Intel-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Matthew Auld <matthew.auld@intel.com>
To: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>,
	intel-gfx@lists.freedesktop.org
Cc: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>,
	"Kenneth Graunke" <kenneth@whitecape.org>,
	dri-devel@lists.freedesktop.org,
	"Daniel Vetter" <daniel.vetter@ffwll.ch>
Subject: Re: [Intel-gfx] [PATCH 03/10] drm/i915/uapi: expose the avail tracking
Date: Thu, 26 May 2022 09:10:35 +0100	[thread overview]
Message-ID: <4acbf57d-fe19-8e6a-2521-5557bf842e54@intel.com> (raw)
In-Reply-To: <db5b82b1-4eb0-0d26-eaa8-b814b0de4dc0@linux.intel.com>

On 26/05/2022 08:58, Tvrtko Ursulin wrote:
> 
> On 25/05/2022 19:43, Matthew Auld wrote:
>> Vulkan would like to have a rough measure of how much device memory can
>> in theory be allocated. Also add unallocated_cpu_visible_size to track
>> the visible portion, in case the device is using small BAR.
> 
> I have concerns here that it isn't useful and could even be 
> counter-productive. If we give unprivileged access it may be considered 
> a side channel, but if we "lie" (report total region size) to 
> unprivileged clients (like in this patch), then they don't co-operate 
> well and end trashing.
> 
> Is Vulkan really sure it wants this and why?

Lionel pointed out: 
https://www.khronos.org/registry/vulkan/specs/1.3-extensions/man/html/VK_EXT_memory_budget.html

Also note that the existing behaviour was to lie. I'm not sure what's 
the best option here.

