[v9] drm/i915: Allocate intel_engine_cs structure only for the enabled engines

Commit Message

[e-mail] Oct. 7, 2016, 9:33 a.m. UTC

From: Akash Goel <[e-mail]>

With the possibility of addition of many more number of rings in future,
the drm_i915_private structure could bloat as an array, of type
intel_engine_cs, is embedded inside it.
	struct intel_engine_cs engine[I915_NUM_ENGINES];
Though this is still fine as generally there is only a single instance of
drm_i915_private structure used, but not all of the possible rings would be
enabled or active on most of the platforms. Some memory can be saved by
allocating intel_engine_cs structure only for the enabled/active engines.
Currently the engine/ring ID is kept static and dev_priv->engine[] is simply
indexed using the enums defined in intel_engine_id.
To save memory and continue using the static engine/ring IDs, 'engine' is
defined as an array of pointers.
	struct intel_engine_cs *engine[I915_NUM_ENGINES];
dev_priv->engine[engine_ID] will be NULL for disabled engine instances.

v2:
- Remove the engine iterator field added in drm_i915_private structure,
  instead pass a local iterator variable to the for_each_engine**
  macros. (Chris)
- Do away with intel_engine_initialized() and instead directly use the
  NULL pointer check on engine pointer. (Chris)

v3:
- Remove for_each_engine_id() macro, as the updated macro for_each_engine()
  can be used in place of it. (Chris)
- Protect the access to Render engine Fault register with a NULL check, as
  engine specific init is done later in Driver load sequence.

v4:
- Use !!dev_priv->engine[VCS] style for the engine check in getparam. (Chris)
- Kill the superfluous init_engine_lists().

v5:
- Cleanup the intel_engines_init() & intel_engines_setup(), with respect to
  allocation of intel_engine_cs structure. (Chris)

v6:
- Rebase.

v7:
- Optimize the for_each_engine_masked() macro. (Chris)
- Change the type of 'iter' local variable to enum intel_engine_id. (Chris)
- Rebase.

v8: Rebase.

v9: Rebase.

Cc: Chris Wilson <[e-mail]>
Signed-off-by: Akash Goel <[e-mail]>
---
 drivers/gpu/drm/i915/i915_cmd_parser.c     |  3 +-
 drivers/gpu/drm/i915/i915_debugfs.c        | 57 ++++++++++++++++++------------
 drivers/gpu/drm/i915/i915_drv.c            | 11 +++---
 drivers/gpu/drm/i915/i915_drv.h            | 22 ++++--------
 drivers/gpu/drm/i915/i915_gem.c            | 29 +++++++--------
 drivers/gpu/drm/i915/i915_gem_context.c    | 15 ++++----
 drivers/gpu/drm/i915/i915_gem_evict.c      |  3 +-
 drivers/gpu/drm/i915/i915_gem_execbuffer.c |  6 ++--
 drivers/gpu/drm/i915/i915_gem_gtt.c        | 14 +++++---
 drivers/gpu/drm/i915/i915_gem_request.c    |  5 +--
 drivers/gpu/drm/i915/i915_gpu_error.c      | 30 ++++++++--------
 drivers/gpu/drm/i915/i915_guc_submission.c | 12 ++++---
 drivers/gpu/drm/i915/i915_irq.c            | 44 ++++++++++++-----------
 drivers/gpu/drm/i915/intel_breadcrumbs.c   |  6 ++--
 drivers/gpu/drm/i915/intel_display.c       |  8 ++---
 drivers/gpu/drm/i915/intel_engine_cs.c     | 22 ++++++++----
 drivers/gpu/drm/i915/intel_guc_loader.c    |  6 ++--
 drivers/gpu/drm/i915/intel_lrc.c           | 10 ++++--
 drivers/gpu/drm/i915/intel_overlay.c       |  2 +-
 drivers/gpu/drm/i915/intel_pm.c            | 17 +++++----
 drivers/gpu/drm/i915/intel_ringbuffer.c    | 14 +++++---
 drivers/gpu/drm/i915/intel_ringbuffer.h    |  6 ----
 22 files changed, 191 insertions(+), 151 deletions(-)

Comments

Joonas Lahtinen Oct. 7, 2016, 10:49 a.m. UTC | #1

On pe, 2016-10-07 at 15:03 +0530, [e-mail] wrote:
> > From: Akash Goel <[e-mail]>
> 
> With the possibility of addition of many more number of rings in future,
> the drm_i915_private structure could bloat as an array, of type
> intel_engine_cs, is embedded inside it.
> 	struct intel_engine_cs engine[I915_NUM_ENGINES];
> Though this is still fine as generally there is only a single instance of
> drm_i915_private structure used, but not all of the possible rings would be
> enabled or active on most of the platforms. Some memory can be saved by
> allocating intel_engine_cs structure only for the enabled/active engines.
> Currently the engine/ring ID is kept static and dev_priv->engine[] is simply
> indexed using the enums defined in intel_engine_id.
> To save memory and continue using the static engine/ring IDs, 'engine' is
> defined as an array of pointers.
> 	struct intel_engine_cs *engine[I915_NUM_ENGINES];
> dev_priv->engine[engine_ID] will be NULL for disabled engine instances.
> 
> v2:
> - Remove the engine iterator field added in drm_i915_private structure,
>   instead pass a local iterator variable to the for_each_engine**
>   macros. (Chris)
> - Do away with intel_engine_initialized() and instead directly use the
>   NULL pointer check on engine pointer. (Chris)
> 
> v3:
> - Remove for_each_engine_id() macro, as the updated macro for_each_engine()
>   can be used in place of it. (Chris)
> - Protect the access to Render engine Fault register with a NULL check, as
>   engine specific init is done later in Driver load sequence.
> 
> v4:
> - Use !!dev_priv->engine[VCS] style for the engine check in getparam. (Chris)
> - Kill the superfluous init_engine_lists().
> 
> v5:
> - Cleanup the intel_engines_init() & intel_engines_setup(), with respect to
>   allocation of intel_engine_cs structure. (Chris)
> 
> v6:
> - Rebase.
> 
> v7:
> - Optimize the for_each_engine_masked() macro. (Chris)
> - Change the type of 'iter' local variable to enum intel_engine_id. (Chris)
> - Rebase.
> 

Would not it be consistent to go with 'id' everywhere rather than
'iter'. Consistency is good, and my vote for 'id' as it's more
descriptive?
 
> @@ -153,9 +163,9 @@ int intel_engines_init(struct drm_device *dev)
>  cleanup:
>  	for (i = 0; i < I915_NUM_ENGINES; i++) {

Use for_each_engine here too.

> diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
> index 936f6f6..08303e3 100644
> --- a/drivers/gpu/drm/i915/intel_lrc.c
> +++ b/drivers/gpu/drm/i915/intel_lrc.c
> @@ -1645,7 +1645,7 @@ void intel_logical_ring_cleanup(struct intel_engine_cs *engine)
>  {
>  	struct drm_i915_private *dev_priv;
>  
> -	if (!intel_engine_initialized(engine))
> +	if (!engine)
>  		return;

Remove this check or make it GEM_BUG_ON(!engine); but I don't think we
need that much paranoia.

> --- a/drivers/gpu/drm/i915/intel_ringbuffer.c
> +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c

<SNIP>

> @@ -2091,7 +2092,7 @@ void intel_engine_cleanup(struct intel_engine_cs *engine)
>  {
> >  	struct drm_i915_private *dev_priv;
>  
> -	if (!intel_engine_initialized(engine))
> +	if (!engine)
>  		return;

Same as above.

With those points fixed;

Reviewed-by: Joonas Lahtinen <[e-mail]>

Regards, Joonas

Akash Goel Oct. 7, 2016, 1:24 p.m. UTC | #2

On Fri, Oct 7, 2016 at 4:19 PM, Joonas Lahtinen
<[e-mail]> wrote:
> On pe, 2016-10-07 at 15:03 +0530, [e-mail] wrote:
>> > From: Akash Goel <[e-mail]>
>>
>> With the possibility of addition of many more number of rings in future,
>> the drm_i915_private structure could bloat as an array, of type
>> intel_engine_cs, is embedded inside it.
>>       struct intel_engine_cs engine[I915_NUM_ENGINES];
>> Though this is still fine as generally there is only a single instance of
>> drm_i915_private structure used, but not all of the possible rings would be
>> enabled or active on most of the platforms. Some memory can be saved by
>> allocating intel_engine_cs structure only for the enabled/active engines.
>> Currently the engine/ring ID is kept static and dev_priv->engine[] is simply
>> indexed using the enums defined in intel_engine_id.
>> To save memory and continue using the static engine/ring IDs, 'engine' is
>> defined as an array of pointers.
>>       struct intel_engine_cs *engine[I915_NUM_ENGINES];
>> dev_priv->engine[engine_ID] will be NULL for disabled engine instances.
>>
>> v2:
>> - Remove the engine iterator field added in drm_i915_private structure,
>>   instead pass a local iterator variable to the for_each_engine**
>>   macros. (Chris)
>> - Do away with intel_engine_initialized() and instead directly use the
>>   NULL pointer check on engine pointer. (Chris)
>>
>> v3:
>> - Remove for_each_engine_id() macro, as the updated macro for_each_engine()
>>   can be used in place of it. (Chris)
>> - Protect the access to Render engine Fault register with a NULL check, as
>>   engine specific init is done later in Driver load sequence.
>>
>> v4:
>> - Use !!dev_priv->engine[VCS] style for the engine check in getparam. (Chris)
>> - Kill the superfluous init_engine_lists().
>>
>> v5:
>> - Cleanup the intel_engines_init() & intel_engines_setup(), with respect to
>>   allocation of intel_engine_cs structure. (Chris)
>>
>> v6:
>> - Rebase.
>>
>> v7:
>> - Optimize the for_each_engine_masked() macro. (Chris)
>> - Change the type of 'iter' local variable to enum intel_engine_id. (Chris)
>> - Rebase.
>>
>
> Would not it be consistent to go with 'id' everywhere rather than
> 'iter'. Consistency is good, and my vote for 'id' as it's more
> descriptive?

Fine will then go with 'id' only.

>
>> @@ -153,9 +163,9 @@ int intel_engines_init(struct drm_device *dev)
>>  cleanup:
>>       for (i = 0; i < I915_NUM_ENGINES; i++) {
>
> Use for_each_engine here too.
>

Fine after using 'for_each_engine' here, the below Null pointer check
on 'engine' can be removed safely.
As you suggested, will not keep BUG_ON also.

>> diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
>> index 936f6f6..08303e3 100644
>> --- a/drivers/gpu/drm/i915/intel_lrc.c
>> +++ b/drivers/gpu/drm/i915/intel_lrc.c
>> @@ -1645,7 +1645,7 @@ void intel_logical_ring_cleanup(struct intel_engine_cs *engine)
>>  {
>>       struct drm_i915_private *dev_priv;
>>
>> -     if (!intel_engine_initialized(engine))
>> +     if (!engine)
>>               return;
>
> Remove this check or make it GEM_BUG_ON(!engine); but I don't think we
> need that much paranoia.
>
>> --- a/drivers/gpu/drm/i915/intel_ringbuffer.c
>> +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
>
> <SNIP>
>
>> @@ -2091,7 +2092,7 @@ void intel_engine_cleanup(struct intel_engine_cs *engine)
>>  {
>> >     struct drm_i915_private *dev_priv;
>>
>> -     if (!intel_engine_initialized(engine))
>> +     if (!engine)
>>               return;
>
> Same as above.
>
> With those points fixed;
>
Chris found an issue inside intel_engine_sync_index(), will add his
suggested fix also in the next version.


