[PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Jan Sommer]

Hello,

I finally got around to port the e1000 driver fixes which are already
present in the 5-freebsd-12 branch of rtems-libbsd also for the master
(and 5) branch.

I created two corresponding tickets for it: 
https://devel.rtems.org/ticket/4052 and
https://devel.rtems.org/ticket/4053.

Could someone please check and push them?

Best regards,

Jan

Jan Sommer (6):
  e1000: Add missing files
  waf: Add path-mappings feature
  i386: Add missing files to build system
  Callout: Redefine callout_reset_on for rtems
  iflib.c: Deactivate use of ifc_cpus
  i386: Delete old machine dependent files

 freebsd/sys/dev/e1000/if_em.c |6 +
 freebsd/sys/i386/include/machine/bus.h|6 -
 freebsd/sys/i386/include/machine/legacyvar.h  |   63 -
 freebsd/sys/i386/include/machine/specialreg.h |6 -
 freebsd/sys/net/iflib.c   | 6827 +
 freebsd/sys/net/iflib_private.h   |   70 +
 freebsd/sys/net/mp_ring.c |  554 ++
 freebsd/sys/net/mp_ring.h |   75 +
 freebsd/sys/sys/callout.h |6 +
 freebsd/sys/x86/include/machine/specialreg.h  | 1143 +++
 freebsd/sys/x86/include/machine/x86_var.h |  145 +
 libbsd.py |   18 +-
 rtemsbsd/include/x86/x86_var.h|  146 +-
 waf_libbsd.py |   13 +-
 14 files changed, 8853 insertions(+), 225 deletions(-)
 delete mode 100644 freebsd/sys/i386/include/machine/bus.h
 delete mode 100644 freebsd/sys/i386/include/machine/legacyvar.h
 delete mode 100644 freebsd/sys/i386/include/machine/specialreg.h
 create mode 100644 freebsd/sys/net/iflib.c
 create mode 100644 freebsd/sys/net/iflib_private.h
 create mode 100644 freebsd/sys/net/mp_ring.c
 create mode 100644 freebsd/sys/net/mp_ring.h
 create mode 100644 freebsd/sys/x86/include/machine/specialreg.h
 create mode 100644 freebsd/sys/x86/include/machine/x86_var.h

-- 
2.17.1

___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Chris Johns]

On 15/8/20 5:57 am, Jan Sommer wrote:
> Hello,
> 
> I finally got around to port the e1000 driver fixes which are already
> present in the 5-freebsd-12 branch of rtems-libbsd also for the master
> (and 5) branch.

Thanks. I will push these soon.

I am not sure if the 5 branch in the rtems-libbsd.git repo is right and if it
should be made from the 5-freebsd-12 branch? I feel it is currently confusing
but either way it has issues.

Chris
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Joel Sherrill]

On Fri, Aug 14, 2020 at 4:42 PM Chris Johns  wrote:

> On 15/8/20 5:57 am, Jan Sommer wrote:
> > Hello,
> >
> > I finally got around to port the e1000 driver fixes which are already
> > present in the 5-freebsd-12 branch of rtems-libbsd also for the master
> > (and 5) branch.
>
> Thanks. I will push these soon.
>
> I am not sure if the 5 branch in the rtems-libbsd.git repo is right and if
> it
> should be made from the 5-freebsd-12 branch? I feel it is currently
> confusing
> but either way it has issues.
>

If I remember correctly, Sebastian told me 5 was FreeBSD master on RTEMS 5.
If that's right, it would have been better with the longer name
5-freebsd-master since
we use 5 as the name for other repos to indicate trusted "5".

--joel

>
> Chris
> ___
> devel mailing list
> devel@rtems.org
> http://lists.rtems.org/mailman/listinfo/devel
>
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Chris Johns]

On 15/8/20 8:31 am, Joel Sherrill wrote:
> On Fri, Aug 14, 2020 at 4:42 PM Chris Johns  > wrote:
> 
> On 15/8/20 5:57 am, Jan Sommer wrote:
> > Hello,
> >
> > I finally got around to port the e1000 driver fixes which are already
> > present in the 5-freebsd-12 branch of rtems-libbsd also for the master
> > (and 5) branch.
> 
> Thanks. I will push these soon.
> 
> I am not sure if the 5 branch in the rtems-libbsd.git repo is right and 
> if it
> should be made from the 5-freebsd-12 branch? I feel it is currently 
> confusing
> but either way it has issues.
> 
> If I remember correctly, Sebastian told me 5 was FreeBSD master on RTEMS 5.
> If that's right, it would have been better with the longer name 
> 5-freebsd-master
> since we use 5 as the name for other repos to indicate trusted "5".