> 
> Regards,
> 
> Tvrtko
> 
>> Testcase: igt@i915_query@query-regions-unallocated
>> Testcase: igt@i915_query@query-regions-sanity-check
>> Signed-off-by: Matthew Auld <matthew.auld@intel.com>
>> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
>> Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
>> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
>> Cc: Jon Bloomfield <jon.bloomfield@intel.com>
>> Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
>> Cc: Jordan Justen <jordan.l.justen@intel.com>
>> Cc: Kenneth Graunke <kenneth@whitecape.org>
>> Cc: Akeem G Abodunrin <akeem.g.abodunrin@intel.com>
>> ---
>>   drivers/gpu/drm/i915/i915_query.c             | 10 +++++-
>>   drivers/gpu/drm/i915/i915_ttm_buddy_manager.c | 20 ++++++++++++
>>   drivers/gpu/drm/i915/i915_ttm_buddy_manager.h |  3 ++
>>   drivers/gpu/drm/i915/intel_memory_region.c    | 14 +++++++++
>>   drivers/gpu/drm/i915/intel_memory_region.h    |  3 ++
>>   include/uapi/drm/i915_drm.h                   | 31 ++++++++++++++++++-
>>   6 files changed, 79 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/i915/i915_query.c 
>> b/drivers/gpu/drm/i915/i915_query.c
>> index 9aa0b28aa6ee..e095c55f4d4b 100644
>> --- a/drivers/gpu/drm/i915/i915_query.c
>> +++ b/drivers/gpu/drm/i915/i915_query.c
>> @@ -502,7 +502,15 @@ static int query_memregion_info(struct 
>> drm_i915_private *i915,
>>           else
>>               info.probed_cpu_visible_size = mr->total;
>> -        info.unallocated_size = mr->avail;
>> +        if (perfmon_capable()) {
>> +            intel_memory_region_avail(mr,
>> +                          &info.unallocated_size,
>> +                          &info.unallocated_cpu_visible_size);
>> +        } else {
>> +            info.unallocated_size = info.probed_size;
>> +            info.unallocated_cpu_visible_size =
>> +                info.probed_cpu_visible_size;
>> +        }
>>           if (__copy_to_user(info_ptr, &info, sizeof(info)))
>>               return -EFAULT;
>> diff --git a/drivers/gpu/drm/i915/i915_ttm_buddy_manager.c 
>> b/drivers/gpu/drm/i915/i915_ttm_buddy_manager.c
>> index a5109548abc0..aa5c91e44438 100644
>> --- a/drivers/gpu/drm/i915/i915_ttm_buddy_manager.c
>> +++ b/drivers/gpu/drm/i915/i915_ttm_buddy_manager.c
>> @@ -365,6 +365,26 @@ u64 i915_ttm_buddy_man_visible_size(struct 
>> ttm_resource_manager *man)
>>       return bman->visible_size;
>>   }
>> +/**
>> + * i915_ttm_buddy_man_visible_size - Query the avail tracking for the 
>> manager.
>> + *
>> + * @man: The buddy allocator ttm manager
>> + * @avail: The total available memory in pages for the entire manager.
>> + * @visible_avail: The total available memory in pages for the CPU 
>> visible
>> + * portion. Note that this will always give the same value as @avail on
>> + * configurations that don't have a small BAR.
>> + */
>> +void i915_ttm_buddy_man_avail(struct ttm_resource_manager *man,
>> +                 u64 *avail, u64 *visible_avail)
>> +{
>> +    struct i915_ttm_buddy_manager *bman = to_buddy_manager(man);
>> +
>> +    mutex_lock(&bman->lock);
>> +    *avail = bman->mm.avail >> PAGE_SHIFT;
>> +    *visible_avail = bman->visible_avail;
>> +    mutex_unlock(&bman->lock);
>> +}
>> +
>>   #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
>>   void i915_ttm_buddy_man_force_visible_size(struct 
>> ttm_resource_manager *man,
>>                          u64 size)
>> diff --git a/drivers/gpu/drm/i915/i915_ttm_buddy_manager.h 
>> b/drivers/gpu/drm/i915/i915_ttm_buddy_manager.h
>> index 52d9586d242c..d64620712830 100644
>> --- a/drivers/gpu/drm/i915/i915_ttm_buddy_manager.h
>> +++ b/drivers/gpu/drm/i915/i915_ttm_buddy_manager.h
>> @@ -61,6 +61,9 @@ int i915_ttm_buddy_man_reserve(struct 
>> ttm_resource_manager *man,
>>   u64 i915_ttm_buddy_man_visible_size(struct ttm_resource_manager *man);
>> +void i915_ttm_buddy_man_avail(struct ttm_resource_manager *man,
>> +                  u64 *avail, u64 *avail_visible);
>> +
>>   #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
>>   void i915_ttm_buddy_man_force_visible_size(struct 
>> ttm_resource_manager *man,
>>                          u64 size);
>> diff --git a/drivers/gpu/drm/i915/intel_memory_region.c 
>> b/drivers/gpu/drm/i915/intel_memory_region.c
>> index e38d2db1c3e3..94ee26e99549 100644
>> --- a/drivers/gpu/drm/i915/intel_memory_region.c
>> +++ b/drivers/gpu/drm/i915/intel_memory_region.c
>> @@ -279,6 +279,20 @@ void intel_memory_region_set_name(struct 
>> intel_memory_region *mem,
>>       va_end(ap);
>>   }
>> +void intel_memory_region_avail(struct intel_memory_region *mr,
>> +                   u64 *avail, u64 *visible_avail)
>> +{
>> +    if (mr->type == INTEL_MEMORY_LOCAL) {
>> +        i915_ttm_buddy_man_avail(mr->region_private,
>> +                     avail, visible_avail);
>> +        *avail <<= PAGE_SHIFT;
>> +        *visible_avail <<= PAGE_SHIFT;
>> +    } else {
>> +        *avail = mr->total;
>> +        *visible_avail = mr->total;
>> +    }
>> +}
>> +
>>   void intel_memory_region_destroy(struct intel_memory_region *mem)
>>   {
>>       int ret = 0;
>> diff --git a/drivers/gpu/drm/i915/intel_memory_region.h 
>> b/drivers/gpu/drm/i915/intel_memory_region.h
>> index 3d8378c1b447..2214f251bec3 100644
>> --- a/drivers/gpu/drm/i915/intel_memory_region.h
>> +++ b/drivers/gpu/drm/i915/intel_memory_region.h
>> @@ -127,6 +127,9 @@ int intel_memory_region_reserve(struct 
>> intel_memory_region *mem,
>>   void intel_memory_region_debug(struct intel_memory_region *mr,
>>                      struct drm_printer *printer);
>> +void intel_memory_region_avail(struct intel_memory_region *mr,
>> +                   u64 *avail, u64 *visible_avail);
>> +
>>   struct intel_memory_region *
>>   i915_gem_ttm_system_setup(struct drm_i915_private *i915,
>>                 u16 type, u16 instance);
>> diff --git a/include/uapi/drm/i915_drm.h b/include/uapi/drm/i915_drm.h
>> index 9df419a45244..e30f31a440b3 100644
>> --- a/include/uapi/drm/i915_drm.h
>> +++ b/include/uapi/drm/i915_drm.h
>> @@ -3228,7 +3228,15 @@ struct drm_i915_memory_region_info {
>>        */
>>       __u64 probed_size;
>> -    /** @unallocated_size: Estimate of memory remaining (-1 = 
>> unknown) */
>> +    /**
>> +     * @unallocated_size: Estimate of memory remaining (-1 = unknown)
>> +     *
>> +     * Requires CAP_PERFMON or CAP_SYS_ADMIN to get reliable accounting.
>> +     * Without this (or if this is an older kernel) the value here will
>> +     * always equal the @probed_size. Note this is only currently 
>> tracked
>> +     * for I915_MEMORY_CLASS_DEVICE regions (for other types the 
>> value here
>> +     * will always equal the @probed_size).
>> +     */
>>       __u64 unallocated_size;
>>       union {
>> @@ -3262,6 +3270,27 @@ struct drm_i915_memory_region_info {
>>                * @probed_size.
>>                */
>>               __u64 probed_cpu_visible_size;
>> +
>> +            /**
>> +             * @unallocated_cpu_visible_size: Estimate of CPU
>> +             * visible memory remaining (-1 = unknown).
>> +             *
>> +             * Note this is only tracked for
>> +             * I915_MEMORY_CLASS_DEVICE regions (for other types the
>> +             * value here will always equal the
>> +             * @probed_cpu_visible_size).
>> +             *
>> +             * Requires CAP_PERFMON or CAP_SYS_ADMIN to get reliable
>> +             * accounting.  Without this the value here will always
>> +             * equal the @probed_cpu_visible_size. Note this is only
>> +             * currently tracked for I915_MEMORY_CLASS_DEVICE
>> +             * regions (for other types the value here will also
>> +             * always equal the @probed_cpu_visible_size).
>> +             *
>> +             * If this is an older kernel the value here will be
>> +             * zero, see also @probed_cpu_visible_size.
>> +             */
>> +            __u64 unallocated_cpu_visible_size;
>>           };
>>       };
>>   };

