Re: [PATCH] vfio/type1: Remove locked page accounting workqueue

2017-04-06 Thread Auger Eric
Hi Alex, On 06/04/2017 16:43, Alex Williamson wrote: > On Thu, 6 Apr 2017 10:23:59 +0200 > Auger Eric wrote: > >> Hi Alex, >> >> On 03/04/2017 22:02, Alex Williamson wrote: >>> If the mmap_sem is contented then the vfio type1 IOMMU backend will >>> defer locked page

Re: [PATCH] vfio/type1: Remove locked page accounting workqueue

2017-04-06 Thread Auger Eric
Hi Alex, On 06/04/2017 16:43, Alex Williamson wrote: > On Thu, 6 Apr 2017 10:23:59 +0200 > Auger Eric wrote: > >> Hi Alex, >> >> On 03/04/2017 22:02, Alex Williamson wrote: >>> If the mmap_sem is contented then the vfio type1 IOMMU backend will >>> defer locked page accounting updates to a

Re: [PATCH] vfio/type1: Remove locked page accounting workqueue

2017-04-06 Thread Alex Williamson
On Thu, 6 Apr 2017 10:23:59 +0200 Auger Eric wrote: > Hi Alex, > > On 03/04/2017 22:02, Alex Williamson wrote: > > If the mmap_sem is contented then the vfio type1 IOMMU backend will > > defer locked page accounting updates to a workqueue task. This has > > a few

Re: [PATCH] vfio/type1: Remove locked page accounting workqueue

2017-04-06 Thread Alex Williamson
On Thu, 6 Apr 2017 10:23:59 +0200 Auger Eric wrote: > Hi Alex, > > On 03/04/2017 22:02, Alex Williamson wrote: > > If the mmap_sem is contented then the vfio type1 IOMMU backend will > > defer locked page accounting updates to a workqueue task. This has > > a few problems and depending on

Re: [PATCH] vfio/type1: Remove locked page accounting workqueue

2017-04-06 Thread Auger Eric
Hi Alex, On 03/04/2017 22:02, Alex Williamson wrote: > If the mmap_sem is contented then the vfio type1 IOMMU backend will > defer locked page accounting updates to a workqueue task. This has > a few problems and depending on which side the user tries to play, > they might be over-penalized for

Re: [PATCH] vfio/type1: Remove locked page accounting workqueue

2017-04-06 Thread Auger Eric
Hi Alex, On 03/04/2017 22:02, Alex Williamson wrote: > If the mmap_sem is contented then the vfio type1 IOMMU backend will > defer locked page accounting updates to a workqueue task. This has > a few problems and depending on which side the user tries to play, > they might be over-penalized for

[PATCH] vfio/type1: Remove locked page accounting workqueue

2017-04-03 Thread Alex Williamson
If the mmap_sem is contented then the vfio type1 IOMMU backend will defer locked page accounting updates to a workqueue task. This has a few problems and depending on which side the user tries to play, they might be over-penalized for unmaps that haven't yet been accounted, or able to race the

[PATCH] vfio/type1: Remove locked page accounting workqueue

2017-04-03 Thread Alex Williamson
If the mmap_sem is contented then the vfio type1 IOMMU backend will defer locked page accounting updates to a workqueue task. This has a few problems and depending on which side the user tries to play, they might be over-penalized for unmaps that haven't yet been accounted, or able to race the