On Thu, 30 Oct 2025, Luca Coelho <[email protected]> wrote: > On Wed, 2025-10-22 at 18:17 +0300, Jani Nikula wrote: >> diff --git a/drivers/gpu/drm/i915/display/intel_display_utils.c >> b/drivers/gpu/drm/i915/display/intel_display_utils.c >> index 13d3999dd580..04d010f7c23e 100644 >> --- a/drivers/gpu/drm/i915/display/intel_display_utils.c >> +++ b/drivers/gpu/drm/i915/display/intel_display_utils.c >> @@ -1,10 +1,15 @@ >> // SPDX-License-Identifier: MIT >> /* Copyright © 2025 Intel Corporation */ >> >> +#include <linux/device.h> >> + >> +#include <drm/drm_device.h> >> + >> #ifdef CONFIG_X86 >> #include <asm/hypervisor.h> >> #endif >> >> +#include "intel_display_core.h" >> #include "intel_display_utils.h" >> >> bool intel_display_run_as_guest(struct intel_display *display) >> @@ -16,3 +21,12 @@ bool intel_display_run_as_guest(struct intel_display >> *display) >> return false; >> #endif >> } >> + >> +bool intel_display_vtd_active(struct intel_display *display) >> +{ >> + if (device_iommu_mapped(display->drm->dev)) >> + return true; >> + >> + /* Running as a guest, we assume the host is enforcing VT'd */ >> + return intel_display_run_as_guest(display); >> +} > > This also looks quite much like an "inlineable" function, no?
Same reason as before, and that display->drm->dev dereference would require pulling in both intel_display_core.h and drm/drm_device.h in the header. BR, Jani. -- Jani Nikula, Intel
