Re: [PATCH v1] i915/i915_driver: move intel_gvt_init() level higher

["Michał Grzelak" <michal.grzelak@intel.com>]

[-- Attachment #1: Type: text/plain, Size: 2955 bytes --]

On Wed, 28 Jan 2026, Jani Nikula wrote:
> On Wed, 28 Jan 2026, Michał Grzelak <michal.grzelak@intel.com> wrote:
>> i915_driver_probe() cannot currently handle error of intel_gvt_init()
>> because it is called from i915_driver_hw_probe(). Nevertheless, it's
>> error path is done in i915_driver_probe(). Fault of
>> i915_driver_hw_probe() thus causes list corruption, since
>> intel_gvt_driver_remove() was called twice.
>>
>> Move intel_gvt_init() up to i915_driver_probe(). Adjust error path of
>> i915_driver_hw_probe() so that intel_gvt_driver_remove() is called
>> always once.
>>
>> Signed-off-by: Michał Grzelak <michal.grzelak@intel.com>
>> ---
>>  drivers/gpu/drm/i915/i915_driver.c | 18 +++++++++---------
>>  1 file changed, 9 insertions(+), 9 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/i915/i915_driver.c b/drivers/gpu/drm/i915/i915_driver.c
>> index 1e087dfe03d0..0f507b8ae2f1 100644
>> --- a/drivers/gpu/drm/i915/i915_driver.c
>> +++ b/drivers/gpu/drm/i915/i915_driver.c
>> @@ -557,10 +557,6 @@ static int i915_driver_hw_probe(struct drm_i915_private *dev_priv)
>>  			drm_dbg(&dev_priv->drm, "can't enable MSI");
>>  	}
>>
>> -	ret = intel_gvt_init(dev_priv);
>> -	if (ret)
>> -		goto err_msi;
>> -
>>  	intel_opregion_setup(display);
>>
>>  	ret = i915_pcode_init(dev_priv);
>> @@ -581,9 +577,6 @@ static int i915_driver_hw_probe(struct drm_i915_private *dev_priv)
>>
>>  err_opregion:
>>  	intel_opregion_cleanup(display);
>> -err_msi:
>> -	if (pdev->msi_enabled)
>> -		pci_disable_msi(pdev);
>
> Why's this being moved?
>

No sane reason, it's me overlooking err_opregion error path.
Will fix it in next version.

BR,
Michał

>>  err_mem_regions:
>>  	intel_memory_regions_driver_release(dev_priv);
>>  err_ggtt:
>> @@ -870,9 +863,13 @@ int i915_driver_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>>  	if (ret < 0)
>>  		goto out_cleanup_mmio;
>>
>> +	ret = intel_gvt_init(i915);
>> +	if (ret)
>> +		goto out_cleanup_hw;
>> +
>>  	ret = intel_display_driver_probe_noirq(display);
>>  	if (ret < 0)
>> -		goto out_cleanup_hw;
>> +		goto out_cleanup_gvt;
>>
>>  	ret = intel_irq_install(i915);
>>  	if (ret)
>> @@ -921,6 +918,10 @@ int i915_driver_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>>  	intel_irq_uninstall(i915);
>>  out_cleanup_modeset:
>>  	intel_display_driver_remove_nogem(display);
>> +out_cleanup_gvt:
>> +	if (pdev->msi_enabled)
>> +		pci_disable_msi(pdev);
>> +	intel_gvt_driver_remove(i915);
>>  out_cleanup_hw:
>>  	i915_driver_hw_remove(i915);
>>  	intel_memory_regions_driver_release(i915);
>> @@ -928,7 +929,6 @@ int i915_driver_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>>  	i915_gem_drain_freed_objects(i915);
>>  	i915_ggtt_driver_late_release(i915);
>>  out_cleanup_mmio:
>> -	intel_gvt_driver_remove(i915);
>>  	i915_driver_mmio_release(i915);
>>  out_runtime_pm_put:
>>  	enable_rpm_wakeref_asserts(&i915->runtime_pm);
>
> -- 
> Jani Nikula, Intel
>