Skip to contentSkip to conversation

drm/doc: Allow new UAPI to be used once it's in the driver's -next.

PATCH Superseded Eric Anholt 4 participants 10 messages Apr 24, 2019 – May 21, 2019

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/2 drm/doc: Allow new UAPI to be used once it's in the driver's -next.

I was trying to figure out if it was permissible to merge the Mesa side of V3D's CSD support yet while it's in drm-misc-next but not drm-next, and developers on #dri-devel IRC had differing opinions of what the requirement was. Propose a clarification here to see if Dave Airlie agrees.

---

Personally, I thought the rule was "has to be in drm-next", but assuming our review processes aren't totally broken, this should be enough.

 Documentation/gpu/drm-uapi.rst | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
Documentation/gpu/drm-uapi.rst +3 -2
index c9fd23efd957..8e5545dfbf82 100644
@@ -92,8 +92,9 @@ leads to a few additional requirements:
   requirements by doing a quick fork.
 
 - The kernel patch can only be merged after all the above requirements are met,
-  but it **must** be merged **before** the userspace patches land. uAPI always flows
-  from the kernel, doing things the other way round risks divergence of the uAPI
+  but it **must** be merged to the driver's -next tree (as documented in
+  MAINTAINERS) **before** the userspace patches land. uAPI always flows from
+  the kernel, doing things the other way round risks divergence of the uAPI
   definitions and header files.
 
 These are fairly steep requirements, but have grown out from years of shared
Signature · 1 line
2.20.1
#1 lore raw

PATCH 2/2 drm/doc: Document expectation that userspace review looks at kernel uAPI.

The point of this review process is that userspace using the new uAPI can actually live with the uAPI being provided, and it's hard to know that without having actually looked into a kernel patch yourself.

---
 Documentation/gpu/drm-uapi.rst | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Documentation/gpu/drm-uapi.rst +3 -1
index 8e5545dfbf82..298424b98d99 100644
@@ -85,7 +85,9 @@ leads to a few additional requirements:
 - The userspace side must be fully reviewed and tested to the standards of that
   userspace project. For e.g. mesa this means piglit testcases and review on the
   mailing list. This is again to ensure that the new interface actually gets the
-  job done.
+  job done.  The userspace-side reviewer should also provide at least an
+  Acked-by on the kernel uAPI patch indicating that they've looked at how the
+  kernel side is implementing the new feature being used.
 
 - The userspace patches must be against the canonical upstream, not some vendor
   fork. This is to make sure that no one cheats on the review and testing
Signature · 1 line
2.20.1
Reviewed-by on v1 2/2
#2 lore raw

Re: [PATCH 2/2] drm/doc: Document expectation that userspace review looks at kernel uAPI.

Quoted 15 lines · Eric Anholt

The point of this review process is that userspace using the new uAPI can actually live with the uAPI being provided, and it's hard to know that without having actually looked into a kernel patch yourself.

---
 Documentation/gpu/drm-uapi.rst | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Documentation/gpu/drm-uapi.rst +0 -0
index 8e5545dfbf82..298424b98d99 100644

Documentation/gpu/drm-uapi.rst leads to a few additional requirements: · line 90

Quoted 7 lines
Documentation/gpu/drm-uapi.rst +3 -1
@@ -85,7 +85,9 @@ leads to a few additional requirements:
 - The userspace side must be fully reviewed and tested to the standards of that
   userspace project. For e.g. mesa this means piglit testcases and review on the
   mailing list. This is again to ensure that the new interface actually gets the
-  job done.
+  job done.  The userspace-side reviewer should also provide at least an
+  Acked-by on the kernel uAPI patch indicating that they've looked at how the
+  kernel side is implementing the new feature being used.

Answers a question that just recently came up on merging new kms properties.

Quoted 6 lines
Documentation/gpu/drm-uapi.rst +0 -0
 - The userspace patches must be against the canonical upstream, not some vendor
   fork. This is to make sure that no one cheats on the review and testing
Signature · 1 line
2.20.1
Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#3 lore raw

Re: [PATCH 2/2] drm/doc: Document expectation that userspace review looks at kernel uAPI.

Quoted 27 lines · On Wed, 24 Apr 2019 21:36:36 +0200 Daniel Vetter <[email protected]> wrote:
Quoted 15 lines · Eric Anholt

The point of this review process is that userspace using the new uAPI can actually live with the uAPI being provided, and it's hard to know that without having actually looked into a kernel patch yourself.

---
 Documentation/gpu/drm-uapi.rst | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Documentation/gpu/drm-uapi.rst +0 -0
index 8e5545dfbf82..298424b98d99 100644

Documentation/gpu/drm-uapi.rst leads to a few additional requirements: · line 90

Quoted 7 lines
Documentation/gpu/drm-uapi.rst +3 -1
@@ -85,7 +85,9 @@ leads to a few additional requirements:
 - The userspace side must be fully reviewed and tested to the standards of that
   userspace project. For e.g. mesa this means piglit testcases and review on the
   mailing list. This is again to ensure that the new interface actually gets the
-  job done.
+  job done.  The userspace-side reviewer should also provide at least an
+  Acked-by on the kernel uAPI patch indicating that they've looked at how the
+  kernel side is implementing the new feature being used.  

