Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #188719 > unrolled thread

Re: Anyone using stretch/buster/sid on ARMv4t ?

Back to article view | Back to linux.debian.user

Contents

  Re: Anyone using stretch/buster/sid on ARMv4t ? Arnd Bergmann <arnd@arndb.de> - 2017-11-07 23:00 +0100
    Re: Anyone using stretch/buster/sid on ARMv4t ? Aaro Koskinen <aaro.koskinen@iki.fi> - 2017-11-08 00:20 +0100
      Re: Anyone using stretch/buster/sid on ARMv4t ? Arnd Bergmann <arnd@arndb.de> - 2017-11-08 00:30 +0100
      Re: Anyone using stretch/buster/sid on ARMv4t ? Adrian Bunk <bunk@debian.org> - 2017-11-08 21:40 +0100
        Re: Anyone using stretch/buster/sid on ARMv4t ? Arnd Bergmann <arnd@arndb.de> - 2017-11-08 23:20 +0100
          Re: Anyone using stretch/buster/sid on ARMv4t ? Adrian Bunk <bunk@debian.org> - 2017-11-10 22:50 +0100

#188719 — Re: Anyone using stretch/buster/sid on ARMv4t ?

To clarify which platforms are affected by moving from ARMv4T to ARMv5TE,
here is what the kernel supports in principle for people that just use the
debian packages but build their own kernels:

ARMv4 (prior to ARMv4t) machines are:

- StrongARM 110 based: RiscPC, EBSA110, Footbridge/NetWinder: these all
  have a minuscule user base, if any
- StrongARM 1100: used in many old PocketPC-like machines
  (Jornada, Zaururs, iPAQ, ...) and a couple of other machines that some
  people still use in production today
- FA526 based: moxart and gemini, both embedded SoCs that are
  still maintained and used to some degree

ARMv4T machines supported in Linux:

- Cirrus Logic EP721x/EP731x is based on ARM720T, the only
  person that was trying to improve support for this has apparently
  moved on to slightly newer hardware
- Atmel AT91RM9200 is still being sold new by Microchip, but all
  AT91 variants that people use in production are the ARM926 (ARMv5E)
  based one/
- TI OMAP15xx was based on ARM925, we support exactly one machine
  with it (HTC Herald phone, a.k.a. T-Mobile Wing), I'd call it dead
- Freescale i.MX1 used an ARM920T, we support two boards, the
  official reference board and the Armadeus APF9328. This is equally
  dead as far as I can tell, and it has only 8MB RAM.
- Micrel/Kendin KS8695 is used in several wireless routers, and based
  on an ARM922T. Not a bad chip, but hasn't seen any work in a while,
  and all users are probably on something like OpenWRT/LEDE due to
  memory limitations, see https://wikidevi.com/wiki/Micrel
  Apparently owned by Microchip and still being sold but not recommended
  for new designs. Who knew?
- Many Samsung S3C24xx chips use an ARM920T: S3C2410,
  S3C2440, S3C2442 and S3C2443, but not S3C2412 and S3C2416
  and S3C245. These were very popular for all kinds of machines
  even a few years ago, and probably many of them running Debian.
  This includes the Openmoko GTA02, and various embedded boards,
  but interest has declined a lot after Samsung discontinued their
  ARM9 based embedded SoCs a few years ago.

ARMv5/ARMv5T (prior to ARMv5TE) based:
- We have some code for the original ARM1020 processor, but don't support
  any machines with it, if they ever existed outside of the lab.

ARMv5TE based:
- All ARM926 based machines: around 20 different SoC families from as
  many companies. This is still fairly popular in many embedded designs
  (Atmel AT91, TI DaVinci, Aspeed, ...), even more so than the newer ARMv6
  based implementations.
- Marvell PJ1 and related: kirkwood, orion, mv78xx0, mmp: some of these,
  in particular kirkwood are still used a lot on older machines like plug
  computers. The chips have been discontinued.
- Intel XScale: ixp4xx, iop, older pxa: The PXA family still has some life in it
  with hobbyists, the rest are rather obsolete these days. These chips have
  been discontinued for even longer.

        Arnd

[toc] | [next] | [standalone]

#188728

Hi,

On Tue, Nov 07, 2017 at 10:36:57PM +0100, Arnd Bergmann wrote:
> - TI OMAP15xx was based on ARM925, we support exactly one machine
>   with it (HTC Herald phone, a.k.a. T-Mobile Wing), I'd call it dead

Check again, Herald is OMAP850. OMAP15xx is frequently tested and used
by OMAP1 hackers (thanks to Amstrad Delta).

