[LLVMdev] Testing an LLVM IR pass

Kristof Beyls kristof.beyls at arm.com
Mon Jul 7 08:48:28 PDT 2014


Hi Jan,

 

This probably doesn’t answer all of your questions directly, but I’ve used the following
command line in the past using LNT to cross-compile the test suite and run the
binaries on qemu:

./bin/lnt runtest nt --sandbox ~/lnt_sandbox --cc ~/build/bin/clang --test-suite ~/test-suite \

  --cxx ~/build/bin/clang \

  --cflag="-target aarch64-none-linux-gnu ..." \

  --llvm-arch=AArch64 --make-param RUNUNDER=~/qemu.sh --make-param USER_MODE_EMULATION=YES

Here, “qemu.sh” is a shell script that gets invoked to run the binary, so a similar approach could
probably be used to run the analysis tool you refer to (i.e. using the RUNUNDER parameter).

 

For remote execution; I’m not sure if it’s clearly documented anywhere, but it
seems lnt has --remote-{user, host, port, client} parameters to specify that.

 

Thanks,

 

Kristof

 

From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On Behalf Of Jan Hoogerbrugge
Sent: 07 July 2014 15:58
To: llvmdev at cs.uiuc.edu
Subject: [LLVMdev] Testing an LLVM IR pass

 

Hi,

I wrote an LLVM IR pass and after testing it on relatively small pieces of code I want to 
test it on the LLVM test suite in projects/test-suite. How can I specify to the test

infrastructure that I want to run my pass during the testing?

I also need to run an analysis tool on produced binaries before I can run it. Is there a way

to tell the test infrastructure to run a tool on produced binaries before executing them?

Is it also possible to build the test programs for another target and run them on a remote

machine?

 

Is running the test-suite with the Makefiles still a good option or should I use the LNT system?

Regards,
Jan.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20140707/fce28041/attachment.html>


More information about the llvm-dev mailing list