2009-11-18 20:02:14

by Geert Uytterhoeven

[permalink] [raw]
Subject: TIF_NOTIFY_RESUME on m68k?

Has anyone already looked into TIF_NOTIFY_RESUME support for m68k?

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=d0420c83f39f79afb82010c2d2cafd150eef651b
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=ee18d64c1f632043a02e6f5ba5e045bb26a5465f

These commits broke CONFIG_KEYS=y on m68k, so technically the missing
support is a regression compared to 2.6.31.

Thanks!

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds


2009-11-18 23:25:38

by David Howells

[permalink] [raw]
Subject: Re: [uClinux-dev] TIF_NOTIFY_RESUME on m68k?

Geert Uytterhoeven <[email protected]> wrote:

> These commits broke CONFIG_KEYS=y on m68k, so technically the missing
> support is a regression compared to 2.6.31.

I couldn't work out how to do it. I mentioned that it needed doing in the
logs in the hope that someone on the arch list would do it for the missing
arches.

David

2009-11-19 16:21:39

by Mikael Pettersson

[permalink] [raw]
Subject: Re: TIF_NOTIFY_RESUME on m68k?

Geert Uytterhoeven writes:
> Has anyone already looked into TIF_NOTIFY_RESUME support for m68k?

You need to change your entry.S code where it now does single-bit tests
for TIF_SIGPENDING with lsl/jmi do_signal_return, to instead test a 2-bit
(or larger, it that gives better code) flag field for zero/non-zero.
TIF_SIGPENDING and TIF_NOTIFY_RESUME should be alone in that field.
In the non-zero case you should call a new do_notify_resume() written in C,
which will then test TIF_SIGPENDING and TIF_NOTIFY_RESUME just like many
other archs do.

BTW, m68k also lacks TIF_RESTORE_SIGMASK support. You could probably use
the one I added to ARM very recently as a model.

Do current 2.6.3x m68k kernels boot on aranym? Last time I tried aranym
it could only boot som ancient hacked 2.4 kernel, for 2.6 kernels I got
disk detection errors due to what looked like byteswapping issues.

2009-11-19 16:34:22

by Andreas Schwab

[permalink] [raw]
Subject: Re: TIF_NOTIFY_RESUME on m68k?

Mikael Pettersson <[email protected]> writes:

> Do current 2.6.3x m68k kernels boot on aranym?

2.6.31 works fine here (from the linux-m68k tree).

Andreas.

--
Andreas Schwab, [email protected]
GPG Key fingerprint = D4E8 DBE3 3813 BB5D FA84 5EC7 45C6 250E 6F00 984E
"And now for something completely different."

2009-11-22 07:05:28

by Michael Schmitz

[permalink] [raw]
Subject: Re: TIF_NOTIFY_RESUME on m68k?

Hi,

> Do current 2.6.3x m68k kernels boot on aranym? Last time I tried aranym
> it could only boot som ancient hacked 2.4 kernel, for 2.6 kernels I got
> disk detection errors due to what looked like byteswapping issues.

Your disk image may need byte swapping (there should be an ARAnyM option for
this). 2.4 probably still could byte-swap IDE disks on Atari, 2.6 cannot.

Michael