Re: virtfs-proxy-helper fails due to bogus libattr test

[Paolo Bonzini]

On 12/01/21 13:29, Daniel P. Berrangé wrote:
> The -O0 might come from a pkg-config file
>
> $BUILD_DIR/meson-logs/meson-log.txt
>
> might show where it comes from.

It's added by Meson for compiler checks.  It's useful for example for 
functions that are declared as const or pure, and would be optimized 
away by the compiler.  However, it does not matter apart from causing a 
warning to show up in the logs.

(The warning is probably caused by -D_FORTIFY_SOURCE=2 in the CFLAGS 
environment variable, though I cannot say that with certainty.  QEMU is 
able to add -D_FORTIFY_SOURCE on its own).

If I do

$ CFLAGS='-O2 -g -D_FORTIFY_SOURCE=2' ../configure --enable-attr

I get the same warning in the logs, and nevertheless the summary reports 
that the test is successful:

               ATTR/XATTR support: YES

as well as in config-host.h

#define CONFIG_ATTR
#undef CONFIG_LIBATTR

Unfortunately, from Olaf's report I cannot understand what the 
virtfs-proxy-helper failure is like (since no "make" log was included) 
or the exact details of the build environment (since no 
meson-logs/meson-log.txt is included).  Olaf, can you expand on that?

Paolo