> Reviewed-by: Joonas Lahtinen <[e-mail]>

Thanks much for the review.

Best regards
Akash

>
> Regards, Joonas
> --
> Joonas Lahtinen
> Open Source Technology Center
> Intel Corporation
> _______________________________________________
> Intel-gfx mailing list
> [e-mail]
> https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[e-mail] Oct. 13, 2016, 4:10 p.m. UTC | #3

On 10/10/2016 6:03 PM, Patchwork wrote:
> == Series Details ==
>
> Series: drm/i915: Allocate intel_engine_cs structure only for the enabled engines (rev3)
> URL   : https://patchwork.freedesktop.org/series/13435/
> State : warning
>
> == Summary ==
>
> Series 13435v3 drm/i915: Allocate intel_engine_cs structure only for the enabled engines
> https://patchwork.freedesktop.org/api/1.0/series/13435/revisions/3/mbox/
>
> Test vgem_basic:
>         Subgroup unload:
>                 pass       -> SKIP       (fi-skl-6260u)
>                 pass       -> SKIP       (fi-skl-6700hq)
>                 skip       -> PASS       (fi-skl-6700k)
>
Checked with Chris about the above failure.
He said that the above unload failure for vgem module can't be
attributed to the patch, most likely a CI framework issue.

Best regards
Akash

> fi-bdw-5557u     total:248  pass:231  dwarn:0   dfail:0   fail:0   skip:17
> fi-bsw-n3050     total:248  pass:204  dwarn:0   dfail:0   fail:0   skip:44
> fi-bxt-t5700     total:248  pass:217  dwarn:0   dfail:0   fail:0   skip:31
> fi-byt-j1900     total:248  pass:214  dwarn:1   dfail:0   fail:1   skip:32
> fi-byt-n2820     total:248  pass:210  dwarn:0   dfail:0   fail:1   skip:37
> fi-hsw-4770      total:248  pass:224  dwarn:0   dfail:0   fail:0   skip:24
> fi-hsw-4770r     total:248  pass:224  dwarn:0   dfail:0   fail:0   skip:24
> fi-ilk-650       total:248  pass:185  dwarn:0   dfail:0   fail:2   skip:61
> fi-ivb-3520m     total:248  pass:221  dwarn:0   dfail:0   fail:0   skip:27
> fi-ivb-3770      total:248  pass:207  dwarn:0   dfail:0   fail:0   skip:41
> fi-kbl-7200u     total:248  pass:222  dwarn:0   dfail:0   fail:0   skip:26
> fi-skl-6260u     total:248  pass:232  dwarn:0   dfail:0   fail:0   skip:16
> fi-skl-6700hq    total:248  pass:223  dwarn:1   dfail:0   fail:0   skip:24
> fi-skl-6700k     total:248  pass:222  dwarn:1   dfail:0   fail:0   skip:25
> fi-skl-6770hq    total:248  pass:231  dwarn:1   dfail:0   fail:1   skip:15
> fi-snb-2520m     total:248  pass:211  dwarn:0   dfail:0   fail:0   skip:37
> fi-snb-2600      total:248  pass:209  dwarn:0   dfail:0   fail:0   skip:39
>
> Results at /archive/results/CI_IGT_test/Patchwork_2652/
>
> f35ed31aea66b3230c366fcba5f3456ae2cb956e drm-intel-nightly: 2016y-10m-10d-11h-28m-51s UTC integration manifest
> 401facf drm/i915: Allocate intel_engine_cs structure only for the enabled engines
>

Tvrtko Ursulin Oct. 13, 2016, 4:28 p.m. UTC | #4

On 13/10/2016 17:10, Goel, Akash wrote:
>
>
> On 10/10/2016 6:03 PM, Patchwork wrote:
>> == Series Details ==
>>
>> Series: drm/i915: Allocate intel_engine_cs structure only for the 
>> enabled engines (rev3)
>> URL   : https://patchwork.freedesktop.org/series/13435/
>> State : warning
>>
>> == Summary ==
>>
>> Series 13435v3 drm/i915: Allocate intel_engine_cs structure only for 
>> the enabled engines
>> https://patchwork.freedesktop.org/api/1.0/series/13435/revisions/3/mbox/
>>
>> Test vgem_basic:
>>         Subgroup unload:
>>                 pass       -> SKIP       (fi-skl-6260u)
>>                 pass       -> SKIP       (fi-skl-6700hq)
>>                 skip       -> PASS       (fi-skl-6700k)
>>
> Checked with Chris about the above failure.
> He said that the above unload failure for vgem module can't be
> attributed to the patch, most likely a CI framework issue.
>

I tried to merge it but due recent changes to error capture it doesn't 
apply any longer. Can you please rebase and resend?

Regards,

Tvrtko

Saarinen, Jani Oct. 13, 2016, 6:47 p.m. UTC | #5

> > == Series Details ==

> >

> > Series: drm/i915: Allocate intel_engine_cs structure only for the enabled

> engines (rev3)

> > URL   : https://patchwork.freedesktop.org/series/13435/

> > State : warning

> >

> > == Summary ==

> >

> > Series 13435v3 drm/i915: Allocate intel_engine_cs structure only for

> > the enabled engines

> > https://patchwork.freedesktop.org/api/1.0/series/13435/revisions/3/mbo

> > x/

> >

> > Test vgem_basic:

> >         Subgroup unload:

> >                 pass       -> SKIP       (fi-skl-6260u)

> >                 pass       -> SKIP       (fi-skl-6700hq)

> >                 skip       -> PASS       (fi-skl-6700k)

> >

> Checked with Chris about the above failure.

> He said that the above unload failure for vgem module can't be attributed to

> the patch, most likely a CI framework issue.

Yes, this test is still behaving badly especially with SKL systems but also with bdw and kbl.

> 

> Best regards

> Akash

> 

> > fi-bdw-5557u     total:248  pass:231  dwarn:0   dfail:0   fail:0   skip:17

> > fi-bsw-n3050     total:248  pass:204  dwarn:0   dfail:0   fail:0   skip:44

> > fi-bxt-t5700     total:248  pass:217  dwarn:0   dfail:0   fail:0   skip:31

> > fi-byt-j1900     total:248  pass:214  dwarn:1   dfail:0   fail:1   skip:32

> > fi-byt-n2820     total:248  pass:210  dwarn:0   dfail:0   fail:1   skip:37

> > fi-hsw-4770      total:248  pass:224  dwarn:0   dfail:0   fail:0   skip:24

> > fi-hsw-4770r     total:248  pass:224  dwarn:0   dfail:0   fail:0   skip:24

> > fi-ilk-650       total:248  pass:185  dwarn:0   dfail:0   fail:2   skip:61

> > fi-ivb-3520m     total:248  pass:221  dwarn:0   dfail:0   fail:0   skip:27

> > fi-ivb-3770      total:248  pass:207  dwarn:0   dfail:0   fail:0   skip:41

> > fi-kbl-7200u     total:248  pass:222  dwarn:0   dfail:0   fail:0   skip:26

> > fi-skl-6260u     total:248  pass:232  dwarn:0   dfail:0   fail:0   skip:16

> > fi-skl-6700hq    total:248  pass:223  dwarn:1   dfail:0   fail:0   skip:24

> > fi-skl-6700k     total:248  pass:222  dwarn:1   dfail:0   fail:0   skip:25

> > fi-skl-6770hq    total:248  pass:231  dwarn:1   dfail:0   fail:1   skip:15

> > fi-snb-2520m     total:248  pass:211  dwarn:0   dfail:0   fail:0   skip:37

> > fi-snb-2600      total:248  pass:209  dwarn:0   dfail:0   fail:0   skip:39

> >

> > Results at /archive/results/CI_IGT_test/Patchwork_2652/

> >

> > f35ed31aea66b3230c366fcba5f3456ae2cb956e drm-intel-nightly:

> > 2016y-10m-10d-11h-28m-51s UTC integration manifest 401facf drm/i915:

> > Allocate intel_engine_cs structure only for the enabled engines

> >


Jani Saarinen

Chris Wilson Oct. 13, 2016, 7:57 p.m. UTC | #6

On Thu, Oct 13, 2016 at 06:47:24PM +0000, Saarinen, Jani wrote:
> > > == Series Details ==
> > >
> > > Series: drm/i915: Allocate intel_engine_cs structure only for the enabled
> > engines (rev3)
> > > URL   : https://patchwork.freedesktop.org/series/13435/
> > > State : warning
> > >
> > > == Summary ==
> > >
> > > Series 13435v3 drm/i915: Allocate intel_engine_cs structure only for
> > > the enabled engines
> > > https://patchwork.freedesktop.org/api/1.0/series/13435/revisions/3/mbo
> > > x/
> > >
> > > Test vgem_basic:
> > >         Subgroup unload:
> > >                 pass       -> SKIP       (fi-skl-6260u)
> > >                 pass       -> SKIP       (fi-skl-6700hq)
> > >                 skip       -> PASS       (fi-skl-6700k)
> > >
> > Checked with Chris about the above failure.
> > He said that the above unload failure for vgem module can't be attributed to
> > the patch, most likely a CI framework issue.
> Yes, this test is still behaving badly especially with SKL systems but also with bdw and kbl.

Are the test run in the order defined by fast-feedback.testlist ?
I intended the vgem unload test to be run as the first vgem testcase to
minimise the chance of a stray module leak. Can we define the order
within CI? Can we put comments into fast-feedback.testlist ?
-Chris

Saarinen, Jani Oct. 14, 2016, 6:03 a.m. UTC | #7

> > > > == Series Details ==

> > > >

> > > > Series: drm/i915: Allocate intel_engine_cs structure only for the

> > > > enabled

> > > engines (rev3)

> > > > URL   : https://patchwork.freedesktop.org/series/13435/

> > > > State : warning

> > > >

> > > > == Summary ==

> > > >

> > > > Series 13435v3 drm/i915: Allocate intel_engine_cs structure only

> > > > for the enabled engines

> > > > https://patchwork.freedesktop.org/api/1.0/series/13435/revisions/3

> > > > /mbo

> > > > x/

> > > >

> > > > Test vgem_basic:

> > > >         Subgroup unload:

> > > >                 pass       -> SKIP       (fi-skl-6260u)

> > > >                 pass       -> SKIP       (fi-skl-6700hq)

> > > >                 skip       -> PASS       (fi-skl-6700k)

> > > >

> > > Checked with Chris about the above failure.

> > > He said that the above unload failure for vgem module can't be

> > > attributed to the patch, most likely a CI framework issue.

> > Yes, this test is still behaving badly especially with SKL systems but also with

> bdw and kbl.

