similar to: [LLVMdev] Cross compiling llvm with arm toolchain

> I've tried arm-2008q3-72, > arm-2009 and arm-2010q1-202 from code-sourcery with little success - > arm2010q1 fails to compile, arm-2009 take an age to link and arm-2008q3 > seems to work but only when compiled with -O0. Does anyone have any > advice on the matter? CS's release series is renowned for being bug-ridden. To be fair, this is probably because CS do

I'm compiling samba-3.6.22 for android with the sourcery arm-2010q1 compiler. The compile itself goes fine, but when I copy smbd to my device it's not recognized as a binary. I get the error sh: ./smbd: No such file or directory file under linux tells me: smbd: ELF 32-bit LSB shared object, ARM, EABI5 version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.16, not

Hello I am new to LLVM and would like to cross-compile LLVM 2.9 on x86 (Ubuntu) to ARM using Code Sourcery LIte toolchain for ARM EABI (2011.03-41). This is the command I used to configure: ./configure --build=i686-pc-linux-gnu --host=arm-none-linux-gnueabi --target=arm-none-linux-gnueabi followed by: make -C llvm.2.9 ENABLE_OPTIMIZED=0 I encountered the following errors: configure:error:

Hello, (cc:llvmdev Keeping on the list to get more eyes on the problem and for archival purposes). Configure will be re-run, yes. LLVM includes some tools which need to be built for the host, even when cross compiling (e.g., tblgen). In a cross-build, these are built in the BuiltTools directory and that's where you're seeing configure being re-run. >From the below snippet it appears

On 18 July 2012 14:33, salvatore benedetto <salvatore.benedetto at gmail.com> wrote: > but I still haven't figure out how to build for cortex-m3 > > clang -march=armv7-m -mfloat-abi=soft <something missing?> testReference.cpp -c -march should have done the trick. You can also try -mcpu=cortex-m3, or try -ccc-host-triple armv7m-none-gnueabi (or -eabi), and possibly

Hey all I am trying to configure all my pjsip transports form a database table. The issue I am running into is that pjsip is auto binding to 0.0.0.0:5060 before it reads my list of transports from the database. This means that my entries for port 5060 are already bound and the settings in the database are not loaded. When loading the transport form the .conf file it works as expected

Hello all, We would like to pull the RTT of registered endpoints from MySQL for use in a webportal. However it doesn't appear asterisk tables this by default like chan_sip did. I've found some information [1] that a modification of sorcery.conf can get it writing to a table. But I'm struggling to figure out what that configuration is. It seems the object is called

>>>>> I am using GCC 4.3.2 ( Debian 4.3.2-1.1) on x86. I will try out a different compiler. I tried out Sourcery G++ Lite 2009q1-203 (GCC 4.3.3). I still can't get lli to run. This time the error is like this: 0 lli 0x00651128 1 lli 0x00651960 2 libc.so.6 0x40212600 __default_sa_restorer_v1 + 0 3 lli 0x00282ef0 4 lli 0x00283cec

Hi, Method RefNameBuilder::buildDuplicateNameMap() has an assert which blocks atoms with empty name. In general it looks reasonable but some toolchains (for example Sourcery CodeBench in both MIPS and ARM editions) can generate an object file contains absolute STT_FILE symbols with empty name. Moreover crt1.o object file from this toolchain has such symbol. I do not know is it a feature or bug

Hi all, I'm running rsync version 2.6.3 protocol version 28 on Solaris 8 and am having a problem. We have some scripts that wrap around rsync and generate a list of files to transfer from the sending system (regardless of whether that directory exists on the receiver). I was expecting rsync to fail when transferring files with a path that doesn't exist on the receiver (in fact

Hi all, It seems that using exten => _*0. is not possible in extensions.conf. I changed disconnect => *0 in features.conf to something else. From what I can tell with the little C knowledge I have is that it's caused by a hardcoded *0 value chan_zap.c. Line 5730 of chan_zap.c (svn rev 1077) shows: } else if (!strcmp(exten, "*0")) { struct ast_channel *nbridge =

Grr why is this so hard? Compiling samba-3.6.24 for android using the sourcery toolchain. Dropbear and busybox went w/o a hitch. I've googled the hell out of this, nothing I've found has helped. Here's my config: ./configure --without-krb5 \ --without-ldap \ --without-ads \ --disable-cups \ --enable-swat=no \ --enable-static=yes \ --disable-shared \ --with-winbind=no \

You will need to install a cross compiler. I think you can find one somewhere in this link: https://sourcery.mentor.com/GNUToolchain/ Once you have a cross compiler, this should do what you want (if you want little-endian code): clang -target mipsel-unknown-linux -ccc-clang-archs mipsel -O3 hello.c -o hello.bc -emit-llvm llc -march=mipsel hello.bc -o hello.s mips-linux-gnu-gcc hello.s -o hello

Thanks Akira, it worked! On Thu, Feb 23, 2012 at 10:01 PM, Akira Hatanaka <ahatanak at gmail.com> wrote: > You will need to install a cross compiler. > I think you can find one somewhere in this link: > https://sourcery.mentor.com/GNUToolchain/ > > Once you have a cross compiler, this should do what you want (if you > want little-endian code): > > clang -target

There is also: 1) an alternate clang driver at http://code.google.com/p/alternate-clang-driver/ . This makes it more or less gcc compatible. 2) there is a way to invoke it all directly with just a call to clang as is with x86/linux but it's not totally finished. i will try and post something about this. the direct form will work if you are on a mips linux machine but there are still some

I have read ELF specification once again. There is no any explicit references to absolute symbols with no name. The only real case when I see such symbol is a local absolute symbol with "STT_FILE" type. The name of "STT_FILE" symbol is a name of the source file associated with the object file and the "STT_FILE" symbol precedes the other STB_LOCAL symbols from this

I have been following the recent posts regarding LLVM ARM backend on the list. I am trying to figure out if I should start using LLVM to compile my project or if I should just stick with cross compilers like those provided by Code Sourcery and wait till a later time. I have the following questions in this regard: 1. Can some one tell me if LLVM ARM is being used by anyone to compile any shipped

Great Job! If I use alternate-clang-driver, will I need mips-gcc still? I mean alternate-clang-driver use as&ld directly or by calling gcc? Regards, Jia Liu On Tue, Feb 28, 2012 at 10:39 AM, reed kotler <rkotler at mips.com> wrote: > There is also: > > 1) an alternate clang driver at > http://code.google.com/p/alternate-clang-driver/ . This makes > it more or less gcc

Hello, I built a llvm toolchain for ARM and I'm trying to build a simple application on ARM using LLVM. When I compiled the code, it has the following message and the binary code has a segmentation fault when it is executed. /tmp/ccJPjMe4.s: Assembler messages: /tmp/ccJPjMe4.s:4250: rdhi, rdlo and rm must all be different /tmp/ccJPjMe4.s:4589: rdhi, rdlo and rm must all be different

Hi, On Wed, Jul 4, 2012 at 9:21 AM, ETANI NORIKO <noriko-e at fc.ritsumei.ac.jp> wrote: > Please advise me how to change a compiler from a host one to a target one. Suppose MIPS toolchain is installed in the $MIPS folder (i.e. mips-linux-gnu-gcc is in the $MIPS/bin folder). Note, if you want to generate little-endian code and/or 64-bit code, you have to create the following links in the