Skip to contentSkip to conversation

hw/arm/virt-acpi-build: Fix IORT id_count

PATCH Superseded Nicolin Chen 2 participants 3 messages Jun 13, 2024 – Jun 17, 2024

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 hw/arm/virt-acpi-build: Fix IORT id_count

The IORT doc defines "Number of IDs" ("id_count" in the virt-acpi-build) to be "the number of IDs in the range minus one". Otherwise, Linux kernel reports "conflicting mapping for input ID" FW_BUG at the overlapped ID.

Fixes 42e0f050e3a5 ("hw/arm/virt-acpi-build: Add IORT support to bypass SMMUv3")
---
 hw/arm/virt-acpi-build.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
hw/arm/virt-acpi-build.c +4 -1
index c3ccfef026..b9343dde0f 100644
@@ -243,7 +243,8 @@ iort_host_bridges(Object *obj, void *opaque)
 
             AcpiIortIdMapping idmap = {
                 .input_base = min_bus << 8,
-                .id_count = (max_bus - min_bus + 1) << 8,
+                /* id_count is the number of IDs in the range minus one */
+                .id_count = ((max_bus - min_bus + 1) << 8) - 1,
             };
             g_array_append_val(idmap_blob, idmap);
         }
@@ -298,7 +299,9 @@ build_iort(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms)
             idmap = &g_array_index(smmu_idmaps, AcpiIortIdMapping, i);
 
             if (next_range.input_base < idmap->input_base) {
+                /* id_count is the number of IDs in the range minus one */
                 next_range.id_count = idmap->input_base - next_range.input_base;
+                next_range.id_count -= 1;
                 g_array_append_val(its_idmaps, next_range);
             }
 
Signature · 1 line
2.43.0
#1 lore raw
Quoted 25 lines · Nicolin Chen

The IORT doc defines "Number of IDs" ("id_count" in the virt-acpi-build) to be "the number of IDs in the range minus one". Otherwise, Linux kernel reports "conflicting mapping for input ID" FW_BUG at the overlapped ID.

Fixes 42e0f050e3a5 ("hw/arm/virt-acpi-build: Add IORT support to bypass SMMUv3")
---
 hw/arm/virt-acpi-build.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
hw/arm/virt-acpi-build.c +2 -1
index c3ccfef026..b9343dde0f 100644
@@ -243,7 +243,8 @@ iort_host_bridges(Object *obj, void *opaque)
 
             AcpiIortIdMapping idmap = {
                 .input_base = min_bus << 8,
-                .id_count = (max_bus - min_bus + 1) << 8,
+                /* id_count is the number of IDs in the range minus one */
+                .id_count = ((max_bus - min_bus + 1) << 8) - 1,
             };
             g_array_append_val(idmap_blob, idmap);
         }

hw/arm/virt-acpi-build.c build_iort(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms) · line 304

Quoted 6 lines
hw/arm/virt-acpi-build.c +2 -0
@@ -298,7 +299,9 @@ build_iort(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms)
             idmap = &g_array_index(smmu_idmaps, AcpiIortIdMapping, i);
 
             if (next_range.input_base < idmap->input_base) {
+                /* id_count is the number of IDs in the range minus one */
                 next_range.id_count = idmap->input_base - next_range.input_base;
+                next_range.id_count -= 1;

I would just add - 1 on the previous line, instead of making it incorrect then correcting it.

hw/arm/virt-acpi-build.c build_iort(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms) · line 306

Quoted 2 lines
hw/arm/virt-acpi-build.c +0 -0
                 g_array_append_val(its_idmaps, next_range);
             }

But the value is used later:

            next_range.input_base = idmap->input_base + idmap->id_count;

Wouldn't that make next_range incorrect?

