From: Rodrigo Vivi <rodrigo.vivi@intel.com>
To: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>
Cc: intel-xe@lists.freedesktop.org
Subject: Re: [Intel-xe] [PATCH] fixup! drm/xe/bo: Evict VRAM to TT rather than to system
Date: Wed, 20 Sep 2023 16:58:24 -0400 [thread overview]
Message-ID: <ZQtc8NUtdAHuHdol@intel.com> (raw)
In-Reply-To: <b954e58e-88ae-fba4-74cb-6f5c5919a4ca@linux.intel.com>
On Wed, Sep 20, 2023 at 10:07:18PM +0200, Thomas Hellström wrote:
>
> On 9/20/23 21:43, Rodrigo Vivi wrote:
> > On Wed, Sep 20, 2023 at 10:22:17AM +0200, Thomas Hellström wrote:
> > > The dma-buf kunit test and to some extent the dma-buf code assumed
> > > that all bos were evicted to system. Fix this by having the dma-buf
> > > code not assume anything about eviction placement and the test
> > > code assume that VRAM bos are evicted to TT.
> > I believe that we should minimize the fixup patches now and use that
> > only for the display ones that still need a lot of work.
> >
> > This could be a simple Fixes: case. And even the commit message is
> > very good for a standalone patch.
>
> Thanks for reviewing, Rodrigo.
>
> So are we done rebasing, so that the commit hashes are stable? Like
oh dang.. good point... because display code will move up that hash will
change. let's go with fixup for this case. sorry for the noise.
>
> Fixes: <stable commit hash>
>
> /Thomas
>
> > Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
> >
> > > The patch applies using automerge conflict resolution.
> > >
> > > This fixes the following errors:
> > >
> > > [ 102.102086] KTAP version 1
> > > [ 102.102102] # Subtest: xe_dma_buf
> > > [ 102.102113] 1..1
> > > [ 102.102377] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.106167] # xe_dma_buf_kunit: EXPECTATION FAILED at drivers/gpu/drm/xe/tests/xe_dma_buf.c:76
> > > Importer wasn't properly evicted.
> > > [ 102.106327] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.107136] # xe_dma_buf_kunit: EXPECTATION FAILED at drivers/gpu/drm/xe/tests/xe_dma_buf.c:98
> > > Expected xe_bo_is_mem_type(exported, mem_type) to be true, but is false
> > > [ 102.107691] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.108099] # xe_dma_buf_kunit: EXPECTATION FAILED at drivers/gpu/drm/xe/tests/xe_dma_buf.c:76
> > > Importer wasn't properly evicted.
> > > [ 102.108158] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.108294] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.108742] # xe_dma_buf_kunit: EXPECTATION FAILED at drivers/gpu/drm/xe/tests/xe_dma_buf.c:76
> > > Importer wasn't properly evicted.
> > > [ 102.108800] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.108908] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.109080] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.109842] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.109941] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.110131] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.110236] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.110402] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.110731] # xe_dma_buf_kunit: EXPECTATION FAILED at drivers/gpu/drm/xe/tests/xe_dma_buf.c:76
> > > Importer wasn't properly evicted.
> > > [ 102.110790] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.111498] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.111793] # xe_dma_buf_kunit: EXPECTATION FAILED at drivers/gpu/drm/xe/tests/xe_dma_buf.c:76
> > > Importer wasn't properly evicted.
> > > [ 102.111851] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.112302] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.112580] # xe_dma_buf_kunit: EXPECTATION FAILED at drivers/gpu/drm/xe/tests/xe_dma_buf.c:76
> > > Importer wasn't properly evicted.
