This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] Fix g++.eh/badalloc.C failure
- From: Richard Henderson <rth at redhat dot com>
- To: Jakub Jelinek <jakub at redhat dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Mon, 17 Dec 2001 12:58:57 -0800
- Subject: Re: [PATCH] Fix g++.eh/badalloc.C failure
- References: <20011217145933.H542@sunsite.ms.mff.cuni.cz>
On Mon, Dec 17, 2001 at 02:59:33PM +0100, Jakub Jelinek wrote:
> badalloc1.C test fails due to recent libstdc++-v3 locale changes, because
> the initial arena to serve malloc requests from is too small (e.g. on ia32
> the first malloc call is allocating a 640 bytes long chunk, there is also a
> malloc(1320) and about 3 other calls over 200 bytes, total > 3K).
> This is all before a C++ program hits main.
> Maybe it would be worth some analysis what could be avoided...
If I add a printf below the fail check, on Alpha I see
It's hard for me to believe we need only 3k on ix86...
Not sure what to do here, since allocating 24k isn't going
to be feasable on embedded targets, and this 24k is really
only needed when linking against shared libstdc++.