Dovetail/Evl confusion on my end

Dovetail/Evl confusion on my end

[Florian Bezdeka]

Hi Philippe,

I just recognized that you updated the source of my rebase branches
meanwhile, so v7.0-dovetail-rebase (for dovetail) and next/v7.0-evl-
rebase for evl.

While syncing the missing pieces I noticed that commit 
ad7e492ff2 ("/dev/{null, zero}: dovetail: drop deprecated .write handler ") 
is part of v7.0-evl-rebase but is not part of v7.0-dovetail-rebase.

Why is that? Any specific reason?

Best regards,
Florian

PS: wip/flo/v7.1-evl-rebase will be updated soon again to bring in the
missing parts.

Re: Dovetail/Evl confusion on my end

[Philippe Gerum]

Florian Bezdeka <florian.bezdeka@siemens.com> writes:

> Hi Philippe,
>
> I just recognized that you updated the source of my rebase branches
> meanwhile, so v7.0-dovetail-rebase (for dovetail) and next/v7.0-evl-
> rebase for evl.
>
> While syncing the missing pieces I noticed that commit 
> ad7e492ff2 ("/dev/{null, zero}: dovetail: drop deprecated .write handler ") 
> is part of v7.0-evl-rebase but is not part of v7.0-dovetail-rebase.
>
> Why is that? Any specific reason?
>

I pushed this patch yesterday, and wanted to have some hindsight about
it for a couple of (testing) days of the evl tree. I believe it is
correct, but given the patched area, better safe than sorry. I plan to
merge it last on top of 7.0-dovetail.

-- 
Philippe.

Re: Dovetail/Evl confusion on my end

[Philippe Gerum]

Philippe Gerum <rpm@xenomai.org> writes:

> Florian Bezdeka <florian.bezdeka@siemens.com> writes:
>
>> Hi Philippe,
>>
>> I just recognized that you updated the source of my rebase branches
>> meanwhile, so v7.0-dovetail-rebase (for dovetail) and next/v7.0-evl-
>> rebase for evl.
>>
>> While syncing the missing pieces I noticed that commit 
>> ad7e492ff2 ("/dev/{null, zero}: dovetail: drop deprecated .write handler ") 
>> is part of v7.0-evl-rebase but is not part of v7.0-dovetail-rebase.
>>
>> Why is that? Any specific reason?
>>
>
> I pushed this patch yesterday, and wanted to have some hindsight about
> it for a couple of (testing) days of the evl tree. I believe it is
> correct, but given the patched area, better safe than sorry. I plan to
> merge it last on top of 7.0-dovetail.

To elaborate a bit more, this patch only helps evl strictly speaking, by
preventing stdout/err redirections through proxies to cause a kernel
warning, which in fact reveals that some portion of the upstream code
(namely the /null and /zero cdevs at least) are still using a deprecated
file_operations handler, which the VFS complains about when using the
kernel interface to write to them.

-- 
Philippe.

Re: Dovetail/Evl confusion on my end

[Philippe Gerum]

Philippe Gerum <rpm@xenomai.org> writes:

> Philippe Gerum <rpm@xenomai.org> writes:
>
>> Florian Bezdeka <florian.bezdeka@siemens.com> writes:
>>
>>> Hi Philippe,
>>>
>>> I just recognized that you updated the source of my rebase branches
>>> meanwhile, so v7.0-dovetail-rebase (for dovetail) and next/v7.0-evl-
>>> rebase for evl.
>>>
>>> While syncing the missing pieces I noticed that commit 
>>> ad7e492ff2 ("/dev/{null, zero}: dovetail: drop deprecated .write handler ") 
>>> is part of v7.0-evl-rebase but is not part of v7.0-dovetail-rebase.
>>>
>>> Why is that? Any specific reason?
>>>
>>
>> I pushed this patch yesterday, and wanted to have some hindsight about
>> it for a couple of (testing) days of the evl tree. I believe it is
>> correct, but given the patched area, better safe than sorry. I plan to
>> merge it last on top of 7.0-dovetail.
>
> To elaborate a bit more, this patch only helps evl strictly speaking, by
> preventing stdout/err redirections through proxies to cause a kernel
> warning, which in fact reveals that some portion of the upstream code
> (namely the /null and /zero cdevs at least) are still using a deprecated
> file_operations handler, which the VFS complains about when using the
> kernel interface to write to them.

The feedback I received from internal users regarding this is
positive, no regression.

-- 
Philippe.

Re: Dovetail/Evl confusion on my end

[Florian Bezdeka]

On Tue, 2026-07-07 at 17:02 +0200, Philippe Gerum wrote:
> Philippe Gerum <rpm@xenomai.org> writes:
> 
> > Philippe Gerum <rpm@xenomai.org> writes:
> > 
> > > Florian Bezdeka <florian.bezdeka@siemens.com> writes:
> > > 
> > > > Hi Philippe,
> > > > 
> > > > I just recognized that you updated the source of my rebase branches
> > > > meanwhile, so v7.0-dovetail-rebase (for dovetail) and next/v7.0-evl-
> > > > rebase for evl.
> > > > 
> > > > While syncing the missing pieces I noticed that commit 
> > > > ad7e492ff2 ("/dev/{null, zero}: dovetail: drop deprecated .write handler ") 
> > > > is part of v7.0-evl-rebase but is not part of v7.0-dovetail-rebase.
> > > > 
> > > > Why is that? Any specific reason?
> > > > 
> > > 
> > > I pushed this patch yesterday, and wanted to have some hindsight about
> > > it for a couple of (testing) days of the evl tree. I believe it is
> > > correct, but given the patched area, better safe than sorry. I plan to
> > > merge it last on top of 7.0-dovetail.
> > 
> > To elaborate a bit more, this patch only helps evl strictly speaking, by
> > preventing stdout/err redirections through proxies to cause a kernel
> > warning, which in fact reveals that some portion of the upstream code
> > (namely the /null and /zero cdevs at least) are still using a deprecated
> > file_operations handler, which the VFS complains about when using the
> > kernel interface to write to them.
> 
> The feedback I received from internal users regarding this is
> positive, no regression.

Ok. I applied it to wip/flo/v7.1-dovetail-rebase now and rebased
wip/flo/v7.1-evl-rebase accordingly.

For full risc-v support [1] has to be applied to evl. I will comment on
that series later. Seems to have style issues. To be able to test
already wip/flo/v7.1-evl-rebase+riscv has been pushed.

Waiting for over-night test results.

[1] https://lore.kernel.org/xenomai/20251010101130.3758633-1-tobias.schaffner@siemens.com/