smdk6400/S3C6400 failure on execution of init process

smdk6400/S3C6400 failure on execution of init process

[Markus Schwarzenberg]

booting linux-2.6.32-rc5 on our smdk6400 (S3C6400) fails somewhere in/after 
the asm call in arch/arm/kernel/sys_arm.c:kernel_execve, when the init 
process is to be run.

The preceding do_execve returned OK, which suggests that the /init file is 
found, executable etc. in rootfs. 

The failure happens for both booting from  sd-card as well as from initramfs.
I've tried several /init applications, from busybox to a simple wait(1); loop
all with the same behavior.

What's the best way to debug this?

Messages (first two lines from additional diagnostics in 
          do_execve [**] / kernel_execve [*] ):

  do_execve (/endless-loop-nonstatic-arm) succeeded
  kernel_execve calling asm
  mmc0: starting CMD18 arg 30868800 flags 000000b5
  mmc0:     blksz 512 blocks 72 flags 00000200 tsac 100 ms nsac 0
  mmc0:     CMD12 arg 00000000 flags 0000049d
  sdhci [sdhci_irq()]: *** mmc0 got interrupt: 0x00000001
  sdhci [sdhci_irq()]: *** mmc0 got interrupt: 0x00000002
  sdhci [sdhci_irq()]: *** mmc0 got interrupt: 0x00000001
  mmc0: req done (CMD18): 0: 00000900 00000000 00000000 00000000
  mmc0:     36864 bytes transferred: 0
  mmc0:     (CMD12): 0: 00000b00 00000000 00000000 00000000
  Kernel panic - not syncing: Attempted to kill init!
  Backtrace: 
  [<c002a9f0>] (dump_backtrace+0x0/0x114) from [<c0275358>]
  (dump_stack+0x18/0x1c)
   r7:c781ff64 r6:c781d900 r5:c781d900 r4:c034201c
  [<c0275340>] (dump_stack+0x0/0x1c) from [<c02753a8>] (panic+0x4c/0x11c)
  [<c027535c>] (panic+0x0/0x11c) from [<c0048990>] (do_exit+0x74/0x5b4)
   r3:c032a428 r2:c781d900 r1:c781fc28 r0:c02df366
  [<c004891c>] (do_exit+0x0/0x5b4) from [<c0048f54>]
  (do_group_exit+0x84/0xb8)
  [<c0048ed0>] (do_group_exit+0x0/0xb8) from [<c0052cc8>]
  (get_signal_to_deliver+0x2b8/0x2e8)
   r4:00000004
  [<c0052a10>] (get_signal_to_deliver+0x0/0x2e8) from [<c00291c0>]
  (do_notify_resume+0x64/0x5b8)
  [<c002915c>] (do_notify_resume+0x0/0x5b8) from [<c0026f2c>]
  (work_pending+0x1c/0x20)


Regards,

Markus Schwarzenberg

[*] additional messages generated in kernel_execve

--- linux-2.6.32-rc5.orig/arch/arm/kernel/sys_arm.c     2009-10-16 02:41:50.000000000 +0200
+++ linux-2.6.32-rc5/arch/arm/kernel/sys_arm.c  2009-10-28 16:11:25.000000000 +0100
@@ -265,6 +265,7 @@
        struct pt_regs regs;
        int ret;
 
+       printk("kernel_execve %s\n", filename);
        memset(&regs, 0, sizeof(struct pt_regs));
        ret = do_execve((char *)filename, (char __user * __user *)argv,
                        (char __user * __user *)envp, &regs);
@@ -275,6 +276,7 @@
         * Save argc to the register structure for userspace.
         */
        regs.ARM_r0 = ret;
+       printk("kernel_execve calling asm\n");
 
        /*
         * We were successful.  We won't be returning to our caller, but
@@ -296,6 +298,7 @@
                : "r0", "r1", "r2", "r3", "ip", "lr", "memory");
 
  out:
+       printk("kernel_execve return\n");
        return ret;
 }
 EXPORT_SYMBOL(kernel_execve);

[**] additional messages generated in do_execve

--- linux-2.6.32-rc5.orig/fs/exec.c  2009-10-16 02:41:50.000000000 +0200
+++ linux-2.6.32-rc5/fs/exec.c  2009-10-28 10:20:46.000000000 +0100
@@ -1362,6 +1362,7 @@
        current->stack_start = current->mm->start_stack;
 
        /* execve succeeded */
+       printk("do_execve (%s) succeeded\n", filename);
        current->fs->in_exec = 0;
        current->in_execve = 0;
        acct_update_integrals(current);
@@ -1392,6 +1393,7 @@
        if (displaced)
                reset_files_struct(displaced);
 out_ret:
+       printk("do_execve (%s) had errors\n", filename);
        return retval;
 }

smdk6400/S3C6400 failure on execution of init process

[Ben Dooks]

On Wed, Oct 28, 2009 at 04:20:37PM +0100, Markus Schwarzenberg wrote:
> 
> booting linux-2.6.32-rc5 on our smdk6400 (S3C6400) fails somewhere in/after 
> the asm call in arch/arm/kernel/sys_arm.c:kernel_execve, when the init 
> process is to be run.

Which is the last kernel version that ran this ok?
 
> The preceding do_execve returned OK, which suggests that the /init file is 
> found, executable etc. in rootfs. 
> 
> The failure happens for both booting from  sd-card as well as from initramfs.
> I've tried several /init applications, from busybox to a simple wait(1); loop
> all with the same behavior.
> 
> What's the best way to debug this?
> 
> Messages (first two lines from additional diagnostics in 
>           do_execve [**] / kernel_execve [*] ):

Does the kernel have vfp support compiled in?
 
