RFC: Aligning data in buffers with a macro?

[David Miller]

From: "Carlos O'Donell" <carlos@systemhalted.org>
Date: Sun, 1 Apr 2012 15:12:48 -0400

> On Sun, Apr 1, 2012 at 12:30 PM, Marek Polacek <polacek@redhat.com> wrote:
>>>  628 # elif _STACK_GROWS_UP
>>>  629           char *guard = (char *) ALIGN_DOWN((uintptr_t) pd -
>>> guardsize, pagesize_m1);
>>>  630 #endif
>>
>> To be honest, that doesn't make much difference to me.  The second
>> variant is not shorter nor more readable I suspect.  Also I have
>> a qualm about these changes--they might be quite error-prone.
> 
> It is my opinion that using a macro with a clear purpose makes the
> code more understandable e.g. Are you aligning up, or down, and to
> what alignment?

I completely agree.

> I do *not* propose we go through and fixup everything.

However anyone who chooses to partake in such a cleanup should be
rewarded with punctual patch review from the rest of us.

GLIBC is less maintainable because of those by-hand alignment
expressions, I can't tell you how many times I had to peer over
the nptl stack alignment code to figure out what exactly it was
doing.

> I would rather learn to read through _ALIGND and _ALIGNU than have to
> do the masking operations mentally to figure out what the author
> intended.

Indeed.