LKM programming - problem with printk();

[apacz]

Hello,
I've got a problem with printk() function. Im trying to print some buffor char by char like:

for&#40;i=0;i<=len;i++&#41; 
     printk&#40;KERN_ALERT "%c", k_space&#91;i&#93;&#41;;

But then i see in /var/log/syslog something like:
Aug 7 11:52:01 dstool kernel: i<1>n<1>s<1>m<1>o<1>d<1><1><1>
Aug 7 11:52:01 dstool kernel: e<1>.<1>o<1><1><1>
With some extra "<1>". Why it appears? When i print k_space as one string i don't have this "<1>". At first i though it was some problem with user space <-> kernel space but it transformed orginall buffor to k_space like:

if ((k_space=(char *)kmalloc(len, GFP_KERNEL))==NULL)
                        return -1;
memcpy_fromfs((void *)k_space, (void *)argv[argc], len);

And it didnt solve the problem. Any ideas what to do with this ?

[scientica]

I think printk maybe atdds some message count. And as you're printing the message char by char I think that's why you get it after each char.
Isn't k_space a 0-terminated string? if so somply do:

printk(KERN_ALERT "%s", k_space);

or you could do this:

char tem_char;
temp_char=k_space[len];  // maybe you could do botthia and above on one line.
k_space[len]='\0'; // terminate the string
printk(KERN_ALERT "%s", k_space);
k_space[len]=temp_char; // restore the char we replaced with a 0-terminator

either way try to unroll the loop - it's faster. (loops contains jumps, and jumps are 'realativle slow' - speaking about lowlevel stuff as branch prediction, the cosst of a jump (predicted or misspredited, etc) ).

[apacz]

Hello,

Isn't k_space a 0-terminated string? if so somply do:

I wrote that if i prinkt it as a one string (same as you wrote) there is no extra crap like "<1>".
It is important for to have access to simple char. I think "<1>" is connected in some way with this char becouse it doesnt appears only with printk. It doesnt allow me to compare it with other simple chars.
Any idea?