Patchwork PS3: Strange issue with kexec and FreeBSD loader

login
register
mail settings
Submitter Phileas Fogg
Date Feb. 22, 2013, 11:41 p.m.
Message ID <51280223.9070109@mail.ru>
Download mbox | patch
Permalink /patch/222687/
State Not Applicable
Headers show

Comments

Benjamin Herrenschmidt - Feb. 22, 2013, 10:45 p.m.
On Sat, 2013-02-23 at 00:41 +0100, Phileas Fogg wrote:
> Benjamin Herrenschmidt wrote:
> > On Fri, 2013-02-22 at 21:49 +0100, Phileas Fogg wrote:
> >> i wanted to let you know that i tested your advice. And let me say, it's was a
> >> damn good advice :) I can boot FreeBSD loader on Linux 3.8 now, no SHA256
> >> checksum failures. And no panics with FreeBSD LiveCD anymore too.
> >>
> >> I just inserted hard_irq_disable() after each local_irq_disable() in
> >> arch/powerpc/kernel/machine_kexec_64.c
> >
> > Awesome ! :-)
> >
> > Care to send a patch with a Signed-off-by: ?
>
> No problem, but as i said it was your idea how to fix the issue with kexec.
> Anyways here is the patch which i tested on my PS3 console with Linux 3.8.
> After applying this patch i can boot any Linux kernel starting with 2.6,
> FreeBSD loader, FreeBSD LiveCD and my own tiny ELF kernels too.
> Even OpenBSD bootloader starts now too :)
> And i don't see any failed SHA256 checksums in the purgatory code.

Thanks, but I still need the Signed-off-by: line before i can apply
it :-) (legal...)

Cheers,
Ben.

> regards
> 
>  From c17cdf38dfe180b4a571827bb547aaf9b678cf29 Mon Sep 17 00:00:00 2001
> From: Phileas Fogg <phileas-fogg@mail.ru>
> Date: Sat, 23 Feb 2013 00:32:19 +0100
> Subject: [PATCH] kexec: disable hard IRQ before kexec
> 
> Disable hard IRQ before kexec a new kernel image.
> Not doing it can result in corrupted data in the memory segments
> reserved for the new kernel.
> ---
>   arch/powerpc/kernel/machine_kexec_64.c | 3 +++
>   1 file changed, 3 insertions(+)
> 
> diff --git a/arch/powerpc/kernel/machine_kexec_64.c 
> b/arch/powerpc/kernel/machine_kexec_64.c
> index 7206701..e08b9d0 100644
> --- a/arch/powerpc/kernel/machine_kexec_64.c
> +++ b/arch/powerpc/kernel/machine_kexec_64.c
> @@ -162,6 +162,7 @@ static int kexec_all_irq_disabled = 0;
>   static void kexec_smp_down(void *arg)
>   {
>   	local_irq_disable();
> +	hard_irq_disable();
>   	mb(); /* make sure our irqs are disabled before we say they are */
>   	get_paca()->kexec_state = KEXEC_STATE_IRQS_OFF;
>   	while(kexec_all_irq_disabled == 0)
> @@ -244,6 +245,7 @@ static void kexec_prepare_cpus(void)
>   	wake_offline_cpus();
>   	smp_call_function(kexec_smp_down, NULL, /* wait */0);
>   	local_irq_disable();
> +	hard_irq_disable();
>   	mb(); /* make sure IRQs are disabled before we say they are */
>   	get_paca()->kexec_state = KEXEC_STATE_IRQS_OFF;
> 
> @@ -281,6 +283,7 @@ static void kexec_prepare_cpus(void)
>   	if (ppc_md.kexec_cpu_down)
>   		ppc_md.kexec_cpu_down(0, 0);
>   	local_irq_disable();
> +	hard_irq_disable();
>   }
> 
>   #endif /* SMP */
Phileas Fogg - Feb. 22, 2013, 11:41 p.m.
Benjamin Herrenschmidt wrote:
> On Fri, 2013-02-22 at 21:49 +0100, Phileas Fogg wrote:
>> i wanted to let you know that i tested your advice. And let me say, it's was a
>> damn good advice :) I can boot FreeBSD loader on Linux 3.8 now, no SHA256
>> checksum failures. And no panics with FreeBSD LiveCD anymore too.
>>
>> I just inserted hard_irq_disable() after each local_irq_disable() in
>> arch/powerpc/kernel/machine_kexec_64.c
>
> Awesome ! :-)
>
> Care to send a patch with a Signed-off-by: ?
>
> Cheers,
> Ben.
>
>

No problem, but as i said it was your idea how to fix the issue with kexec.
Anyways here is the patch which i tested on my PS3 console with Linux 3.8.
After applying this patch i can boot any Linux kernel starting with 2.6,
FreeBSD loader, FreeBSD LiveCD and my own tiny ELF kernels too.
Even OpenBSD bootloader starts now too :)
And i don't see any failed SHA256 checksums in the purgatory code.

regards

 From c17cdf38dfe180b4a571827bb547aaf9b678cf29 Mon Sep 17 00:00:00 2001
From: Phileas Fogg <phileas-fogg@mail.ru>
Date: Sat, 23 Feb 2013 00:32:19 +0100
Subject: [PATCH] kexec: disable hard IRQ before kexec

Disable hard IRQ before kexec a new kernel image.
Not doing it can result in corrupted data in the memory segments
reserved for the new kernel.
---
  arch/powerpc/kernel/machine_kexec_64.c | 3 +++
  1 file changed, 3 insertions(+)

Patch

diff --git a/arch/powerpc/kernel/machine_kexec_64.c 
b/arch/powerpc/kernel/machine_kexec_64.c
index 7206701..e08b9d0 100644
--- a/arch/powerpc/kernel/machine_kexec_64.c
+++ b/arch/powerpc/kernel/machine_kexec_64.c
@@ -162,6 +162,7 @@  static int kexec_all_irq_disabled = 0;
  static void kexec_smp_down(void *arg)
  {
  	local_irq_disable();
+	hard_irq_disable();
  	mb(); /* make sure our irqs are disabled before we say they are */
  	get_paca()->kexec_state = KEXEC_STATE_IRQS_OFF;
  	while(kexec_all_irq_disabled == 0)
@@ -244,6 +245,7 @@  static void kexec_prepare_cpus(void)
  	wake_offline_cpus();
  	smp_call_function(kexec_smp_down, NULL, /* wait */0);
  	local_irq_disable();
+	hard_irq_disable();
  	mb(); /* make sure IRQs are disabled before we say they are */
  	get_paca()->kexec_state = KEXEC_STATE_IRQS_OFF;

@@ -281,6 +283,7 @@  static void kexec_prepare_cpus(void)
  	if (ppc_md.kexec_cpu_down)
  		ppc_md.kexec_cpu_down(0, 0);
  	local_irq_disable();
+	hard_irq_disable();
  }

  #endif /* SMP */