A.

[toc] | [prev] | [next] | [standalone]

#188731

On Tue, Nov 7, 2017 at 11:52 PM, Aaro Koskinen <aaro.koskinen@iki.fi> wrote:
> Hi,
>
> On Tue, Nov 07, 2017 at 10:36:57PM +0100, Arnd Bergmann wrote:
>> - TI OMAP15xx was based on ARM925, we support exactly one machine
>>   with it (HTC Herald phone, a.k.a. T-Mobile Wing), I'd call it dead
>
> Check again, Herald is OMAP850. OMAP15xx is frequently tested and used
> by OMAP1 hackers (thanks to Amstrad Delta).

Ah right, it did look odd when I looked it up, apparently I got the
wrong Kconfig
symbol at some point and wasn't overly familiar with the family to know better.

OMAP15xx is indeed used on half the OMAP1 machines we support.

      Arnd

[toc] | [prev] | [next] | [standalone]

#188773

On Wed, Nov 08, 2017 at 12:52:05AM +0200, Aaro Koskinen wrote:
>
>... OMAP15xx is frequently tested and used
> by OMAP1 hackers (thanks to Amstrad Delta).

But is anyone still using Debian on OMAP15xx?

Looking at the amount of RAM in the Amstrad Delta,
I'd be surprised if anyone is actually using
Debian on that hardware.
 
> A.

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

[toc] | [prev] | [next] | [standalone]

#188784

On Wed, Nov 8, 2017 at 9:18 PM, Adrian Bunk <bunk@debian.org> wrote:
> On Wed, Nov 08, 2017 at 12:52:05AM +0200, Aaro Koskinen wrote:
>>
>>... OMAP15xx is frequently tested and used
>> by OMAP1 hackers (thanks to Amstrad Delta).
>
> But is anyone still using Debian on OMAP15xx?
>
> Looking at the amount of RAM in the Amstrad Delta,
> I'd be surprised if anyone is actually using
> Debian on that hardware.

It's 128MB, right? How unrealistic is that these days?

All the ARMv4T chips I checked (at91rm9200, imx1, ep73xx, ep93xx,
ks8695, omap1510 and s3c24xx) with the exception of the s3c2443
(which we barely support) only support SDRAM, not DDR, so they
are already limited to 64MB (512Mbit) per memory chip, and typically
256MB as the absolute maximum on a 32-bit interface with four x8
chips.

       Arnd

[toc] | [prev] | [next] | [standalone]

#188866

On Wed, Nov 08, 2017 at 10:56:03PM +0100, Arnd Bergmann wrote:
> On Wed, Nov 8, 2017 at 9:18 PM, Adrian Bunk <bunk@debian.org> wrote:
> > On Wed, Nov 08, 2017 at 12:52:05AM +0200, Aaro Koskinen wrote:
> >>
> >>... OMAP15xx is frequently tested and used
> >> by OMAP1 hackers (thanks to Amstrad Delta).
> >
> > But is anyone still using Debian on OMAP15xx?
> >
> > Looking at the amount of RAM in the Amstrad Delta,
> > I'd be surprised if anyone is actually using
> > Debian on that hardware.
> 
> It's 128MB, right?

AFAIK 32MB (but I could be wrong on that).

> How unrealistic is that these days?

The stretch installation manual says at least 31MB RAM [1],
but since it says the same for jessie says the same it is
possible that noone ever tried stretch with that few RAM.

> All the ARMv4T chips I checked (at91rm9200, imx1, ep73xx, ep93xx,
> ks8695, omap1510 and s3c24xx) with the exception of the s3c2443
> (which we barely support) only support SDRAM, not DDR, so they
> are already limited to 64MB (512Mbit) per memory chip, and typically
> 256MB as the absolute maximum on a 32-bit interface with four x8
> chips.

Neo FreeRunner with 128MB RAM would still be fine if there would
be a working recent kernel and people actually running current
Debian on it.

So far I haven't heard from a single person running stretch/buster/sid 
on ARMv4t, and I'm becoming confident that there are no users left.[2]

>        Arnd

cu
Adrian

[1] https://www.debian.org/releases/stable/armel/ch02s05.html.en
[2] there might still be people running Linux on such hardware and
    it might be possible to get Debian running if someone wants to,
    but actual users who would also report breakages seem gone

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.user

csiph-web