Date: Fri, 18 Aug 2017 01:42:41 +0000 From: bugzilla-noreply@freebsd.org To: freebsd-bugs@FreeBSD.org Subject: [Bug 221337] -fsanitize=address (asan) fails on i386 Message-ID: <bug-221337-8-uqDQMewTy9@https.bugs.freebsd.org/bugzilla/> In-Reply-To: <bug-221337-8@https.bugs.freebsd.org/bugzilla/> References: <bug-221337-8@https.bugs.freebsd.org/bugzilla/>
next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D221337 --- Comment #8 from Ed Maste <emaste@freebsd.org> --- Some possibly useful references: This (GetThreadStackAndTls) originally arrived in https://reviews.llvm.org/rL156542. FreeBSD support was added in https://reviews.llvm.org/rL203235. Thread sanitizer algorithm description at https://github.com/google/sanitizers/wiki/ThreadSanitizerAlgorithm I think the use of dtv[2] is indeed strange - AFAICT the intent is that it should be returning the address and size of the entire TLS block -- at leas= t, the contiguous block of "initially available modules." I suppose there are a few fairly straightforward changes/hacks that could be tried: 1. Change the FreeBSD GetTls in sanitizer_linux_libcdep.cc to set *addr =3D= 0 and *size =3D 0. My guess is that this will just exclude all tls accesses from = the sanitizer. If that works I'd be happy enough with it as an initial stopgap = on FreeBSD/i386 (asan that does not detect errors with tls is better than asan that does not work at all). 2. Make TLS_TCB_ALIGN 8 for i386 in libc tls.c 3. Round addr down to be 8-byte aligned (increasing size as necessary) All of this goop is to implement a function void GetThreadStackAndTls(bool main, uptr *stk_addr, uptr *stk_size, uptr *tls_addr, uptr *tls_size) to fe= tch the extent of the current thread's stack and tls regions; ideally we can replace this with a clean and consistent FreeBSD-specific interface. --=20 You are receiving this mail because: You are the assignee for the bug.=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-221337-8-uqDQMewTy9>