Re: [PATCH 2/2] perf tools: make the GTK4 report browser actually loadable at runtime

[sashiko-bot@kernel.org]

> perf report --gtk dlopen()s libperf-gtk.so and expects it to resolve
> symbols back against the running perf binary (callchain_param,
> symbol_conf, evsel__name, and friends all live in perf itself, not in
> the plugin). Two things stood in the way of that after the GTK 2 to
> GTK 4 port:
> 
> perf never passed -rdynamic, so none of its own symbols were in its
> dynamic symbol table for a dlopen()ed plugin to find at all. Add
> -rdynamic to LDFLAGS when GTK4 support is enabled.
> 
> annotated_source__hist_entry() was a static inline in annotate.h, so
> ui/gtk/annotate.c calling it pulled hashmap__find()'s expansion,
> hashmap_find(), directly into libperf-gtk.so as an undefined symbol.
> The only hashmap_find perf links against in the common case is
> libbpf's internal one (tools/lib/bpf/hashmap.c), which libbpf
> compiles with -fvisibility=hidden; a hidden symbol can never be
> exported to a dlopen()ed plugin no matter what LDFLAGS perf itself
> gets. Move annotated_source__hist_entry() out of the header and into
> annotate.c, as an ordinary exported function, so the plugin depends
> on it the same way it already depends on evsel__group_desc() and
> friends, instead of reaching for hashmap_find directly.
> 
> With both fixes, a default 'make GTK4=1' build (libbpf statically
> linked, as usual) can dlopen() libperf-gtk.so and open the report
> browser without needing NO_LIBBPF=1 or manual LDFLAGS. Verified with
> perf report --gtk against a real perf.data on an actual GTK4 desktop
> session.
> 
> Signed-off-by: Matt Turner <mattst88@gmail.com>

Sashiko has reviewed this patch and found no issues. It looks great!

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260906-perf-gtk2-v1-0-7564bf8523a9@gmail.com?part=2