intel-gfx.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: Intel-gfx@lists.freedesktop.org
Subject: Re: [Intel-gfx] [PATCH] drm/i915: Align engine->uabi_class/instance with i915_drm.h
Date: Wed, 15 Jan 2020 15:28:19 +0000	[thread overview]
Message-ID: <b2917343-c654-2300-9716-7747d8b88be8@linux.intel.com> (raw)
In-Reply-To: <20200115152437.13207-1-tvrtko.ursulin@linux.intel.com>


On 15/01/2020 15:24, Tvrtko Ursulin wrote:
> From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> 
> In our ABI we have defined I915_ENGINE_CLASS_INVALID_NONE and
> I915_ENGINE_CLASS_INVALID_VIRTUAL as negative values which creates
> implicit coupling with type widths used in, also ABI, struct
> i915_engine_class_instance.
> 
> When for instance we export engine->uabi_class
> I915_ENGINE_CLASS_INVALID_VIRTUAL from our our tracepoints, because the
> type of the former is u8 in contrast to u16 defined in the ABI, 254 will
> be returned instead of 65534 which userspace would legitimately expect.
> 
> Therefore we need to align the type used to store engine ABI class and
> instance.
> 
> I did not find any other user visible inconsistency apart from the
> tracepoints so I think importance of the fix is low.

Alternatives:

1.
Embed struct i915_engine_class_instance in struct intel_engine_cs, but 
downside is more churn.

2.
Only tweak the tracepoints to cast back and forth, but is it possible to 
cast from unsigned to signed and get a negative number?

3.
Do nothing, does anyone cares?

Regards,

Tvrtko

> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> Cc: Chris Wilson <chris@chris-wilson.co.uk>
> ---
>   drivers/gpu/drm/i915/gem/i915_gem_busy.c     | 12 ++++++------
>   drivers/gpu/drm/i915/gt/intel_engine_types.h |  4 ++--
>   2 files changed, 8 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_busy.c b/drivers/gpu/drm/i915/gem/i915_gem_busy.c
> index 3d4f5775a4ba..25235ef630c1 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_busy.c
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_busy.c
> @@ -9,16 +9,16 @@
>   #include "i915_gem_ioctls.h"
>   #include "i915_gem_object.h"
>   
> -static __always_inline u32 __busy_read_flag(u8 id)
> +static __always_inline u32 __busy_read_flag(u16 id)
>   {
> -	if (id == (u8)I915_ENGINE_CLASS_INVALID)
> +	if (id == (u16)I915_ENGINE_CLASS_INVALID)
>   		return 0xffff0000u;
>   
>   	GEM_BUG_ON(id >= 16);
>   	return 0x10000u << id;
>   }
>   
> -static __always_inline u32 __busy_write_id(u8 id)
> +static __always_inline u32 __busy_write_id(u16 id)
>   {
>   	/*
>   	 * The uABI guarantees an active writer is also amongst the read
> @@ -29,14 +29,14 @@ static __always_inline u32 __busy_write_id(u8 id)
>   	 * last_read - hence we always set both read and write busy for
>   	 * last_write.
>   	 */
> -	if (id == (u8)I915_ENGINE_CLASS_INVALID)
> +	if (id == (u16)I915_ENGINE_CLASS_INVALID)
>   		return 0xffffffffu;
>   
>   	return (id + 1) | __busy_read_flag(id);
>   }
>   
>   static __always_inline unsigned int
> -__busy_set_if_active(const struct dma_fence *fence, u32 (*flag)(u8 id))
> +__busy_set_if_active(const struct dma_fence *fence, u32 (*flag)(u16 id))
>   {
>   	const struct i915_request *rq;
>   
> @@ -57,7 +57,7 @@ __busy_set_if_active(const struct dma_fence *fence, u32 (*flag)(u8 id))
>   		return 0;
>   
>   	/* Beware type-expansion follies! */
> -	BUILD_BUG_ON(!typecheck(u8, rq->engine->uabi_class));
> +	BUILD_BUG_ON(!typecheck(u16, rq->engine->uabi_class));
>   	return flag(rq->engine->uabi_class);
>   }
>   
> diff --git a/drivers/gpu/drm/i915/gt/intel_engine_types.h b/drivers/gpu/drm/i915/gt/intel_engine_types.h
> index 00287515e7af..350da59e605b 100644
> --- a/drivers/gpu/drm/i915/gt/intel_engine_types.h
> +++ b/drivers/gpu/drm/i915/gt/intel_engine_types.h
> @@ -278,8 +278,8 @@ struct intel_engine_cs {
>   	u8 class;
>   	u8 instance;
>   
> -	u8 uabi_class;
> -	u8 uabi_instance;
> +	u16 uabi_class;
> +	u16 uabi_instance;
>   
>   	u32 uabi_capabilities;
>   	u32 context_size;
> 
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

  reply	other threads:[~2020-01-15 15:28 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-01-15 15:24 [Intel-gfx] [PATCH] drm/i915: Align engine->uabi_class/instance with i915_drm.h Tvrtko Ursulin
2020-01-15 15:28 ` Tvrtko Ursulin [this message]
2020-01-15 15:35   ` Chris Wilson
2020-01-15 16:29     ` Tvrtko Ursulin
2020-01-15 18:06 ` [Intel-gfx] ✓ Fi.CI.BAT: success for " Patchwork
2020-01-15 18:07 ` [Intel-gfx] ✗ Fi.CI.BUILD: warning " Patchwork
2020-01-16 13:45 ` [Intel-gfx] [PATCH v2] " Tvrtko Ursulin
2020-01-16 15:50 ` [Intel-gfx] ✓ Fi.CI.BAT: success for drm/i915: Align engine->uabi_class/instance with i915_drm.h (rev2) Patchwork
2020-01-16 15:50 ` [Intel-gfx] ✗ Fi.CI.BUILD: warning " Patchwork
2020-01-18  5:12 ` [Intel-gfx] ✗ Fi.CI.IGT: failure for drm/i915: Align engine->uabi_class/instance with i915_drm.h Patchwork
2020-01-19 11:15 ` [Intel-gfx] ✓ Fi.CI.IGT: success for drm/i915: Align engine->uabi_class/instance with i915_drm.h (rev2) 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=b2917343-c654-2300-9716-7747d8b88be8@linux.intel.com \
    --to=tvrtko.ursulin@linux.intel.com \
    --cc=Intel-gfx@lists.freedesktop.org \
    /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;
as well as URLs for NNTP newsgroup(s).