Re: recalc_sigpending() / recalc_sigpending_tsk() ?

[Linus Torvalds]

On Thu, 28 Feb 2002, David Woodhouse wrote:
>
> It seems that the name of the function was changed to recalc_sigpending_tsk()
> and a new function called recalc_sigpending() was added.

Yes.

> Was there a reason for doing this, rather than just introducing the new
> function with a different name, such as recalc_sigpending_cur()? It breaks
> 2.4 source compatibility in a way that seems entirely gratuitous.

I don't care. I care 1000% more about clean code than about backwards
source level compatibility.

99% of all users did it for the current task, and for the current task
only. And the non-local users were all in low-level core code (and should
_not_ exist anywhere else - if some driver is playing around with another
tasks signal state, that driver is so incredibly fundamentally broken that
I don't even want to hear about it)

In short, the 2.5.x interface is the correct one.

> Before I have to go and do something evil in my compatmac.h to work round
> this, is there any chance of putting the original recalc_sigpending() back?

Not a chance in hell. The backwards compatibility looks like a trivial
one-liner:

   compat-2.4.h:
	#define recalc_sigpending() recalc_sigpending(current)

so what are you complaining about?

		Linus