[PATCH v5 0/6] igvm: Supply MADT via IGVM parameter

[PATCH v5 0/6] igvm: Supply MADT via IGVM parameter

[Oliver Steffen]

When launching using an IGVM file, supply a copy of the MADT (part of the ACPI
tables) via an IGVM parameter (IGVM_VHY_MADT) to the guest, in addition to the
regular fw_cfg mechanism.

The IGVM parameter can be consumed by Coconut SVSM [1], instead of relying on
the fw_cfg interface, which has caused problems before due to unexpected access
[2,3]. Using IGVM parameters is the default way for Coconut SVSM; switching
over would allow removing specialized code paths for QEMU in Coconut.

Coconut SVSM needs to know the SMP configuration, but does not look at
any other ACPI data, nor does it interact with the PCI bus settings.
Since the MADT is static and not linked with other ACPI tables, it can
be supplied stand-alone like this.

Generating the MADT twice (IGVM processing and ACPI table building) seems
acceptable, since there is no infrastructure to obtain the MADT out of the ACPI
table memory area during IGVM processing.

In any case OVMF, which runs after SVSM has already been initialized, will
continue reading all ACPI tables via fw_cfg and provide fixed up ACPI data to
the OS as before.

This series makes ACPI table building more generic by making the BIOS linker
optional. This allows the MADT to be generated outside of the ACPI build
context. A new function (acpi_build_madt_standalone()) is added for that. With
that, the IGVM MADT parameter field can be filled with the MADT data during
processing of the IGVM file.

[1] https://github.com/coconut-svsm/svsm/pull/858
[2] https://gitlab.com/qemu-project/qemu/-/issues/2882
[3] https://github.com/coconut-svsm/svsm/issues/646

v5:
- Make qigvm_find_parameter_entry() more generic and use everywhere possible;
  It now takes the index as input, instead of the full parameter struct
- Consistently use early returns when looking up parameter entries
- Emit a warning message if no parameter area can be found for a given index

v4:
- Add ACPI table checksum calculation without BIOS linker, used
  for the standalone MADT.
- Don't pass ConfidentialGuestState into the IGVM backend anymore.
  Not needed, since we already have the full MachineState there now.
- Move the NULL check patch out into a new series (to be posted).
- Address remaining cleanup comments.

v3:
- Pass the machine state into IGVM file processing context instead of MADT data
- Generate MADT from inside the IGVM backend
- Refactor: Extract common code for finding IGVM parameter from IGVM parameter handlers
- Add NULL pointer check for igvm_get_buffer()

v2:
- Provide more context in the message of the main commit
- Document the madt parameter of IgvmCfgClass::process()
- Document why no MADT data is provided the process call in sev.c

Based-on: <20251118122133.1695767-1-kraxel@redhat.com>
Signed-off-by: Oliver Steffen <osteffen@redhat.com>

Oliver Steffen (6):
  hw/acpi: Make BIOS linker optional
  hw/acpi: Add standalone function to build MADT
  igvm: Add common function for finding parameter entries
  igvm: Refactor qigvm_parameter_insert
  igvm: Pass machine state to IGVM file processing
  igvm: Fill MADT IGVM parameter field

 backends/igvm-cfg.c       |   2 +-
 backends/igvm.c           | 255 +++++++++++++++++++++++---------------
 hw/acpi/aml-build.c       |  29 ++++-
 hw/i386/acpi-build.c      |   9 ++
 hw/i386/acpi-build.h      |   2 +
 include/system/igvm-cfg.h |   3 +-
 include/system/igvm.h     |   5 +-
 target/i386/sev.c         |   3 +-
 8 files changed, 197 insertions(+), 111 deletions(-)

-- 
2.52.0

[PATCH v5 1/6] hw/acpi: Make BIOS linker optional

[Oliver Steffen]

Make the BIOS linker optional in acpi_table_end() and calculate the ACPI
table checksum directly if no linker is provided.

