From: Matthew Brost <matthew.brost@intel.com>
To: Oak Zeng <oak.zeng@intel.com>
Cc: <igt-dev@lists.freedesktop.org>,
<Thomas.Hellstrom@linux.intel.com>, <kamil.konieczny@intel.com>,
<zbigniew.kempczynski@intel.com>, <ashutosh.dixit@intel.com>,
<juha-pekka.heikkila@intel.com>, <rodrigo.vivi@intel.com>
Subject: Re: [i-g-t v6 4/4] tests/intel/xe_exec_fault_mode: Test scratch page under fault mode
Date: Mon, 17 Mar 2025 12:07:47 -0700 [thread overview]
Message-ID: <Z9hzA7jrIrwt7Ayp@lstrano-desk.jf.intel.com> (raw)
In-Reply-To: <20250224174010.594192-5-oak.zeng@intel.com>
On Mon, Feb 24, 2025 at 12:40:10PM -0500, Oak Zeng wrote:
> On certain HW (such as lunarlake and battlemage), driver now allows
> scratch page be enabled under fault mode. Test this functionality
>
> v1: Move the test to separate function.
> Write to scratch page before rebind (Matt)
>
> Signed-off-by: Oak Zeng <oak.zeng@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
> ---
> tests/intel/xe_exec_fault_mode.c | 68 ++++++++++++++++++++++++++++++++
> 1 file changed, 68 insertions(+)
>
> diff --git a/tests/intel/xe_exec_fault_mode.c b/tests/intel/xe_exec_fault_mode.c
> index ae40e099b..6cc27102a 100644
> --- a/tests/intel/xe_exec_fault_mode.c
> +++ b/tests/intel/xe_exec_fault_mode.c
> @@ -17,6 +17,7 @@
> #include "igt.h"
> #include "lib/igt_syncobj.h"
> #include "lib/intel_reg.h"
> +#include "lib/xe/xe_util.h"
> #include "xe_drm.h"
>
> #include "xe/xe_ioctl.h"
> @@ -36,6 +37,64 @@
> #define INVALID_VA (0x1 << 8)
> #define ENABLE_SCRATCH (0x1 << 9)
>
> +
> +/**
> + * SUBTEST: scratch-fault
> + * Description: Test scratch page functionality
> + * Test category: functionality test
> + */
> +static void test_scratch(int fd, struct drm_xe_engine_class_instance *eci)
> +{
> + size_t bb_size = xe_bb_size(fd, PAGE_ALIGN_UFENCE);
> + uint32_t vm;
> +
> + struct xe_buffer dst_buf = {
> + .fd = fd,
> + .size = bb_size,
> + .gpu_addr = 0x1a0000,
> + .placement = vram_if_possible(fd, eci->gt_id),
> + .flag = DRM_XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM,
> + };
> +
> + struct xe_cmdbuf cmdbuf = {
> + .buf = {
> + .fd = fd,
> + .gpu_addr = 0x10a0000,
> + .placement = vram_if_possible(fd, eci->gt_id),
> + .flag = DRM_XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM,
> + },
> + .cmd_size = bb_size,
> + };
> +
> + vm = xe_vm_create(fd, DRM_XE_VM_CREATE_FLAG_LR_MODE |
> + DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE |
> + DRM_XE_VM_CREATE_FLAG_FAULT_MODE, 0);
> +
> + dst_buf.vm = vm;
> + cmdbuf.buf.vm = vm;
> + /* Submit a command to write 0x1a0000
> + * Since 0x1a0000 is mapped scratch page, cmd execution should still
> + * be successful. Write is either be dropped by HW (NULL PTE case) or
> + * written to scratch page.
> + */
> + xe_cmdbuf_create(&cmdbuf, eci);
> + xe_cmdbuf_insert_store(&cmdbuf, dst_buf.gpu_addr, 0xc0ffee);
> + xe_cmdbuf_insert_bbe(&cmdbuf);
> + xe_cmdbuf_submit(&cmdbuf);
> +
> + /* Create a buffer object, vm_bind it to 0x1a0000, then re-submit
> + * the command buffer. This should write to the buffer object.
