This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] Add Prefer_MAP_32BIT_EXEC for Silvermont
- From: Florian Weimer <fweimer at redhat dot com>
- To: Adhemerval Zanella <adhemerval dot zanella at linaro dot org>
- Cc: "H.J. Lu" <hjl dot tools at gmail dot com>, GNU C Library <libc-alpha at sourceware dot org>
- Date: Sat, 12 Dec 2015 15:50:32 +0100
- Subject: Re: [PATCH] Add Prefer_MAP_32BIT_EXEC for Silvermont
- Authentication-results: sourceware.org; auth=none
- References: <20151211143706 dot GA7868 at intel dot com> <alpine dot DEB dot 2 dot 10 dot 1512111539300 dot 17023 at digraph dot polyomino dot org dot uk> <CAMe9rOqbqyFw3CMa35vwOEefdFq1xK2Q9hX8GXoGMKVZ-A2y0g at mail dot gmail dot com> <566AF894 dot 4060300 at linaro dot org> <CAMe9rOr-LypZXvq4Y4uwE_JybYoTXctZXMLjo4TH517NnC6omg at mail dot gmail dot com> <566B01BE dot 1070703 at linaro dot org>
On 12/11/2015 06:02 PM, Adhemerval Zanella wrote:
> Another issue is this is basically limiting ALSR really hard on x86_64.
Looking at my Fedora 22 system, it seems that most (all?) processes map
their DSOs into a single 31-bit window. I don't know why it's
implemented this way and if it's also about silicon limitations.
Non-relocatable and PIE executables and the vDSO are stilled mapped
outside of this window, though. A conservative fix would change that
aspect only.
I'm worried that mapping a lot of things into the first 2 GiB of address
space breaks application expectations. LuaJIT applications might be
restricted to smaller-sized heaps, and Hotspot might not be able to use
unscaled compressed oops, resulting in a performance loss.
Florian