Problems appear when we have variants in names because management around the
repos needs to handle special cases ...

https://git.rtems.org/rtems-release/tree/rtems-release-package-start?h=5#n109

On the other hand the difference in one repo like libbsd has created a special
case in the release scripts because 5 was being picked up and released without
error until I caught it testing. Thus the need for the mapping and that rippled
out in to other issues.

I am not sure what the answer is but any solution or change needs to be worked
through our procedures.

Chris
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Joel Sherrill]

On Fri, Aug 14, 2020, 7:04 PM Chris Johns  wrote:

> On 15/8/20 8:31 am, Joel Sherrill wrote:
> > On Fri, Aug 14, 2020 at 4:42 PM Chris Johns  > > wrote:
> >
> > On 15/8/20 5:57 am, Jan Sommer wrote:
> > > Hello,
> > >
> > > I finally got around to port the e1000 driver fixes which are
> already
> > > present in the 5-freebsd-12 branch of rtems-libbsd also for the
> master
> > > (and 5) branch.
> >
> > Thanks. I will push these soon.
> >
> > I am not sure if the 5 branch in the rtems-libbsd.git repo is right
> and if it
> > should be made from the 5-freebsd-12 branch? I feel it is currently
> confusing
> > but either way it has issues.
> >
> > If I remember correctly, Sebastian told me 5 was FreeBSD master on RTEMS
> 5.
> > If that's right, it would have been better with the longer name
> 5-freebsd-master
> > since we use 5 as the name for other repos to indicate trusted "5".
>
> Problems appear when we have variants in names because management around
> the
> repos needs to handle special cases ...
>
>
> https://git.rtems.org/rtems-release/tree/rtems-release-package-start?h=5#n109
>
> On the other hand the difference in one repo like libbsd has created a
> special
> case in the release scripts because 5 was being picked up and released
> without
> error until I caught it testing. Thus the need for the mapping and that
> rippled
> out in to other issues.
>
> I am not sure what the answer is but any solution or change needs to be
> worked
> through our procedures.
>

+1

I rather like the idea of combining RTEMS version with the bsd version in
the branch name. I know it is a bit of a pain in scripts..even explaining
it to people is odd. But a regular pattern would help and not having simple
numbers would help avoid mistakes.

>
> Chris
>
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

RE: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Jan.Sommer]

Hi Chris,

> -Original Message-
> From: Chris Johns 
> Sent: Friday, August 14, 2020 11:43 PM
> To: Sommer, Jan ; devel@rtems.org
> Subject: Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5
> 
> On 15/8/20 5:57 am, Jan Sommer wrote:
> > Hello,
> >
> > I finally got around to port the e1000 driver fixes which are already
> > present in the 5-freebsd-12 branch of rtems-libbsd also for the master
> > (and 5) branch.
> 
> Thanks. I will push these soon.
> 

As mentioned over at the users list, could you please push these changes?
They should enable compilation for i386 again and we could establish Ethernet 
connections with Intel network devices.
The patches should apply to libbsd master and 5 branch, as they are both based 
on FreeBSD master.

> I am not sure if the 5 branch in the rtems-libbsd.git repo is right and if it
> should be made from the 5-freebsd-12 branch? I feel it is currently confusing
> but either way it has issues.
> 

Maybe, we can continue to discuss that outside of this patch set? Renaming a 
branch later should be no trouble.
An option could be to rename 5 -> 5-freebsd-master, then the mapping between 
rtems and FreeBSD would always be explicit.

