[C++ PATCH] Fix 14179 - memory-hog

["Giovanni Bajo" <giovannibajo at libero dot it>]

Hello,

we were doing something quite silly in reshape_init for arrays: using
INTEGER_CST to track indices for the sole purposes of counting up from zero to
the declared size of the array. For an array of 8 million of elements like that
in c++/14179, it means that we allocate approximately 8 million of totally
useless trees that we throw away the first time GC collects.

This patch fixes it, by using a couple of unsigned HOST_WIDE_INTEGER (low+high)
instead. This mimics INTEGER_CSTs but without building trees every time. I
guess we cannot do it with a single HWI for cases like crosscompiling a 32bit
application with arrays with more than 65536 elements on a 16bit host, so this
is why I am using a couple of them (just like INTEGER_CST).

While I was at it, I also noticed that designated initializers larger than the
array size were not diagnosed. I know the extension is pretty much borked, but
it was an easy fix so I put it in.

With a reduced testcases with half million elements, we save approximately
13Mb. On the full testcase, the projected gain is around 200Mb. Together with
Nathan's cache for INTEGER_CSTs, I think this solves the memory-hog on
mainline. I will update the PR with updated measurements when/if this patch
goes in.

Bootstrapped on i686-pc-linux-gnu, OK for mainline?

OK for 3.4 and 3.3 too? The patch does cure a regression, and it looks pretty
safe to me. The diff looks more complex than it needs because of some
refactoring and extracting the code into a new function -- if needed, I can
prepare a stripped down version with minimal intrusiveness (assuming it's an
English word...).

Giovanni Bajo


cp/
        PR c++/14179
        * decl.c (reshape_init): Extract array handling into...
        (reshape_init_array): New function. Use integers instead of trees
        for indices. Handle out-of-range designated initializers.

Attachment: patch-pr14179-2.txt
Description: Text document