This makes it possible to call for example
acpi_build_madt() from outside the ACPI table builder context.

Signed-off-by: Oliver Steffen <osteffen@redhat.com>
---
 hw/acpi/aml-build.c | 29 +++++++++++++++++++++++++++--
 1 file changed, 27 insertions(+), 2 deletions(-)

diff --git a/hw/acpi/aml-build.c b/hw/acpi/aml-build.c
index dad4cfcc7d..6a3650076f 100644
--- a/hw/acpi/aml-build.c
+++ b/hw/acpi/aml-build.c
@@ -1730,6 +1730,25 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
 }
 
+static uint8_t calculate_acpi_checksum(const gchar *data, size_t len)
+{
+    size_t i;
+    uint8_t sum = 0;
+
+    for (i = 0; i < len; ++i) {
+        sum += (uint8_t)data[i];
+    }
+
+    return sum;
+}
+
+static void update_acpi_checksum(gchar *data, size_t start_offset,
+                                 size_t table_len, size_t checksum_offset)
+{
+    uint8_t sum = calculate_acpi_checksum(&data[start_offset], table_len);
+    data[checksum_offset] = 0xff - sum + 1;
+}
+
 void acpi_table_end(BIOSLinker *linker, AcpiTable *desc)
 {
     /*
@@ -1748,8 +1767,14 @@ void acpi_table_end(BIOSLinker *linker, AcpiTable *desc)
      */
     memcpy(len_ptr, &table_len_le, sizeof table_len_le);
 
-    bios_linker_loader_add_checksum(linker, ACPI_BUILD_TABLE_FILE,
-        desc->table_offset, table_len, desc->table_offset + checksum_offset);
+    if (linker != NULL) {
+        bios_linker_loader_add_checksum(linker, ACPI_BUILD_TABLE_FILE,
+                                        desc->table_offset, table_len,
+                                        desc->table_offset + checksum_offset);
+    } else {
+        update_acpi_checksum(desc->array->data, desc->table_offset,
+                             table_len, desc->table_offset + checksum_offset);
+    }
 }
 
 void *acpi_data_push(GArray *table_data, unsigned size)
-- 
2.52.0

[PATCH v5 0/6] igvm: Supply MADT via IGVM parameter

[Oliver Steffen]

When launching using an IGVM file, supply a copy of the MADT (part of the ACPI
tables) via an IGVM parameter (IGVM_VHY_MADT) to the guest, in addition to the
regular fw_cfg mechanism.

The IGVM parameter can be consumed by Coconut SVSM [1], instead of relying on
the fw_cfg interface, which has caused problems before due to unexpected access
[2,3]. Using IGVM parameters is the default way for Coconut SVSM; switching
over would allow removing specialized code paths for QEMU in Coconut.

Coconut SVSM needs to know the SMP configuration, but does not look at
any other ACPI data, nor does it interact with the PCI bus settings.
Since the MADT is static and not linked with other ACPI tables, it can
be supplied stand-alone like this.

Generating the MADT twice (IGVM processing and ACPI table building) seems
acceptable, since there is no infrastructure to obtain the MADT out of the ACPI
table memory area during IGVM processing.

In any case OVMF, which runs after SVSM has already been initialized, will
continue reading all ACPI tables via fw_cfg and provide fixed up ACPI data to
the OS as before.

This series makes ACPI table building more generic by making the BIOS linker
optional. This allows the MADT to be generated outside of the ACPI build
context. A new function (acpi_build_madt_standalone()) is added for that. With
that, the IGVM MADT parameter field can be filled with the MADT data during
processing of the IGVM file.

