Re: [kvm-unit-tests 5/5] scripts: pretty print stack traces

[Paolo Bonzini <pbonzini@redhat.com>]

On 03/03/2016 10:35, Andrew Jones wrote:
> wrt to enabling/disabling it though, how about we create a new configure
> command line option instead. One that defaults to on, if the architecture
> supports pretty-printing, but can be switched off. 

Which architectures don't?  But I like a
--enable/disable-pretty-print-stacks option.

> It should be possible, just another "temp_file", I think. But then we
> expect the standalone test to be executed somewhere with pyhon (likely)
> and addr2line (not as likely). Also, without the source available, then
> it probably isn't all the useful to the user executing the test, and
> anybody they send the results to can copy+paste the backtrace line
> directly into addr2line themselves.
> 
> Long story short, I vote we don't add it to mkstandalone.

I agree.  Something like

diff --git a/scripts/mkstandalone.sh b/scripts/mkstandalone.sh
index c7e78f8..d6607b8 100755
--- a/scripts/mkstandalone.sh
+++ b/scripts/mkstandalone.sh
@@ -41,6 +41,10 @@ generate_test ()
 	echo "#!/bin/bash"
 	echo "export STANDALONE=yes"
 	echo "export HOST=\$(uname -m | sed -e s/i.86/i386/ | sed -e
's/arm.*/arm/')"
+
+	# do not require addr2line on the test machine
+	echo "export PRETTY_PRINT_STACKS=no"
+
 	config_export ARCH
 	config_export ARCH_NAME
 	config_export PROCESSOR

if we go with your suggestion of configurable pretty printing.

> Now, one addition that may make sense, wrt mkstandalone, is that we output
> what HEAD was when mkstandalone was used to build the tests. That would be
> useful just by itself to know what code is running, and extra useful when
> using addr2line.

That makes sense.  Can you send a patch (Drew)?

Paolo