public inbox for igt-dev@lists.freedesktop.org
 help / color / mirror / Atom feed
From: "Yadav, Arvind" <arvind.yadav@intel.com>
To: "Sharma, Nishit" <nishit.sharma@intel.com>,
	"Gurram, Pravalika" <pravalika.gurram@intel.com>,
	"igt-dev@lists.freedesktop.org" <igt-dev@lists.freedesktop.org>
Cc: "Brost, Matthew" <matthew.brost@intel.com>,
	"Ghimiray, Himal Prasad" <himal.prasad.ghimiray@intel.com>,
	"thomas.hellstrom@linux.intel.com"
	<thomas.hellstrom@linux.intel.com>
Subject: Re: [PATCH i-g-t v3 5/8] tests/intel/xe_madvise: Add dontneed-before-exec subtest
Date: Mon, 23 Feb 2026 15:15:35 +0530	[thread overview]
Message-ID: <640d2e40-ef5a-4cc5-acdb-b468b8b537cd@intel.com> (raw)
In-Reply-To: <100574ec-a137-44f9-a4db-7a7e32bbc4fd@intel.com>


On 23-02-2026 11:36, Sharma, Nishit wrote:
>
> On 2/23/2026 10:40 AM, Gurram, Pravalika wrote:
>>
>>> -----Original Message-----
>>> From: Yadav, Arvind <arvind.yadav@intel.com>
>>> Sent: Tuesday, 17 February, 2026 08:04 AM
>>> To: igt-dev@lists.freedesktop.org
>>> Cc: Brost, Matthew <matthew.brost@intel.com>; Ghimiray, Himal Prasad
>>> <himal.prasad.ghimiray@intel.com>; thomas.hellstrom@linux.intel.com;
>>> Sharma, Nishit <nishit.sharma@intel.com>; Gurram, Pravalika
>>> <pravalika.gurram@intel.com>
>>> Subject: [PATCH i-g-t v3 5/8] tests/intel/xe_madvise: Add 
>>> dontneed-before-
>>> exec subtest
>>>
>>> This test validates GPU execution behavior when a data BO is purged 
>>> before
>>> submission. The test creates a batch that writes to a data BO, 
>>> purges the data
>>> BO (while keeping the batch BO valid to avoid GPU reset), then 
>>> submits for
>>> execution. With VM_CREATE_FLAG_SCRATCH_PAGE, the GPU write may
>>> succeed by landing on scratch memory instead of the purged BO,
>>> demonstrating graceful handling of purged memory during GPU operations.
>>>
>>> Cc: Nishit Sharma <nishit.sharma@intel.com>
>>> Cc: Pravalika Gurram <pravalika.gurram@intel.com>
>>> Cc: Matthew Brost <matthew.brost@intel.com>
>>> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
>>> Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
>>> Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
>>> ---
>>>   tests/intel/xe_madvise.c | 140
>>> +++++++++++++++++++++++++++++++++++++++
>>>   1 file changed, 140 insertions(+)
>>>
>>> diff --git a/tests/intel/xe_madvise.c b/tests/intel/xe_madvise.c index
>>> 5c6a24553..42cf1b3e5 100644
>>> --- a/tests/intel/xe_madvise.c
>>> +++ b/tests/intel/xe_madvise.c
>>> @@ -19,7 +19,11 @@
>>>
>>>   /* Purgeable test constants */
>>>   #define PURGEABLE_ADDR        0x1a0000
>>> +#define PURGEABLE_BATCH_ADDR    0x3c0000
>>>   #define PURGEABLE_BO_SIZE    4096
>>> +#define PURGEABLE_FENCE_VAL    0xbeef
>>> +#define PURGEABLE_TEST_PATTERN    0xc0ffee
>>> +#define PURGEABLE_DEAD_PATTERN    0xdead
>>>
>>>   /**
>>>    * trigger_memory_pressure - Fill VRAM + 25% to force purgeable 
>>> reclaim @@
>>> -144,6 +148,62 @@ static void purgeable_setup_simple_bo(int fd, 
>>> uint32_t
>>> *vm, uint32_t *bo,
>>>       xe_wait_ufence(fd, &sync_val, 1, 0, NSEC_PER_SEC); }
>>>
>>> +/**
>>> + * purgeable_setup_batch_and_data - Setup VM with batch and data BOs
>>> +for GPU exec
>>> + * @fd: DRM file descriptor
>>> + * @vm: Output VM handle
>>> + * @bind_engine: Output bind engine handle
>>> + * @batch_bo: Output batch BO handle
>>> + * @data_bo: Output data BO handle
>>> + * @batch: Output batch buffer pointer
>>> + * @data: Output data buffer pointer
>>> + * @batch_addr: Batch virtual address
>>> + * @data_addr: Data virtual address
>>> + * @batch_size: Batch buffer size
>>> + * @data_size: Data buffer size
>>> + *
>>> + * Helper to create VM, bind engine, batch and data BOs, and bind 
>>> them.