[1] https://github.com/coconut-svsm/svsm/pull/858
[2] https://gitlab.com/qemu-project/qemu/-/issues/2882
[3] https://github.com/coconut-svsm/svsm/issues/646

v5:
- Rebase on latest IGVM rework series from Gerd
- Make qigvm_find_parameter_entry() more generic and use everywhere possible;
  It now takes the index as input, instead of the full parameter struct
- Consistently use early returns when looking up parameter entries
- Emit a warning message if no parameter area can be found for a given index

v4:
- Add ACPI table checksum calculation without BIOS linker, used
  for the standalone MADT.
- Don't pass ConfidentialGuestState into the IGVM backend anymore.
  Not needed, since we already have the full MachineState there now.
- Move the NULL check patch out into a new series (to be posted).
- Address remaining cleanup comments.

v3:
- Pass the machine state into IGVM file processing context instead of MADT data
- Generate MADT from inside the IGVM backend
- Refactor: Extract common code for finding IGVM parameter from IGVM parameter handlers
- Add NULL pointer check for igvm_get_buffer()

v2:
- Provide more context in the message of the main commit
- Document the MADT parameter of IgvmCfgClass::process()
- Document why no MADT data is provided the process call in sev.c

Based-on: <20260126123755.357378-1-kraxel@redhat.com>
Signed-off-by: Oliver Steffen <osteffen@redhat.com>

Oliver Steffen (6):
  hw/acpi: Make BIOS linker optional
  hw/acpi: Add standalone function to build MADT
  igvm: Add common function for finding parameter entries
  igvm: Refactor qigvm_parameter_insert
  igvm: Pass machine state to IGVM file processing
  igvm: Fill MADT IGVM parameter field

 backends/igvm-cfg.c       |   2 +-
 backends/igvm.c           | 255 +++++++++++++++++++++++---------------
 hw/acpi/aml-build.c       |  29 ++++-
 hw/i386/acpi-build.c      |   9 ++
 hw/i386/acpi-build.h      |   2 +
 include/system/igvm-cfg.h |   3 +-
 include/system/igvm.h     |   5 +-
 target/i386/sev.c         |   3 +-
 8 files changed, 197 insertions(+), 111 deletions(-)

-- 
2.52.0

[PATCH v5 1/6] hw/acpi: Make BIOS linker optional

[Oliver Steffen]

Make the BIOS linker optional in acpi_table_end() and calculate the ACPI
table checksum directly if no linker is provided.

This makes it possible to call for example
acpi_build_madt() from outside the ACPI table builder context.

Signed-off-by: Oliver Steffen <osteffen@redhat.com>
---
 hw/acpi/aml-build.c | 29 +++++++++++++++++++++++++++--
 1 file changed, 27 insertions(+), 2 deletions(-)

diff --git a/hw/acpi/aml-build.c b/hw/acpi/aml-build.c
index dad4cfcc7d..6a3650076f 100644
--- a/hw/acpi/aml-build.c
+++ b/hw/acpi/aml-build.c
@@ -1730,6 +1730,25 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
 }
 
+static uint8_t calculate_acpi_checksum(const gchar *data, size_t len)
+{
+    size_t i;
+    uint8_t sum = 0;
+
+    for (i = 0; i < len; ++i) {
+        sum += (uint8_t)data[i];
+    }
+
+    return sum;
+}
+
+static void update_acpi_checksum(gchar *data, size_t start_offset,
+                                 size_t table_len, size_t checksum_offset)
+{
+    uint8_t sum = calculate_acpi_checksum(&data[start_offset], table_len);
+    data[checksum_offset] = 0xff - sum + 1;
+}
+
 void acpi_table_end(BIOSLinker *linker, AcpiTable *desc)
 {
     /*
@@ -1748,8 +1767,14 @@ void acpi_table_end(BIOSLinker *linker, AcpiTable *desc)
      */
     memcpy(len_ptr, &table_len_le, sizeof table_len_le);
 