> > > [ 102.112638] # xe_dma_buf_kunit: running xe_test_dmabuf_import_same_driver
> > > [ 102.113292] not ok 1 xe_dma_buf_kunit
> > > [ 102.113296] not ok 1 xe_dma_buf
> > >
> > > Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
> > > ---
> > > drivers/gpu/drm/xe/tests/xe_dma_buf.c | 5 ++---
> > > drivers/gpu/drm/xe/xe_dma_buf.c | 11 ++++-------
> > > 2 files changed, 6 insertions(+), 10 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/xe/tests/xe_dma_buf.c b/drivers/gpu/drm/xe/tests/xe_dma_buf.c
> > > index 1c3f4bc72b99..1c4d8751be69 100644
> > > --- a/drivers/gpu/drm/xe/tests/xe_dma_buf.c
> > > +++ b/drivers/gpu/drm/xe/tests/xe_dma_buf.c
> > > @@ -72,7 +72,7 @@ static void check_residency(struct kunit *test, struct xe_bo *exported,
> > > }
> > > /* Verify that also importer has been evicted to SYSTEM */
> > > - if (!xe_bo_is_mem_type(imported, XE_PL_SYSTEM)) {
> > > + if (exported != imported && !xe_bo_is_mem_type(imported, XE_PL_SYSTEM)) {
> > > KUNIT_FAIL(test, "Importer wasn't properly evicted.\n");
> > > return;
> > > }
> > > @@ -91,8 +91,7 @@ static void check_residency(struct kunit *test, struct xe_bo *exported,
> > > * possible, saving a migration step as the transfer is just
> > > * likely as fast from system memory.
> > > */
> > > - if (params->force_different_devices &&
> > > - params->mem_mask & XE_BO_CREATE_SYSTEM_BIT)
> > > + if (params->mem_mask & XE_BO_CREATE_SYSTEM_BIT)
> > > KUNIT_EXPECT_TRUE(test, xe_bo_is_mem_type(exported, XE_PL_TT));
> > > else
> > > KUNIT_EXPECT_TRUE(test, xe_bo_is_mem_type(exported, mem_type));
> > > diff --git a/drivers/gpu/drm/xe/xe_dma_buf.c b/drivers/gpu/drm/xe/xe_dma_buf.c
> > > index bd8d51e7f93f..cfde3be3b0dc 100644
> > > --- a/drivers/gpu/drm/xe/xe_dma_buf.c
> > > +++ b/drivers/gpu/drm/xe/xe_dma_buf.c
> > > @@ -98,13 +98,10 @@ static struct sg_table *xe_dma_buf_map(struct dma_buf_attachment *attach,
> > > return ERR_PTR(-EOPNOTSUPP);
> > > if (!xe_bo_is_pinned(bo)) {
> > > - if (!attach->peer2peer ||
> > > - bo->ttm.resource->mem_type == XE_PL_SYSTEM) {
> > > - if (xe_bo_can_migrate(bo, XE_PL_TT))
> > > - r = xe_bo_migrate(bo, XE_PL_TT);
> > > - else
> > > - r = xe_bo_validate(bo, NULL, false);
> > > - }
> > > + if (!attach->peer2peer)
> > > + r = xe_bo_migrate(bo, XE_PL_TT);
> > > + else
> > > + r = xe_bo_validate(bo, NULL, false);
> > > if (r)
> > > return ERR_PTR(r);
> > > }
> > > --
> > > 2.41.0
> > >
prev parent reply other threads:[~2023-09-20 21:11 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-09-20 8:22 [Intel-xe] [PATCH] fixup! drm/xe/bo: Evict VRAM to TT rather than to system Thomas Hellström
2023-09-20 8:54 ` [Intel-xe] ✓ CI.Patch_applied: success for " Patchwork
2023-09-20 8:54 ` [Intel-xe] ✓ CI.checkpatch: " Patchwork
2023-09-20 8:55 ` [Intel-xe] ✓ CI.KUnit: " Patchwork
2023-09-20 9:02 ` [Intel-xe] ✓ CI.Build: " Patchwork
2023-09-20 9:02 ` [Intel-xe] ✓ CI.Hooks: " Patchwork
2023-09-20 9:04 ` [Intel-xe] ✓ CI.checksparse: " Patchwork
2023-09-20 9:37 ` [Intel-xe] ✓ CI.BAT: " Patchwork
2023-09-20 19:43 ` [Intel-xe] [PATCH] " Rodrigo Vivi
2023-09-20 20:07 ` Thomas Hellström
2023-09-20 20:58 ` Rodrigo Vivi [this message]
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=ZQtc8NUtdAHuHdol@intel.com \
--to=rodrigo.vivi@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--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