> 

> Are the test run in the order defined by fast-feedback.testlist ?

> I intended the vgem unload test to be run as the first vgem testcase to 

> minimise the chance of a stray module leak. Can we define the order within

> CI? Can we put comments into fast-feedback.testlist ?

My understanding, yes, we are running on that order.
Adding comment, no I think no, Petri, Tomi?

> -Chris

> 

> --

> Chris Wilson, Intel Open Source Technology Centre




Jani Saarinen
Intel Finland Oy - BIC 0357606-4 - Westendinkatu 7, 02160 Espoo

[e-mail] Oct. 14, 2016, 9:02 a.m. UTC | #8

On 10/13/2016 10:50 PM, Patchwork wrote:
> == Series Details ==
>
> Series: drm/i915: Allocate intel_engine_cs structure only for the enabled engines (rev4)
> URL   : https://patchwork.freedesktop.org/series/13435/
> State : warning
>
> == Summary ==
>
> Series 13435v4 drm/i915: Allocate intel_engine_cs structure only for the enabled engines
> https://patchwork.freedesktop.org/api/1.0/series/13435/revisions/4/mbox/
>
> Test kms_pipe_crc_basic:
>         Subgroup nonblocking-crc-pipe-b:
>                 pass       -> DMESG-WARN (fi-ilk-650)

Have filed a new BZ: https://bugs.freedesktop.org/show_bug.cgi?id=98251

Most likely the above failure isn't related with the concerned patch.

Best regards
Akash

 >         Subgroup read-crc-pipe-b-frame-sequence:
 >                 dmesg-warn -> PASS       (fi-ilk-650)
 > Test vgem_basic:
 >         Subgroup unload:
 >                 skip       -> PASS       (fi-skl-6770hq)
 >
> fi-bdw-5557u     total:246  pass:231  dwarn:0   dfail:0   fail:0   skip:15
> fi-bsw-n3050     total:246  pass:204  dwarn:0   dfail:0   fail:0   skip:42
> fi-bxt-t5700     total:246  pass:216  dwarn:0   dfail:0   fail:0   skip:30
> fi-byt-j1900     total:246  pass:212  dwarn:2   dfail:0   fail:1   skip:31
> fi-byt-n2820     total:246  pass:210  dwarn:0   dfail:0   fail:1   skip:35
> fi-hsw-4770      total:246  pass:223  dwarn:0   dfail:0   fail:0   skip:23
> fi-hsw-4770r     total:246  pass:224  dwarn:0   dfail:0   fail:0   skip:22
> fi-ilk-650       total:246  pass:183  dwarn:1   dfail:0   fail:2   skip:60
> fi-ivb-3520m     total:246  pass:221  dwarn:0   dfail:0   fail:0   skip:25
> fi-ivb-3770      total:246  pass:221  dwarn:0   dfail:0   fail:0   skip:25
> fi-kbl-7200u     total:246  pass:222  dwarn:0   dfail:0   fail:0   skip:24
> fi-skl-6260u     total:246  pass:232  dwarn:0   dfail:0   fail:0   skip:14
> fi-skl-6700hq    total:246  pass:223  dwarn:0   dfail:0   fail:0   skip:23
> fi-skl-6700k     total:246  pass:221  dwarn:1   dfail:0   fail:0   skip:24
> fi-skl-6770hq    total:246  pass:229  dwarn:1   dfail:0   fail:1   skip:15
> fi-snb-2520m     total:246  pass:210  dwarn:0   dfail:0   fail:0   skip:36
> fi-snb-2600      total:246  pass:209  dwarn:0   dfail:0   fail:0   skip:37
>
> Results at /archive/results/CI_IGT_test/Patchwork_2708/
>
> dbcf6fbb541e70fac7db669631958eab2e4e0d9c drm-intel-nightly: 2016y-10m-13d-15h-31m-19s UTC integration manifest
> 391ff6c drm/i915: Allocate intel_engine_cs structure only for the enabled engines
>

Tvrtko Ursulin Oct. 14, 2016, 9:04 a.m. UTC | #9

On 14/10/2016 10:02, Goel, Akash wrote:
>
>
> On 10/13/2016 10:50 PM, Patchwork wrote:
>> == Series Details ==
>>
>> Series: drm/i915: Allocate intel_engine_cs structure only for the 
>> enabled engines (rev4)
>> URL   : https://patchwork.freedesktop.org/series/13435/
>> State : warning
>>
>> == Summary ==
>>
>> Series 13435v4 drm/i915: Allocate intel_engine_cs structure only for 
>> the enabled engines
>> https://patchwork.freedesktop.org/api/1.0/series/13435/revisions/4/mbox/
>>
>> Test kms_pipe_crc_basic:
>>         Subgroup nonblocking-crc-pipe-b:
>>                 pass       -> DMESG-WARN (fi-ilk-650)
>
> Have filed a new BZ: https://bugs.freedesktop.org/show_bug.cgi?id=98251
>
> Most likely the above failure isn't related with the concerned patch.
>

Agreed.

In the future also please put a fail log snippet here when replying and 
a bug title.

> Best regards
> Akash
>
> >         Subgroup read-crc-pipe-b-frame-sequence:
> >                 dmesg-warn -> PASS       (fi-ilk-650)
> > Test vgem_basic:
> >         Subgroup unload:
> >                 skip       -> PASS       (fi-skl-6770hq)
> >
>> fi-bdw-5557u     total:246  pass:231 dwarn:0   dfail:0   fail:0   
>> skip:15
>> fi-bsw-n3050     total:246  pass:204  dwarn:0   dfail:0 fail:0   skip:42
>> fi-bxt-t5700     total:246  pass:216  dwarn:0   dfail:0 fail:0   skip:30
>> fi-byt-j1900     total:246  pass:212  dwarn:2   dfail:0 fail:1   skip:31
>> fi-byt-n2820     total:246  pass:210  dwarn:0   dfail:0 fail:1   skip:35
>> fi-hsw-4770      total:246  pass:223  dwarn:0   dfail:0 fail:0   skip:23
>> fi-hsw-4770r     total:246  pass:224  dwarn:0   dfail:0 fail:0   skip:22
>> fi-ilk-650       total:246  pass:183  dwarn:1   dfail:0 fail:2   skip:60
>> fi-ivb-3520m     total:246  pass:221  dwarn:0   dfail:0 fail:0   skip:25
>> fi-ivb-3770      total:246  pass:221  dwarn:0   dfail:0 fail:0   skip:25
>> fi-kbl-7200u     total:246  pass:222  dwarn:0   dfail:0 fail:0   skip:24
>> fi-skl-6260u     total:246  pass:232  dwarn:0   dfail:0 fail:0   skip:14
>> fi-skl-6700hq    total:246  pass:223  dwarn:0   dfail:0 fail:0   skip:23
>> fi-skl-6700k     total:246  pass:221  dwarn:1   dfail:0 fail:0   skip:24
>> fi-skl-6770hq    total:246  pass:229  dwarn:1   dfail:0 fail:1   skip:15
>> fi-snb-2520m     total:246  pass:210  dwarn:0   dfail:0 fail:0   skip:36
>> fi-snb-2600      total:246  pass:209  dwarn:0   dfail:0 fail:0   skip:37
>>
>> Results at /archive/results/CI_IGT_test/Patchwork_2708/
>>
>> dbcf6fbb541e70fac7db669631958eab2e4e0d9c drm-intel-nightly: 
>> 2016y-10m-13d-15h-31m-19s UTC integration manifest
>> 391ff6c drm/i915: Allocate intel_engine_cs structure only for the 
>> enabled engines
>>

Merged to dinq, thanks for the patch and review!

Regards,

Tvrtko

Petri Latvala Oct. 14, 2016, 10:37 a.m. UTC | #10

>> Are the test run in the order defined by fast-feedback.testlist ?
>> I intended the vgem unload test to be run as the first vgem testcase to
>> minimise the chance of a stray module leak. Can we define the order within
>> CI? Can we put comments into fast-feedback.testlist ?
> My understanding, yes, we are running on that order.
> Adding comment, no I think no, Petri, Tomi?
>

Order: Yes, that order. I'm waiting for an opportune moment to test a 
patch to sleep-and-retry in vgem unload to prune out the cases of just 
having soon-to-finish work left over. Just need to tune the amount of 
retries and sleeping.

We can also order vgem unload test to be first (along with 
drv_module_reload) depending on whether we want to catch those stray 
module leaks. Comments on that tradeoff?

Comments in the test list: Not yet. Piglit patch for them just sent to 
piglit mailing list.



Petri Latvala

Chris Wilson Oct. 14, 2016, 11:01 a.m. UTC | #11

On Fri, Oct 14, 2016 at 01:37:46PM +0300, Petri Latvala wrote:
> 
> >>Are the test run in the order defined by fast-feedback.testlist ?
> >>I intended the vgem unload test to be run as the first vgem testcase to
> >>minimise the chance of a stray module leak. Can we define the order within
> >>CI? Can we put comments into fast-feedback.testlist ?
> >My understanding, yes, we are running on that order.
> >Adding comment, no I think no, Petri, Tomi?
> >
> 
> Order: Yes, that order. I'm waiting for an opportune moment to test
> a patch to sleep-and-retry in vgem unload to prune out the cases of
> just having soon-to-finish work left over. Just need to tune the
> amount of retries and sleeping.
> 
> We can also order vgem unload test to be first (along with
> drv_module_reload) depending on whether we want to catch those stray
> module leaks. Comments on that tradeoff?

