qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH 08/15] hw/intc/arm_gic: Make ICCBPR/GICC_BPR ban


From: Greg Bellows
Subject: Re: [Qemu-devel] [PATCH 08/15] hw/intc/arm_gic: Make ICCBPR/GICC_BPR banked
Date: Tue, 9 Sep 2014 18:10:21 -0500



On 22 August 2014 05:29, Fabian Aggeler <address@hidden> wrote:
This register is banked in GICs with Security Extensions. Storing the
non-secure copy of BPR in the abpr, which is an alias to the non-secure
copy for secure access. ABPR itself is only accessible from secure state
if the GIC implements Security Extensions.

Signed-off-by: Fabian Aggeler <address@hidden>
---
 hw/intc/arm_gic.c                | 25 +++++++++++++++++++++----
 include/hw/intc/arm_gic_common.h |  8 +++++---
 2 files changed, 26 insertions(+), 7 deletions(-)

diff --git a/hw/intc/arm_gic.c b/hw/intc/arm_gic.c
index 7f7fac3..57021fd 100644
--- a/hw/intc/arm_gic.c
+++ b/hw/intc/arm_gic.c
@@ -792,7 +792,12 @@ static uint32_t gic_cpu_read(GICState *s, int cpu, int offset)
     case 0x04: /* Priority mask */
         return s->priority_mask[cpu];
     case 0x08: /* Binary Point */
-        return s->bpr[cpu];
+        if (s->security_extn && ns_access()) {
+            /* BPR is banked. Non-secure copy stored in ABPR. */
+            return s->abpr[cpu];
+        } else {
+            return s->bpr[cpu];
+        }
     case 0x0c: /* Acknowledge */
         return gic_acknowledge_irq(s, cpu);
     case 0x14: /* Running Priority */
@@ -800,7 +805,14 @@ static uint32_t gic_cpu_read(GICState *s, int cpu, int offset)
     case 0x18: /* Highest Pending Interrupt */
         return s->current_pending[cpu];
     case 0x1c: /* Aliased Binary Point */
-        return s->abpr[cpu];
+        if ((s->security_extn && ns_access())) {
+            /* If Security Extensions are present ABPR is a secure register,
+             * only accessible from secure state.
+             */
+            return 0;
+        } else {
+            return s->abpr[cpu];
+        }
     case 0xd0: case 0xd4: case 0xd8: case 0xdc:
         return s->apr[(offset - 0xd0) / 4][cpu];
     default:
@@ -819,12 +831,17 @@ static void gic_cpu_write(GICState *s, int cpu, int offset, uint32_t value)
         s->priority_mask[cpu] = (value & 0xff);
         break;
     case 0x08: /* Binary Point */
-        s->bpr[cpu] = (value & 0x7);
+        if (s->security_extn && ns_access()) {
+            /* BPR is banked. Non-secure copy stored in ABPR. */
+            s->abpr[cpu] = (value & 0x7);
+        } else {
+            s->bpr[cpu] = (value & 0x7);
+        }
         break;
     case 0x10: /* End Of Interrupt */
         return gic_complete_irq(s, cpu, value & 0x3ff);
     case 0x1c: /* Aliased Binary Point */
-        if (s->revision >= 2) {
+        if (s->revision >= 2 && !(s->security_extn && ns_access())) {

According to to the v2 spec, this register is present in GICv1 if the security extensions are present but always in GICv2 (reason for the previous condition).  I think this needs to be rewritten to be :

if ((s->revision >= 2 && !s->security_extn) || (s->security_extn && !ns_access()) {
s->abpr[cpu] = (value & 0x7);
}
 
             s->abpr[cpu] = (value & 0x7);
         }
         break;
diff --git a/include/hw/intc/arm_gic_common.h b/include/hw/intc/arm_gic_common.h
index a912972..c547418 100644
--- a/include/hw/intc/arm_gic_common.h
+++ b/include/hw/intc/arm_gic_common.h
@@ -78,9 +78,11 @@ typedef struct GICState {
     uint16_t running_priority[GIC_NCPU];
     uint16_t current_pending[GIC_NCPU];

-    /* We present the GICv2 without security extensions to a guest and
-     * therefore the guest can configure the GICC_CTLR to configure group 1
-     * binary point in the abpr.
+    /* If we present the GICv2 without security extensions to a guest,
+     * the guest can configure the GICC_CTLR to configure group 1 binary point
+     * in the abpr.
+     * For a GIC with Security Extensions we use use bpr for the
+     * secure copy and abpr as storage for the non-secure copy of the register.
      */
     uint8_t  bpr[GIC_NCPU];
     uint8_t  abpr[GIC_NCPU];
--
1.8.3.2


Not sure if it occurs elsewhere, but these changes don't account for the GICC_CTLR.CBPR settings which affects the read value.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]