Package: llvm-3.0 Version: 3.0-10 Severity: normal Tags: patch Hi,
clang 3.0 fails with -march=native on an Core i7 Ivy Bridge CPU on [amd64]: # clang -c -v -march=native /tmp/x.c -m64 Debian clang version 3.0-6.1 (tags/RELEASE_30/final) (based on LLVM 3.0) Target: x86_64-pc-linux-gnu Thread model: posix "/usr/bin/clang" -cc1 -triple x86_64-pc-linux-gnu -emit-obj -mrelax-all -disable-free -disable-llvm-verifier -main-file-name x.c -mrelocation-model static -mdisable-fp-elim -masm-verbose -mconstructor-aliases -munwind-tables -target-cpu i686 -target-linker-version 2.22 -momit-leaf-frame-pointer -v -coverage-file x.o -resource-dir /usr/bin/../lib/clang/3.0 -fmodule-cache-path /var/tmp/clang-module-cache -internal-isystem /usr/local/include -internal-isystem /usr/bin/../lib/clang/3.0/include -internal-externc-isystem /usr/include/x86_64-linux-gnu -internal-externc-isystem /usr/include -ferror-limit 19 -fmessage-length 194 -fgnu-runtime -fobjc-runtime-has-arc -fobjc-runtime-has-weak -fobjc-fragile-abi -fdiagnostics-show-option -fcolor-diagnostics -o x.o -x c /tmp/x.c error: unknown target CPU 'i686' Hmm, well, that one only exists for -m32. # cat /proc/cpuinfo processor : 0 vendor_id : GenuineIntel cpu family : 6 model : 58 model name : Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz stepping : 9 microcode : 0x12 ... I cherry-picked a few patches from llvm-3.2 from their git-svn repository that * add detection for a few newer CPUs * use a default for unknown CPUs that takes into account whether the CPU is 64-bit capable, so this would fall back to -march=x86-64 which would actually work. After applying these patches to llvm-3.0 clang-3.0 started to behave sane again. (But due to lack of AVX support I'll have to move to llvm-3.2/clang-3.2 anyway.) If help is needed, I could take care of a NMU and try to get this unblocked for wheezy. Andreas -- ------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------ Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher Geschaeftsfuehrung: Prof. Dr. Achim Bachem (Vorsitzender), Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt ------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------
>From aff59685a9ff59edb525a1be2120e9d0206f3bb8 Mon Sep 17 00:00:00 2001 From: Evan Cheng <[email protected]> Date: Mon, 23 Apr 2012 22:41:39 +0000 Subject: [PATCH] Add a missing cpu subtype. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@155402 91177308-0d34-0410-b5e6-96231b3b80d8 --- lib/Support/Host.cpp | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/lib/Support/Host.cpp b/lib/Support/Host.cpp index 0f06964..a5e0cb0 100644 --- a/lib/Support/Host.cpp +++ b/lib/Support/Host.cpp @@ -230,6 +230,10 @@ std::string sys::getHostCPUName() { case 45: return "corei7-avx"; + // Ivy Bridge: + case 58: + return "core-avx-i"; + case 28: // Intel Atom processor. All processors are manufactured using // the 45 nm process return "atom"; -- 1.7.10.4
>From 0d38d3a0005c60f721899475902026a91ef65fad Mon Sep 17 00:00:00 2001 From: Bob Wilson <[email protected]> Date: Wed, 9 May 2012 17:47:03 +0000 Subject: [PATCH] Use the cpuid 64 bit flag to pick the default CPU name for an unknown model. For the Family 6 switch in sys::getHostCPUName, an unrecognized model was reported as "i686". That's a really bad default since it means that new CPUs will be treated as if they can only use 32-bit code. This just looks at the cpuid extended feature flag for 64 bit support, and if that is set, it uses a default x86-64 cpu. Similar logic is already used for the Family 15 code. <rdar://problem/11314502> git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@156486 91177308-0d34-0410-b5e6-96231b3b80d8 --- lib/Support/Host.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) --- a/lib/Support/Host.cpp +++ b/lib/Support/Host.cpp @@ -231,7 +231,7 @@ // the 45 nm process return "atom"; - default: return "i686"; + default: return (Em64T) ? "x86-64" : "i686"; } case 15: { switch (Model) {
>From 4335e3495d5cd6784a99749879114caf22b9d398 Mon Sep 17 00:00:00 2001 From: Benjamin Kramer <[email protected]> Date: Wed, 26 Sep 2012 18:21:47 +0000 Subject: [PATCH] Add support for detecting some corei7-class Xeons. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164714 91177308-0d34-0410-b5e6-96231b3b80d8 --- lib/Support/Host.cpp | 2 ++ 1 file changed, 2 insertions(+) diff --git a/lib/Support/Host.cpp b/lib/Support/Host.cpp index a13b9e2..9ee3f2d 100644 --- a/lib/Support/Host.cpp +++ b/lib/Support/Host.cpp @@ -234,6 +234,8 @@ std::string sys::getHostCPUName() { case 37: // Intel Core i7, laptop version. case 44: // Intel Core i7 processor and Intel Xeon processor. All // processors are manufactured using the 32 nm process. + case 46: // Nehalem EX + case 47: // Westmere EX return "corei7"; // SandyBridge: -- 1.7.10.4