>>> + */
>>> +static void purgeable_setup_batch_and_data(int fd, uint32_t *vm,
>>> +                       uint32_t *bind_engine,
>>> +                       uint32_t *batch_bo,
>>> +                       uint32_t *data_bo,
>>> +                       uint32_t **batch,
>>> +                       uint32_t **data,
>>> +                       uint64_t batch_addr,
>>> +                       uint64_t data_addr,
>>> +                       size_t batch_size,
>>> +                       size_t data_size)
>>> +{
>>> +    struct drm_xe_sync sync = {
>>> +        .type = DRM_XE_SYNC_TYPE_USER_FENCE,
>>> +        .flags = DRM_XE_SYNC_FLAG_SIGNAL,
>>> +        .timeline_value = PURGEABLE_FENCE_VAL,
>>> +    };
>>> +    uint64_t vm_sync = 0;
>>> +
>>> +    *vm = xe_vm_create(fd, DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE,
>>> 0);
>>> +    *bind_engine = xe_bind_exec_queue_create(fd, *vm, 0);
>>> +
>>> +    /* Create and bind batch BO */
>>> +    *batch_bo = xe_bo_create(fd, *vm, batch_size, vram_if_possible(fd,
>>> 0),
>>> +
>>> DRM_XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM);
>>> +    *batch = xe_bo_map(fd, *batch_bo, batch_size);
>>> +
>>> +    sync.addr = to_user_pointer(&vm_sync);
>>> +    xe_vm_bind_async(fd, *vm, *bind_engine, *batch_bo, 0, batch_addr,
>>> batch_size, &sync, 1);
>>> +    xe_wait_ufence(fd, &vm_sync, PURGEABLE_FENCE_VAL, 0,
>>> NSEC_PER_SEC);
>>> +
>>> +    /* Create and bind data BO */
>>> +    *data_bo = xe_bo_create(fd, *vm, data_size, 
>>> vram_if_possible(fd, 0),
>>> +
>>>     DRM_XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM);
>>> +    *data = xe_bo_map(fd, *data_bo, data_size);
>>> +
>>> +    vm_sync = 0;
>>> +    xe_vm_bind_async(fd, *vm, *bind_engine, *data_bo, 0, data_addr,
>>> data_size, &sync, 1);
>>> +    xe_wait_ufence(fd, &vm_sync, PURGEABLE_FENCE_VAL, 0,
>>> NSEC_PER_SEC); }
>>> +
>>>   /**
>>>    * SUBTEST: dontneed-before-mmap
>>>    * Description: Mark BO as DONTNEED before mmap, verify mmap fails or
>>> SIGBUS on access @@ -244,6 +304,80 @@ static void
>>> test_dontneed_after_mmap(int fd, struct drm_xe_engine_class_instance
>>>       xe_vm_destroy(fd, vm);
>>>   }
>>>
>>> +/**
>>> + * SUBTEST: dontneed-before-exec
>>> + * Description: Mark BO as DONTNEED before GPU exec, verify GPU
