[BUG?] do_generic_direct_write

["J.A. Magallon" <jamagallon@able.es>]

On 06.27, Marcelo Tosatti wrote:
> 
> Hello,
> 
> Here goes -pre2 with a big number of changes, including the new aic7xxx
> driver.
> 
> I wont accept any big changes after -pre4: I want 2.4.22 timecycle to be
> short.
> 

Correct me if I'm wrong. I found this just by chance:

mm/filemap.c:
ssize_t
do_generic_direct_write(struct file *file,const char *buf,size_t count, loff_t *ppos)
...

    if (!file->f_flags & O_DIRECT)
        BUG();

This fails to trigger the BUG() just when it should. This should be:

    if (!(file->f_flags & O_DIRECT))
        BUG();

To check my C, I tried this:

#include <stdio.h>

#define O_D 2

int main()
{
    int flags;

    flags = O_D +4;
    printf("%d\n",!flags & O_D);
    printf("%d\n",!(flags & O_D));
    flags = 1+4;
    printf("%d\n",!flags & O_D);
    printf("%d\n",!(flags & O_D));

    return 0;
}

Output:

werewolf:~# ./kk
0
0
0
1

So it gives 0 when 'flags' != 0, not regarding if O_D is there or not....

Correct ?

One other job for the Stanford checker... ?

-- 
J.A. Magallon <jamagallon@able.es>      \                 Software is like sex:
werewolf.able.es                         \           It's better when it's free
Mandrake Linux release 9.2 (Cooker) for i586
Linux 2.4.21-jam1 (gcc 3.3 (Mandrake Linux 9.2 3.3-2mdk))