is alpha jensen support dead?

is alpha jensen support dead?

[Christoph Hellwig]

Hi all,

it seems like the Alpha Jense build (the only one using pci-noop.c)
and thus being a different build than all the later PCI capable
system has been broken since at least:

commit 6aca0503847f6329460b15b3ab2b0e30bb752793
Author: Christian Borntraeger <borntraeger@de.ibm.com>
Date:   Tue Feb 2 21:46:33 2016 -0800

    alpha/dma: use common noop dma ops

which switches pci-noop.c to use generic code, but fat fingered a symbol
and didn't wire up the Kconfig.

Is there any value in keeping it alive?  Especially as there probably
isn't any build coverage..

Btw, how well is alpha working these days?  It looks like there hasn't
been any maintainer activity for about two years.

Re: is alpha jensen support dead?

[Bob Tracy]

On Sun, May 21, 2017 at 10:58:48AM +0200, Christoph Hellwig wrote:
> Btw, how well is alpha working these days?  It looks like there hasn't
> been any maintainer activity for about two years.

Speaking for myself, the PWS 433au I've got continues to function
admirably as my IPv6 gateway.  Debian unstable and experimental updates
continue to apply and work within the limits of my ability to test them.
I'm building and running the release candidates and final versions of
the kernel.org tree with the unstable tool chain.  Recently there was a
FTBFS due to relocation issues during the final vmlinux link on alpha:
that has been fixed, and what's left of the alpha community is pretty
quick to respond and identify what's going on.

In summary, alpha is still a viable platform for *me*.

--Bob

Re: is alpha jensen support dead?

[Bob Tracy]

On Sun, May 21, 2017 at 10:58:48AM +0200, Christoph Hellwig wrote:
> Btw, how well is alpha working these days?  It looks like there hasn't
> been any maintainer activity for about two years.

Speaking for myself, the PWS 433au I've got continues to function
admirably as my IPv6 gateway.  Debian unstable and experimental updates
continue to apply and work within the limits of my ability to test them.
I'm building and running the release candidates and final versions of
the kernel.org tree with the unstable tool chain.  Recently there was a
FTBFS due to relocation issues during the final vmlinux link on alpha:
that has been fixed, and what's left of the alpha community is pretty
quick to respond and identify what's going on.

In summary, alpha is still a viable platform for *me*.

--Bob

Re: is alpha jensen support dead?

[Christoph Hellwig]

On Sun, May 21, 2017 at 01:11:28PM -0500, Bob Tracy wrote:
> Speaking for myself, the PWS 433au I've got continues to function
> admirably as my IPv6 gateway.  Debian unstable and experimental updates
> continue to apply and work within the limits of my ability to test them.
> I'm building and running the release candidates and final versions of
> the kernel.org tree with the unstable tool chain.  Recently there was a
> FTBFS due to relocation issues during the final vmlinux link on alpha:
> that has been fixed, and what's left of the alpha community is pretty
> quick to respond and identify what's going on.
> 
> In summary, alpha is still a viable platform for *me*.

Great!  Still makes me wonder about Jensen, which appears to be
the only platform that requires a different build than the normal
generic PCI-enabled kernel.  And I suspect it's pretty much dead
at this point..

Re: is alpha jensen support dead?

[Maciej W. Rozycki]

On Mon, 22 May 2017, Christoph Hellwig wrote:

> Great!  Still makes me wonder about Jensen, which appears to be
> the only platform that requires a different build than the normal
> generic PCI-enabled kernel.  And I suspect it's pretty much dead
> at this point..

 Jensen is EISA-only, which is likely why support wasn't folded into the 
generic configuration.

  Maciej

Re: is alpha jensen support dead?

[Maciej W. Rozycki]

On Mon, 22 May 2017, Christoph Hellwig wrote:

> Great!  Still makes me wonder about Jensen, which appears to be
> the only platform that requires a different build than the normal
> generic PCI-enabled kernel.  And I suspect it's pretty much dead
> at this point..

 Jensen is EISA-only, which is likely why support wasn't folded into the 
generic configuration.

  Maciej

Re: is alpha jensen support dead?

[Matt Turner]

On Sun, May 21, 2017 at 1:58 AM, Christoph Hellwig <hch@lst.de> wrote:
> Hi all,
>
> it seems like the Alpha Jense build (the only one using pci-noop.c)
> and thus being a different build than all the later PCI capable
> system has been broken since at least:
>
> commit 6aca0503847f6329460b15b3ab2b0e30bb752793
> Author: Christian Borntraeger <borntraeger@de.ibm.com>
> Date:   Tue Feb 2 21:46:33 2016 -0800
>
>     alpha/dma: use common noop dma ops
>
> which switches pci-noop.c to use generic code, but fat fingered a symbol
> and didn't wire up the Kconfig.
>
> Is there any value in keeping it alive?  Especially as there probably
> isn't any build coverage..
>
> Btw, how well is alpha working these days?  It looks like there hasn't
> been any maintainer activity for about two years.

I haven't had time for alpha stuff in quite a while.

I've never even had a Jensen, so it's never been important to me personally.

Re: is alpha jensen support dead?