Cheers,

   Jan

> Chris
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Chris Johns]

On 9/9/20 4:55 pm, jan.som...@dlr.de wrote:
> Hi Chris,
> 
>> -Original Message-
>> From: Chris Johns 
>> Sent: Friday, August 14, 2020 11:43 PM
>> To: Sommer, Jan ; devel@rtems.org
>> Subject: Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 
>> 5
>>
>> On 15/8/20 5:57 am, Jan Sommer wrote:
>>> Hello,
>>>
>>> I finally got around to port the e1000 driver fixes which are already
>>> present in the 5-freebsd-12 branch of rtems-libbsd also for the master
>>> (and 5) branch.
>>
>> Thanks. I will push these soon.
>>
> 
> As mentioned over at the users list, could you please push these changes?

Yes I will tomorrow if no one else does it.

> They should enable compilation for i386 again and we could establish Ethernet 
> connections with Intel network devices.
> The patches should apply to libbsd master and 5 branch, as they are both 
> based on FreeBSD master.
> 
>> I am not sure if the 5 branch in the rtems-libbsd.git repo is right and if it
>> should be made from the 5-freebsd-12 branch? I feel it is currently confusing
>> but either way it has issues.
>>
> 
> Maybe, we can continue to discuss that outside of this patch set? Renaming a 
> branch later should be no trouble.
> An option could be to rename 5 -> 5-freebsd-master, then the mapping between 
> rtems and FreeBSD would always be explicit.

Yes this is one Joel mentioned the other day and I missed the context in
relation to these patches. I think moving 5 to 5-freebsd-master makes sense.
What we have is confusing.

Chris
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Gedare Bloom]

On Wed, Sep 9, 2020 at 2:16 AM Chris Johns  wrote:
>
> On 9/9/20 4:55 pm, jan.som...@dlr.de wrote:
> > Hi Chris,
> >
> >> -Original Message-
> >> From: Chris Johns 
> >> Sent: Friday, August 14, 2020 11:43 PM
> >> To: Sommer, Jan ; devel@rtems.org
> >> Subject: Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master 
> >> and 5
> >>
> >> On 15/8/20 5:57 am, Jan Sommer wrote:
> >>> Hello,
> >>>
> >>> I finally got around to port the e1000 driver fixes which are already
> >>> present in the 5-freebsd-12 branch of rtems-libbsd also for the master
> >>> (and 5) branch.
> >>
> >> Thanks. I will push these soon.
> >>
> >
> > As mentioned over at the users list, could you please push these changes?
>
> Yes I will tomorrow if no one else does it.
>
> > They should enable compilation for i386 again and we could establish 
> > Ethernet connections with Intel network devices.
> > The patches should apply to libbsd master and 5 branch, as they are both 
> > based on FreeBSD master.
> >
> >> I am not sure if the 5 branch in the rtems-libbsd.git repo is right and if 
> >> it
> >> should be made from the 5-freebsd-12 branch? I feel it is currently 
> >> confusing
> >> but either way it has issues.
> >>
> >
> > Maybe, we can continue to discuss that outside of this patch set? Renaming 
> > a branch later should be no trouble.
> > An option could be to rename 5 -> 5-freebsd-master, then the mapping 
> > between rtems and FreeBSD would always be explicit.
>
> Yes this is one Joel mentioned the other day and I missed the context in
> relation to these patches. I think moving 5 to 5-freebsd-master makes sense.
> What we have is confusing.
>
+1

Would be good to get Christian or Sebastian to comment/agree :)

> Chris
> ___
> devel mailing list
> devel@rtems.org
> http://lists.rtems.org/mailman/listinfo/devel
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Sebastian Huber]

On 09/09/2020 17:28, Gedare Bloom wrote:


I am not sure if the 5 branch in the rtems-libbsd.git repo is right and if it
should be made from the 5-freebsd-12 branch? I feel it is currently confusing
but either way it has issues.


Maybe, we can continue to discuss that outside of this patch set? Renaming a 
branch later should be no trouble.
An option could be to rename 5 -> 5-freebsd-master, then the mapping between 
rtems and FreeBSD would always be explicit.

Yes this is one Joel mentioned the other day and I missed the context in
relation to these patches. I think moving 5 to 5-freebsd-master makes sense.
What we have is confusing.


