Messages in this thread | | | Date | Tue, 9 Apr 2013 16:50:54 +0200 | From | Oleg Nesterov <> | Subject | Re: [PATCH 4/4] uprobes/tracing: generalize struct uprobe_trace_entry_head |
| |
On 04/08, Steven Rostedt wrote: > > On Fri, 2013-03-29 at 19:15 +0100, Oleg Nesterov wrote: > > > > @@ -491,20 +502,19 @@ static int uprobe_trace_func(struct trace_uprobe *tu, struct pt_regs *regs) > > struct uprobe_trace_entry_head *entry; > > struct ring_buffer_event *event; > > struct ring_buffer *buffer; > > - u8 *data; > > + void *data; > > int size, i; > > struct ftrace_event_call *call = &tu->call; > > > > - size = sizeof(*entry) + tu->size; > > - > > + size = SIZEOF_TRACE_ENTRY(1) + tu->size; > > That '1' is confusing. When I first looked at this code, I thought it > was a bug as it should have been '0' (thinking of arrays). And then I > realized that you want the entry *after* the element. > > Instead of '1' and I assume '2' for ret probes, how about defining an > enum: > > enum { > UPROBE_NORM = 1, > UPROBE_RET = 2, > }; > > and then you can have; > > size = SIZEOF_TRACE_ENTRY(UPROBE_NORM); > > and later: > > size = SIZEOF_TRACE_ENTRY(UPROBE_RET); > > Same goes for DATAOF_TRACE_ENTRY(). > > This would make it a lot easier to understand and review, and much less > bug prone when we have to deal with two different types of > uprobe_trace_entry_head's.
OK, will do.
Or. Instead of enum we can use "bool is_return". So, instead of
if (is_ret_probe(tu)) size = SIZEOF_TRACE_ENTRY(UPROBE_ENTRY_RETPROBE); else size = SIZEOF_TRACE_ENTRY(UPROBE_ENTRY_NORMAL);
we can do
size = SIZEOF_TRACE_ENTRY(is_ret_probe(tu));
What do you like more?
Oleg.
| |