linux-kernel - Re: avoiding duplicate icache flushing of shared maps on nommu

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Sat, 28 Nov 2009 13:53:35 -0500
From:	Mike Frysinger <vapier.adi@...il.com>
To:	David Howells <dhowells@...hat.com>,
	Paul Mundt <lethal@...ux-sh.org>,
	Bernd Schmidt <bernds_cb1@...nline.de>,
	Jie Zhang <jie.zhang@...log.com>
Cc:	Linux kernel mailing list <linux-kernel@...r.kernel.org>,
	uclinux-dist-devel <uclinux-dist-devel@...ckfin.uclinux.org>
Subject: Re: avoiding duplicate icache flushing of shared maps on nommu

On Sat, Nov 28, 2009 at 10:40, Mike Frysinger wrote:
> when working with FDPIC, there are many shared maps of read only text
> regions (the C library, applet packages like busybox, ...) between
> applications.  but the current mm/nommu.c:do_mmap_pgoff() function
> will issue an icache flush whenever a vma is added to a mm instead of
> only doing it when the map is initially created.  am i missing
> something obvious here, or would a change like below be OK ?  this
> easily cuts the number of icache flushes during boot by 50% if not
> more.

for some actual numbers, my default boot iflushes 18,562,124 bytes.
with this fix, it's down to 4,528,580 bytes (a 75% shrink).  with the
stack fix, it's down to 989,636 bytes (a 95% shrink).
-mike
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/