I planned on having vgem/unload be the first vgem test. It probably
makes sense to have a second one at the end to catch a leak over the
run.
-Chris

Patch

diff --git a/drivers/gpu/drm/i915/i915_cmd_parser.c b/drivers/gpu/drm/i915/i915_cmd_parser.c
index 70980f8..bbcd2ec 100644
--- a/drivers/gpu/drm/i915/i915_cmd_parser.c
+++ b/drivers/gpu/drm/i915/i915_cmd_parser.c
@@ -1308,10 +1308,11 @@  int intel_engine_cmd_parser(struct intel_engine_cs *engine,
 int i915_cmd_parser_get_version(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	bool active = false;
 
 	/* If the command parser is not enabled, report 0 - unsupported */
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		if (intel_engine_needs_cmd_parser(engine)) {
 			active = true;
 			break;
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 20689f1..f9fd949 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -150,7 +150,7 @@  describe_obj(struct seq_file *m, struct drm_i915_gem_object *obj)
 		   obj->base.size / 1024,
 		   obj->base.read_domains,
 		   obj->base.write_domain);
-	for_each_engine_id(engine, dev_priv, id)
+	for_each_engine(engine, dev_priv, id)
 		seq_printf(m, "%x ",
 			   i915_gem_active_get_seqno(&obj->last_read[id],
 						     &obj->base.dev->struct_mutex));
@@ -332,11 +332,12 @@  static void print_batch_pool_stats(struct seq_file *m,
 	struct drm_i915_gem_object *obj;
 	struct file_stats stats;
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int j;
 
 	memset(&stats, 0, sizeof(stats));
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		for (j = 0; j < ARRAY_SIZE(engine->batch_pool.cache_list); j++) {
 			list_for_each_entry(obj,
 					    &engine->batch_pool.cache_list[j],
@@ -605,6 +606,7 @@  static int i915_gem_batch_pool_info(struct seq_file *m, void *data)
 	struct drm_device *dev = &dev_priv->drm;
 	struct drm_i915_gem_object *obj;
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int total = 0;
 	int ret, j;
 
@@ -612,7 +614,7 @@  static int i915_gem_batch_pool_info(struct seq_file *m, void *data)
 	if (ret)
 		return ret;
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		for (j = 0; j < ARRAY_SIZE(engine->batch_pool.cache_list); j++) {
 			int count;
 
@@ -664,8 +666,9 @@  static int i915_gem_request_info(struct seq_file *m, void *data)
 {
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
 	struct drm_device *dev = &dev_priv->drm;
-	struct intel_engine_cs *engine;
 	struct drm_i915_gem_request *req;
+	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int ret, any;
 
 	ret = mutex_lock_interruptible(&dev->struct_mutex);
@@ -673,7 +676,7 @@  static int i915_gem_request_info(struct seq_file *m, void *data)
 		return ret;
 
 	any = 0;
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		int count;
 
 		count = 0;
@@ -719,8 +722,9 @@  static int i915_gem_seqno_info(struct seq_file *m, void *data)
 {
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		i915_ring_seqno_info(m, engine);
 
 	return 0;
@@ -731,6 +735,7 @@  static int i915_interrupt_info(struct seq_file *m, void *data)
 {
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int i, pipe;
 
 	intel_runtime_pm_get(dev_priv);
@@ -899,7 +904,7 @@  static int i915_interrupt_info(struct seq_file *m, void *data)
 		seq_printf(m, "Graphics Interrupt mask:		%08x\n",
 			   I915_READ(GTIMR));
 	}
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		if (INTEL_GEN(dev_priv) >= 6) {
 			seq_printf(m,
 				   "Graphics Interrupt mask (%s):	%08x\n",
@@ -947,7 +952,7 @@  static int i915_hws_info(struct seq_file *m, void *data)
 	const u32 *hws;
 	int i;
 
-	engine = &dev_priv->engine[(uintptr_t)node->info_ent->data];
+	engine = dev_priv->engine[(uintptr_t)node->info_ent->data];
 	hws = engine->status_page.page_addr;
 	if (hws == NULL)
 		return 0;
@@ -1334,7 +1339,7 @@  static int i915_hangcheck_info(struct seq_file *m, void *unused)
 
 	intel_runtime_pm_get(dev_priv);
 
-	for_each_engine_id(engine, dev_priv, id) {
+	for_each_engine(engine, dev_priv, id) {
 		acthd[id] = intel_engine_get_active_head(engine);
 		seqno[id] = intel_engine_get_seqno(engine);
 	}
@@ -1350,7 +1355,7 @@  static int i915_hangcheck_info(struct seq_file *m, void *unused)
 	} else
 		seq_printf(m, "Hangcheck inactive\n");
 
-	for_each_engine_id(engine, dev_priv, id) {
+	for_each_engine(engine, dev_priv, id) {
 		struct intel_breadcrumbs *b = &engine->breadcrumbs;
 		struct rb_node *rb;
 
@@ -1949,6 +1954,7 @@  static int i915_context_status(struct seq_file *m, void *unused)
 	struct drm_device *dev = &dev_priv->drm;
 	struct intel_engine_cs *engine;
 	struct i915_gem_context *ctx;
+	enum intel_engine_id iter;
 	int ret;
 
 	ret = mutex_lock_interruptible(&dev->struct_mutex);
@@ -1975,7 +1981,7 @@  static int i915_context_status(struct seq_file *m, void *unused)
 		seq_putc(m, ctx->remap_slice ? 'R' : 'r');
 		seq_putc(m, '\n');
 
-		for_each_engine(engine, dev_priv) {
+		for_each_engine(engine, dev_priv, iter) {
 			struct intel_context *ce = &ctx->engine[engine->id];
 
 			seq_printf(m, "%s: ", engine->name);
@@ -2042,6 +2048,7 @@  static int i915_dump_lrc(struct seq_file *m, void *unused)
 	struct drm_device *dev = &dev_priv->drm;
 	struct intel_engine_cs *engine;
 	struct i915_gem_context *ctx;
+	enum intel_engine_id iter;
 	int ret;
 
 	if (!i915.enable_execlists) {
@@ -2054,7 +2061,7 @@  static int i915_dump_lrc(struct seq_file *m, void *unused)
 		return ret;
 
 	list_for_each_entry(ctx, &dev_priv->context_list, link)
-		for_each_engine(engine, dev_priv)
+		for_each_engine(engine, dev_priv, iter)
 			i915_dump_lrc_obj(m, ctx, engine);
 
 	mutex_unlock(&dev->struct_mutex);
@@ -2163,14 +2170,15 @@  static int per_file_ctx(int id, void *ptr, void *data)
 static void gen8_ppgtt_info(struct seq_file *m,
 			    struct drm_i915_private *dev_priv)
 {
-	struct intel_engine_cs *engine;
 	struct i915_hw_ppgtt *ppgtt = dev_priv->mm.aliasing_ppgtt;
+	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int i;
 
 	if (!ppgtt)
 		return;
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		seq_printf(m, "%s\n", engine->name);
 		for (i = 0; i < 4; i++) {
 			u64 pdp = I915_READ(GEN8_RING_PDP_UDW(engine, i));
@@ -2185,11 +2193,12 @@  static void gen6_ppgtt_info(struct seq_file *m,
 			    struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
 	if (IS_GEN6(dev_priv))
 		seq_printf(m, "GFX_MODE: 0x%08x\n", I915_READ(GFX_MODE));
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		seq_printf(m, "%s\n", engine->name);
 		if (IS_GEN7(dev_priv))
 			seq_printf(m, "GFX_MODE: 0x%08x\n",
@@ -2258,9 +2267,10 @@  out_unlock:
 static int count_irq_waiters(struct drm_i915_private *i915)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int count = 0;
 
-	for_each_engine(engine, i915)
+	for_each_engine(engine, i915, iter)
 		count += intel_engine_has_waiter(engine);
 
 	return count;
@@ -2423,7 +2433,7 @@  static void i915_guc_client_info(struct seq_file *m,
 	seq_printf(m, "\tFailed doorbell: %u\n", client->b_fail);
 	seq_printf(m, "\tLast submission result: %d\n", client->retcode);
 
-	for_each_engine_id(engine, dev_priv, id) {
+	for_each_engine(engine, dev_priv, id) {
 		u64 submissions = client->submissions[id];
 		tot += submissions;
 		seq_printf(m, "\tSubmissions: %llu %s\n",
@@ -2466,7 +2476,7 @@  static int i915_guc_info(struct seq_file *m, void *data)
 	seq_printf(m, "GuC last action error code: %d\n", guc.action_err);
 
 	seq_printf(m, "\nGuC submissions:\n");
-	for_each_engine_id(engine, dev_priv, id) {
+	for_each_engine(engine, dev_priv, id) {
 		u64 submissions = guc.submissions[id];
 		total += submissions;
 		seq_printf(m, "\t%-24s: %10llu, last seqno 0x%08x\n",
@@ -3087,8 +3097,9 @@  static int i915_engine_info(struct seq_file *m, void *unused)
 {
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		struct intel_breadcrumbs *b = &engine->breadcrumbs;
 		struct drm_i915_gem_request *rq;
 		struct rb_node *rb;
@@ -3236,7 +3247,7 @@  static int i915_semaphore_status(struct seq_file *m, void *unused)
 		page = i915_gem_object_get_page(dev_priv->semaphore->obj, 0);
 
 		seqno = (uint64_t *)kmap_atomic(page);
-		for_each_engine_id(engine, dev_priv, id) {
+		for_each_engine(engine, dev_priv, id) {
 			uint64_t offset;
 
 			seq_printf(m, "%s\n", engine->name);
@@ -3261,7 +3272,7 @@  static int i915_semaphore_status(struct seq_file *m, void *unused)
 		kunmap_atomic(seqno);
 	} else {
 		seq_puts(m, "  Last signal:");
-		for_each_engine(engine, dev_priv)
+		for_each_engine(engine, dev_priv, id)
 			for (j = 0; j < num_rings; j++)
 				seq_printf(m, "0x%08x\n",
 					   I915_READ(engine->semaphore.mbox.signal[j]));
@@ -3269,7 +3280,7 @@  static int i915_semaphore_status(struct seq_file *m, void *unused)
 	}
 
 	seq_puts(m, "\nSync seqno:\n");
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, id) {
 		for (j = 0; j < num_rings; j++)
 			seq_printf(m, "  0x%08x ",
 				   engine->semaphore.sync_seqno[j]);
@@ -3325,7 +3336,7 @@  static int i915_wa_registers(struct seq_file *m, void *unused)
 	intel_runtime_pm_get(dev_priv);
 
 	seq_printf(m, "Workarounds applied: %d\n", workarounds->count);
-	for_each_engine_id(engine, dev_priv, id)
+	for_each_engine(engine, dev_priv, id)
 		seq_printf(m, "HW whitelist count for %s: %d\n",
 			   engine->name, workarounds->hw_whitelist_count[id]);
 	for (i = 0; i < workarounds->count; ++i) {
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 89d3222..5788593 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -255,16 +255,16 @@  static int i915_getparam(struct drm_device *dev, void *data,
 		value = dev_priv->overlay ? 1 : 0;
 		break;
 	case I915_PARAM_HAS_BSD:
-		value = intel_engine_initialized(&dev_priv->engine[VCS]);
+		value = !!dev_priv->engine[VCS];
 		break;
 	case I915_PARAM_HAS_BLT:
-		value = intel_engine_initialized(&dev_priv->engine[BCS]);
+		value = !!dev_priv->engine[BCS];
 		break;
 	case I915_PARAM_HAS_VEBOX:
-		value = intel_engine_initialized(&dev_priv->engine[VECS]);
+		value = !!dev_priv->engine[VECS];
 		break;
 	case I915_PARAM_HAS_BSD2:
-		value = intel_engine_initialized(&dev_priv->engine[VCS2]);
+		value = !!dev_priv->engine[VCS2];
 		break;
 	case I915_PARAM_HAS_EXEC_CONSTANTS:
 		value = INTEL_GEN(dev_priv) >= 4;
@@ -1731,10 +1731,11 @@  int i915_resume_switcheroo(struct drm_device *dev)
 static void disable_engines_irq(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
 	/* Ensure irq handler finishes, and not run again. */
 	disable_irq(dev_priv->drm.irq);
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		tasklet_kill(&engine->irq_tasklet);
 }
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index a219a35..3efd5a6 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1804,7 +1804,7 @@  struct drm_i915_private {
 
 	struct pci_dev *bridge_dev;
 	struct i915_gem_context *kernel_context;
-	struct intel_engine_cs engine[I915_NUM_ENGINES];
+	struct intel_engine_cs *engine[I915_NUM_ENGINES];
 	struct i915_vma *semaphore;
 	u32 next_seqno;
 
