[PATCH] init: Scream bloody murder if interrupts are enabled too early

[PATCH] init: Scream bloody murder if interrupts are enabled too early

[Steven Rostedt]

[ Andrew, can you pull this patch into -mm? ]

As I was testing a lot of my code recently, and having several
"successes", I accidentally noticed in the dmesg this little line:

[    0.000000] start_kernel(): bug: interrupts were enabled *very* early, fixing it


Sure enough, one of my patches two commits ago enabled interrupts early.
The sad part here is that I never noticed it, and I ran several tests
with ktest too, and ktest did not notice this line.

What ktest looks for (and so does many other automated testing scripts)
is a back trace produced by a WARN_ON() or BUG(). As a back trace was
never produced, my buggy patch could have slipped into linux-next, or
even worse, mainline.

Adding a WARN_ON() makes this bug a little more obvious:

[    0.000000] PID hash table entries: 4096 (order: 3, 32768 bytes)
[    0.000000] __ex_table already sorted, skipping sort
[    0.000000] Checking aperture...
[    0.000000] No AGP bridge found
[    0.000000] Calgary: detecting Calgary via BIOS EBDA area
[    0.000000] Calgary: Unable to locate Rio Grande table in EBDA - bailing!
[    0.000000] Memory: 2003192k/2054848k available (4792k kernel code, 460k absent, 51196k reserved, 6277k data, 1152k init)
[    0.000000] ------------[ cut here ]------------
[    0.000000] WARNING: at /home/rostedt/work/git/linux-trace.git/init/main.c:543 start_kernel+0x215/0x40e()
[    0.000000] Hardware name: To Be Filled By O.E.M.
[    0.000000] Modules linked in:
[    0.000000] Pid: 0, comm: swapper/0 Not tainted 3.8.0-test+ #275
[    0.000000] Call Trace:
[    0.000000]  [<ffffffff8103781d>] warn_slowpath_common+0x83/0x9b
[    0.000000]  [<ffffffff8103784f>] warn_slowpath_null+0x1a/0x1c
[    0.000000]  [<ffffffff81ae5a30>] start_kernel+0x215/0x40e
[    0.000000]  [<ffffffff81ae5623>] ? repair_env_string+0x56/0x56
[    0.000000]  [<ffffffff81ae5312>] x86_64_start_reservations+0x10e/0x112
[    0.000000]  [<ffffffff81ae5120>] ? early_idt_handlers+0x120/0x120
[    0.000000]  [<ffffffff81ae5418>] x86_64_start_kernel+0x102/0x111
[    0.000000] ---[ end trace 1ca9c978643fe899 ]---
[    0.000000] start_kernel(): bug: interrupts were enabled *very* early, fixing it
[    0.000000] Preemptible hierarchical RCU implementation.
[    0.000000]  RCU restricting CPUs from NR_CPUS=8 to nr_cpu_ids=4.
[    0.000000] NR_IRQS:4352 nr_irqs:712 16
[    0.000000] Console: colour VGA+ 80x25
[    0.000000] console [ttyS0] enabled, bootconsole disabled
[    0.000000] console [ttyS0] enabled, bootconsole disabled

Do you see it?

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

diff --git a/init/main.c b/init/main.c
index 63534a1..f0fe0a5 100644
--- a/init/main.c
+++ b/init/main.c
@@ -539,7 +539,7 @@ asmlinkage void __init start_kernel(void)
 	 * fragile until we cpu_idle() for the first time.
 	 */
 	preempt_disable();
-	if (!irqs_disabled()) {
+	if (WARN_ON(!irqs_disabled())) {
 		printk(KERN_WARNING "start_kernel(): bug: interrupts were "
 				"enabled *very* early, fixing it\n");
 		local_irq_disable();

Re: [PATCH] init: Scream bloody murder if interrupts are enabled too early

[Ard van Breemen]

Hi,

On Thu, Mar 07, 2013 at 05:19:03PM -0500, Steven Rostedt wrote:
> diff --git a/init/main.c b/init/main.c
> index 63534a1..f0fe0a5 100644
> --- a/init/main.c
> +++ b/init/main.c
> @@ -539,7 +539,7 @@ asmlinkage void __init start_kernel(void)
>  	 * fragile until we cpu_idle() for the first time.
>  	 */
>  	preempt_disable();
> -	if (!irqs_disabled()) {
> +	if (WARN_ON(!irqs_disabled())) {
>  		printk(KERN_WARNING "start_kernel(): bug: interrupts were "
>  				"enabled *very* early, fixing it\n");
>  		local_irq_disable();
> 

The original code was a few versions before the WARN macros.
I think it's better to just change the printk to something like:
WARN(1,"interrupts were enabled *very* early, fixing it\n")

This will print the cause of the warning and then dump stack,
instead of having an unrelated message under the stackdump.

