[PATCH] dma/idxd: remove experimental on whole driver

[PATCH] dma/idxd: remove experimental on whole driver

[Bruce Richardson]

The idxd DMA driver was marked as experimental in MAINTAINERS file.
Remove this designation as it's been stable for a long time now.

Fixes: e33ad06eaeeb ("dma/idxd: add skeleton for VFIO based DSA device")
Cc: stable@dpdk.org

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
---

Putting the fixline as the original start of the driver implementation,
since there is no good commit point to use to identify the stabilization
of the driver. Since this change should be good to backport to all
currently supported LTS releases, the orignal addition of the driver
works fine as a commit being fixed.
---
 MAINTAINERS | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/MAINTAINERS b/MAINTAINERS
index 482bc7df76..51bf6b32bd 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1368,7 +1368,7 @@ F: doc/guides/compressdevs/features/zsda.ini
 DMAdev Drivers
 --------------
 
-Intel IDXD - EXPERIMENTAL
+Intel IDXD
 M: Bruce Richardson <bruce.richardson@intel.com>
 M: Kevin Laatz <kevin.laatz@intel.com>
 F: drivers/dma/idxd/
-- 
2.53.0

Re: [PATCH] dma/idxd: remove experimental on whole driver

[David Marchand]

On Thu, 8 Oct 2026 at 17:44, Bruce Richardson
<bruce.richardson@intel.com> wrote:
>
> The idxd DMA driver was marked as experimental in MAINTAINERS file.
> Remove this designation as it's been stable for a long time now.
>
> Fixes: e33ad06eaeeb ("dma/idxd: add skeleton for VFIO based DSA device")
> Cc: stable@dpdk.org
>
> Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
> ---
>
> Putting the fixline as the original start of the driver implementation,
> since there is no good commit point to use to identify the stabilization
> of the driver. Since this change should be good to backport to all
> currently supported LTS releases, the orignal addition of the driver
> works fine as a commit being fixed.

No opinion on the backporting aspect.
At least the driver looks stable for a while, so the patch lgtm.


-- 
David Marchand

Re: [PATCH] dma/idxd: remove experimental on whole driver

[Bruce Richardson]

On Fri, Oct 09, 2026 at 10:55:49AM +0200, David Marchand wrote:
> On Thu, 8 Oct 2026 at 17:44, Bruce Richardson
> <bruce.richardson@intel.com> wrote:
> >
> > The idxd DMA driver was marked as experimental in MAINTAINERS file.
> > Remove this designation as it's been stable for a long time now.
> >
> > Fixes: e33ad06eaeeb ("dma/idxd: add skeleton for VFIO based DSA device")
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
> > ---
> >
> > Putting the fixline as the original start of the driver implementation,
> > since there is no good commit point to use to identify the stabilization
> > of the driver. Since this change should be good to backport to all
> > currently supported LTS releases, the orignal addition of the driver
> > works fine as a commit being fixed.
> 
> No opinion on the backporting aspect.
> At least the driver looks stable for a while, so the patch lgtm.
> 
Given that the driver has been around since 2021 and the oldest supported
LTS is 23.11, I felt it worthwhile to backport. However, if LTS
maintainers want to avoid this patch, I'm not too concerned :-)

Re: [PATCH] dma/idxd: remove experimental on whole driver

[Kevin Traynor]

On 10/9/26 10:03, Bruce Richardson wrote:
> On Fri, Oct 09, 2026 at 10:55:49AM +0200, David Marchand wrote:
>> On Thu, 8 Oct 2026 at 17:44, Bruce Richardson
>> <bruce.richardson@intel.com> wrote:
>>>
>>> The idxd DMA driver was marked as experimental in MAINTAINERS file.
>>> Remove this designation as it's been stable for a long time now.
>>>
>>> Fixes: e33ad06eaeeb ("dma/idxd: add skeleton for VFIO based DSA device")
>>> Cc: stable@dpdk.org
>>>
>>> Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
>>> ---
>>>
>>> Putting the fixline as the original start of the driver implementation,
>>> since there is no good commit point to use to identify the stabilization
>>> of the driver. Since this change should be good to backport to all
>>> currently supported LTS releases, the orignal addition of the driver
>>> works fine as a commit being fixed.
>>
>> No opinion on the backporting aspect.
>> At least the driver looks stable for a while, so the patch lgtm.
>>
> Given that the driver has been around since 2021 and the oldest supported
> LTS is 23.11, I felt it worthwhile to backport. However, if LTS
> maintainers want to avoid this patch, I'm not too concerned :-)

We're saying the main branch is non-experimental _now_ but it has 
diverged from what is in LTS branches e.g. [0]

I think it would be better not to backport this patch to LTS as the 
driver was released as experimental in those releases and the docs 
reflect that.

