public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
From: "kcc at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org> To: gcc-bugs@gcc.gnu.org Subject: [Bug sanitizer/55309] gcc's address-sanitizer 66% slower than clang's Date: Tue, 12 Feb 2013 07:03:00 -0000 [thread overview] Message-ID: <bug-55309-4-vNIPyosuoe@http.gcc.gnu.org/bugzilla/> (raw) In-Reply-To: <bug-55309-4@http.gcc.gnu.org/bugzilla/> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55309 --- Comment #33 from Kostya Serebryany <kcc at gcc dot gnu.org> 2013-02-12 07:02:40 UTC --- (In reply to comment #31) > If the mapping is so flexible, how can you detect mismatches? Different scale > or shadow offsets are ABI incompatible... We don't detect mismatches. This has never been a problem for our users (who build everything from scratch) but we do see it as a coming problem as asan is becoming more popular. (in reply to comment from another bug) > Perhaps instead of global vars defined outside of libasan (which e.g. requires > GOT accesses to those vars in libasan) Accessing these vars was never a perf problem (we run asan with perf regularly) > , it might be better to have the scale > and offset as arguments of __asan_init? We did this in the very early version, but it did not work in general. Consider you are linking your program with a third-party object not built with asan. It may have constructor functions called before main and before __asan_init, and those functions call malloc which has to call __asan_init, but can not pass arguments. In some cases we can use .preinit_array to call __asan_init there, but that is not always available (?). We were (and still are) thinking about encoding the abi version in the name of the init function, e.g. __asan_init_v_123. It will help us detect abi mismatches when two objects are instrumented with different generations of asan. This doesn't solve the problem of using different offsets though. > Then you could easily test at runtime, > whether all compilation units agree on the same offset/scale, and complain if > they don't. Then __asan_mapping_offset and __asan_mapping_scale or how are the > vars called could be hidden attribute, used with PC relative addressing and > avoid one extra indirection, and more importantly have better runtime checking > of mismatches.
next prev parent reply other threads:[~2013-02-12 7:03 UTC|newest] Thread overview: 59+ messages / expand[flat|nested] mbox.gz Atom feed top 2012-11-13 10:01 [Bug other/55309] New: " markus at trippelsdorf dot de 2012-11-13 21:10 ` [Bug other/55309] " konstantin.s.serebryany at gmail dot com 2012-11-13 21:31 ` markus at trippelsdorf dot de 2012-11-14 7:52 ` jakub at gcc dot gnu.org 2012-11-14 16:38 ` jakub at gcc dot gnu.org 2012-11-14 17:03 ` markus at trippelsdorf dot de 2013-02-05 9:22 ` [Bug sanitizer/55309] " kcc at gcc dot gnu.org 2013-02-05 9:43 ` kcc at gcc dot gnu.org 2013-02-05 9:56 ` jakub at gcc dot gnu.org 2013-02-05 10:31 ` kcc at gcc dot gnu.org 2013-02-05 10:42 ` kcc at gcc dot gnu.org 2013-02-05 10:55 ` jakub at gcc dot gnu.org 2013-02-05 11:18 ` markus at trippelsdorf dot de 2013-02-05 11:24 ` jakub at gcc dot gnu.org 2013-02-05 11:26 ` jakub at gcc dot gnu.org 2013-02-05 12:23 ` kcc at gcc dot gnu.org 2013-02-06 10:56 ` dodji at gcc dot gnu.org 2013-02-06 11:19 ` kcc at gcc dot gnu.org 2013-02-06 12:25 ` kcc at gcc dot gnu.org 2013-02-06 12:39 ` rguenth at gcc dot gnu.org 2013-02-06 12:43 ` kcc at gcc dot gnu.org 2013-02-06 12:49 ` jakub at gcc dot gnu.org 2013-02-06 15:03 ` dodji at gcc dot gnu.org 2013-02-07 5:02 ` kcc at gcc dot gnu.org 2013-02-07 17:01 ` jakub at gcc dot gnu.org 2013-02-07 17:18 ` dvyukov at google dot com 2013-02-08 6:31 ` kcc at gcc dot gnu.org 2013-02-08 9:02 ` jakub at gcc dot gnu.org 2013-02-08 9:13 ` kcc at gcc dot gnu.org 2013-02-08 9:25 ` jakub at gcc dot gnu.org 2013-02-11 14:43 ` kcc at gcc dot gnu.org 2013-02-11 15:03 ` jakub at gcc dot gnu.org 2013-02-12 6:48 ` kcc at gcc dot gnu.org 2013-02-12 7:03 ` kcc at gcc dot gnu.org [this message] 2013-02-12 8:40 ` jakub at gcc dot gnu.org 2013-02-12 8:47 ` dvyukov at google dot com 2013-02-12 8:59 ` kcc at gcc dot gnu.org 2013-02-12 11:18 ` kcc at gcc dot gnu.org 2013-02-12 11:31 ` kcc at gcc dot gnu.org 2013-02-12 11:42 ` jakub at gcc dot gnu.org 2013-02-12 14:00 ` howarth at nitro dot med.uc.edu 2013-02-12 14:12 ` jakub at gcc dot gnu.org 2013-02-12 14:42 ` howarth at nitro dot med.uc.edu 2013-02-22 7:11 ` kcc at gcc dot gnu.org 2013-02-22 8:31 ` Joost.VandeVondele at mat dot ethz.ch 2013-02-22 8:36 ` kcc at gcc dot gnu.org 2013-02-22 13:09 ` jakub at gcc dot gnu.org 2013-02-22 13:52 ` kcc at gcc dot gnu.org 2013-02-22 13:55 ` Joost.VandeVondele at mat dot ethz.ch 2013-02-22 14:30 ` kcc at gcc dot gnu.org 2013-02-22 14:54 ` kcc at gcc dot gnu.org 2013-02-22 15:01 ` jakub at gcc dot gnu.org 2013-02-22 15:04 ` jakub at gcc dot gnu.org 2013-02-22 15:06 ` kcc at gcc dot gnu.org 2013-02-22 15:14 ` jakub at gcc dot gnu.org 2013-02-22 16:11 ` joseph at codesourcery dot com 2013-02-26 7:43 ` kcc at gcc dot gnu.org 2013-02-28 11:32 ` kcc at gcc dot gnu.org 2014-01-27 8:22 ` trippels at gcc dot gnu.org
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=bug-55309-4-vNIPyosuoe@http.gcc.gnu.org/bugzilla/ \ --to=gcc-bugzilla@gcc.gnu.org \ --cc=gcc-bugs@gcc.gnu.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).