Re: [PATCH 1/2] setup_arg_pages can insert overlapping vma

[Hugh Dickins]

On Wed, 24 Nov 2004, Martin Schwidefsky wrote:
> 
> In principle the patch is fine (it works and fixes a problem).
> 
> I tried to find out why s390 needs its own setup_arg_pages32 function.
> After I've compared the function with the common setup_arg_pages several
> times and still couldn't find a reason for it I just removed the function.
> Still works fine. I think the reason to introduce setup_arg_pages32 has
> been the STACK_TOP define which needs to reflect the smaller address
> space for 31 bit programs. Since the change that made TASK_SIZE look
> at the TIF_31BIT bit to return the correct value for 31 and 64 bit
> programs STACK_TOP is just correct as well. 
> 
> In short, just remove setup_arg_pages32.

Tell me I'm being silly, Martin, but wouldn't it be wiser to stick with
setup_arg_pages32 (with Nanhai's patch) for 2.6.10, then remove it after?

I'm cautious here because about six months ago I sent Andrew a patch
which eliminated setup_arg_pages32 (or equiv) from the three arches,
but the x86_64 one just wouldn't boot on Andrew's machine.  Both hch
and I spent hours staring at the code, neither of us could work out why.

Now, s390 may be greatly superior ;) but all the setup_arg_pages32
redefining is twisty weird stuff - good to be rid of it, but right now?

Hugh

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/