Re: [PATCH v4 4/4] selftest/bpf: Test a perf bpf program that suppresses side effects.

On Thu, Jan 18, 2024 at 4:14 PM Kyle Huey me@kylehuey.com wrote:

Acked-by: Song Liu song@kernel.org

with a couple nitpicks below.

[...]

+int sigio_count, sigtrap_count;

+static void handle_sigio(int sig __always_unused) +{

•   ++sigio_count;
  

+}

+static void handle_sigtrap(int signum __always_unused,

•                      siginfo_t *info,
  

•                      void *ucontext __always_unused)
  

+{

•   ASSERT_EQ(info->si_code, TRAP_PERF, "wrong si_code");
  

nit: I would just call it "si_code", not "wrong si_code".

•   ++sigtrap_count;
  

+}

[...]

•   if (!ASSERT_OK(sigaction(SIGTRAP, &action, &previous_sigtrap), "sigaction"))
  

•           return;
  

•   previous_sigio = signal(SIGIO, handle_sigio);
  

•   if (!ASSERT_NEQ(previous_sigio, SIG_ERR, "signal"))
  

•           goto cleanup;
  

nit: If we goto cleanup here, we will do

signal(SIGIO, SIG_ERR);

This is a no-op, so it is not a real issue. But it is not very clean.

[...]