Answers a question that just recently came up on merging new kms properties.

Quoted 1 line

Hi,

for the record, I personally will not be able to provide such Acked-by tag according to kernel review rules, because I am completely unfamiliar with kernel DRM internals and cannot review kernel code at all. This might make people expecting Weston to prove their uAPI disappointed, since there are very few Weston reviewers available.

If you meant something else, please word it to that you actually meant.

Thanks,
pq

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#4 lore raw

Re: [PATCH 2/2] drm/doc: Document expectation that userspace review looks at kernel uAPI.

Quoted 34 lines · Pekka Paalanen
Quoted 27 lines · On Wed, 24 Apr 2019 21:36:36 +0200 Daniel Vetter <[email protected]> wrote:
Quoted 15 lines · Eric Anholt

The point of this review process is that userspace using the new uAPI can actually live with the uAPI being provided, and it's hard to know that without having actually looked into a kernel patch yourself.

---
 Documentation/gpu/drm-uapi.rst | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Documentation/gpu/drm-uapi.rst +0 -0
index 8e5545dfbf82..298424b98d99 100644

Documentation/gpu/drm-uapi.rst leads to a few additional requirements: · line 90

Quoted 7 lines
Documentation/gpu/drm-uapi.rst +3 -1
@@ -85,7 +85,9 @@ leads to a few additional requirements:
 - The userspace side must be fully reviewed and tested to the standards of that
   userspace project. For e.g. mesa this means piglit testcases and review on the
   mailing list. This is again to ensure that the new interface actually gets the
-  job done.
+  job done.  The userspace-side reviewer should also provide at least an
+  Acked-by on the kernel uAPI patch indicating that they've looked at how the
+  kernel side is implementing the new feature being used.  

Answers a question that just recently came up on merging new kms properties.

Quoted 1 line

Hi,

Quoted 7 lines

for the record, I personally will not be able to provide such Acked-by tag according to kernel review rules, because I am completely unfamiliar with kernel DRM internals and cannot review kernel code at all. This might make people expecting Weston to prove their uAPI disappointed, since there are very few Weston reviewers available.

If you meant something else, please word it to that you actually meant.

Hm right, that wording is putting a bit too high a bar. We want the userspace view point here, not force userspace people to review kernel code. I'll try to clarify this a bit better.
-Daniel

Signature · 3 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
Reviewed-by on v1 1/2
#5 lore raw
In reply to Eric Anholt
Quoted 6 lines · Eric Anholt

I was trying to figure out if it was permissible to merge the Mesa side of V3D's CSD support yet while it's in drm-misc-next but not drm-next, and developers on #dri-devel IRC had differing opinions of what the requirement was. Propose a clarification here to see if Dave Airlie agrees.

Quoted 6 lines

---

Personally, I thought the rule was "has to be in drm-next", but assuming our review processes aren't totally broken, this should be enough.

Yeah if you end up with a revert on your hands the process failed much harder and you get to keep the pieces no matter what. Not sure we should clarify whether you need a stable sha1 or not (helps with cross referencing uapi header updates), but imo good as is. And matches what I've been doing/recommending past few years.

Quoted 27 lines
 Documentation/gpu/drm-uapi.rst | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
Documentation/gpu/drm-uapi.rst +3 -2
index c9fd23efd957..8e5545dfbf82 100644
@@ -92,8 +92,9 @@ leads to a few additional requirements:
   requirements by doing a quick fork.
 
 - The kernel patch can only be merged after all the above requirements are met,
-  but it **must** be merged **before** the userspace patches land. uAPI always flows
-  from the kernel, doing things the other way round risks divergence of the uAPI
+  but it **must** be merged to the driver's -next tree (as documented in
+  MAINTAINERS) **before** the userspace patches land. uAPI always flows from
+  the kernel, doing things the other way round risks divergence of the uAPI
   definitions and header files.
 
 These are fairly steep requirements, but have grown out from years of shared
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Signature · 3 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
maintainer
#6 lore raw
Quoted 34 lines · Daniel Vetter
Quoted 6 lines · Eric Anholt

I was trying to figure out if it was permissible to merge the Mesa side of V3D's CSD support yet while it's in drm-misc-next but not drm-next, and developers on #dri-devel IRC had differing opinions of what the requirement was. Propose a clarification here to see if Dave Airlie agrees.

Quoted 6 lines

---

Personally, I thought the rule was "has to be in drm-next", but assuming our review processes aren't totally broken, this should be enough.

