[15/15] ARM: mark footbridge as deprecated

Commit Message

Arnd Bergmann Dec. 4, 2024, 10:29 a.m. UTC

From: Arnd Bergmann <[e-mail]>

Along with RiscPC and SA1100, these are the last remaining Intel StrongARM
machines. The Corel NetWinder used to be particular popular in the late
1990s, but was discontinued during the bankruptcy of rebel.com in 2001.
The other machine is the DEC (later Intel) EBSA285 evaluation board that
was made in small numbers in 1997 for software developers.

The footbridge/netwinder platform was the main target for the first Debian
2.0 "Hamm" release on the Arm architecture back in 1998, but was dropped
in Debian 6.0 "Squeeze" in 2011, which only supported ARMv4T and higher
with the EABI based ports as ARMv4 hardware had fallen already out of
use by that time.

Link: http://netwinder.org/
Cc: Linus Walleij <[e-mail]>
Cc: Russell King <[e-mail]>
Cc: Ralph Siemsen <[e-mail]>
Signed-off-by: Arnd Bergmann <[e-mail]>
---
 arch/arm/mach-footbridge/Kconfig | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

Comments

Linus Walleij Dec. 4, 2024, 2:29 p.m. UTC | #1

On Wed, Dec 4, 2024 at 11:30 AM Arnd Bergmann <[e-mail]> wrote:

> From: Arnd Bergmann <[e-mail]>
>
> Along with RiscPC and SA1100, these are the last remaining Intel StrongARM
> machines. The Corel NetWinder used to be particular popular in the late
> 1990s, but was discontinued during the bankruptcy of rebel.com in 2001.
> The other machine is the DEC (later Intel) EBSA285 evaluation board that
> was made in small numbers in 1997 for software developers.

IIRC David Rusling at DEC was sending this board out to interested
developers.

> The footbridge/netwinder platform was the main target for the first Debian
> 2.0 "Hamm" release on the Arm architecture back in 1998, but was dropped
> in Debian 6.0 "Squeeze" in 2011, which only supported ARMv4T and higher
> with the EABI based ports as ARMv4 hardware had fallen already out of
> use by that time.
>
> Link: http://netwinder.org/
> Cc: Linus Walleij <[e-mail]>
> Cc: Russell King <[e-mail]>
> Cc: Ralph Siemsen <[e-mail]>
> Signed-off-by: Arnd Bergmann <[e-mail]>

I am booting it occasionally, last time to test my KCFI patches.
But admittedly that is only to test it for other SA110 users. I remember
that Christoph had problems consolidating the DMA used in this
machine as well so it is standing in the way of useful work.

Apart from Ralph and me I know Marc Z has been known to boot
this machine, so paging him as well.

My machine is using an OABI RedHat MontaVista Hard Hat
derivative but the default userspace doesn't even have a /sys
directory so it's quite dated. I have a minimal homebrew
busybox userspace that I actually use for tests, admittedly
built with an OpenWrt compiler using the --fix-v4bx trick.

So from my point of view this system is on life support:
Acked-by: Linus Walleij <[e-mail]>

Yours,
Linus Walleij

Russell King (Oracle) Dec. 4, 2024, 2:37 p.m. UTC | #2

On Wed, Dec 04, 2024 at 03:29:09PM +0100, Linus Walleij wrote:
> On Wed, Dec 4, 2024 at 11:30 AM Arnd Bergmann <[e-mail]> wrote:
> 
> > From: Arnd Bergmann <[e-mail]>
> >
> > Along with RiscPC and SA1100, these are the last remaining Intel StrongARM
> > machines. The Corel NetWinder used to be particular popular in the late
> > 1990s, but was discontinued during the bankruptcy of rebel.com in 2001.
> > The other machine is the DEC (later Intel) EBSA285 evaluation board that
> > was made in small numbers in 1997 for software developers.
> 
> IIRC David Rusling at DEC was sending this board out to interested
> developers.
> 
> > The footbridge/netwinder platform was the main target for the first Debian
> > 2.0 "Hamm" release on the Arm architecture back in 1998, but was dropped
> > in Debian 6.0 "Squeeze" in 2011, which only supported ARMv4T and higher
> > with the EABI based ports as ARMv4 hardware had fallen already out of
> > use by that time.
> >
> > Link: http://netwinder.org/
> > Cc: Linus Walleij <[e-mail]>
> > Cc: Russell King <[e-mail]>
> > Cc: Ralph Siemsen <[e-mail]>
> > Signed-off-by: Arnd Bergmann <[e-mail]>
> 
> I am booting it occasionally, last time to test my KCFI patches.
> But admittedly that is only to test it for other SA110 users. I remember
> that Christoph had problems consolidating the DMA used in this
> machine as well so it is standing in the way of useful work.

What problems with DMA? The fact that it uses an offset between
physical and PCI bus space should be nothing that causes a problem
for the kernel. As maintainer for Footbridge, I've heard nothing
about there being any issue.

