Re: [PATCH 3/3] x86: signal: unify signal_{32|64}.c

From: Ingo Molnar
Date: Tue Nov 25 2008 - 23:15:20 EST



* Hiroshi Shimamoto <h-shimamoto@xxxxxxxxxxxxx> wrote:

> From: Hiroshi Shimamoto <h-shimamoto@xxxxxxxxxxxxx>
>
> Impact: cleanup
>
> Add #ifdef directive for 32-bit only code and unify signal_{32|64}.c
>
> Signed-off-by: Hiroshi Shimamoto <h-shimamoto@xxxxxxxxxxxxx>
> ---
> arch/x86/kernel/Makefile | 2 +-
> arch/x86/kernel/signal.c | 915 +++++++++++++++++++++++++++++++++++++++++++
> arch/x86/kernel/signal_32.c | 909 ------------------------------------------
> arch/x86/kernel/signal_64.c | 801 -------------------------------------
> 4 files changed, 916 insertions(+), 1711 deletions(-)
> create mode 100644 arch/x86/kernel/signal.c
> delete mode 100644 arch/x86/kernel/signal_32.c
> delete mode 100644 arch/x86/kernel/signal_64.c

yay! :-)

applied to tip/x86/signal, thanks Hiroshi!

One small detail, i inserted yet another commit in the middle:

5ceb40d: x86: signal: unify signal_{32|64}.c
e5fa2d0: x86: signal: unify signal_{32|64}.c, prepare
bfeb91a: x86: signal: cosmetic unification of __setup_sigframe() and __setup_rt_sigframe()
2601657: x86: signal: move {setup|restore}_sigcontext()

the e5fa2d0 commit (see it below) contains all the non-mechanic
modifications that you did in the final unification commit. It's best
to keep this separate - because a big unification commit is hard to
review and debug, should it cause any problems.

the 5ceb40d commit can thus do pure unification:

arch/x86/kernel/signal.c | 915 +++++++++++++++++++++++++++++++++++++++++++
arch/x86/kernel/signal_32.c | 915 -------------------------------------------
arch/x86/kernel/signal_64.c | 915 -------------------------------------------

... and the end result is still the same as with your series.

Thanks,

Ingo

------------------>