-    bios_linker_loader_add_checksum(linker, ACPI_BUILD_TABLE_FILE,
-        desc->table_offset, table_len, desc->table_offset + checksum_offset);
+    if (linker != NULL) {
+        bios_linker_loader_add_checksum(linker, ACPI_BUILD_TABLE_FILE,
+                                        desc->table_offset, table_len,
+                                        desc->table_offset + checksum_offset);
+    } else {
+        update_acpi_checksum(desc->array->data, desc->table_offset,
+                             table_len, desc->table_offset + checksum_offset);
+    }
 }
 
 void *acpi_data_push(GArray *table_data, unsigned size)
-- 
2.52.0

[PATCH v5 0/6] igvm: Supply MADT via IGVM parameter

[Oliver Steffen]

When launching using an IGVM file, supply a copy of the MADT (part of the ACPI
tables) via an IGVM parameter (IGVM_VHY_MADT) to the guest, in addition to the
regular fw_cfg mechanism.

The IGVM parameter can be consumed by Coconut SVSM [1], instead of relying on
the fw_cfg interface, which has caused problems before due to unexpected access
[2,3]. Using IGVM parameters is the default way for Coconut SVSM; switching
over would allow removing specialized code paths for QEMU in Coconut.

Coconut SVSM needs to know the SMP configuration, but does not look at
any other ACPI data, nor does it interact with the PCI bus settings.
Since the MADT is static and not linked with other ACPI tables, it can
be supplied stand-alone like this.

Generating the MADT twice (IGVM processing and ACPI table building) seems
acceptable, since there is no infrastructure to obtain the MADT out of the ACPI
table memory area during IGVM processing.

In any case OVMF, which runs after SVSM has already been initialized, will
continue reading all ACPI tables via fw_cfg and provide fixed up ACPI data to
the OS as before.

This series makes ACPI table building more generic by making the BIOS linker
optional. This allows the MADT to be generated outside of the ACPI build
context. A new function (acpi_build_madt_standalone()) is added for that. With
that, the IGVM MADT parameter field can be filled with the MADT data during
processing of the IGVM file.

[1] https://github.com/coconut-svsm/svsm/pull/858
[2] https://gitlab.com/qemu-project/qemu/-/issues/2882
[3] https://github.com/coconut-svsm/svsm/issues/646

v5:
- Rebase on latest IGVM rework series from Gerd
- Make qigvm_find_parameter_entry() more generic and use everywhere possible;
  It now takes the index as input, instead of the full parameter struct
- Consistently use early returns when looking up parameter entries
- Emit a warning message if no parameter area can be found for a given index

v4:
- Add ACPI table checksum calculation without BIOS linker, used
  for the standalone MADT.
- Don't pass ConfidentialGuestState into the IGVM backend anymore.
  Not needed, since we already have the full MachineState there now.
- Move the NULL check patch out into a new series (to be posted).
- Address remaining cleanup comments.

v3:
- Pass the machine state into IGVM file processing context instead of MADT data
- Generate MADT from inside the IGVM backend
- Refactor: Extract common code for finding IGVM parameter from IGVM parameter handlers
- Add NULL pointer check for igvm_get_buffer()

v2:
- Provide more context in the message of the main commit
- Document the MADT parameter of IgvmCfgClass::process()
- Document why no MADT data is provided the process call in sev.c

Based-on: <20260126123755.357378-1-kraxel@redhat.com>
Signed-off-by: Oliver Steffen <osteffen@redhat.com>

Oliver Steffen (6):
  hw/acpi: Make BIOS linker optional
  hw/acpi: Add standalone function to build MADT
  igvm: Add common function for finding parameter entries
  igvm: Refactor qigvm_parameter_insert
  igvm: Pass machine state to IGVM file processing
  igvm: Fill MADT IGVM parameter field

 backends/igvm-cfg.c       |   2 +-
 backends/igvm.c           | 255 +++++++++++++++++++++++---------------
 hw/acpi/aml-build.c       |  29 ++++-
 hw/i386/acpi-build.c      |   9 ++
 hw/i386/acpi-build.h      |   2 +
 include/system/igvm-cfg.h |   3 +-
 include/system/igvm.h     |   5 +-
 target/i386/sev.c         |   3 +-
 8 files changed, 197 insertions(+), 111 deletions(-)

-- 
2.52.0

[PATCH v5 1/6] hw/acpi: Make BIOS linker optional

[Oliver Steffen]

Make the BIOS linker optional in acpi_table_end() and calculate the ACPI
table checksum directly if no linker is provided.