Marc Zyngier Dec. 4, 2024, 2:44 p.m. UTC | #3

On Wed, 04 Dec 2024 14:29:09 +0000,
Linus Walleij <[e-mail]> wrote:
> 
> On Wed, Dec 4, 2024 at 11:30 AM Arnd Bergmann <[e-mail]> wrote:
> 
> > From: Arnd Bergmann <[e-mail]>
> >
> > Along with RiscPC and SA1100, these are the last remaining Intel StrongARM
> > machines. The Corel NetWinder used to be particular popular in the late
> > 1990s, but was discontinued during the bankruptcy of rebel.com in 2001.
> > The other machine is the DEC (later Intel) EBSA285 evaluation board that
> > was made in small numbers in 1997 for software developers.
> 
> IIRC David Rusling at DEC was sending this board out to interested
> developers.
> 
> > The footbridge/netwinder platform was the main target for the first Debian
> > 2.0 "Hamm" release on the Arm architecture back in 1998, but was dropped
> > in Debian 6.0 "Squeeze" in 2011, which only supported ARMv4T and higher
> > with the EABI based ports as ARMv4 hardware had fallen already out of
> > use by that time.
> >
> > Link: http://netwinder.org/
> > Cc: Linus Walleij <[e-mail]>
> > Cc: Russell King <[e-mail]>
> > Cc: Ralph Siemsen <[e-mail]>
> > Signed-off-by: Arnd Bergmann <[e-mail]>
> 
> I am booting it occasionally, last time to test my KCFI patches.
> But admittedly that is only to test it for other SA110 users. I remember
> that Christoph had problems consolidating the DMA used in this
> machine as well so it is standing in the way of useful work.
> 
> Apart from Ralph and me I know Marc Z has been known to boot
> this machine, so paging him as well.

I won't be crying. It's probably been a couple of years since I last
switched it on, and it is IMO ready to join the DEC Shark on the shelf
of "used to be great" HW.

FWIW,

Acked-by: Marc Zyngier <[e-mail]>

	M.

Linus Walleij Dec. 4, 2024, 3:14 p.m. UTC | #4

On Wed, Dec 4, 2024 at 3:37 PM Russell King (Oracle)
<[e-mail]> wrote:

> > I am booting it occasionally, last time to test my KCFI patches.
> > But admittedly that is only to test it for other SA110 users. I remember
> > that Christoph had problems consolidating the DMA used in this
> > machine as well so it is standing in the way of useful work.
>
> What problems with DMA? The fact that it uses an offset between
> physical and PCI bus space should be nothing that causes a problem
> for the kernel. As maintainer for Footbridge, I've heard nothing
> about there being any issue.

No issue other than testing and iterating the patches I think.

I was thinking of this patchset to do dma_direct_set_offset():
https://lore.kernel.org/linux-arm-kernel/[e-mail]/

It seems to have stalled some years ago, but maybe there was
some other reason.

IIRC Christoph was onto me to test it on the NetWinder, but I
couldn't locate a good DMA capable device to test it with.
(Would be the soundblaster but then I needed some userspace...)

Yours,
Linus Walleij

Russell King (Oracle) Dec. 4, 2024, 3:30 p.m. UTC | #5

On Wed, Dec 04, 2024 at 04:14:07PM +0100, Linus Walleij wrote:
> On Wed, Dec 4, 2024 at 3:37 PM Russell King (Oracle)
> <[e-mail]> wrote:
> 
> > > I am booting it occasionally, last time to test my KCFI patches.
> > > But admittedly that is only to test it for other SA110 users. I remember
> > > that Christoph had problems consolidating the DMA used in this
> > > machine as well so it is standing in the way of useful work.
> >
> > What problems with DMA? The fact that it uses an offset between
> > physical and PCI bus space should be nothing that causes a problem
> > for the kernel. As maintainer for Footbridge, I've heard nothing
> > about there being any issue.
> 
> No issue other than testing and iterating the patches I think.
> 
> I was thinking of this patchset to do dma_direct_set_offset():
> https://lore.kernel.org/linux-arm-kernel/[e-mail]/
> 
> It seems to have stalled some years ago, but maybe there was
> some other reason.
> 
> IIRC Christoph was onto me to test it on the NetWinder, but I
> couldn't locate a good DMA capable device to test it with.
> (Would be the soundblaster but then I needed some userspace...)

It isn't soundblaster, but waveartist, and needs special setup to be
functional (which was all in my OSS waveartist driver.)

Patch

diff --git a/arch/arm/mach-footbridge/Kconfig b/arch/arm/mach-footbridge/Kconfig
index 78189997caa1..1ac37cadb201 100644
--- a/arch/arm/mach-footbridge/Kconfig
+++ b/arch/arm/mach-footbridge/Kconfig
@@ -1,6 +1,6 @@ 
 # SPDX-License-Identifier: GPL-2.0-only
 menuconfig ARCH_FOOTBRIDGE
-	bool "FootBridge Implementations"
+	bool "FootBridge Implementations (DEPRECATED)"
 	depends on ARCH_MULTI_V4 && !(ARCH_MULTI_V4T || ARCH_MULTI_V5)
 	depends on !(ARCH_MOXART || ARCH_GEMINI || ARCH_SA1100)
 	depends on ATAGS
@@ -12,7 +12,10 @@  menuconfig ARCH_FOOTBRIDGE
 	select NEED_MACH_MEMORY_H
 	help
 	  Support for systems based on the DC21285 companion chip
-	  ("FootBridge"), such as the Simtec CATS and the Rebel NetWinder.
+	  ("FootBridge"), used in the EBSA285 and the Rebel NetWinder.
+
+	  Support for these machines will go away in 2025,
+	  unless there are any remaining users that speak up.
 
 if ARCH_FOOTBRIDGE