>>> +behavior with SCRATCH_PAGE
>>> + * Test category: functionality test
>>> + */
>>> +static void test_dontneed_before_exec(int fd, struct
>>> +drm_xe_engine_class_instance *hwe) {
>>> +    uint32_t vm, exec_queue, bo, batch_bo, bind_engine;
>>> +    uint64_t data_addr = PURGEABLE_ADDR;
>>> +    uint64_t batch_addr = PURGEABLE_BATCH_ADDR;
>>> +    size_t data_size = PURGEABLE_BO_SIZE;
>>> +    size_t batch_size = PURGEABLE_BO_SIZE;
>>> +    struct drm_xe_sync sync[1] = {
>>> +        { .type = DRM_XE_SYNC_TYPE_USER_FENCE,
>>> +          .flags = DRM_XE_SYNC_FLAG_SIGNAL,
>>> +          .timeline_value = PURGEABLE_FENCE_VAL },
>>> +    };
>>> +    struct drm_xe_exec exec = {
>>> +        .num_batch_buffer = 1,
>>> +        .num_syncs = 1,
>>> +        .syncs = to_user_pointer(sync),
>>> +    };
>>> +    uint32_t *data, *batch;
>>> +    uint64_t vm_sync = 0;
>>> +    int b, ret;
>>> +
>>> +    purgeable_setup_batch_and_data(fd, &vm, &bind_engine,
>>> &batch_bo,
>>> +                       &bo, &batch, &data, batch_addr,
>>> +                       data_addr, batch_size, data_size);
>>> +
>>> +    /* Prepare batch */
>>> +    b = 0;
>>> +    batch[b++] = MI_STORE_DWORD_IMM_GEN4;
>>> +    batch[b++] = data_addr;
>>> +    batch[b++] = data_addr >> 32;
>>> +    batch[b++] = PURGEABLE_DEAD_PATTERN;
>>> +    batch[b++] = MI_BATCH_BUFFER_END;
>>> +
>>> +    /* Phase 1: Purge data BO, batch BO still valid */
>>> +    igt_assert(purgeable_mark_and_verify_purged(fd, vm, data_addr,
>>> +data_size));
> Resource cleanup?


Noted,

>>> +
>>> +    exec_queue = xe_exec_queue_create(fd, vm, hwe, 0);
>>> +    exec.exec_queue_id = exec_queue;
>>> +    exec.address = batch_addr;
>>> +
>>> +    vm_sync = 0;
>>> +    sync[0].addr = to_user_pointer(&vm_sync);
>>> +
>>> +    /*
>>> +     * VM has SCRATCH_PAGE — exec may succeed with the GPU write
>>> +     * landing on scratch instead of the purged data BO.
>>> +     */
>>> +    ret = __xe_exec(fd, &exec);
>>> +    if (ret == 0) {
>>> +        int64_t timeout = NSEC_PER_SEC;
>>> +
>>> +        __xe_wait_ufence(fd, &vm_sync, PURGEABLE_FENCE_VAL,
>>> +                 exec_queue, &timeout);
>>> +    }
>>> +
>>> +    /*
>>> +     * Don't purge the batch BO — GPU would fetch zeroed scratch
>>> +     * instructions and trigger an engine reset.