This makes it possible to call for example
acpi_build_madt() from outside the ACPI table builder context.

Signed-off-by: Oliver Steffen <osteffen@redhat.com>
---
 hw/acpi/aml-build.c | 29 +++++++++++++++++++++++++++--
 1 file changed, 27 insertions(+), 2 deletions(-)

diff --git a/hw/acpi/aml-build.c b/hw/acpi/aml-build.c
index dad4cfcc7d..6a3650076f 100644
--- a/hw/acpi/aml-build.c
+++ b/hw/acpi/aml-build.c
@@ -1730,6 +1730,25 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
 }
 
+static uint8_t calculate_acpi_checksum(const gchar *data, size_t len)
+{
+    size_t i;
+    uint8_t sum = 0;
+
+    for (i = 0; i < len; ++i) {
+        sum += (uint8_t)data[i];
+    }
+
+    return sum;
+}
+
+static void update_acpi_checksum(gchar *data, size_t start_offset,
+                                 size_t table_len, size_t checksum_offset)
+{
+    uint8_t sum = calculate_acpi_checksum(&data[start_offset], table_len);
+    data[checksum_offset] = 0xff - sum + 1;
+}
+
 void acpi_table_end(BIOSLinker *linker, AcpiTable *desc)
 {
     /*
@@ -1748,8 +1767,14 @@ void acpi_table_end(BIOSLinker *linker, AcpiTable *desc)
      */
     memcpy(len_ptr, &table_len_le, sizeof table_len_le);
 
-    bios_linker_loader_add_checksum(linker, ACPI_BUILD_TABLE_FILE,
-        desc->table_offset, table_len, desc->table_offset + checksum_offset);
+    if (linker != NULL) {
+        bios_linker_loader_add_checksum(linker, ACPI_BUILD_TABLE_FILE,
+                                        desc->table_offset, table_len,
+                                        desc->table_offset + checksum_offset);
+    } else {
+        update_acpi_checksum(desc->array->data, desc->table_offset,
+                             table_len, desc->table_offset + checksum_offset);
+    }
 }
 
 void *acpi_data_push(GArray *table_data, unsigned size)
-- 
2.52.0

Re: [PATCH v5 1/6] hw/acpi: Make BIOS linker optional

[Luigi Leonardi]

Hi Oliver,

On Tue, Jan 27, 2026 at 11:02:52AM +0100, Oliver Steffen wrote:
>Make the BIOS linker optional in acpi_table_end() and calculate the ACPI
>table checksum directly if no linker is provided.
>
>This makes it possible to call for example
>acpi_build_madt() from outside the ACPI table builder context.
>
>Signed-off-by: Oliver Steffen <osteffen@redhat.com>
>---
> hw/acpi/aml-build.c | 29 +++++++++++++++++++++++++++--
> 1 file changed, 27 insertions(+), 2 deletions(-)
>
>diff --git a/hw/acpi/aml-build.c b/hw/acpi/aml-build.c
>index dad4cfcc7d..6a3650076f 100644
>--- a/hw/acpi/aml-build.c
>+++ b/hw/acpi/aml-build.c
>@@ -1730,6 +1730,25 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
>     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
> }
>
>+static uint8_t calculate_acpi_checksum(const gchar *data, size_t len)
>+{
>+    size_t i;
>+    uint8_t sum = 0;
>+
>+    for (i = 0; i < len; ++i) {
>+        sum += (uint8_t)data[i];
>+    }
>+
>+    return sum;
>+}
>+

In `hw/acpi/core.c` there is a `acpi_checksum` function that does 
exactly this. Can we reuse this to reduce code duplication? Currently 
that function is marked as static.

Thanks for all the work so far.

Luigi