RT and kernel debugger ( 2.6.12rc6 + RT > 48-00 )

RT and kernel debugger ( 2.6.12rc6 + RT > 48-00 )

[Serge Noiraud]

Hi,

	I would like to know what kernel debugger you propose over the RT
patch. I used to test kgdb, but since spinlock modification, it doesn't
work anymore.

Does someone work over RT to port a kernel debugger ?

Re: RT and kernel debugger ( 2.6.12rc6 + RT > 48-00 )

[Ingo Molnar]

* Serge Noiraud <serge.noiraud@bull.net> wrote:

> 	I would like to know what kernel debugger you propose over the 
> RT patch. I used to test kgdb, but since spinlock modification, it 
> doesn't work anymore.
> 
> Does someone work over RT to port a kernel debugger ?

someone had kgdb working - but i'm not using it myself. IIRC it was 
mostly an issue of changing the proper spinlocks to raw_spinlock_t.

	Ingo

Re: RT and kernel debugger ( 2.6.12rc6 + RT > 48-00 )

[Lee Revell]

On Mon, 2005-06-13 at 16:31 +0200, Serge Noiraud wrote:
> Hi,
> 
> 	I would like to know what kernel debugger you propose over the RT
> patch. I used to test kgdb, but since spinlock modification, it doesn't
> work anymore.
> 
> Does someone work over RT to port a kernel debugger ?

I got kdb to work a few months ago.  I might have only been using
PREEMPT_DESKTOP + irq threading + hard/softirq preemption.  All that was
needed are some trivial adjustments to get the patch to apply.

Lee

RE: RT and kernel debugger ( 2.6.12rc6 + RT > 48-00 )

[kus Kusche Klaus]

I was one of those who tried to get kgdb working.

Here I described how far I came:
http://www.ussg.iu.edu/hypermail/linux/kernel/0505.1/0700.html

-- 
Klaus Kusche                 (Software Development - Control Systems)
KEBA AG             Gewerbepark Urfahr, A-4041 Linz, Austria (Europe)
Tel: +43 / 732 / 7090-3120                 Fax: +43 / 732 / 7090-6301
E-Mail: kus@keba.com                                WWW: www.keba.com


> -----Original Message-----
> From: linux-kernel-owner@vger.kernel.org 
> [mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of Serge Noiraud
> Sent: Monday, June 13, 2005 4:31 PM
> To: linux-kernel
> Cc: Ingo Molnar
> Subject: RT and kernel debugger ( 2.6.12rc6 + RT > 48-00 )
> 
> 
> Hi,
> 
> 	I would like to know what kernel debugger you propose 
> over the RT
> patch. I used to test kgdb, but since spinlock modification, 
> it doesn't
> work anymore.
> 
> Does someone work over RT to port a kernel debugger ?
> 
> -
> 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/
> 

Re: RT and kernel debugger ( 2.6.12rc6 + RT > 48-00 )

[Ingo Molnar]

* kus Kusche Klaus <kus@keba.com> wrote:

> I was one of those who tried to get kgdb working.
> 
> Here I described how far I came:
> http://www.ussg.iu.edu/hypermail/linux/kernel/0505.1/0700.html

does ethernet debugging work if you disable the netpoll WARN_ON() that 
triggers?

	Ingo

Re: RT and kernel debugger ( 2.6.12rc6 + RT > 48-00 )

[Tom Rini]

On Tue, Jun 14, 2005 at 09:20:15AM +0200, Ingo Molnar wrote:
> 
> * kus Kusche Klaus <kus@keba.com> wrote:
> 
> > I was one of those who tried to get kgdb working.
> > 
> > Here I described how far I came:
> > http://www.ussg.iu.edu/hypermail/linux/kernel/0505.1/0700.html
> 
> does ethernet debugging work if you disable the netpoll WARN_ON() that 
> triggers?

With the KGDB at http://kgdb.linsyssoft.com/cvs.htm the following is all
that's needed for PREEMPT_RT to work:
--- linux-2.6.10.orig/drivers/net/kgdb_eth.c
+++ linux-2.6.10/drivers/net/kgdb_eth.c
@@ -55,22 +55,49 @@ static int eth_getDebugChar(void)
 {
 	int chr;
 
-	while (atomic_read(&in_count) == 0)
+#ifdef CONFIG_PREEMPT_RT
+	/*
+	 * A bit hairy. Netpoll API users uses mutexes (indirectly) and
+	 * thus must have interrupts enabled:
+	 */
+	local_irq_enable();
+#endif
+
+	while (atomic_read(&in_count) == 0) {
+		WARN_ON_RT(irqs_disabled());
 		netpoll_poll(&np);
+		WARN_ON_RT(irqs_disabled());
+	}
 
 	chr = in_buf[in_tail++];
 	in_tail &= (IN_BUF_SIZE - 1);
 	atomic_dec(&in_count);
+#ifdef CONFIG_PREEMPT_RT
+	local_irq_disable();
+#endif
 	return chr;
 }
 
 static void eth_flushDebugChar(void)
 {
+#ifdef CONFIG_PREEMPT_RT
+	/*
+	 * A bit hairy. Netpoll API users uses mutexes (indirectly) and
+	 * thus must have interrupts enabled:
+	 */
+	local_irq_enable();
+#endif
+
 	if (out_count && np.dev) {
+		WARN_ON_RT(irqs_disabled());
 		netpoll_send_udp(&np, out_buf, out_count);
+		WARN_ON_RT(irqs_disabled());
 		memset(out_buf, 0, sizeof(out_buf));
 		out_count = 0;
 	}
+#ifdef CONFIG_PREEMPT_RT
+	local_irq_disable();
+#endif
 }
 
 static void eth_putDebugChar(int chr)
--- linux-2.6.10.orig/kernel/kgdb.c
+++ linux-2.6.10/kernel/kgdb.c
@@ -85,7 +85,7 @@ struct kgdb_arch *kgdb_ops = &arch_kgdb_
 
 static const char hexchars[] = "0123456789abcdef";
 
-static spinlock_t slavecpulocks[NR_CPUS];
+static raw_spinlock_t slavecpulocks[NR_CPUS];
 static volatile int procindebug[NR_CPUS];
 atomic_t kgdb_setting_breakpoint;
 struct task_struct *kgdb_usethread, *kgdb_contthread;

-- 
Tom Rini
http://gate.crashing.org/~trini/

RE: RT and kernel debugger ( 2.6.12rc6 + RT > 48-00 )

[kus Kusche Klaus]

> * kus Kusche Klaus <kus@keba.com> wrote:
> > I was one of those who tried to get kgdb working.
> > 
> > Here I described how far I came:
> > http://www.ussg.iu.edu/hypermail/linux/kernel/0505.1/0700.html
> 
> does ethernet debugging work if you disable the netpoll 
> WARN_ON() that 
> triggers?

Sorry, I can't try at the moment:
That installation doesn't exist any longer,
and debuggers aren't on the current agenda of our RT linux analysis.

I'll try if I've some spare time, but no promises.

-- 
Klaus Kusche                 (Software Development - Control Systems)
KEBA AG             Gewerbepark Urfahr, A-4041 Linz, Austria (Europe)
Tel: +43 / 732 / 7090-3120                 Fax: +43 / 732 / 7090-6301
E-Mail: kus@keba.com                                WWW: www.keba.com