Re: [PATCH v6 2/3] perf tests: avoid storing an absolute path in perf binary

[Jiri Olsa <jolsa@redhat.com>]

On Fri, May 21, 2021 at 10:27:52AM -0700, Denys Zagorui wrote:

SNIP

>  	free(cmd);
> +out:
> +	free(pythonpath);
>  	return ret;
>  }
> diff --git a/tools/perf/util/util.c b/tools/perf/util/util.c
> index 3bba74e431ed..53f3dbf02f58 100644
> --- a/tools/perf/util/util.c
> +++ b/tools/perf/util/util.c
> @@ -22,6 +22,7 @@
>  #include "cap.h"
>  #include "strlist.h"
>  #include "string2.h"
> +#include <libgen.h>
>  
>  /*
>   * XXX We need to find a better place for these things...
> @@ -388,3 +389,14 @@ char *perf_exe(char *buf, int len)
>  	}
>  	return strcpy(buf, "perf");
>  }
> +
> +char *perf_exe_path(void)
> +{
> +	char buf[PATH_MAX];
> +	char *dname;
> +
> +	perf_exe(buf, PATH_MAX);
> +	dname = dirname(buf);

I think this is equally bad.. once you get out of perf_exe_path,
dname might be screwed

> +
> +	return dname;

just return strdup(dname) in here?

jirka