>>> +     */
>>> +
>>> +    munmap(data, data_size);
>>> +    munmap(batch, batch_size);
>>> +    gem_close(fd, bo);
>>> +    gem_close(fd, batch_bo);
>>> +    xe_exec_queue_destroy(fd, bind_engine);
>>> +    xe_exec_queue_destroy(fd, exec_queue);
>>> +    xe_vm_destroy(fd, vm);
>>> +}
>>> +
>>>   int igt_main()
>>>   {
>>>       struct drm_xe_engine_class_instance *hwe; @@ -266,6 +400,12 @@
>>> int igt_main()
>>>               break;
>>>           }
>>>
>>> +    igt_subtest("dontneed-before-exec")
>>> +        xe_for_each_engine(fd, hwe) {
>>> +            test_dontneed_before_exec(fd, hwe);
>>> +            break;
>>> +        }
>>> +
>>>       igt_fixture() {
>>>           xe_device_put(fd);
>>>           drm_close_driver(fd);
>> Reviewed-by: Pravalika Gurram <pravalika.gurram@intel.com>
>>> -- 
>>> 2.43.0

  reply	other threads:[~2026-02-23  9:45 UTC|newest]

Thread overview: 38+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-02-17  2:34 [PATCH i-g-t v3 0/8] tests/xe: Add purgeable memory madvise tests for system allocator Arvind Yadav
2026-02-17  2:34 ` [PATCH i-g-t v3 1/8] drm-uapi/xe_drm: Add UAPI support for purgeable buffer objects Arvind Yadav
2026-02-18 11:22   ` Kamil Konieczny
2026-02-23  4:29   ` Gurram, Pravalika
2026-02-23  9:33     ` Yadav, Arvind
2026-02-17  2:34 ` [PATCH i-g-t v3 2/8] lib/xe: Add purgeable memory ioctl support Arvind Yadav
2026-02-23  4:30   ` Gurram, Pravalika
2026-02-17  2:34 ` [PATCH i-g-t v3 3/8] tests/intel/xe_madvise: Add dontneed-before-mmap subtest Arvind Yadav
2026-02-19  4:58   ` Gurram, Pravalika
2026-02-20  6:06     ` Yadav, Arvind
2026-02-23  5:09   ` Gurram, Pravalika
2026-02-23  5:49     ` Sharma, Nishit
2026-02-23  9:36       ` Yadav, Arvind
2026-02-17  2:34 ` [PATCH i-g-t v3 4/8] tests/intel/xe_madvise: Add dontneed-after-mmap subtest Arvind Yadav
2026-02-19  5:01   ` Gurram, Pravalika
2026-02-20  6:13     ` Yadav, Arvind
2026-02-23  5:10   ` Gurram, Pravalika
2026-02-23  5:52     ` Sharma, Nishit
2026-02-23  9:37       ` Yadav, Arvind
2026-02-17  2:34 ` [PATCH i-g-t v3 5/8] tests/intel/xe_madvise: Add dontneed-before-exec subtest Arvind Yadav
2026-02-23  5:10   ` Gurram, Pravalika
2026-02-23  6:06     ` Sharma, Nishit
2026-02-23  9:45       ` Yadav, Arvind [this message]
2026-02-17  2:34 ` [PATCH i-g-t v3 6/8] tests/intel/xe_madvise: Add dontneed-after-exec subtest Arvind Yadav
2026-02-23  5:11   ` Gurram, Pravalika
2026-02-23  7:04     ` Sharma, Nishit
2026-02-24  4:07       ` Yadav, Arvind
2026-02-17  2:34 ` [PATCH i-g-t v3 7/8] tests/intel/xe_madvise: Add per-vma-tracking subtest Arvind Yadav
2026-02-23  5:11   ` Gurram, Pravalika
2026-02-23  7:17     ` Sharma, Nishit
2026-02-24  4:08       ` Yadav, Arvind
2026-02-17  2:34 ` [PATCH i-g-t v3 8/8] tests/intel/xe_madvise: Add per-vma-protection subtest Arvind Yadav
2026-02-23  5:12   ` Gurram, Pravalika
2026-02-23  7:25     ` Sharma, Nishit
2026-02-17  3:25 ` ✓ Xe.CI.BAT: success for tests/xe: Add purgeable memory madvise tests for system allocator (rev3) Patchwork
2026-02-17  4:22 ` ✓ Xe.CI.FULL: " Patchwork
2026-02-17  8:57 ` ✓ i915.CI.BAT: " Patchwork
2026-02-17 10:50 ` ✗ i915.CI.Full: failure " 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=640d2e40-ef5a-4cc5-acdb-b468b8b537cd@intel.com \
    --to=arvind.yadav@intel.com \
    --cc=himal.prasad.ghimiray@intel.com \
    --cc=igt-dev@lists.freedesktop.org \
    --cc=matthew.brost@intel.com \
    --cc=nishit.sharma@intel.com \
    --cc=pravalika.gurram@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