On Thu, 11 Jun 2020 at 16:43, Richard Henderson <richard.hender...@linaro.org> wrote: > > On 6/11/20 7:45 AM, Peter Maydell wrote: > > In commit 37bfce81b10450071 we accidentally introduced a leak of a TCG > > temporary in do_2shift_env_64(); free it. > > > > Signed-off-by: Peter Maydell <peter.mayd...@linaro.org> > > --- > > My test setup wasn't looking for temporary-leak warnings (they are > > not as easy to get at as they used to be because they only appear > > if you enable qemu_log tracing for some other purpose). This is the > > only one that snuck through, though. > > --- > > target/arm/translate-neon.inc.c | 1 + > > 1 file changed, 1 insertion(+) > > > > diff --git a/target/arm/translate-neon.inc.c > > b/target/arm/translate-neon.inc.c > > index 7c4888a80c9..f2c241a87e9 100644 > > --- a/target/arm/translate-neon.inc.c > > +++ b/target/arm/translate-neon.inc.c > > @@ -1329,6 +1329,7 @@ static bool do_2shift_env_64(DisasContext *s, > > arg_2reg_shift *a, > > neon_load_reg64(tmp, a->vm + pass); > > fn(tmp, cpu_env, tmp, constimm); > > neon_store_reg64(tmp, a->vd + pass); > > + tcg_temp_free_i64(tmp); > > Huh. I thought all the a32 stores magically freed their inputs. I guess > that's just the general-purpose registers.
Confusingly, neon_load_reg()/neon_store_reg() do the "create new TCG temp for the load, free the store input" thing, but neon_load_reg64()/neon_store_reg64()/neon_load_reg32()/neon_store_reg32() do not. I'd kind of like to get rid of this weird proliferation of load and store functions, but neon_load_reg()/neon_store_reg() are also the only ones which take a (regno, pass) pair of inputs, so it's not a completely trivial substitution. thanks -- PMM