But: IANAM(acro)E(xpert), the real incantation might be
different.
But you are absolutely right that we should scream and shout.
This time in a way that's more in line with the others.

Regards,
Ard van Breemen.

Re: [PATCH] init: Scream bloody murder if interrupts are enabled too early

[Steven Rostedt]

On Fri, 2013-03-08 at 10:03 +0100, Ard van Breemen wrote:

> The original code was a few versions before the WARN macros.
> I think it's better to just change the printk to something like:
> WARN(1,"interrupts were enabled *very* early, fixing it\n")
> 

Yeah, I thought about updating to that, but I was lazy. I was hoping
nobody would notice. ;-)

OK, I'll send out -v2 as soon as I finish testing it.

Thanks,

-- Steve

Re: [PATCH] init: Scream bloody murder if interrupts are enabled too early

[Paul E. McKenney]

On Thu, Mar 07, 2013 at 05:19:03PM -0500, Steven Rostedt wrote:
> [ Andrew, can you pull this patch into -mm? ]
> 
> As I was testing a lot of my code recently, and having several
> "successes", I accidentally noticed in the dmesg this little line:
> 
> [    0.000000] start_kernel(): bug: interrupts were enabled *very* early, fixing it
> 
> 
> Sure enough, one of my patches two commits ago enabled interrupts early.
> The sad part here is that I never noticed it, and I ran several tests
> with ktest too, and ktest did not notice this line.
> 
> What ktest looks for (and so does many other automated testing scripts)
> is a back trace produced by a WARN_ON() or BUG(). As a back trace was
> never produced, my buggy patch could have slipped into linux-next, or
> even worse, mainline.
> 
> Adding a WARN_ON() makes this bug a little more obvious:
> 
> [    0.000000] PID hash table entries: 4096 (order: 3, 32768 bytes)
> [    0.000000] __ex_table already sorted, skipping sort
> [    0.000000] Checking aperture...
> [    0.000000] No AGP bridge found
> [    0.000000] Calgary: detecting Calgary via BIOS EBDA area
> [    0.000000] Calgary: Unable to locate Rio Grande table in EBDA - bailing!
> [    0.000000] Memory: 2003192k/2054848k available (4792k kernel code, 460k absent, 51196k reserved, 6277k data, 1152k init)
> [    0.000000] ------------[ cut here ]------------
> [    0.000000] WARNING: at /home/rostedt/work/git/linux-trace.git/init/main.c:543 start_kernel+0x215/0x40e()
> [    0.000000] Hardware name: To Be Filled By O.E.M.
> [    0.000000] Modules linked in:
> [    0.000000] Pid: 0, comm: swapper/0 Not tainted 3.8.0-test+ #275
> [    0.000000] Call Trace:
> [    0.000000]  [<ffffffff8103781d>] warn_slowpath_common+0x83/0x9b
> [    0.000000]  [<ffffffff8103784f>] warn_slowpath_null+0x1a/0x1c
> [    0.000000]  [<ffffffff81ae5a30>] start_kernel+0x215/0x40e
> [    0.000000]  [<ffffffff81ae5623>] ? repair_env_string+0x56/0x56
> [    0.000000]  [<ffffffff81ae5312>] x86_64_start_reservations+0x10e/0x112
> [    0.000000]  [<ffffffff81ae5120>] ? early_idt_handlers+0x120/0x120
> [    0.000000]  [<ffffffff81ae5418>] x86_64_start_kernel+0x102/0x111
> [    0.000000] ---[ end trace 1ca9c978643fe899 ]---
> [    0.000000] start_kernel(): bug: interrupts were enabled *very* early, fixing it
> [    0.000000] Preemptible hierarchical RCU implementation.
> [    0.000000]  RCU restricting CPUs from NR_CPUS=8 to nr_cpu_ids=4.
> [    0.000000] NR_IRQS:4352 nr_irqs:712 16
> [    0.000000] Console: colour VGA+ 80x25
> [    0.000000] console [ttyS0] enabled, bootconsole disabled
> [    0.000000] console [ttyS0] enabled, bootconsole disabled
> 
> Do you see it?

Yes, but only after looking through it three times.

> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

> diff --git a/init/main.c b/init/main.c
> index 63534a1..f0fe0a5 100644
> --- a/init/main.c
> +++ b/init/main.c
> @@ -539,7 +539,7 @@ asmlinkage void __init start_kernel(void)
>  	 * fragile until we cpu_idle() for the first time.
>  	 */
>  	preempt_disable();
> -	if (!irqs_disabled()) {
> +	if (WARN_ON(!irqs_disabled())) {
>  		printk(KERN_WARNING "start_kernel(): bug: interrupts were "
>  				"enabled *very* early, fixing it\n");
>  		local_irq_disable();
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>