@@ -2121,19 +2121,11 @@  static inline struct drm_i915_private *guc_to_i915(struct intel_guc *guc)
 }
 
 /* Simple iterator over all initialised engines */
-#define for_each_engine(engine__, dev_priv__) \
-	for ((engine__) = &(dev_priv__)->engine[0]; \
-	     (engine__) < &(dev_priv__)->engine[I915_NUM_ENGINES]; \
-	     (engine__)++) \
-		for_each_if (intel_engine_initialized(engine__))
-
-/* Iterator with engine_id */
-#define for_each_engine_id(engine__, dev_priv__, id__) \
-	for ((engine__) = &(dev_priv__)->engine[0], (id__) = 0; \
-	     (engine__) < &(dev_priv__)->engine[I915_NUM_ENGINES]; \
-	     (engine__)++) \
-		for_each_if (((id__) = (engine__)->id, \
-			      intel_engine_initialized(engine__)))
+#define for_each_engine(engine__, dev_priv__, iter__) \
+	for ((iter__) = 0; \
+	     (iter__) < I915_NUM_ENGINES; \
+	     (iter__)++) \
+		for_each_if ((engine__) = (dev_priv__)->engine[(iter__)])
 
 #define __mask_next_bit(mask) ({					\
 	int __idx = ffs(mask) - 1;					\
@@ -2144,7 +2136,7 @@  static inline struct drm_i915_private *guc_to_i915(struct intel_guc *guc)
 /* Iterator over subset of engines selected by mask */
 #define for_each_engine_masked(engine__, dev_priv__, mask__, tmp__) \
 	for (tmp__ = mask__ & INTEL_INFO(dev_priv__)->ring_mask;	\
-	     tmp__ ? (engine__ = &(dev_priv__)->engine[__mask_next_bit(tmp__)]), 1 : 0; )
+	     tmp__ ? (engine__ = (dev_priv__)->engine[__mask_next_bit(tmp__)]), 1 : 0; )
 
 enum hdmi_force_audio {
 	HDMI_AUDIO_OFF_DVI = -2,	/* no aux data for HDMI-DVI converter */
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index a89a889..adc7f09 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2622,10 +2622,11 @@  static void i915_gem_reset_engine(struct intel_engine_cs *engine)
 void i915_gem_reset(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
 	i915_gem_retire_requests(dev_priv);
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		i915_gem_reset_engine(engine);
 
 	i915_gem_restore_fences(&dev_priv->drm);
@@ -2673,12 +2674,13 @@  static void i915_gem_cleanup_engine(struct intel_engine_cs *engine)
 void i915_gem_set_wedged(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
 	lockdep_assert_held(&dev_priv->drm.struct_mutex);
 	set_bit(I915_WEDGED, &dev_priv->gpu_error.flags);
 
 	i915_gem_context_lost(dev_priv);
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		i915_gem_cleanup_engine(engine);
 	mod_delayed_work(dev_priv->wq, &dev_priv->gt.idle_work, 0);
 
@@ -2717,6 +2719,7 @@  i915_gem_idle_work_handler(struct work_struct *work)
 		container_of(work, typeof(*dev_priv), gt.idle_work.work);
 	struct drm_device *dev = &dev_priv->drm;
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	bool rearm_hangcheck;
 
 	if (!READ_ONCE(dev_priv->gt.awake))
@@ -2739,7 +2742,7 @@  i915_gem_idle_work_handler(struct work_struct *work)
 	if (dev_priv->gt.active_engines)
 		goto out_unlock;
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		i915_gem_batch_pool_fini(&engine->batch_pool);
 
 	GEM_BUG_ON(!dev_priv->gt.awake);
@@ -2932,9 +2935,10 @@  int i915_gem_wait_for_idle(struct drm_i915_private *dev_priv,
 			   unsigned int flags)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int ret;
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		if (engine->last_context == NULL)
 			continue;
 
@@ -3174,7 +3178,7 @@  i915_gem_object_flush_gtt_write_domain(struct drm_i915_gem_object *obj)
 	 */
 	wmb();
 	if (INTEL_GEN(dev_priv) >= 6 && !HAS_LLC(dev_priv))
-		POSTING_READ(RING_ACTHD(dev_priv->engine[RCS].mmio_base));
+		POSTING_READ(RING_ACTHD(dev_priv->engine[RCS]->mmio_base));
 
 	intel_fb_obj_flush(obj, false, write_origin(obj, I915_GEM_DOMAIN_GTT));
 
@@ -4349,6 +4353,7 @@  i915_gem_init_hw(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int ret;
 
 	/* Double layer security blanket, see i915_gem_init() */
@@ -4392,7 +4397,7 @@  i915_gem_init_hw(struct drm_device *dev)
 	}
 
 	/* Need to do basic initialisation of all rings first: */
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		ret = engine->init_hw(engine);
 		if (ret)
 			goto out;
@@ -4491,17 +4496,12 @@  i915_gem_cleanup_engines(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		dev_priv->gt.cleanup_engine(engine);
 }
 
-static void
-init_engine_lists(struct intel_engine_cs *engine)
-{
-	INIT_LIST_HEAD(&engine->request_list);
-}
-
 void
 i915_gem_load_init_fences(struct drm_i915_private *dev_priv)
 {
@@ -4538,7 +4538,6 @@  void
 i915_gem_load_init(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = to_i915(dev);
-	int i;
 
 	dev_priv->objects =
 		kmem_cache_create("i915_gem_object",
@@ -4562,8 +4561,6 @@  i915_gem_load_init(struct drm_device *dev)
 	INIT_LIST_HEAD(&dev_priv->mm.unbound_list);
 	INIT_LIST_HEAD(&dev_priv->mm.bound_list);
 	INIT_LIST_HEAD(&dev_priv->mm.fence_list);
-	for (i = 0; i < I915_NUM_ENGINES; i++)
-		init_engine_lists(&dev_priv->engine[i]);
 	INIT_DELAYED_WORK(&dev_priv->gt.retire_work,
 			  i915_gem_retire_work_handler);
 	INIT_DELAYED_WORK(&dev_priv->gt.idle_work,
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index df10f4e9..3641955 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -474,10 +474,11 @@  int i915_gem_context_init(struct drm_device *dev)
 void i915_gem_context_lost(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
 	lockdep_assert_held(&dev_priv->drm.struct_mutex);
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		if (engine->last_context) {
 			i915_gem_context_unpin(engine->last_context, engine);
 			engine->last_context = NULL;
@@ -492,13 +493,13 @@  void i915_gem_context_lost(struct drm_i915_private *dev_priv)
 			if (!i915_gem_context_is_default(ctx))
 				continue;
 
-			for_each_engine(engine, dev_priv)
+			for_each_engine(engine, dev_priv, iter)
 				ctx->engine[engine->id].initialised = false;
 
 			ctx->remap_slice = ALL_L3_SLICES(dev_priv);
 		}
 
-		for_each_engine(engine, dev_priv) {
+		for_each_engine(engine, dev_priv, iter) {
 			struct intel_context *kce =
 				&dev_priv->kernel_context->engine[engine->id];
 
@@ -563,6 +564,7 @@  mi_set_context(struct drm_i915_gem_request *req, u32 hw_flags)
 	struct drm_i915_private *dev_priv = req->i915;
 	struct intel_ring *ring = req->ring;
 	struct intel_engine_cs *engine = req->engine;
+	enum intel_engine_id iter;
 	u32 flags = hw_flags | MI_MM_SPACE_GTT;
 	const int num_rings =
 		/* Use an extended w/a on ivb+ if signalling from other rings */
@@ -605,7 +607,7 @@  mi_set_context(struct drm_i915_gem_request *req, u32 hw_flags)
 
 			intel_ring_emit(ring,
 					MI_LOAD_REGISTER_IMM(num_rings));
-			for_each_engine(signaller, dev_priv) {
+			for_each_engine(signaller, dev_priv, iter) {
 				if (signaller == engine)
 					continue;
 
@@ -634,7 +636,7 @@  mi_set_context(struct drm_i915_gem_request *req, u32 hw_flags)
 
 			intel_ring_emit(ring,
 					MI_LOAD_REGISTER_IMM(num_rings));
-			for_each_engine(signaller, dev_priv) {
+			for_each_engine(signaller, dev_priv, iter) {
 				if (signaller == engine)
 					continue;
 
@@ -929,8 +931,9 @@  int i915_switch_context(struct drm_i915_gem_request *req)
 int i915_gem_switch_to_kernel_context(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		struct drm_i915_gem_request *req;
 		int ret;
 
diff --git a/drivers/gpu/drm/i915/i915_gem_evict.c b/drivers/gpu/drm/i915/i915_gem_evict.c
index 5b6f81c..189b3be8 100644
--- a/drivers/gpu/drm/i915/i915_gem_evict.c
+++ b/drivers/gpu/drm/i915/i915_gem_evict.c
@@ -37,8 +37,9 @@  static bool
 gpu_is_idle(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		if (intel_engine_is_active(engine))
 			return false;
 	}
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index e88786e..b70a7ab 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -1599,12 +1599,12 @@  eb_select_engine(struct drm_i915_private *dev_priv,
 			return NULL;
 		}
 
-		engine = &dev_priv->engine[_VCS(bsd_idx)];
+		engine = dev_priv->engine[_VCS(bsd_idx)];
 	} else {
-		engine = &dev_priv->engine[user_ring_map[user_ring_id]];
+		engine = dev_priv->engine[user_ring_map[user_ring_id]];
 	}
 
-	if (!intel_engine_initialized(engine)) {
+	if (!engine) {
 		DRM_DEBUG("execbuf with invalid ring: %u\n", user_ring_id);
 		return NULL;
 	}
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index 0bb4232..488db38 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -1728,8 +1728,9 @@  static void gen8_ppgtt_enable(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		u32 four_level = USES_FULL_48BIT_PPGTT(dev) ? GEN8_GFX_PPGTT_48B : 0;
 		I915_WRITE(RING_MODE_GEN7(engine),
 			   _MASKED_BIT_ENABLE(GFX_PPGTT_ENABLE | four_level));
@@ -1740,6 +1741,7 @@  static void gen7_ppgtt_enable(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	uint32_t ecochk, ecobits;
 
 	ecobits = I915_READ(GAC_ECO_BITS);
@@ -1754,7 +1756,7 @@  static void gen7_ppgtt_enable(struct drm_device *dev)
 	}
 	I915_WRITE(GAM_ECOCHK, ecochk);
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		/* GFX_MODE is per-ring on gen7+ */
 		I915_WRITE(RING_MODE_GEN7(engine),
 			   _MASKED_BIT_ENABLE(GFX_PPGTT_ENABLE));
@@ -2239,11 +2241,12 @@  static bool needs_idle_maps(struct drm_i915_private *dev_priv)
 void i915_check_and_clear_faults(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
 	if (INTEL_INFO(dev_priv)->gen < 6)
 		return;
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		u32 fault_reg;
 		fault_reg = I915_READ(RING_FAULT_REG(engine));
 		if (fault_reg & RING_FAULT_VALID) {
@@ -2260,7 +2263,10 @@  void i915_check_and_clear_faults(struct drm_i915_private *dev_priv)
 				   fault_reg & ~RING_FAULT_VALID);
 		}
 	}
-	POSTING_READ(RING_FAULT_REG(&dev_priv->engine[RCS]));
+
+	/* Engine specific init may not have been done till this point. */
+	if (dev_priv->engine[RCS])
+		POSTING_READ(RING_FAULT_REG(dev_priv->engine[RCS]));
 }
 
 static void i915_ggtt_flush(struct drm_i915_private *dev_priv)
diff --git a/drivers/gpu/drm/i915/i915_gem_request.c b/drivers/gpu/drm/i915/i915_gem_request.c
index 40978bc..c51151e 100644
--- a/drivers/gpu/drm/i915/i915_gem_request.c
+++ b/drivers/gpu/drm/i915/i915_gem_request.c
@@ -256,10 +256,11 @@  static int i915_gem_check_wedge(struct drm_i915_private *dev_priv)
 static int i915_gem_init_seqno(struct drm_i915_private *dev_priv, u32 seqno)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int ret;
 
