Re: [PATCH bpf-next] ftrace: Use hash argument for tmp_ops in update_ftrace_direct_mod

[Jiri Olsa]

On Thu, Mar 12, 2026 at 12:24:24PM +0000, bot+bpf-ci@xxxxxxxxxx wrote:
> > diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
> > index 8df69e702706..71dcbfeac86c 100644
> > --- a/kernel/trace/ftrace.c
> > +++ b/kernel/trace/ftrace.c
> > @@ -6608,7 +6608,7 @@ int update_ftrace_direct_mod(struct ftrace_ops *ops, struct ftrace_hash *hash, b
> >
> >  	/* Enable the tmp_ops to have the same functions as the direct ops */
> >  	ftrace_ops_init(&tmp_ops);
> > -	tmp_ops.func_hash = ops->func_hash;
> > +	tmp_ops.func_hash->filter_hash = hash;
> 
> The comment above says "Enable the tmp_ops to have the same functions
> as the direct ops", but after this change tmp_ops only gets the
> modified subset via the hash argument, not all of the direct ops
> functions.  Should the comment be updated to reflect this?
> 
> Also, the commit message has a minor grammar issue:
> 
> > which contains only the modified sites is always a subset of the
> > ops->func_hash.
> 
> Reads like it is missing an "and" between "sites" and "is".

yep, sent v2 

jirka