[PATCH v3 0/2] igvm: fix byte_offset handling in parameter directives

[Luigi Leonardi <leonardi@redhat.com>]

Several IGVM directives (memory map, vp-count, environment info, MADT,
device tree) _can_ place their data at a byte_offset within a shared
parameter area, letting multiple parameters be packed into one area.

Patch 1 introduces qigvm_get_param_data(), which looks up a
parameter area and validates that byte_offset falls within it in one
step, and uses it in the vp-count and environment-info handlers, which
already relied on byte_offset.

Patch 2 fixes the memory map, MADT and device tree handlers, which
ignored byte_offset entirely and always wrote at the start of the
buffer, potentially corrupting earlier data when several parameters
share an area.

Signed-off-by: Luigi Leonardi <leonardi@redhat.com>
---
Changes in v3:
- Renamed the function to `qigvm_get_param_data`. [Stefano]
- `qigvm_get_param_data` now returns the data pointer directly [Stefano]
- Added fixes tags. [Stefano]
- Renamed variables to param_data and param_size [Stefano]
- Link to v2: https://lore.kernel.org/qemu-devel/20260904-fix_offset-v2-0-f5bb4cf6d4b0@redhat.com

Changes in v2:
- Inverted commit order: first I introduce the helper and use it where
  necessary, then in commit 2 I handle the `offset`. [Stefano]
- The helper now returns data and size, that already consider the byte
  offset [Stefano]
- Link to v1: https://lore.kernel.org/qemu-devel/20260902-fix_offset-v1-0-04b18f7595b2@redhat.com

---
Luigi Leonardi (2):
      igvm: validate byte_offset before using it in parameter directives
      igvm: honor byte_offset when writing memory map, MADT and device tree

 backends/igvm.c                | 96 ++++++++++++++++++++++++++++++++----------
 include/system/igvm-internal.h |  6 +++
 target/i386/igvm.c             | 13 +++---
 3 files changed, 86 insertions(+), 29 deletions(-)
---
base-commit: d2e570cc0f97b936902a5b1b86b73c0f5998b475
change-id: 20260902-fix_offset-a268cb12aafc

Best regards,
-- 
Luigi Leonardi <leonardi@redhat.com>