 	/* Carefully retire all requests without writing to the rings */
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		ret = intel_engine_idle(engine,
 					I915_WAIT_INTERRUPTIBLE |
 					I915_WAIT_LOCKED);
@@ -276,7 +277,7 @@  static int i915_gem_init_seqno(struct drm_i915_private *dev_priv, u32 seqno)
 	}
 
 	/* Finally reset hw state */
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		intel_engine_init_seqno(engine, seqno);
 
 	return 0;
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c
index b5b5869..8176b86 100644
--- a/drivers/gpu/drm/i915/i915_gpu_error.c
+++ b/drivers/gpu/drm/i915/i915_gpu_error.c
@@ -464,7 +464,7 @@  int i915_error_state_to_str(struct drm_i915_error_state_buf *m,
 
 			len += scnprintf(buf + len, sizeof(buf), "%s%s",
 					 first ? "" : ", ",
-					 dev_priv->engine[j].name);
+					 dev_priv->engine[j]->name);
 			first = 0;
 		}
 		scnprintf(buf + len, sizeof(buf), ")");
@@ -482,7 +482,7 @@  int i915_error_state_to_str(struct drm_i915_error_state_buf *m,
 
 		obj = ee->batchbuffer;
 		if (obj) {
-			err_puts(m, dev_priv->engine[i].name);
+			err_puts(m, dev_priv->engine[i]->name);
 			if (ee->pid != -1)
 				err_printf(m, " (submitted by %s [%d])",
 					   ee->comm,
@@ -496,14 +496,14 @@  int i915_error_state_to_str(struct drm_i915_error_state_buf *m,
 		obj = ee->wa_batchbuffer;
 		if (obj) {
 			err_printf(m, "%s (w/a) --- gtt_offset = 0x%08x\n",
-				   dev_priv->engine[i].name,
+				   dev_priv->engine[i]->name,
 				   lower_32_bits(obj->gtt_offset));
 			print_error_obj(m, obj);
 		}
 
 		if (ee->num_requests) {
 			err_printf(m, "%s --- %d requests\n",
-				   dev_priv->engine[i].name,
+				   dev_priv->engine[i]->name,
 				   ee->num_requests);
 			for (j = 0; j < ee->num_requests; j++) {
 				err_printf(m, "  pid %d, seqno 0x%08x, emitted %ld, head 0x%08x, tail 0x%08x\n",
@@ -517,10 +517,10 @@  int i915_error_state_to_str(struct drm_i915_error_state_buf *m,
 
 		if (IS_ERR(ee->waiters)) {
 			err_printf(m, "%s --- ? waiters [unable to acquire spinlock]\n",
-				   dev_priv->engine[i].name);
+				   dev_priv->engine[i]->name);
 		} else if (ee->num_waiters) {
 			err_printf(m, "%s --- %d waiters\n",
-				   dev_priv->engine[i].name,
+				   dev_priv->engine[i]->name,
 				   ee->num_waiters);
 			for (j = 0; j < ee->num_waiters; j++) {
 				err_printf(m, " seqno 0x%08x for %s [%d]\n",
@@ -532,7 +532,7 @@  int i915_error_state_to_str(struct drm_i915_error_state_buf *m,
 
 		if ((obj = ee->ringbuffer)) {
 			err_printf(m, "%s --- ringbuffer = 0x%08x\n",
-				   dev_priv->engine[i].name,
+				   dev_priv->engine[i]->name,
 				   lower_32_bits(obj->gtt_offset));
 			print_error_obj(m, obj);
 		}
@@ -546,7 +546,7 @@  int i915_error_state_to_str(struct drm_i915_error_state_buf *m,
 				hws_page = &obj->pages[LRC_PPHWSP_PN][0];
 			}
 			err_printf(m, "%s --- HW Status = 0x%08llx\n",
-				   dev_priv->engine[i].name, hws_offset);
+				   dev_priv->engine[i]->name, hws_offset);
 			offset = 0;
 			for (elt = 0; elt < PAGE_SIZE/16; elt += 4) {
 				err_printf(m, "[%04x] %08x %08x %08x %08x\n",
@@ -563,12 +563,12 @@  int i915_error_state_to_str(struct drm_i915_error_state_buf *m,
 		if (obj) {
 			u64 wa_ctx_offset = obj->gtt_offset;
 			u32 *wa_ctx_page = &obj->pages[0][0];
-			struct intel_engine_cs *engine = &dev_priv->engine[RCS];
+			struct intel_engine_cs *engine = dev_priv->engine[RCS];
 			u32 wa_ctx_size = (engine->wa_ctx.indirect_ctx.size +
 					   engine->wa_ctx.per_ctx.size);
 
 			err_printf(m, "%s --- WA ctx batch buffer = 0x%08llx\n",
-				   dev_priv->engine[i].name, wa_ctx_offset);
+				   dev_priv->engine[i]->name, wa_ctx_offset);
 			offset = 0;
 			for (elt = 0; elt < wa_ctx_size; elt += 4) {
 				err_printf(m, "[%04x] %08x %08x %08x %08x\n",
@@ -583,7 +583,7 @@  int i915_error_state_to_str(struct drm_i915_error_state_buf *m,
 
 		if ((obj = ee->ctx)) {
 			err_printf(m, "%s --- HW Context = 0x%08x\n",
-				   dev_priv->engine[i].name,
+				   dev_priv->engine[i]->name,
 				   lower_32_bits(obj->gtt_offset));
 			print_error_obj(m, obj);
 		}
@@ -918,7 +918,7 @@  static void gen8_record_semaphore_state(struct drm_i915_error_state *error,
 	if (!error->semaphore)
 		return;
 
-	for_each_engine_id(to, dev_priv, id) {
+	for_each_engine(to, dev_priv, id) {
 		int idx;
 		u16 signal_offset;
 		u32 *tmp;
@@ -1173,14 +1173,14 @@  static void i915_gem_record_rings(struct drm_i915_private *dev_priv,
 		i915_error_object_create(dev_priv, dev_priv->semaphore);
 
 	for (i = 0; i < I915_NUM_ENGINES; i++) {
-		struct intel_engine_cs *engine = &dev_priv->engine[i];
+		struct intel_engine_cs *engine = dev_priv->engine[i];
 		struct drm_i915_error_engine *ee = &error->engine[i];
 		struct drm_i915_gem_request *request;
 
 		ee->pid = -1;
 		ee->engine_id = -1;
 
-		if (!intel_engine_initialized(engine))
+		if (!engine)
 			continue;
 
 		ee->engine_id = i;
@@ -1603,7 +1603,7 @@  void i915_get_engine_instdone(struct drm_i915_private *dev_priv,
 			      enum intel_engine_id engine_id,
 			      struct intel_instdone *instdone)
 {
-	u32 mmio_base = dev_priv->engine[engine_id].mmio_base;
+	u32 mmio_base = dev_priv->engine[engine_id]->mmio_base;
 	int slice;
 	int subslice;
 
diff --git a/drivers/gpu/drm/i915/i915_guc_submission.c b/drivers/gpu/drm/i915/i915_guc_submission.c
index 43358e1..b09327c 100644
--- a/drivers/gpu/drm/i915/i915_guc_submission.c
+++ b/drivers/gpu/drm/i915/i915_guc_submission.c
@@ -905,6 +905,7 @@  static void guc_addon_create(struct intel_guc *guc)
 	struct guc_policies *policies;
 	struct guc_mmio_reg_state *reg_state;
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	struct page *page;
 	u32 size;
 
@@ -932,10 +933,10 @@  static void guc_addon_create(struct intel_guc *guc)
 	 * so its address won't change after we've told the GuC where
 	 * to find it.
 	 */
-	engine = &dev_priv->engine[RCS];
+	engine = dev_priv->engine[RCS];
 	ads->golden_context_lrca = engine->status_page.ggtt_offset;
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		ads->eng_state_size[engine->guc_id] = intel_lr_context_size(engine);
 
 	/* GuC scheduling policies */
@@ -948,7 +949,7 @@  static void guc_addon_create(struct intel_guc *guc)
 	/* MMIO reg state */
 	reg_state = (void *)policies + sizeof(struct guc_policies);
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		reg_state->mmio_white_list[engine->guc_id].mmio_start =
 			engine->mmio_base + GUC_MMIO_WHITE_LIST_START;
 
@@ -1002,9 +1003,10 @@  int i915_guc_submission_init(struct drm_i915_private *dev_priv)
 int i915_guc_submission_enable(struct drm_i915_private *dev_priv)
 {
 	struct intel_guc *guc = &dev_priv->guc;
+	struct drm_i915_gem_request *request;
 	struct i915_guc_client *client;
 	struct intel_engine_cs *engine;
-	struct drm_i915_gem_request *request;
+	enum intel_engine_id iter;
 
 	/* client for execbuf submission */
 	client = guc_client_alloc(dev_priv,
@@ -1021,7 +1023,7 @@  int i915_guc_submission_enable(struct drm_i915_private *dev_priv)
 	guc_init_doorbell_hw(guc);
 
 	/* Take over from manual control of ELSP (execlists) */
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		engine->submit_request = i915_guc_submit;
 
 		/* Replay the current set of previously submitted requests */
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index bd6c8b0..c4fb4bc 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -1058,8 +1058,9 @@  static u32 vlv_wa_c0_ei(struct drm_i915_private *dev_priv, u32 pm_iir)
 static bool any_waiters(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		if (intel_engine_has_waiter(engine))
 			return true;
 
