[lkml]   [2010]   [Jan]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH 11/12] perf lock: New subcommand "perf lock", for analyzing lock statistics
On 2010年01月22日 22:39, Hitoshi Mitake wrote:

Hi Ingo,
I have a trivial but difficult problem on this 11th patch.

> +static void
> +process_lock_acquire_event(void *data,
> + struct event *event __used,
> + int cpu __used,
> + u64 timestamp __used,
> + struct thread *thread __used)
> +{
> + struct trace_acquire_event acquire_event;
> + u64 tmp; /* this is required for casting... */
> +
> + tmp = raw_field_value(event, "lockdep_addr", data);
> + acquire_event.addr = (void *)tmp; # <- type of addr is void *

In the above function,
the statement:
raw_field_value(event, "lockdep_addr", data);
reads member of event which is type of void *.

But, direct assigning like
acquire_event.addr = (void *)raw_field_value(event, "lockdep_addr",
produces warning and it is treated as error on build environment of perf.
GCC says "cast from function call of type ‘long long unsigned int’ to
non-matching type ‘void *’".
# Why is tmp to acquire_event.addr allowed? It's mysterious...

So I declared "u64 tmp" as temporary place for casting.
But next, this tmp causes compile error on x86_32.
Because it is u64. tmp should be u32 on x86_32.

I have to switch type of tmp u64 or u32 based on environment.
Don't you know some good preprocessor flag or some other avoiding methods?
I seeked but not found...

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at

 \ /
  Last update: 2010-01-28 07:53    [W:0.197 / U:1.916 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site