Hello everybody,
Here's the latest version of my script.
I no longer advertise the patches as they are pretty big - for 2.5.64 it's
109022L and 3970383C.704649
bzip2'ed still 700KB - so I'm posting my script instead.
As some weeks ago it searches for occurrences of printk() without KERNEL_* -
specifications (eg KERNEL_DEBUG) and changes these. It tries to be
intelligent about that, ie it tracks if the last printk() in this function
had a \n at the end or not - continued lines shouldn't be messed up with <?>
characters.
This script couldn't parse the files (functions)
./fs/umsdos/mangle.c (umsdos_manglename)
./fs/affs/super.c (affs_write_super)
./sound/oss/emu10k1/audio.c (emu10k1_audio_ioctl)
./sound/oss/sb_ess.c (ess_init)
./sound/isa/sb/sb16.c (snd_sb16_probe)
./sound/isa/opti9xx/opti92x-ad1848.c (snd_card_opti9xx_probe)
and
./lib/zlib_deflate/deftree.c
- the first few because of #ifdef tricks with { or } and the last because of a
#define.
In
./lib/zlib_deflate/deftree.c
it correctly inserts a KERNEL_DEBUG - but as this printk() is the first in
this function AND in a loop you'll have to use this patch:
diff -u crypto/tcrypt.c.orig crypto/tcrypt.c
--- crypto/tcrypt.c.orig Wed Mar 5 04:29:33 2003
+++ crypto/tcrypt.c Mon Mar 10 15:43:16 2003
@@ -55,6 +55,8 @@
static void
hexdump(unsigned char *buf, unsigned int len)
{
+ printk(KERNEL_DEBUG);
+
while (len--)
printk("%02x", *buf++);
to make my script work.
Usage is like
find . -iname "*.c" | xargs perl -i.bak ~/perl/change_printk.pl
so parts of the tree can be changed as well as the full tree.
Please let me know if you're using this script (just for my pride :-) and let
me also know every problem you encounter.
Regards,
Phil