@@ -1257,20 +1258,20 @@  static void ilk_gt_irq_handler(struct drm_i915_private *dev_priv,
 			       u32 gt_iir)
 {
 	if (gt_iir & GT_RENDER_USER_INTERRUPT)
-		notify_ring(&dev_priv->engine[RCS]);
+		notify_ring(dev_priv->engine[RCS]);
 	if (gt_iir & ILK_BSD_USER_INTERRUPT)
-		notify_ring(&dev_priv->engine[VCS]);
+		notify_ring(dev_priv->engine[VCS]);
 }
 
 static void snb_gt_irq_handler(struct drm_i915_private *dev_priv,
 			       u32 gt_iir)
 {
 	if (gt_iir & GT_RENDER_USER_INTERRUPT)
-		notify_ring(&dev_priv->engine[RCS]);
+		notify_ring(dev_priv->engine[RCS]);
 	if (gt_iir & GT_BSD_USER_INTERRUPT)
-		notify_ring(&dev_priv->engine[VCS]);
+		notify_ring(dev_priv->engine[VCS]);
 	if (gt_iir & GT_BLT_USER_INTERRUPT)
-		notify_ring(&dev_priv->engine[BCS]);
+		notify_ring(dev_priv->engine[BCS]);
 
 	if (gt_iir & (GT_BLT_CS_ERROR_INTERRUPT |
 		      GT_BSD_CS_ERROR_INTERRUPT |
@@ -1340,21 +1341,21 @@  static void gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
 				u32 gt_iir[4])
 {
 	if (gt_iir[0]) {
-		gen8_cs_irq_handler(&dev_priv->engine[RCS],
+		gen8_cs_irq_handler(dev_priv->engine[RCS],
 				    gt_iir[0], GEN8_RCS_IRQ_SHIFT);
-		gen8_cs_irq_handler(&dev_priv->engine[BCS],
+		gen8_cs_irq_handler(dev_priv->engine[BCS],
 				    gt_iir[0], GEN8_BCS_IRQ_SHIFT);
 	}
 
 	if (gt_iir[1]) {
-		gen8_cs_irq_handler(&dev_priv->engine[VCS],
+		gen8_cs_irq_handler(dev_priv->engine[VCS],
 				    gt_iir[1], GEN8_VCS1_IRQ_SHIFT);
-		gen8_cs_irq_handler(&dev_priv->engine[VCS2],
+		gen8_cs_irq_handler(dev_priv->engine[VCS2],
 				    gt_iir[1], GEN8_VCS2_IRQ_SHIFT);
 	}
 
 	if (gt_iir[3])
-		gen8_cs_irq_handler(&dev_priv->engine[VECS],
+		gen8_cs_irq_handler(dev_priv->engine[VECS],
 				    gt_iir[3], GEN8_VECS_IRQ_SHIFT);
 
 	if (gt_iir[2] & dev_priv->pm_rps_events)
@@ -1598,7 +1599,7 @@  static void gen6_rps_irq_handler(struct drm_i915_private *dev_priv, u32 pm_iir)
 
 	if (HAS_VEBOX(dev_priv)) {
 		if (pm_iir & PM_VEBOX_USER_INTERRUPT)
-			notify_ring(&dev_priv->engine[VECS]);
+			notify_ring(dev_priv->engine[VECS]);
 
 		if (pm_iir & PM_VEBOX_CS_ERROR_INTERRUPT)
 			DRM_DEBUG("Command parser error, pm_iir 0x%08x\n", pm_iir);
@@ -2840,9 +2841,10 @@  semaphore_wait_to_signaller_ring(struct intel_engine_cs *engine, u32 ipehr,
 {
 	struct drm_i915_private *dev_priv = engine->i915;
 	struct intel_engine_cs *signaller;
+	enum intel_engine_id iter;
 
 	if (INTEL_GEN(dev_priv) >= 8) {
-		for_each_engine(signaller, dev_priv) {
+		for_each_engine(signaller, dev_priv, iter) {
 			if (engine == signaller)
 				continue;
 
@@ -2852,7 +2854,7 @@  semaphore_wait_to_signaller_ring(struct intel_engine_cs *engine, u32 ipehr,
 	} else {
 		u32 sync_bits = ipehr & MI_SEMAPHORE_SYNC_MASK;
 
-		for_each_engine(signaller, dev_priv) {
+		for_each_engine(signaller, dev_priv, iter) {
 			if(engine == signaller)
 				continue;
 
@@ -2973,8 +2975,9 @@  static int semaphore_passed(struct intel_engine_cs *engine)
 static void semaphore_clear_deadlocks(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		engine->hangcheck.deadlock = 0;
 }
 
@@ -3101,6 +3104,7 @@  static void i915_hangcheck_elapsed(struct work_struct *work)
 		container_of(work, typeof(*dev_priv),
 			     gpu_error.hangcheck_work.work);
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	unsigned int hung = 0, stuck = 0;
 	int busy_count = 0;
 #define BUSY 1
@@ -3120,7 +3124,7 @@  static void i915_hangcheck_elapsed(struct work_struct *work)
 	 */
 	intel_uncore_arm_unclaimed_mmio_detection(dev_priv);
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		bool busy = intel_engine_has_waiter(engine);
 		u64 acthd;
 		u32 seqno;
@@ -4011,7 +4015,7 @@  static irqreturn_t i8xx_irq_handler(int irq, void *arg)
 		new_iir = I915_READ16(IIR); /* Flush posted writes */
 
 		if (iir & I915_USER_INTERRUPT)
-			notify_ring(&dev_priv->engine[RCS]);
+			notify_ring(dev_priv->engine[RCS]);
 
 		for_each_pipe(dev_priv, pipe) {
 			int plane = pipe;
@@ -4208,7 +4212,7 @@  static irqreturn_t i915_irq_handler(int irq, void *arg)
 		new_iir = I915_READ(IIR); /* Flush posted writes */
 
 		if (iir & I915_USER_INTERRUPT)
-			notify_ring(&dev_priv->engine[RCS]);
+			notify_ring(dev_priv->engine[RCS]);
 
 		for_each_pipe(dev_priv, pipe) {
 			int plane = pipe;
@@ -4440,9 +4444,9 @@  static irqreturn_t i965_irq_handler(int irq, void *arg)
 		new_iir = I915_READ(IIR); /* Flush posted writes */
 
 		if (iir & I915_USER_INTERRUPT)
-			notify_ring(&dev_priv->engine[RCS]);
+			notify_ring(dev_priv->engine[RCS]);
 		if (iir & I915_BSD_USER_INTERRUPT)
-			notify_ring(&dev_priv->engine[VCS]);
+			notify_ring(dev_priv->engine[VCS]);
 
 		for_each_pipe(dev_priv, pipe) {
 			if (pipe_stats[pipe] & PIPE_START_VBLANK_INTERRUPT_STATUS &&
diff --git a/drivers/gpu/drm/i915/intel_breadcrumbs.c b/drivers/gpu/drm/i915/intel_breadcrumbs.c
index 9bad14d..5e3bd38 100644
--- a/drivers/gpu/drm/i915/intel_breadcrumbs.c
+++ b/drivers/gpu/drm/i915/intel_breadcrumbs.c
@@ -592,6 +592,7 @@  void intel_engine_fini_breadcrumbs(struct intel_engine_cs *engine)
 unsigned int intel_kick_waiters(struct drm_i915_private *i915)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	unsigned int mask = 0;
 
 	/* To avoid the task_struct disappearing beneath us as we wake up
@@ -599,7 +600,7 @@  unsigned int intel_kick_waiters(struct drm_i915_private *i915)
 	 * RCU lock, i.e. as we call wake_up_process() we must be holding the
 	 * rcu_read_lock().
 	 */
-	for_each_engine(engine, i915)
+	for_each_engine(engine, i915, iter)
 		if (unlikely(intel_engine_wakeup(engine)))
 			mask |= intel_engine_flag(engine);
 
@@ -609,9 +610,10 @@  unsigned int intel_kick_waiters(struct drm_i915_private *i915)
 unsigned int intel_kick_signalers(struct drm_i915_private *i915)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	unsigned int mask = 0;
 
-	for_each_engine(engine, i915) {
+	for_each_engine(engine, i915, iter) {
 		if (unlikely(READ_ONCE(engine->breadcrumbs.first_signal))) {
 			wake_up_process(engine->breadcrumbs.signaler);
 			mask |= intel_engine_flag(engine);
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index a366656..df5d9f8 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -12242,19 +12242,19 @@  static int intel_crtc_page_flip(struct drm_crtc *crtc,
 		work->flip_count = I915_READ(PIPE_FLIPCOUNT_G4X(pipe)) + 1;
 
 	if (IS_VALLEYVIEW(dev) || IS_CHERRYVIEW(dev)) {
-		engine = &dev_priv->engine[BCS];
+		engine = dev_priv->engine[BCS];
 		if (fb->modifier[0] != old_fb->modifier[0])
 			/* vlv: DISPLAY_FLIP fails to change tiling */
 			engine = NULL;
 	} else if (IS_IVYBRIDGE(dev) || IS_HASWELL(dev)) {
-		engine = &dev_priv->engine[BCS];
+		engine = dev_priv->engine[BCS];
 	} else if (INTEL_INFO(dev)->gen >= 7) {
 		engine = i915_gem_active_get_engine(&obj->last_write,
 						    &obj->base.dev->struct_mutex);
 		if (engine == NULL || engine->id != RCS)
-			engine = &dev_priv->engine[BCS];
+			engine = dev_priv->engine[BCS];
 	} else {
-		engine = &dev_priv->engine[RCS];
+		engine = dev_priv->engine[RCS];
 	}
 
 	mmio_flip = use_mmio_flip(engine, obj);
diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
index d00ec80..9a028bb 100644
--- a/drivers/gpu/drm/i915/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/intel_engine_cs.c
@@ -82,12 +82,17 @@  static const struct engine_info {
 	},
 };
 
-static struct intel_engine_cs *
+static int
 intel_engine_setup(struct drm_i915_private *dev_priv,
 		   enum intel_engine_id id)
 {
 	const struct engine_info *info = &intel_engines[id];
-	struct intel_engine_cs *engine = &dev_priv->engine[id];
+	struct intel_engine_cs *engine;
+
+	GEM_BUG_ON(dev_priv->engine[id]);
+	engine = kzalloc(sizeof(*engine), GFP_KERNEL);
+	if (!engine)
+		return -ENOMEM;
 
 	engine->id = id;
 	engine->i915 = dev_priv;
@@ -97,7 +102,8 @@  intel_engine_setup(struct drm_i915_private *dev_priv,
 	engine->mmio_base = info->mmio_base;
 	engine->irq_shift = info->irq_shift;
 
-	return engine;
+	dev_priv->engine[id] = engine;
+	return 0;
 }
 
 /**
@@ -131,7 +137,11 @@  int intel_engines_init(struct drm_device *dev)
 		if (!init)
 			continue;
 
-		ret = init(intel_engine_setup(dev_priv, i));
+		ret = intel_engine_setup(dev_priv, i);
+		if (ret)
+			goto cleanup;
+
+		ret = init(dev_priv->engine[i]);
 		if (ret)
 			goto cleanup;
 