[Matt Turner]

On Sun, May 21, 2017 at 1:58 AM, Christoph Hellwig <hch@lst.de> wrote:
> Hi all,
>
> it seems like the Alpha Jense build (the only one using pci-noop.c)
> and thus being a different build than all the later PCI capable
> system has been broken since at least:
>
> commit 6aca0503847f6329460b15b3ab2b0e30bb752793
> Author: Christian Borntraeger <borntraeger@de.ibm.com>
> Date:   Tue Feb 2 21:46:33 2016 -0800
>
>     alpha/dma: use common noop dma ops
>
> which switches pci-noop.c to use generic code, but fat fingered a symbol
> and didn't wire up the Kconfig.
>
> Is there any value in keeping it alive?  Especially as there probably
> isn't any build coverage..
>
> Btw, how well is alpha working these days?  It looks like there hasn't
> been any maintainer activity for about two years.

I haven't had time for alpha stuff in quite a while.

I've never even had a Jensen, so it's never been important to me personally.

Re: is alpha jensen support dead?

[Michael Ellerman]

Christoph Hellwig <hch@lst.de> writes:

> Hi all,
>
> it seems like the Alpha Jense build (the only one using pci-noop.c)
> and thus being a different build than all the later PCI capable
> system has been broken since at least:
>
> commit 6aca0503847f6329460b15b3ab2b0e30bb752793
> Author: Christian Borntraeger <borntraeger@de.ibm.com>
> Date:   Tue Feb 2 21:46:33 2016 -0800
>
>     alpha/dma: use common noop dma ops
>
> which switches pci-noop.c to use generic code, but fat fingered a symbol
> and didn't wire up the Kconfig.
>
> Is there any value in keeping it alive?  Especially as there probably
> isn't any build coverage..

We still build the defconfig FWIW:

  http://kisskb.ellerman.id.au/kisskb/target/2499/
  http://kisskb.ellerman.id.au/kisskb/target/2494/

cheers

Re: is alpha jensen support dead?

[Michael Ellerman]

Christoph Hellwig <hch@lst.de> writes:

> Hi all,
>
> it seems like the Alpha Jense build (the only one using pci-noop.c)
> and thus being a different build than all the later PCI capable
> system has been broken since at least:
>
> commit 6aca0503847f6329460b15b3ab2b0e30bb752793
> Author: Christian Borntraeger <borntraeger@de.ibm.com>
> Date:   Tue Feb 2 21:46:33 2016 -0800
>
>     alpha/dma: use common noop dma ops
>
> which switches pci-noop.c to use generic code, but fat fingered a symbol
> and didn't wire up the Kconfig.
>
> Is there any value in keeping it alive?  Especially as there probably
> isn't any build coverage..

We still build the defconfig FWIW:

  http://kisskb.ellerman.id.au/kisskb/target/2499/
  http://kisskb.ellerman.id.au/kisskb/target/2494/

cheers

Re: is alpha jensen support dead?

[Christian Borntraeger]

On 05/21/2017 10:58 AM, Christoph Hellwig wrote:
> Hi all,
> 
> it seems like the Alpha Jense build (the only one using pci-noop.c)
> and thus being a different build than all the later PCI capable
> system has been broken since at least:
> 
> commit 6aca0503847f6329460b15b3ab2b0e30bb752793
> Author: Christian Borntraeger <borntraeger@de.ibm.com>
> Date:   Tue Feb 2 21:46:33 2016 -0800
> 
>     alpha/dma: use common noop dma ops
> 

Hi Christoph,

are you providing a fix (or a jensen removal) or do you expect a fix from me?


> which switches pci-noop.c to use generic code, but fat fingered a symbol
> and didn't wire up the Kconfig.
> 
> Is there any value in keeping it alive?  Especially as there probably
> isn't any build coverage..
> 
> Btw, how well is alpha working these days?  It looks like there hasn't
> been any maintainer activity for about two years.
> 

Re: is alpha jensen support dead?

[Christian Borntraeger]

On 05/21/2017 10:58 AM, Christoph Hellwig wrote:
> Hi all,
> 
> it seems like the Alpha Jense build (the only one using pci-noop.c)
> and thus being a different build than all the later PCI capable
> system has been broken since at least:
> 
> commit 6aca0503847f6329460b15b3ab2b0e30bb752793
> Author: Christian Borntraeger <borntraeger@de.ibm.com>
> Date:   Tue Feb 2 21:46:33 2016 -0800
> 
>     alpha/dma: use common noop dma ops
> 

Hi Christoph,

are you providing a fix (or a jensen removal) or do you expect a fix from me?


> which switches pci-noop.c to use generic code, but fat fingered a symbol
> and didn't wire up the Kconfig.
> 
> Is there any value in keeping it alive?  Especially as there probably
> isn't any build coverage..
> 
> Btw, how well is alpha working these days?  It looks like there hasn't
> been any maintainer activity for about two years.
> 

Re: is alpha jensen support dead?

[Christoph Hellwig]

On Wed, May 24, 2017 at 11:16:01AM +0200, Christian Borntraeger wrote:
> Hi Christoph,
> 
> are you providing a fix (or a jensen removal) or do you expect a fix from me?

Mostly just looking into the area for a series around dma_mapping_ops.

Without build coverage I don't really dare to touch jensen, but I'll see
if I can come up with a fix.