
<html><head><meta http-equiv="Content-Type" content="text/html charset=utf-8"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;" class="">Hi Evgenii,<div class=""><br class=""></div><div class="">Thanks for the inputs.</div><div class=""><br class=""></div><div class="">I’ve disabled shrink-wrapping for functions with the sanitize like attributes in <span style="font-family: Menlo; font-size: 11px;" class="">Committed revision 253116.</span></div><div class="">My understanding is that the frame information must be correct at the location of the crash, which can happen anywhere, not just at what shrink-wrapping considers as needing the frame to be lowered.</div><div class=""><br class=""></div><div class="">Does that make sense or did I just create a workaround for something I didn’t understand?</div><div class=""><br class=""></div><div class="">Thanks,</div><div class="">-Quentin</div><div class=""><br class=""><div><blockquote type="cite" class=""><div class="">On Nov 12, 2015, at 12:59 PM, Evgenii Stepanov <<a href="mailto:eugenis@google.com" class="">eugenis@google.com</a>> wrote:</div><br class="Apple-interchange-newline"><div class=""><div class="">Looks like debug location for prologue code could be broken or<br class="">missing. See EntryDebugLocation in<br class="">FunctionStackPoisoner::poisonStack() in AddressSanitizer.cpp.<br class=""><br class="">On Thu, Nov 12, 2015 at 12:32 PM, Quentin Colombet <<a href="mailto:qcolombet@apple.com" class="">qcolombet@apple.com</a>> wrote:<br class=""><blockquote type="cite" class="">Hi,<br class=""><br class="">I’ve had a quick look and the generated code looks good to me.<br class="">The prologue code in the failing example is needed only just before the call to the asan function (see attached assembly for x86).<br class=""><br class="">I am guessing that ASAN has some heuristics to find the prologue code in the disassembly and that fails for the new code.<br class="">Note that the debugger just get it right!<br class=""><br class="">Could you advise on what to do?<br class=""><br class="">Thanks,<br class="">Q<br class=""><br class=""><br class=""><br class=""><blockquote type="cite" class="">On Nov 12, 2015, at 9:45 AM, Quentin Colombet <<a href="mailto:qcolombet@apple.com" class="">qcolombet@apple.com</a>> wrote:<br class=""><br class="">Hi Renato,<br class=""><br class="">Thanks for the follow-up.<br class=""><br class="">I’ll try to have a look as well, unless you beat me at it :).<br class=""><br class="">Cheers,<br class="">-Quentin<br class=""><blockquote type="cite" class="">On Nov 12, 2015, at 5:35 AM, Renato Golin <<a href="mailto:renato.golin@linaro.org" class="">renato.golin@linaro.org</a>> wrote:<br class=""><br class="">On 11 November 2015 at 23:31, Quentin Colombet via llvm-commits<br class=""><<a href="mailto:llvm-commits@lists.llvm.org" class="">llvm-commits@lists.llvm.org</a>> wrote:<br class=""><blockquote type="cite" class="">Author: qcolombet<br class="">Date: Wed Nov 11 17:31:46 2015<br class="">New Revision: 252825<br class=""><br class="">URL: <a href="http://llvm.org/viewvc/llvm-project?rev=252825&view=rev" class="">http://llvm.org/viewvc/llvm-project?rev=252825&view=rev</a><br class="">Log:<br class="">[ARM] Enable shrink-wrapping by default.<br class=""><br class="">Differential Revision: <a href="http://reviews.llvm.org/D14357" class="">http://reviews.llvm.org/D14357</a><br class=""><br class=""><a href="rdar://problem/21942589" class="">rdar://problem/21942589</a><br class=""></blockquote><br class="">Quentin,<br class=""><br class="">After bisecting, this seems to have caused:<br class=""><br class=""><a href="http://lab.llvm.org:8011/builders/clang-cmake-armv7-a15-full/builds/9014" class="">http://lab.llvm.org:8011/builders/clang-cmake-armv7-a15-full/builds/9014</a><br class=""><br class="">http://lab.llvm.org:8011/builders/clang-cmake-armv7-a15-full/builds/9014/steps/ninja%20check%201/logs/FAIL%3A%20AddressSanitizer-armhf-linux%3A%3Anull_deref.cc<br class=""><br class="">I'm copying sanitizer folks to understand what's the change here, but<br class="">I'm guessing the sanitizer instrumentation didn't get inserted right,<br class="">with the prologue changed.<br class=""><br class="">I'll revert for now...<br class=""><br class="">cheers,<br class="">--renato<br class=""></blockquote><br class=""></blockquote><br class=""><br class=""></blockquote></div></div></blockquote></div><br class=""></div></body></html>