@@ -153,9 +163,9 @@  int intel_engines_init(struct drm_device *dev)
 cleanup:
 	for (i = 0; i < I915_NUM_ENGINES; i++) {
 		if (i915.enable_execlists)
-			intel_logical_ring_cleanup(&dev_priv->engine[i]);
+			intel_logical_ring_cleanup(dev_priv->engine[i]);
 		else
-			intel_engine_cleanup(&dev_priv->engine[i]);
+			intel_engine_cleanup(dev_priv->engine[i]);
 	}
 
 	return ret;
diff --git a/drivers/gpu/drm/i915/intel_guc_loader.c b/drivers/gpu/drm/i915/intel_guc_loader.c
index 7ace96b..fd49bfb 100644
--- a/drivers/gpu/drm/i915/intel_guc_loader.c
+++ b/drivers/gpu/drm/i915/intel_guc_loader.c
@@ -100,12 +100,13 @@  const char *intel_guc_fw_status_repr(enum intel_guc_fw_status status)
 static void guc_interrupts_release(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int irqs;
 
 	/* tell all command streamers NOT to forward interrupts or vblank to GuC */
 	irqs = _MASKED_FIELD(GFX_FORWARD_VBLANK_MASK, GFX_FORWARD_VBLANK_NEVER);
 	irqs |= _MASKED_BIT_DISABLE(GFX_INTERRUPT_STEERING);
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		I915_WRITE(RING_MODE_GEN7(engine), irqs);
 
 	/* route all GT interrupts to the host */
@@ -117,12 +118,13 @@  static void guc_interrupts_release(struct drm_i915_private *dev_priv)
 static void guc_interrupts_capture(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	int irqs;
 	u32 tmp;
 
 	/* tell all command streamers to forward interrupts (but not vblank) to GuC */
 	irqs = _MASKED_BIT_ENABLE(GFX_INTERRUPT_STEERING);
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		I915_WRITE(RING_MODE_GEN7(engine), irqs);
 
 	/* route USER_INTERRUPT to Host, all others are sent to GuC. */
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 936f6f6..08303e3 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -1645,7 +1645,7 @@  void intel_logical_ring_cleanup(struct intel_engine_cs *engine)
 {
 	struct drm_i915_private *dev_priv;
 
-	if (!intel_engine_initialized(engine))
+	if (!engine)
 		return;
 
 	/*
@@ -1674,13 +1674,16 @@  void intel_logical_ring_cleanup(struct intel_engine_cs *engine)
 
 	lrc_destroy_wa_ctx_obj(engine);
 	engine->i915 = NULL;
+	dev_priv->engine[engine->id] = NULL;
+	kfree(engine);
 }
 
 void intel_execlists_enable_submission(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		engine->submit_request = execlists_submit_request;
 }
 
@@ -2148,6 +2151,7 @@  void intel_lr_context_resume(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
 	struct i915_gem_context *ctx;
+	enum intel_engine_id iter;
 
 	/* Because we emit WA_TAIL_DWORDS there may be a disparity
 	 * between our bookkeeping in ce->ring->head and ce->ring->tail and
@@ -2160,7 +2164,7 @@  void intel_lr_context_resume(struct drm_i915_private *dev_priv)
 	 * simplicity, we just zero everything out.
 	 */
 	list_for_each_entry(ctx, &dev_priv->context_list, link) {
-		for_each_engine(engine, dev_priv) {
+		for_each_engine(engine, dev_priv, iter) {
 			struct intel_context *ce = &ctx->engine[engine->id];
 			u32 *reg;
 
diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c
index a24bc8c..272eb93 100644
--- a/drivers/gpu/drm/i915/intel_overlay.c
+++ b/drivers/gpu/drm/i915/intel_overlay.c
@@ -233,7 +233,7 @@  static int intel_overlay_do_wait_request(struct intel_overlay *overlay,
 static struct drm_i915_gem_request *alloc_request(struct intel_overlay *overlay)
 {
 	struct drm_i915_private *dev_priv = overlay->i915;
-	struct intel_engine_cs *engine = &dev_priv->engine[RCS];
+	struct intel_engine_cs *engine = dev_priv->engine[RCS];
 
 	return i915_gem_request_alloc(engine, dev_priv->kernel_context);
 }
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 7f1748a..8755580 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -5359,6 +5359,7 @@  static void gen9_enable_rps(struct drm_i915_private *dev_priv)
 static void gen9_enable_rc6(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	uint32_t rc6_mask = 0;
 
 	/* 1a: Software RC state - RC0 */
@@ -5380,7 +5381,7 @@  static void gen9_enable_rc6(struct drm_i915_private *dev_priv)
 		I915_WRITE(GEN6_RC6_WAKE_RATE_LIMIT, 54 << 16);
 	I915_WRITE(GEN6_RC_EVALUATION_INTERVAL, 125000); /* 12500 * 1280ns */
 	I915_WRITE(GEN6_RC_IDLE_HYSTERSIS, 25); /* 25 * 1280ns */
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		I915_WRITE(RING_MAX_IDLE(engine->mmio_base), 10);
 
 	if (HAS_GUC(dev_priv))
@@ -5425,6 +5426,7 @@  static void gen9_enable_rc6(struct drm_i915_private *dev_priv)
 static void gen8_enable_rps(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	uint32_t rc6_mask = 0;
 
 	/* 1a: Software RC state - RC0 */
@@ -5441,7 +5443,7 @@  static void gen8_enable_rps(struct drm_i915_private *dev_priv)
 	I915_WRITE(GEN6_RC6_WAKE_RATE_LIMIT, 40 << 16);
 	I915_WRITE(GEN6_RC_EVALUATION_INTERVAL, 125000); /* 12500 * 1280ns */
 	I915_WRITE(GEN6_RC_IDLE_HYSTERSIS, 25); /* 25 * 1280ns */
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		I915_WRITE(RING_MAX_IDLE(engine->mmio_base), 10);
 	I915_WRITE(GEN6_RC_SLEEP, 0);
 	if (IS_BROADWELL(dev_priv))
@@ -5501,6 +5503,7 @@  static void gen8_enable_rps(struct drm_i915_private *dev_priv)
 static void gen6_enable_rps(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	u32 rc6vids, rc6_mask = 0;
 	u32 gtfifodbg;
 	int rc6_mode;
@@ -5534,7 +5537,7 @@  static void gen6_enable_rps(struct drm_i915_private *dev_priv)
 	I915_WRITE(GEN6_RC_EVALUATION_INTERVAL, 125000);
 	I915_WRITE(GEN6_RC_IDLE_HYSTERSIS, 25);
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		I915_WRITE(RING_MAX_IDLE(engine->mmio_base), 10);
 
 	I915_WRITE(GEN6_RC_SLEEP, 0);
@@ -5983,6 +5986,7 @@  static void valleyview_cleanup_gt_powersave(struct drm_i915_private *dev_priv)
 static void cherryview_enable_rps(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	u32 gtfifodbg, val, rc6_mode = 0, pcbr;
 
 	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
@@ -6009,7 +6013,7 @@  static void cherryview_enable_rps(struct drm_i915_private *dev_priv)
 	I915_WRITE(GEN6_RC_EVALUATION_INTERVAL, 125000); /* 12500 * 1280ns */
 	I915_WRITE(GEN6_RC_IDLE_HYSTERSIS, 25); /* 25 * 1280ns */
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		I915_WRITE(RING_MAX_IDLE(engine->mmio_base), 10);
 	I915_WRITE(GEN6_RC_SLEEP, 0);
 
@@ -6071,6 +6075,7 @@  static void cherryview_enable_rps(struct drm_i915_private *dev_priv)
 static void valleyview_enable_rps(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 	u32 gtfifodbg, val, rc6_mode = 0;
 
 	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
@@ -6110,7 +6115,7 @@  static void valleyview_enable_rps(struct drm_i915_private *dev_priv)
 	I915_WRITE(GEN6_RC_EVALUATION_INTERVAL, 125000);
 	I915_WRITE(GEN6_RC_IDLE_HYSTERSIS, 25);
 
-	for_each_engine(engine, dev_priv)
+	for_each_engine(engine, dev_priv, iter)
 		I915_WRITE(RING_MAX_IDLE(engine->mmio_base), 10);
 
 	I915_WRITE(GEN6_RC6_THRESHOLD, 0x557);
@@ -6793,7 +6798,7 @@  static void __intel_autoenable_gt_powersave(struct work_struct *work)
 	if (READ_ONCE(dev_priv->rps.enabled))
 		goto out;
 
-	rcs = &dev_priv->engine[RCS];
+	rcs = dev_priv->engine[RCS];
 	if (rcs->last_context)
 		goto out;
 
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 35f359e..ad0cdd1 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -1228,7 +1228,7 @@  static int gen8_rcs_signal(struct drm_i915_gem_request *req)
 	if (ret)
 		return ret;
 
-	for_each_engine_id(waiter, dev_priv, id) {
+	for_each_engine(waiter, dev_priv, id) {
 		u64 gtt_offset = req->engine->semaphore.signal_ggtt[id];
 		if (gtt_offset == MI_SEMAPHORE_SYNC_INVALID)
 			continue;
@@ -1265,7 +1265,7 @@  static int gen8_xcs_signal(struct drm_i915_gem_request *req)
 	if (ret)
 		return ret;
 
-	for_each_engine_id(waiter, dev_priv, id) {
+	for_each_engine(waiter, dev_priv, id) {
 		u64 gtt_offset = req->engine->semaphore.signal_ggtt[id];
 		if (gtt_offset == MI_SEMAPHORE_SYNC_INVALID)
 			continue;
@@ -1292,6 +1292,7 @@  static int gen6_signal(struct drm_i915_gem_request *req)
 	struct intel_ring *ring = req->ring;
 	struct drm_i915_private *dev_priv = req->i915;
 	struct intel_engine_cs *engine;
+	enum intel_engine_id id;
 	int ret, num_rings;
 
 	num_rings = INTEL_INFO(dev_priv)->num_rings;
@@ -1299,7 +1300,7 @@  static int gen6_signal(struct drm_i915_gem_request *req)
 	if (ret)
 		return ret;
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, id) {
 		i915_reg_t mbox_reg;
 
 		if (!(BIT(engine->hw_id) & GEN6_SEMAPHORES_MASK))
@@ -2091,7 +2092,7 @@  void intel_engine_cleanup(struct intel_engine_cs *engine)
 {
 	struct drm_i915_private *dev_priv;
 
-	if (!intel_engine_initialized(engine))
+	if (!engine)
 		return;
 
 	dev_priv = engine->i915;
@@ -2120,13 +2121,16 @@  void intel_engine_cleanup(struct intel_engine_cs *engine)
 	intel_ring_context_unpin(dev_priv->kernel_context, engine);
 
 	engine->i915 = NULL;
+	dev_priv->engine[engine->id] = NULL;
+	kfree(engine);
 }
 
 void intel_legacy_submission_resume(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
+	enum intel_engine_id iter;
 
-	for_each_engine(engine, dev_priv) {
+	for_each_engine(engine, dev_priv, iter) {
 		engine->buffer->head = engine->buffer->tail;
 		engine->buffer->last_retired_head = -1;
 	}
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 498931f0..60a8d9c 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -394,12 +394,6 @@  struct intel_engine_cs {
 	u32 (*get_cmd_length_mask)(u32 cmd_header);
 };
 
-static inline bool
-intel_engine_initialized(const struct intel_engine_cs *engine)
-{
-	return engine->i915 != NULL;
-}
-
 static inline unsigned
 intel_engine_flag(const struct intel_engine_cs *engine)
 {