drm/i915/gvt: fix incorrect cache entry for guest page mapping

Submitted by Xiaolin Zhang on July 17, 2019, 3:56 p.m.

Details

Message ID 1563378987-21880-1-git-send-email-xiaolin.zhang@intel.com
State New
Headers show
Series "drm/i915/gvt: fix incorrect cache entry for guest page mapping" ( rev: 1 ) in Intel GVT devel

Not browsing as part of any series.

Commit Message

Xiaolin Zhang July 17, 2019, 3:56 p.m.
GPU hang observed during the guest OCL conformance test which is caused
by THP GTT feature used durning the test.

It was observed the same GFN with different size (4K and 2M) requested
from the guest in GVT. So during the guest page dma map stage, it is
required to unmap first with orginal size and then remap again with
requested size.

Fixes: b901b252b6cf ("drm/i915/gvt: Add 2M huge gtt support")
Signed-off-by: Xiaolin Zhang <xiaolin.zhang@intel.com>
---
 drivers/gpu/drm/i915/gvt/kvmgt.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

Patch hide | download patch | download mbox

diff --git a/drivers/gpu/drm/i915/gvt/kvmgt.c b/drivers/gpu/drm/i915/gvt/kvmgt.c
index a68addf..4a7cf86 100644
--- a/drivers/gpu/drm/i915/gvt/kvmgt.c
+++ b/drivers/gpu/drm/i915/gvt/kvmgt.c
@@ -1911,6 +1911,18 @@  static int kvmgt_dma_map_guest_page(unsigned long handle, unsigned long gfn,
 		ret = __gvt_cache_add(info->vgpu, gfn, *dma_addr, size);
 		if (ret)
 			goto err_unmap;
+	} else if (entry->size != size) {
+		/* the same gfn with different size: unmap and re-map */
+		gvt_dma_unmap_page(vgpu, gfn, entry->dma_addr, entry->size);
+		__gvt_cache_remove_entry(vgpu, entry);
+
+		ret = gvt_dma_map_page(vgpu, gfn, dma_addr, size);
+		if (ret)
+			goto err_unlock;
+
+		ret = __gvt_cache_add(info->vgpu, gfn, *dma_addr, size);
+		if (ret)
+			goto err_unmap;
 	} else {
 		kref_get(&entry->ref);
 		*dma_addr = entry->dma_addr;

Comments

On Wed, Jul 17, 2019 at 11:56:27PM +0800, Xiaolin Zhang wrote:
> GPU hang observed during the guest OCL conformance test which is caused
> by THP GTT feature used durning the test.
> 
> It was observed the same GFN with different size (4K and 2M) requested
> from the guest in GVT. So during the guest page dma map stage, it is
> required to unmap first with orginal size and then remap again with
> requested size.
> 
> Fixes: b901b252b6cf ("drm/i915/gvt: Add 2M huge gtt support")
> Signed-off-by: Xiaolin Zhang <xiaolin.zhang@intel.com>
> ---
>  drivers/gpu/drm/i915/gvt/kvmgt.c | 12 ++++++++++++
>  1 file changed, 12 insertions(+)

<formletter>

This is not the correct way to submit patches for inclusion in the
stable kernel tree.  Please read:
    https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
for how to do this properly.

</formletter>
On 07/17/2019 03:55 PM, Greg KH wrote:
> On Wed, Jul 17, 2019 at 11:56:27PM +0800, Xiaolin Zhang wrote:
>> GPU hang observed during the guest OCL conformance test which is caused
>> by THP GTT feature used durning the test.
>>
>> It was observed the same GFN with different size (4K and 2M) requested
>> from the guest in GVT. So during the guest page dma map stage, it is
>> required to unmap first with orginal size and then remap again with
>> requested size.
>>
>> Fixes: b901b252b6cf ("drm/i915/gvt: Add 2M huge gtt support")
>> Signed-off-by: Xiaolin Zhang <xiaolin.zhang@intel.com>
>> ---
>>  drivers/gpu/drm/i915/gvt/kvmgt.c | 12 ++++++++++++
>>  1 file changed, 12 insertions(+)
> <formletter>
>
> This is not the correct way to submit patches for inclusion in the
> stable kernel tree.  Please read:
>     https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
> for how to do this properly.
>
> </formletter>
>
Greg, Thanks great information to point out this I am not fully aware.
will resend to correct this.

-Xiaolin