I also note that

static void build_iort_id_mapping(GArray *table_data, uint32_t input_base,

                                  uint32_t id_count, uint32_t out_ref)

{
/* Table 4 ID mapping format */ build_append_int_noprefix(table_data, input_base, 4); /* Input base */ build_append_int_noprefix(table_data, id_count, 4); /* Number of IDs */ build_append_int_noprefix(table_data, input_base, 4); /* Output base */ build_append_int_noprefix(table_data, out_ref, 4); /* Output Reference */ /* Flags */
build_append_int_noprefix(table_data, 0 /* Single mapping (disabled) */, 4);
}

That comment

    /* Table 4 ID mapping format */

really should be before the function and it should mention the spec it's from - specifically the earliest spec including the relevant table.

Quoted 3 lines
Signature · 2 lines
-- 
2.43.0
#2 lore raw
Quoted 28 lines · Michael S. Tsirkin
Quoted 27 lines · Nicolin Chen

The IORT doc defines "Number of IDs" ("id_count" in the virt-acpi-build) to be "the number of IDs in the range minus one". Otherwise, Linux kernel reports "conflicting mapping for input ID" FW_BUG at the overlapped ID.

Fixes 42e0f050e3a5 ("hw/arm/virt-acpi-build: Add IORT support to bypass SMMUv3")
---
 hw/arm/virt-acpi-build.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
hw/arm/virt-acpi-build.c +2 -1
index c3ccfef026..b9343dde0f 100644
@@ -243,7 +243,8 @@ iort_host_bridges(Object *obj, void *opaque)
 
             AcpiIortIdMapping idmap = {
                 .input_base = min_bus << 8,
-                .id_count = (max_bus - min_bus + 1) << 8,
+                /* id_count is the number of IDs in the range minus one */
+                .id_count = ((max_bus - min_bus + 1) << 8) - 1,
             };
             g_array_append_val(idmap_blob, idmap);
         }
@@ -298,7 +299,9 @@ build_iort(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms)
             idmap = &g_array_index(smmu_idmaps, AcpiIortIdMapping, i);
 
Quoted 7 lines

hw/arm/virt-acpi-build.c build_iort(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms) · line 304

Quoted 4 lines
hw/arm/virt-acpi-build.c +2 -0
             if (next_range.input_base < idmap->input_base) {
+                /* id_count is the number of IDs in the range minus one */
                 next_range.id_count = idmap->input_base - next_range.input_base;
+                next_range.id_count -= 1;

I would just add - 1 on the previous line, instead of making it incorrect then correcting it.

OK. Let's do this then:

-                next_range.id_count = idmap->input_base - next_range.input_base;
+                /* id_count is the number of IDs in the range minus one */
+                next_range.id_count = idmap->input_base -

+ next_range.input_base - 1;

Quoted 10 lines

hw/arm/virt-acpi-build.c build_iort(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms) · line 306

Quoted 3 lines
hw/arm/virt-acpi-build.c +0 -0
                 g_array_append_val(its_idmaps, next_range);
             }

But the value is used later:

            next_range.input_base = idmap->input_base + idmap->id_count;

Wouldn't that make next_range incorrect?

Ah, missed that. Thanks!

- next_range.input_base = idmap->input_base + idmap->id_count;
+ next_range.input_base = idmap->input_base + idmap->id_count + 1;

Quoted 16 lines

I also note that

static void build_iort_id_mapping(GArray *table_data, uint32_t input_base,

                                  uint32_t id_count, uint32_t out_ref)

{
/* Table 4 ID mapping format */ build_append_int_noprefix(table_data, input_base, 4); /* Input base */ build_append_int_noprefix(table_data, id_count, 4); /* Number of IDs */ build_append_int_noprefix(table_data, input_base, 4); /* Output base */ build_append_int_noprefix(table_data, out_ref, 4); /* Output Reference */ /* Flags */
build_append_int_noprefix(table_data, 0 /* Single mapping (disabled) */, 4);
}

Quoted 5 lines

That comment

    /* Table 4 ID mapping format */

really should be before the function and it should mention the spec it's from - specifically the earliest spec including the relevant table.

Well, it's an inline-like function called by build_iort() only, where there is a function header mentioning the doc.

Should this function have to repeat?

Anyway, that would be a different patch. I'll submit a v2 first.

Thank you
Nicolin