Yeah if you end up with a revert on your hands the process failed much harder and you get to keep the pieces no matter what. Not sure we should clarify whether you need a stable sha1 or not (helps with cross referencing uapi header updates), but imo good as is. And matches what I've been doing/recommending past few years.

Quoted 11 lines
 Documentation/gpu/drm-uapi.rst | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
Documentation/gpu/drm-uapi.rst +0 -0
index c9fd23efd957..8e5545dfbf82 100644
@@ -92,8 +92,9 @@ leads to a few additional requirements:
   requirements by doing a quick fork.
 
Quoted 7 lines

Documentation/gpu/drm-uapi.rst leads to a few additional requirements: · line 98

Quoted 7 lines
Documentation/gpu/drm-uapi.rst +3 -2
 - The kernel patch can only be merged after all the above requirements are met,
-  but it **must** be merged **before** the userspace patches land. uAPI always flows
-  from the kernel, doing things the other way round risks divergence of the uAPI
+  but it **must** be merged to the driver's -next tree (as documented in
+  MAINTAINERS) **before** the userspace patches land. uAPI always flows from
+  the kernel, doing things the other way round risks divergence of the uAPI
   definitions and header files.

I'd rather restrict this to drm-next and drm-misc-next, I frankly don't trust driver trees here to have the review practices in place.

I trust drm-misc-next to have at least had someone unrelated look over the new api.

Dave.

#7 lore raw

PATCH v2 1/2 drm/doc: Allow new UAPI to be used once it's in drm-next/drm-misc-next.

I was trying to figure out if it was permissible to merge the Mesa side of V3D's CSD support yet while it's in drm-misc-next but not drm-next, and developers on #dri-devel IRC had differing opinions of what the requirement was.

v2: Restrict to just drm-next or drm-misc-next on airlied's request.

---
 Documentation/gpu/drm-uapi.rst | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
Documentation/gpu/drm-uapi.rst +3 -3
index c9fd23efd957..b7a96dc02d21 100644
@@ -92,9 +92,9 @@ leads to a few additional requirements:
   requirements by doing a quick fork.
 
 - The kernel patch can only be merged after all the above requirements are met,
-  but it **must** be merged **before** the userspace patches land. uAPI always flows
-  from the kernel, doing things the other way round risks divergence of the uAPI
-  definitions and header files.
+  but it **must** be merged to either drm-next or drm-misc-next **before** the
+  userspace patches land. uAPI always flows from the kernel, doing things the
+  other way round risks divergence of the uAPI definitions and header files.
 
 These are fairly steep requirements, but have grown out from years of shared
 pain and experience with uAPI added hastily, and almost always regretted about
Signature · 1 line
2.20.1
Reviewed-by on v2 1/2
#8 lore raw

Re: [PATCH 1/2 v2] drm/doc: Allow new UAPI to be used once it's in drm-next/drm-misc-next.

Quoted 6 lines · Eric Anholt

I was trying to figure out if it was permissible to merge the Mesa side of V3D's CSD support yet while it's in drm-misc-next but not drm-next, and developers on #dri-devel IRC had differing opinions of what the requirement was.

v2: Restrict to just drm-next or drm-misc-next on airlied's request.

Personally I think that's a bit too strict (if people want to screw up, they will be able to anyway). But since I'm all for clearer rules where possible, this has my support too.

Quoted 30 lines
---
 Documentation/gpu/drm-uapi.rst | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
Documentation/gpu/drm-uapi.rst +3 -3
index c9fd23efd957..b7a96dc02d21 100644
@@ -92,9 +92,9 @@ leads to a few additional requirements:
   requirements by doing a quick fork.
 
 - The kernel patch can only be merged after all the above requirements are met,
-  but it **must** be merged **before** the userspace patches land. uAPI always flows
-  from the kernel, doing things the other way round risks divergence of the uAPI
-  definitions and header files.
+  but it **must** be merged to either drm-next or drm-misc-next **before** the
+  userspace patches land. uAPI always flows from the kernel, doing things the
+  other way round risks divergence of the uAPI definitions and header files.
 
 These are fairly steep requirements, but have grown out from years of shared
 pain and experience with uAPI added hastily, and almost always regretted about
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Signature · 3 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
author Applied
#9 lore raw

Re: [PATCH 1/2 v2] drm/doc: Allow new UAPI to be used once it's in drm-next/drm-misc-next.

Quoted 6 lines · Daniel Vetter
Quoted 5 lines · Eric Anholt

I was trying to figure out if it was permissible to merge the Mesa side of V3D's CSD support yet while it's in drm-misc-next but not drm-next, and developers on #dri-devel IRC had differing opinions of what the requirement was.

Quoted 7 lines
Quoted 1 line

v2: Restrict to just drm-next or drm-misc-next on airlied's request.

Personally I think that's a bit too strict (if people want to screw up, they will be able to anyway). But since I'm all for clearer rules where possible, this has my support too.

Pushed to drm-misc-next now.