On Tue, 28 Nov 2000, Tigran Aivazian wrote:
> Hi,
>
> A simple experiment:
>
> a) insert a code like this in the open() routine of some driver:
>
> struct files_struct *files = current->files;
> extern int count_open_files(struct files_struct *, int);
>
> printk(KERN_ERR "%s has %d files open\n",
> current->comm, count_open_files(files, files->max_fdset));
>
> b) write a program that opens that device and sleeps indefinitely. Run it,
> you will see:
>
> op has 32 files open
>
> c) # lsof -p 659 | nl
> 1 COMMAND PID USER FD TYPE DEVICE SIZE NODE NAME
> 2 op 659 tigran cwd DIR 3,6 4096 225730
> /home/tigran/C/files
> 3 op 659 tigran rtd DIR 3,6 4096 2 /
> 4 op 659 tigran txt REG 3,6 14142 225732
> /home/tigran/C/files/op
> 5 op 659 tigran mem REG 3,6 434945 295227
> /lib/ld-2.1.92.so
> 6 op 659 tigran mem REG 3,6 4776568 295234
> /lib/libc-2.1.92.so
> 7 op 659 tigran 0u CHR 4,9 348027 /dev/tty9
> 8 op 659 tigran 1u CHR 4,9 348027 /dev/tty9
> 9 op 659 tigran 2u CHR 4,9 348027 /dev/tty9
> 10 op 659 tigran 3r CHR 10,184 98313
> /dev/cpu/microcode
>
> so, we see that the process has only 9 files open and yet
> count_open_files() claims there are 32. Is this a bug of
> count_open_files() or is this a minimal granularity (because there are 32
> builtin descriptors in files_struct->fd_array[])?
>
> I know that there is no problem due to the way it is called in
> copy_files() -- it would only be above 32. But for what I want to use it,
> I need the _correct_ number of open file descriptors and not some "rounded
> up to 32" one.
>
> (kernel assumed test12-pre2)
>
> Regards,
> Tigran
>
Yes. This is probably related to the previously-reported MOD_INC_USE_COUNT
macro in the drivers. The open count is now handled in the call to
the driver's open(). The MOD_INC_USE_COUNT and the MOD_DEC_USE_COUNT
should now be defined to do nothing for kernels that already handle
the counts in the calls.
Cheers,
Dick Johnson
Penguin : Linux version 2.4.0 on an i686 machine (799.54 BogoMips).
"Memory is like gasoline. You use it up when you are running. Of
course you get it all back when you reboot..."; Actual explanation
obtained from the Micro$oft help desk.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Nov 30 2000 - 21:00:19 EST