From: Matthew Brost <matthew.brost@intel.com>
To: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>
Cc: Oak Zeng <oak.zeng@intel.com>, <intel-xe@lists.freedesktop.org>,
<joonas.lahtinen@linux.intel.com>
Subject: Re: [PATCH 3/3] drm/xe: Allow scratch page under fault mode for certain platform
Date: Wed, 29 Jan 2025 08:41:07 -0800 [thread overview]
Message-ID: <Z5paI4u5C1Oxr2zJ@lstrano-desk.jf.intel.com> (raw)
In-Reply-To: <ca61d5a98bb2e6bf778912a10f189512bae4ef1e.camel@linux.intel.com>
On Wed, Jan 29, 2025 at 09:52:54AM +0100, Thomas Hellström wrote:
> On Tue, 2025-01-28 at 17:21 -0500, Oak Zeng wrote:
> > Normally scratch page is not allowed when a vm is operate under page
> > fault mode, i.e., in the existing codes,
> > DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE
> > and DRM_XE_VM_CREATE_FLAG_FAULT_MODE are mutual exclusive. The reason
> > is fault mode relies on recoverable page to work, while scratch page
> > can mute recoverable page fault.
> >
> > On xe2 and xe3, HW prefetching can cause page fault interrupt. Due to
> > page fault interrupt overhead (i.e., need Guc and KMD involved to fix
> > the page fault), HW prefetching can be slowed by many orders of
> > magnitude.
>
> The problem on xe2 and xe3 as I understand it is not the overhead but
> the fact that OOB prefetching causes pagefaults and we can't resolve
> those.
>
> >
> > Fix this problem by allowing scratch page under fault mode for xe2
> > and xe3.
> > With scratch page in place, HW prefetching could always hit scratch
> > page
> > instead of causing interrupt.
> >
> > Signed-off-by: Oak Zeng <oak.zeng@intel.com>
> > ---
> > drivers/gpu/drm/xe/xe_vm.c | 9 +++++----
> > 1 file changed, 5 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
> > index 196d347c6ac0..3346f88f284a 100644
> > --- a/drivers/gpu/drm/xe/xe_vm.c
> > +++ b/drivers/gpu/drm/xe/xe_vm.c
> > @@ -1732,6 +1732,11 @@ int xe_vm_create_ioctl(struct drm_device *dev,
> > void *data,
> > if (XE_IOCTL_DBG(xe, args->extensions))
> > return -EINVAL;
> >
> > + if (XE_IOCTL_DBG(xe, args->flags &
> > DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE &&
> > + args->flags &
> > DRM_XE_VM_CREATE_FLAG_FAULT_MODE &&
> > + !(IS_LUNARLAKE(xe) || IS_BATTLEMAGE(xe) ||
> > IS_PANTHERLAKE(xe))))
> > + return -EINVAL;
> > +
>
> Could we perhaps add a bit in the xe->info structure?
>
+1. This seems like a good idea.
Matt
> /Thomas
>
>
> > if (XE_WA(xe_root_mmio_gt(xe), 14016763929))
> > args->flags |= DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE;
> >
> > @@ -1745,10 +1750,6 @@ int xe_vm_create_ioctl(struct drm_device *dev,
> > void *data,
> > if (XE_IOCTL_DBG(xe, args->flags &
> > ~ALL_DRM_XE_VM_CREATE_FLAGS))
> > return -EINVAL;
> >
> > - if (XE_IOCTL_DBG(xe, args->flags &
> > DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE &&
> > - args->flags &
> > DRM_XE_VM_CREATE_FLAG_FAULT_MODE))
> > - return -EINVAL;
> > -
> > if (XE_IOCTL_DBG(xe, !(args->flags &
> > DRM_XE_VM_CREATE_FLAG_LR_MODE) &&
> > args->flags &
> > DRM_XE_VM_CREATE_FLAG_FAULT_MODE))
> > return -EINVAL;
>
next prev parent reply other threads:[~2025-01-29 16:40 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-01-28 22:21 [PATCH 1/3] drm/xe: Add a function to zap page table by address range Oak Zeng
2025-01-28 22:21 ` [PATCH 2/3] drm/xe: Clear scratch page before vm_bind Oak Zeng
2025-01-28 23:05 ` Cavitt, Jonathan
2025-01-28 23:19 ` Matthew Brost
2025-01-29 5:04 ` Matthew Brost
2025-01-29 8:21 ` Thomas Hellström
2025-01-29 21:01 ` Zeng, Oak
2025-01-30 1:36 ` Matthew Brost
2025-01-30 8:36 ` Thomas Hellström
2025-01-30 15:18 ` Zeng, Oak
2025-01-28 22:21 ` [PATCH 3/3] drm/xe: Allow scratch page under fault mode for certain platform Oak Zeng
2025-01-28 23:05 ` Cavitt, Jonathan
2025-01-29 8:52 ` Thomas Hellström
2025-01-29 16:41 ` Matthew Brost [this message]
2025-01-28 23:04 ` [PATCH 1/3] drm/xe: Add a function to zap page table by address range Cavitt, Jonathan
2025-01-28 23:12 ` ✓ CI.Patch_applied: success for series starting with [1/3] " Patchwork
2025-01-28 23:12 ` ✓ CI.checkpatch: " Patchwork
2025-01-28 23:13 ` ✗ CI.KUnit: failure " Patchwork
2025-01-28 23:38 ` [PATCH 1/3] " Matthew Brost
2025-01-29 8:48 ` Thomas Hellström
-- strict thread matches above, loose matches on Subject: below --
2025-02-04 18:45 [PATCH 1/3] drm/xe: Introduced needs_scratch bit in device descriptor Oak Zeng
2025-02-04 18:45 ` [PATCH 3/3] drm/xe: Allow scratch page under fault mode for certain platform Oak Zeng
2025-02-05 13:14 ` Thomas Hellström
2025-02-06 1:54 ` Zeng, Oak
2025-02-06 9:29 ` Thomas Hellström
2025-02-06 15:14 ` Zeng, Oak
2025-02-25 22:05 ` Matthew Brost
2025-02-06 2:11 [PATCH 1/3] drm/xe: Introduced needs_scratch bit in device descriptor Oak Zeng
2025-02-06 2:11 ` [PATCH 3/3] drm/xe: Allow scratch page under fault mode for certain platform Oak Zeng
2025-02-06 21:38 [PATCH 1/3] drm/xe: Introduced needs_scratch bit in device descriptor Oak Zeng
2025-02-06 21:38 ` [PATCH 3/3] drm/xe: Allow scratch page under fault mode for certain platform Oak Zeng
2025-02-13 2:23 [PATCH 1/3] drm/xe: Introduced needs_scratch bit in device descriptor Oak Zeng
2025-02-13 2:23 ` [PATCH 3/3] drm/xe: Allow scratch page under fault mode for certain platform Oak Zeng
2025-02-25 22:10 ` Matthew Brost
2025-02-26 22:12 ` Zeng, Oak
2025-02-27 0:22 ` Matthew Brost
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=Z5paI4u5C1Oxr2zJ@lstrano-desk.jf.intel.com \
--to=matthew.brost@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=joonas.lahtinen@linux.intel.com \
--cc=oak.zeng@intel.com \
--cc=thomas.hellstrom@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox