[Intel-gfx] [PATCH i-g-t] lib: Incrementally mlock()
Caz Yokoyama
Caz.Yokoyama at intel.com
Wed Feb 27 17:34:57 UTC 2019
inline.
v2 patches fixes
- address calculation bug
- not killed
However, the test does not runs faster.
-caz
On Wed, 2019-02-27 at 16:29 +0000, Chris Wilson wrote:
> As we already have the previous portion of the mmap mlocked, we only
> need to mlock() the fresh portion for testing available memory.
>
> v2: Fixup the uint64_t pointer arithmetric and only use a single mmap
> to
> avoid subsequent mlock fail (for reasons unknown, but bet on mm/).
>
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Caz Yokoyama <Caz.Yokoyama at intel.com>
> ---
> lib/igt_aux.h | 2 +-
> lib/intel_os.c | 40 +++++++++++++++++++++--------------
> ----
> tests/eviction_common.c | 17 +++++++++--------
> tests/i915/i915_suspend.c | 17 +++--------------
> 4 files changed, 35 insertions(+), 41 deletions(-)
>
> diff --git a/lib/igt_aux.h b/lib/igt_aux.h
> index fb02c026a..09b6246bf 100644
> --- a/lib/igt_aux.h
> +++ b/lib/igt_aux.h
> @@ -194,7 +194,7 @@ void intel_purge_vm_caches(int fd);
> uint64_t intel_get_avail_ram_mb(void);
> uint64_t intel_get_total_ram_mb(void);
> uint64_t intel_get_total_swap_mb(void);
> -size_t intel_get_total_pinnable_mem(void);
> +void *intel_get_total_pinnable_mem(size_t *pinned);
>
> int __intel_check_memory(uint64_t count, uint64_t size, unsigned
> mode,
> uint64_t *out_required, uint64_t *out_total);
> diff --git a/lib/intel_os.c b/lib/intel_os.c
> index e1e31e230..dd93bea1a 100644
> --- a/lib/intel_os.c
> +++ b/lib/intel_os.c
> @@ -227,11 +227,9 @@ intel_get_total_swap_mb(void)
> *
> * Returns: Amount of memory that can be safely pinned, in bytes.
> */
> -size_t
> -intel_get_total_pinnable_mem(void)
> +void *intel_get_total_pinnable_mem(size_t *total)
> {
> uint64_t *can_mlock, pin, avail;
> - size_t ret;
>
> pin = (intel_get_total_ram_mb() + 1) << 20;
> avail = (intel_get_avail_ram_mb() + 1) << 20;
> @@ -245,34 +243,40 @@ intel_get_total_pinnable_mem(void)
> */
> *can_mlock = (avail >> 1) + (avail >> 2);
> if (mlock(can_mlock, *can_mlock)) {
> - *can_mlock = 0;
> - goto out;
> + munmap(can_mlock, pin);
> + return MAP_FAILED;
> }
>
> for (uint64_t inc = 1024 << 20; inc >= 4 << 10; inc >>= 2) {
> - igt_debug("Testing mlock %'"PRIu64" B (%'"PRIu64"
> MiB)\n",
> - *can_mlock, *can_mlock >> 20);
> + uint64_t locked = *can_mlock;
> +
> + igt_debug("Testing mlock %'"PRIu64"B (%'"PRIu64"MiB) +
> %'"PRIu64"B\n",
> + locked, locked >> 20, inc);
>
> igt_fork(child, 1) {
> - for (uint64_t bytes = *can_mlock;
> - bytes <= pin;
> - bytes += inc) {
> - if (mlock(can_mlock, bytes))
> + uint64_t bytes = *can_mlock;
> +
> + while (bytes <= pin) {
> + if (mlock((void *)can_mlock + bytes,
> inc))
> break;
>
> - *can_mlock = bytes;
> + *can_mlock = bytes += inc;
> __sync_synchronize();
> }
> }
> __igt_waitchildren();
> - igt_assert(!mlock(can_mlock, *can_mlock));
> - }
>
> -out:
> - ret = *can_mlock;
> - munmap(can_mlock, pin);
> + if (*can_mlock > locked + inc) { /* Weird bit of mm/
> lore */
> + *can_mlock -= inc;
>
Are you able to explain this? Is this for when child is killed during
updating *can_mlock?