+1

Would be good to get Christian or Sebastian to comment/agree :)


Yes, the branch name "5" makes no sense. It should be "5-freebsd-master".
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Sebastian Huber]

On 10/09/2020 12:13, Sebastian Huber wrote:


On 09/09/2020 17:28, Gedare Bloom wrote:

I am not sure if the 5 branch in the rtems-libbsd.git repo is 
right and if it
should be made from the 5-freebsd-12 branch? I feel it is 
currently confusing

but either way it has issues.

Maybe, we can continue to discuss that outside of this patch set? 
Renaming a branch later should be no trouble.
An option could be to rename 5 -> 5-freebsd-master, then the 
mapping between rtems and FreeBSD would always be explicit.
Yes this is one Joel mentioned the other day and I missed the 
context in
relation to these patches. I think moving 5 to 5-freebsd-master 
makes sense.

What we have is confusing.


+1

Would be good to get Christian or Sebastian to comment/agree :)


Yes, the branch name "5" makes no sense. It should be "5-freebsd-master".
Actually, I think we should delete this branch or is there someone 
willing to maintain it? The libbsd branch recommended for RTEMS 5 is 
5-freebsd-12.

___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Chris Johns]

On 10/9/20 9:34 pm, Sebastian Huber wrote:
> On 10/09/2020 12:13, Sebastian Huber wrote:
>> On 09/09/2020 17:28, Gedare Bloom wrote:
>> I am not sure if the 5 branch in the rtems-libbsd.git repo is right and 
>> if it
>> should be made from the 5-freebsd-12 branch? I feel it is currently 
>> confusing
>> but either way it has issues.
>>
> Maybe, we can continue to discuss that outside of this patch set? Renaming
> a branch later should be no trouble.
> An option could be to rename 5 -> 5-freebsd-master, then the mapping
> between rtems and FreeBSD would always be explicit.
 Yes this is one Joel mentioned the other day and I missed the context in
 relation to these patches. I think moving 5 to 5-freebsd-master makes 
 sense.
 What we have is confusing.

>>> +1
>>>
>>> Would be good to get Christian or Sebastian to comment/agree :)
>>>
>> Yes, the branch name "5" makes no sense. It should be "5-freebsd-master".
> Actually, I think we should delete this branch or is there someone willing to
> maintain it? 

If that is the case why have 5-freebsd-12? Why not make it 5 and then all the
repos are the same?

> The libbsd branch recommended for RTEMS 5 is 5-freebsd-12.

Do we maintain FreeBSD related fixes off the freebsd-12 release branch only?

Chris
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel

Re: [PATCH v1 0/6] [libbsd] Fix e1000 driver for i386 in master and 5

[Sebastian Huber]

On 11/09/2020 07:38, Chris Johns wrote:


On 10/9/20 9:34 pm, Sebastian Huber wrote:

On 10/09/2020 12:13, Sebastian Huber wrote:

On 09/09/2020 17:28, Gedare Bloom wrote:

I am not sure if the 5 branch in the rtems-libbsd.git repo is right and if it
should be made from the 5-freebsd-12 branch? I feel it is currently confusing
but either way it has issues.


Maybe, we can continue to discuss that outside of this patch set? Renaming
a branch later should be no trouble.
An option could be to rename 5 -> 5-freebsd-master, then the mapping
between rtems and FreeBSD would always be explicit.

Yes this is one Joel mentioned the other day and I missed the context in
relation to these patches. I think moving 5 to 5-freebsd-master makes sense.
What we have is confusing.


+1

Would be good to get Christian or Sebastian to comment/agree :)


Yes, the branch name "5" makes no sense. It should be "5-freebsd-master".

Actually, I think we should delete this branch or is there someone willing to
maintain it?

If that is the case why have 5-freebsd-12? Why not make it 5 and then all the
repos are the same?
Ok, this is also an option. Delete the 5 branch and then rename 
5-freebsd-12 to 5.



The libbsd branch recommended for RTEMS 5 is 5-freebsd-12.

Do we maintain FreeBSD related fixes off the freebsd-12 release branch only?
From my point of view yes. This branch was used for the 5.1 release 
archive and it should be also used for the 5.2 release archive.

___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel