(trivia) remove useless typecast around `jif' variable
in fs/proc/proc_misc.c:show_stat() routine (which handles /proc/stat file), there's a local variable named 'jif' of type unsigned long, declared as following:
static int show_stat(struct seq_file *p, void *v)
{
unsigned long jif;But later on, it's explicitly casted to (unsigned long) in printf(). Remove the useless cast, to let the compiler to do the work for us in case we'll want to use different type here (say, u64 or something).
fs/proc/proc_misc.c renamed from fs/proc/proc_misc.c.orig +1 -1
@@ -521,15 +521,15 @@ static int show_stat(struct seq_file *p,seq_printf (p,"\nctxt %llu\n" "btime %lu\n" "processes %lu\n" "procs_running %lu\n" "procs_blocked %lu\n" ,nr_context_switches (), - (unsigned long )jif, + jif, total_forks,nr_running (),nr_iowait ());return 0 ; }
Quoted 3 lines · Michael Tokarev
Date: Mon, 5 Mar 2007 19:12:51 +0300 (MSK)
From: Michael Tokarev <[email protected]>
To: undisclosed-recipients: ;
I have no clue what you sent it to, so I added linux-kernel again.
Quoted 6 lines
Subject: (trivia) remove useless typecast around `jif' variable
in fs/proc/proc_misc.c:show_stat() routine (which handles
/proc/statfile), there's a local variable named 'jif' of type unsigned long, declared as following:
Quoted 7 lines
static int
show_stat(struct seq_file *p, void *v){ unsigned long jif;But later on, it's explicitly casted to (unsigned long) in
printf(). Remove the useless cast, to let the compiler to do the work for us in case we'll want to use different type here (say, u64 or something).
In case we wanted to use different types, we would also have to change the accompanying %lu into %llu. Only changing jif to u64 will cause a problem, as the compiler does _not_ automatically promote/demote types in varargs that already have a certain size. In other words,
int foo(unsigned long x)
{
printk("%llu\n", x); /* and */
printk("%lu\n", x);
}will throw a warning (rightfully if you ask me).
Jan
Quoted 7 lines · Jan Engelhardt
Quoted 3 lines · Michael Tokarev
Date: Mon, 5 Mar 2007 19:12:51 +0300 (MSK)
From: Michael Tokarev <[email protected]>
To: undisclosed-recipients: ;I have no clue what you sent it to, so I added linux-kernel again.
Oops.. I forgot to add the To: header (it was only sent to lkml). ][
Quoted 5 lines
In case we wanted to use different types, we would also have to change the accompanying %lu into %llu. Only changing jif to u64 will cause a problem, as the compiler does _not_ automatically promote/demote types in varargs that already have a certain size. In other words,
Sure thing, the change will be needed. But the thing is - with the cast in place, compiler will be completely silent, while w/o the cast it will produce a warning (or at least it's able to).
Generally, casts are bad, that's the point. Especially redundrand ones like here.
As a side note, I dislike when people remove casts from functions returning void*. For example,
struct somestruct *foo;
foo = (struct somestruct *)kmalloc(sizeof(struct somestruct));
With the cast in place, the compiler will warn if somestruct will be changed to something else, but without the cast, the compiler will happily accept the (now wrong) line.
[]
Quoted 1 line
will throw a warning (rightfully if you ask me).
Yes, that's what i was referring to when said "let the compiler do some work for us".
/mjt
A message that is not in the archive (sent off-list or not yet received): 663650480-1173118082-cardhu_blackberry.rim.net-718749525-@bxe041-cell01.bisx.prod.on.blackberry
Quoted 1 line · [email protected] wrote:
Yes sounds correct. How many other places are being casted as well?
Care to find them? :)
This one took my attention by a chance. I looked at various other files, at least in fs/proc/, but don't see any other places like this.
/mjt