If the condition is not met, parent does not mlock. Is this OK?
> + igt_debug("Claiming mlock %'"PRIu64"B
> (%'"PRIu64"MiB)\n",
> + *can_mlock, *can_mlock >> 20);
> + igt_assert(!mlock((void *)can_mlock + locked,
>
> + *can_mlock - locked));
> + }
> + }
>
> - return ret;
> + *total = pin;
>
*total = *can_mlock?
Also you don't unmap. I mean map and unmap are not paired in the same
function. Not elegant. But c'est la vie.
-caz
> + return can_mlock;
> }
>
> static uint64_t vfs_file_max(void)
> diff --git a/tests/eviction_common.c b/tests/eviction_common.c
> index 321772ba7..a3b9e4167 100644
> --- a/tests/eviction_common.c
> +++ b/tests/eviction_common.c
> @@ -133,23 +133,24 @@ static void mlocked_evictions(int fd, struct
> igt_eviction_test_ops *ops,
> uint64_t surface_size,
> uint64_t surface_count)
> {
> + uint64_t sz, pin, total;
> void *mem;
> - uint64_t sz, pin_total;
>
> intel_require_memory(surface_count, surface_size, CHECK_RAM);
>
> - sz = surface_size*surface_count;
> - pin_total = intel_get_total_pinnable_mem();
> - igt_require(pin_total > sz);
> -
> - mem = mmap(NULL, pin_total, PROT_READ, MAP_SHARED | MAP_ANON,
> -1, 0);
> + mem = intel_get_total_pinnable_mem(&total);
> igt_assert(mem != MAP_FAILED);
> + pin = *(uint64_t *)mem;
> + igt_assert(!munlock(mem, pin));
> +
> + sz = surface_size * surface_count;
> + igt_require(pin > sz);
>
> igt_fork(child, 1) {
> uint32_t *bo;
> uint64_t n;
> int ret;
> - size_t lock = pin_total - sz;
> + size_t lock = pin - sz;
>
> bo = malloc(surface_count * sizeof(*bo));
> igt_assert(bo);
> @@ -186,7 +187,7 @@ static void mlocked_evictions(int fd, struct
> igt_eviction_test_ops *ops,
> }
> igt_waitchildren();
>
> - munmap(mem, pin_total);
> + munmap(mem, total);
> }
>
> static int swapping_evictions(int fd, struct igt_eviction_test_ops
> *ops,
> diff --git a/tests/i915/i915_suspend.c b/tests/i915/i915_suspend.c
> index 84cb3b490..cd7cf9675 100644
> --- a/tests/i915/i915_suspend.c
> +++ b/tests/i915/i915_suspend.c
> @@ -163,25 +163,14 @@ test_sysfs_reader(bool hibernate)
> static void
> test_shrink(int fd, unsigned int mode)
> {
> - void *mem;
> size_t size;
> + void *mem;
>
> gem_quiescent_gpu(fd);
> intel_purge_vm_caches(fd);
>
> - size = intel_get_total_pinnable_mem();
> - igt_require(size > 64 << 20);
> - size -= 64 << 20;
> -
> - mem = mmap(NULL, size, PROT_READ, MAP_SHARED | MAP_ANON, -1,
> 0);
> -
> - intel_purge_vm_caches(fd);
> -
> - igt_debug("Locking %'zu B (%'zu MiB)\n",
> - size, size >> 20);
> - igt_assert(!mlock(mem, size));
> - igt_info("Locked %'zu B (%'zu MiB)\n",
> - size, size >> 20);
> + mem = intel_get_total_pinnable_mem(&size);
> + igt_assert(mem != MAP_FAILED);
>
> intel_purge_vm_caches(fd);
> igt_system_suspend_autoresume(mode, SUSPEND_TEST_NONE);
More information about the Intel-gfx
mailing list