> + * Check the buffer object to see if the write was successful.
> + */
> + xe_buffer_create(&dst_buf);
> + xe_cmdbuf_submit(&cmdbuf);
> + igt_assert_eq(*(uint64_t *)dst_buf.cpu_addr, 0xc0ffee);
> +
> + xe_cmdbuf_destroy(&cmdbuf);
> + xe_buffer_destroy(&dst_buf);
> + xe_vm_destroy(fd, vm);
> +}
> +
> /**
> * SUBTEST: invalid-va
> * Description: Access invalid va and check for EIO through user fence.
> @@ -458,6 +517,7 @@ igt_main
> { NULL },
> };
> int fd;
> + uint32_t dev_id;
>
> igt_fixture {
> struct timespec tv = {};
> @@ -466,6 +526,7 @@ igt_main
> int timeout = igt_run_in_simulation() ? 20 : 2;
>
> fd = drm_open_driver(DRIVER_XE);
> + dev_id = intel_get_drm_devid(fd);
> do {
> if (ret)
> usleep(5000);
> @@ -508,6 +569,13 @@ igt_main
> xe_for_each_engine(fd, hwe)
> test_exec(fd, hwe, 1, 1, ENABLE_SCRATCH | INVALID_VA);
>
> + igt_describe("Check if scratch page works under fault mode");
> + igt_subtest("scratch-fault") {
> + igt_skip_on(!IS_LUNARLAKE(dev_id) && !IS_BATTLEMAGE(dev_id));
> + xe_for_each_engine(fd, hwe)
> + test_scratch(fd, hwe);
> + }
> +
> igt_fixture {
> drm_close_driver(fd);
> }
> --
> 2.26.3
>
next prev parent reply other threads:[~2025-03-17 19:07 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-02-24 17:40 [i-g-t v6 0/4] xe: Test scratch page under fault mode Oak Zeng
2025-02-24 17:40 ` [i-g-t v6 1/4] lib/xe: Fix a comment error Oak Zeng
2025-03-11 0:03 ` Dixit, Ashutosh
2025-03-11 0:36 ` Dixit, Ashutosh
2025-03-11 6:56 ` Zbigniew Kempczyński
2025-02-24 17:40 ` [i-g-t v6 2/4] lib/xe/xe_util: Introduce helper functions Oak Zeng
2025-02-24 17:40 ` [i-g-t v6 3/4] tests/intel/xe_vm: Exclude invalid_flags tests from LNL and BMG Oak Zeng
2025-03-11 0:33 ` Dixit, Ashutosh
2025-03-19 21:51 ` Zeng, Oak
2025-02-24 17:40 ` [i-g-t v6 4/4] tests/intel/xe_exec_fault_mode: Test scratch page under fault mode Oak Zeng
2025-03-17 19:07 ` Matthew Brost [this message]
2025-02-27 3:25 ` ✓ Xe.CI.BAT: success for xe: Test scratch page under fault mode (rev5) Patchwork
2025-02-27 3:39 ` ✓ i915.CI.BAT: " Patchwork
2025-02-27 4:24 ` ✗ Xe.CI.Full: failure " Patchwork
2025-02-27 8:30 ` ✗ i915.CI.Full: " Patchwork
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=Z9hzA7jrIrwt7Ayp@lstrano-desk.jf.intel.com \
--to=matthew.brost@intel.com \
--cc=Thomas.Hellstrom@linux.intel.com \
--cc=ashutosh.dixit@intel.com \
--cc=igt-dev@lists.freedesktop.org \
--cc=juha-pekka.heikkila@intel.com \
--cc=kamil.konieczny@intel.com \
--cc=oak.zeng@intel.com \
--cc=rodrigo.vivi@intel.com \
--cc=zbigniew.kempczynski@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