12:32, Roger Pau Monne wrote:
>>> EOIs are always executed in guest vCPU context, so there's no reason to
>>> pass a domain parameter around as can be fetched from current->domain.
>>>
>>> No functional change intended.
>>>
>>> Signed-off-by: Roger Pau Monné <[e-mail]>
>>> Reviewed-by: Paul Durrant <[e-mail]>
>>> ---
>>> Changes since v1:
>>>  - New in this version.
>>
>> Just to make it explicit - possibly same thing as with patch 1,
>> depending on how exactly the issue there gets taken care of.
> 
> I don't think we have any scenario ATM where we allow EOI'ing of
> interrupts from a different vCPU context, neither I can see us
> allowing such in the future, but I don't want this discussion to block
> the series, hence I'm going to drop this patch together with patch 1.

Well, having seen also your reply to Andrew wrt patch 1, I just wanted
to clarify that dropping the two patches isn't the only option. Making
Viridian code resilient in this regard might be another one.

Jan

Patch

diff --git a/xen/arch/x86/hvm/vioapic.c b/xen/arch/x86/hvm/vioapic.c
index 91e5f892787..dcc2de76489 100644
--- a/xen/arch/x86/hvm/vioapic.c
+++ b/xen/arch/x86/hvm/vioapic.c
@@ -284,7 +284,7 @@  static void vioapic_write_redirent(
              */
             ASSERT(prev_level);
             ASSERT(!top_word);
-            hvm_dpci_eoi(d, gsi);
+            hvm_dpci_eoi(gsi);
     }
 
     if ( is_hardware_domain(d) && unmasked )
@@ -541,7 +541,7 @@  void vioapic_update_EOI(unsigned int vector)
             if ( is_iommu_enabled(d) )
             {
                 spin_unlock(&d->arch.hvm.irq_lock);
-                hvm_dpci_eoi(d, vioapic->base_gsi + pin);
+                hvm_dpci_eoi(vioapic->base_gsi + pin);
                 spin_lock(&d->arch.hvm.irq_lock);
             }
 
diff --git a/xen/arch/x86/hvm/vpic.c b/xen/arch/x86/hvm/vpic.c
index f465b7f9979..a69aecad912 100644
--- a/xen/arch/x86/hvm/vpic.c
+++ b/xen/arch/x86/hvm/vpic.c
@@ -235,8 +235,7 @@  static void vpic_ioport_write(
                 unsigned int pin = __scanbit(pending, 8);
 
                 ASSERT(pin < 8);
-                hvm_dpci_eoi(current->domain,
-                             hvm_isa_irq_to_gsi((addr >> 7) ? (pin | 8) : pin));
+                hvm_dpci_eoi(hvm_isa_irq_to_gsi((addr >> 7) ? (pin | 8) : pin));
                 __clear_bit(pin, &pending);
             }
             return;
@@ -285,8 +284,7 @@  static void vpic_ioport_write(
                 /* Release lock and EOI the physical interrupt (if any). */
                 vpic_update_int_output(vpic);
                 vpic_unlock(vpic);
-                hvm_dpci_eoi(current->domain,
-                             hvm_isa_irq_to_gsi((addr >> 7) ? (pin | 8) : pin));
+                hvm_dpci_eoi(hvm_isa_irq_to_gsi((addr >> 7) ? (pin | 8) : pin));
                 return; /* bail immediately */
             case 6: /* Set Priority                */
                 vpic->priority_add = (val + 1) & 7;
diff --git a/xen/drivers/passthrough/x86/hvm.c b/xen/drivers/passthrough/x86/hvm.c
index 2f6c81b1e2c..a9256e7ef51 100644
--- a/xen/drivers/passthrough/x86/hvm.c
+++ b/xen/drivers/passthrough/x86/hvm.c
@@ -911,8 +911,9 @@  static void hvm_gsi_eoi(struct domain *d, unsigned int gsi)
     hvm_pirq_eoi(pirq);
 }
 
-void hvm_dpci_eoi(struct domain *d, unsigned int guest_gsi)
+void hvm_dpci_eoi(unsigned int guest_gsi)
 {
+    struct domain *d = current->domain;
     const struct hvm_irq_dpci *hvm_irq_dpci;
     const struct hvm_girq_dpci_mapping *girq;
 
diff --git a/xen/include/asm-x86/hvm/io.h b/xen/include/asm-x86/hvm/io.h
index 8b8392ec59e..4f294232fb7 100644
--- a/xen/include/asm-x86/hvm/io.h
+++ b/xen/include/asm-x86/hvm/io.h
@@ -101,7 +101,7 @@  bool handle_mmio_with_translation(unsigned long gla, unsigned long gpfn,
                                   struct npfec);
 bool handle_pio(uint16_t port, unsigned int size, int dir);
 void hvm_interrupt_post(struct vcpu *v, int vector, int type);
-void hvm_dpci_eoi(struct domain *d, unsigned int guest_irq);
+void hvm_dpci_eoi(unsigned int guest_irq);
 void msix_write_completion(struct vcpu *);
 
 #ifdef CONFIG_HVM