Messages in this thread |  | | | Subject | Re: printk regression? | | From | Joe Perches <> | | Date | Thu, 02 Jul 2009 08:29:16 -0700 |
On Thu, 2009-07-02 at 07:29 -0500, Michael S. Zick wrote: > On Thu July 2 2009, Joe Perches wrote: > > On Wed, 2009-07-01 at 21:23 -0700, Linus Torvalds wrote: > > > On Wed, 1 Jul 2009, Yinghai Lu wrote: > > > > [ 75.690022] <7>printing local APIC contents on CPU#0/0: > > > Yes. This is because the io_apic code should be fixed. > > > It does: > > > printk("\n" KERN_DEBUG "printing local APIC > > > and that "\n" at the beginning should just be deleted. The log-level > > > should be at the beginning of the printk, not in the middle. > > There's at least 72 of them: > > $ grep -Pr --include=*.[ch] "\bprintk.*\\\n.*KERN_" * > That should qualify it as a documented feature. ;) > "To convert the log level marker into visible garbage, do..."
Before some changes to the printk code, that used to be the only way to get a single printk call with multiple line output to have a KERN_<level> on all output lines.
ie: printk(KERN_INFO "Line 1\n" KERN_INFO "Line 2\n");
That style doesn't show up in the grep I posted because most of them look like:
printk(KERN_INFO "Line 1\n" KERN_INFO "Line 2\n");
An example: here's a bit of kernel/module.c:
printk(KERN_WARNING "%s: '%s'->init suspiciously returned %d, " "it should follow 0/-E convention\n" KERN_WARNING "%s: loading module anyway...\n",
|  |