That it is changing in main is positive :-) but it doesn't seem worth 
evaluating each LTS branch and backports (and any other code context) to 
say if could now say that about the LTS branch too and retro-fitting the 
docs.

[0]
main branch:
$ git log --oneline v23.11..HEAD -- ./drivers/dma/idxd/
47cdad82a1 eal: improve pointer add/sub/align macros
11e237109b dma/idxd: remove next pointer in bus specific device
fbd34e543e dma/idxd: remove specific bus type
be29c42523 bus: implement probe in EAL
76622feba9 bus: refactor device probe
2267d1c26b bus: factorize driver lookup
89127f66f5 bus: consolidate device lookup
b4f0974a99 bus: factorize device list
3e1681fe47 bus: factorize driver list
53045e1e5a dma/idxd: clear device at scan
bba2dca922 bus: factorize device selection
e2287bb112 bus: factorize devargs lookup
ab797d5bcf dma/idxd: mark portal pointer as volatile
62516462cb dma/idxd: use DPDK basename function
fedffb112f dma/idxd: add device ids for new HW versions
fd51012de5 drivers: use portable variadic macros
aa8ed903d2 dma/idxd: set GRPCFG traffic class for better performance
91b026fb46 dma/idxd: fix free function mismatch in device probe
2b843cac23 drivers: use per line logging in helpers
f665790a5d drivers: remove redundant newline from logs
d569af3d4e dma/idxd: fix setup with Ubuntu 24.04
e12a0166c8 drivers: use stdatomic API
27595cd830 drivers: move alignment attribute on types for MSVC
83cfa2cbf3 dma/idxd: verify strdup return

23.11 branch:
$ git log --oneline v23.11..HEAD -- ./drivers/dma/idxd/
5fe3dbe571 dma/idxd: fix free function mismatch in device probe
e56ba1eea6 drivers: remove redundant newline from logs
7353cb767f dma/idxd: fix setup with Ubuntu 24.04
df74839ea1 dma/idxd: verify strdup return

Re: [PATCH] dma/idxd: remove experimental on whole driver

[Bruce Richardson]

On Fri, Oct 09, 2026 at 10:52:50AM +0100, Kevin Traynor wrote:
> On 10/9/26 10:03, Bruce Richardson wrote:
> > On Fri, Oct 09, 2026 at 10:55:49AM +0200, David Marchand wrote:
> > > On Thu, 8 Oct 2026 at 17:44, Bruce Richardson
> > > <bruce.richardson@intel.com> wrote:
> > > > 
> > > > The idxd DMA driver was marked as experimental in MAINTAINERS file.
> > > > Remove this designation as it's been stable for a long time now.
> > > > 
> > > > Fixes: e33ad06eaeeb ("dma/idxd: add skeleton for VFIO based DSA device")
> > > > Cc: stable@dpdk.org
> > > > 
> > > > Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
> > > > ---
> > > > 
> > > > Putting the fixline as the original start of the driver implementation,
> > > > since there is no good commit point to use to identify the stabilization
> > > > of the driver. Since this change should be good to backport to all
> > > > currently supported LTS releases, the orignal addition of the driver
> > > > works fine as a commit being fixed.
> > > 
> > > No opinion on the backporting aspect.
> > > At least the driver looks stable for a while, so the patch lgtm.
> > > 
> > Given that the driver has been around since 2021 and the oldest supported
> > LTS is 23.11, I felt it worthwhile to backport. However, if LTS
> > maintainers want to avoid this patch, I'm not too concerned :-)
> 
> We're saying the main branch is non-experimental _now_ but it has diverged
> from what is in LTS branches e.g. [0]
> 
> I think it would be better not to backport this patch to LTS as the driver
> was released as experimental in those releases and the docs reflect that.
> 
> That it is changing in main is positive :-) but it doesn't seem worth
> evaluating each LTS branch and backports (and any other code context) to say
> if could now say that about the LTS branch too and retro-fitting the docs.
> 
Sure, no problem.