  reply	other threads:[~2022-05-26  8:10 UTC|newest]

Thread overview: 34+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-05-25 18:43 [Intel-gfx] [PATCH 00/10] small BAR uapi bits Matthew Auld
2022-05-25 18:43 ` [Intel-gfx] [PATCH 01/10] drm/doc: add rfc section for small BAR uapi Matthew Auld
2022-06-16 11:18   ` Thomas Hellström (Intel)
2022-05-25 18:43 ` [Intel-gfx] [PATCH 02/10] drm/i915/uapi: add probed_cpu_visible_size Matthew Auld
2022-06-01 12:36   ` Das, Nirmoy
2022-06-16 11:22   ` Thomas Hellström (Intel)
2022-05-25 18:43 ` [Intel-gfx] [PATCH 03/10] drm/i915/uapi: expose the avail tracking Matthew Auld
2022-05-26  2:44   ` kernel test robot
2022-05-26  7:58   ` Tvrtko Ursulin
2022-05-26  8:10     ` Matthew Auld [this message]
2022-05-26  8:33       ` Tvrtko Ursulin
2022-05-30 17:05         ` Matthew Auld
2022-05-25 18:43 ` [Intel-gfx] [PATCH 04/10] drm/i915: remove intel_memory_region avail Matthew Auld
2022-06-17 12:16   ` Thomas Hellström
2022-05-25 18:43 ` [Intel-gfx] [PATCH 05/10] drm/i915/uapi: apply ALLOC_GPU_ONLY by default Matthew Auld
2022-05-25 18:43 ` [Intel-gfx] [PATCH 06/10] drm/i915/uapi: add NEEDS_CPU_ACCESS hint Matthew Auld
2022-06-01 12:30   ` Das, Nirmoy
2022-06-17 14:30   ` Thomas Hellström (Intel)
2022-05-25 18:43 ` [Intel-gfx] [PATCH 07/10] drm/i915/error: skip non-mappable pages Matthew Auld
2022-06-01 12:30   ` Das, Nirmoy
2022-06-17 14:26   ` Thomas Hellström (Intel)
2022-05-25 18:43 ` [Intel-gfx] [PATCH 08/10] drm/i915/uapi: disable capturing objects on recoverable contexts Matthew Auld
2022-05-26  0:08   ` kernel test robot
2022-06-17 12:28   ` Thomas Hellström (Intel)
2022-05-25 18:43 ` [Intel-gfx] [PATCH 09/10] drm/i915: turn on small BAR support Matthew Auld
2022-06-17 12:33   ` Thomas Hellström
2022-06-21  8:38     ` Matthew Auld
2022-06-21  9:05       ` Das, Nirmoy
2022-06-21  9:34         ` Thomas Hellström
2022-05-25 18:43 ` [Intel-gfx] [PATCH 10/10] HAX: force small BAR on dg2 Matthew Auld
2022-05-25 19:25 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for small BAR uapi bits Patchwork
2022-05-25 19:25 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2022-05-25 20:16 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2022-05-26 10:58 ` [Intel-gfx] ✓ Fi.CI.IGT: " 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=4acbf57d-fe19-8e6a-2521-5557bf842e54@intel.com \
    --to=matthew.auld@intel.com \
    --cc=daniel.vetter@ffwll.ch \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=kenneth@whitecape.org \
    --cc=thomas.hellstrom@linux.intel.com \
    --cc=tvrtko.ursulin@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