>   do_execve (/endless-loop-nonstatic-arm) succeeded
>   kernel_execve calling asm
>   mmc0: starting CMD18 arg 30868800 flags 000000b5
>   mmc0:     blksz 512 blocks 72 flags 00000200 tsac 100 ms nsac 0
>   mmc0:     CMD12 arg 00000000 flags 0000049d
>   sdhci [sdhci_irq()]: *** mmc0 got interrupt: 0x00000001
>   sdhci [sdhci_irq()]: *** mmc0 got interrupt: 0x00000002
>   sdhci [sdhci_irq()]: *** mmc0 got interrupt: 0x00000001
>   mmc0: req done (CMD18): 0: 00000900 00000000 00000000 00000000
>   mmc0:     36864 bytes transferred: 0
>   mmc0:     (CMD12): 0: 00000b00 00000000 00000000 00000000

The SD card read seems to have gone ok.

>   Kernel panic - not syncing: Attempted to kill init!
>   Backtrace: 
>   [<c002a9f0>] (dump_backtrace+0x0/0x114) from [<c0275358>]
>   (dump_stack+0x18/0x1c)
>    r7:c781ff64 r6:c781d900 r5:c781d900 r4:c034201c
>   [<c0275340>] (dump_stack+0x0/0x1c) from [<c02753a8>] (panic+0x4c/0x11c)
>   [<c027535c>] (panic+0x0/0x11c) from [<c0048990>] (do_exit+0x74/0x5b4)
>    r3:c032a428 r2:c781d900 r1:c781fc28 r0:c02df366
>   [<c004891c>] (do_exit+0x0/0x5b4) from [<c0048f54>]
>   (do_group_exit+0x84/0xb8)
>   [<c0048ed0>] (do_group_exit+0x0/0xb8) from [<c0052cc8>]
>   (get_signal_to_deliver+0x2b8/0x2e8)
>    r4:00000004
>   [<c0052a10>] (get_signal_to_deliver+0x0/0x2e8) from [<c00291c0>]
>   (do_notify_resume+0x64/0x5b8)
>   [<c002915c>] (do_notify_resume+0x0/0x5b8) from [<c0026f2c>]
>   (work_pending+0x1c/0x20)
> 
> 
> Regards,
> 
> Markus Schwarzenberg
> 
> [*] additional messages generated in kernel_execve
> 
> --- linux-2.6.32-rc5.orig/arch/arm/kernel/sys_arm.c     2009-10-16 02:41:50.000000000 +0200
> +++ linux-2.6.32-rc5/arch/arm/kernel/sys_arm.c  2009-10-28 16:11:25.000000000 +0100
> @@ -265,6 +265,7 @@
>         struct pt_regs regs;
>         int ret;
>  
> +       printk("kernel_execve %s\n", filename);
>         memset(&regs, 0, sizeof(struct pt_regs));
>         ret = do_execve((char *)filename, (char __user * __user *)argv,
>                         (char __user * __user *)envp, &regs);
> @@ -275,6 +276,7 @@
>          * Save argc to the register structure for userspace.
>          */
>         regs.ARM_r0 = ret;
> +       printk("kernel_execve calling asm\n");
>  
>         /*
>          * We were successful.  We won't be returning to our caller, but
> @@ -296,6 +298,7 @@
>                 : "r0", "r1", "r2", "r3", "ip", "lr", "memory");
>  
>   out:
> +       printk("kernel_execve return\n");
>         return ret;
>  }
>  EXPORT_SYMBOL(kernel_execve);
> 
> [**] additional messages generated in do_execve
> 
> --- linux-2.6.32-rc5.orig/fs/exec.c  2009-10-16 02:41:50.000000000 +0200
> +++ linux-2.6.32-rc5/fs/exec.c  2009-10-28 10:20:46.000000000 +0100
> @@ -1362,6 +1362,7 @@
>         current->stack_start = current->mm->start_stack;
>  
>         /* execve succeeded */
> +       printk("do_execve (%s) succeeded\n", filename);
>         current->fs->in_exec = 0;
>         current->in_execve = 0;
>         acct_update_integrals(current);
> @@ -1392,6 +1393,7 @@
>         if (displaced)
>                 reset_files_struct(displaced);
>  out_ret:
> +       printk("do_execve (%s) had errors\n", filename);
>         return retval;
>  }
> 
> 
> 
> 
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

-- 
-- 
Ben

Q:      What's a light-year?
A:      One-third less calories than a regular year.

smdk6400/S3C6400 failure on execution of init process

[Rabin Vincent]

On Wed, Oct 28, 2009 at 04:20:37PM +0100, Markus Schwarzenberg wrote:
> 
> booting linux-2.6.32-rc5 on our smdk6400 (S3C6400) fails somewhere in/after 
> the asm call in arch/arm/kernel/sys_arm.c:kernel_execve, when the init 
> process is to be run.
> 
> The preceding do_execve returned OK, which suggests that the /init file is 
> found, executable etc. in rootfs. 
> 
> The failure happens for both booting from  sd-card as well as from initramfs.
> I've tried several /init applications, from busybox to a simple wait(1); loop
> all with the same behavior.
> 
> What's the best way to debug this?

As a first step, enable CONFIG_DEBUG_USER in your kernel and pass
"user_debug=31" on the kernel command line.  You'll get more info about
why init's being killed.

Rabin

smdk6400/S3C6400 failure on execution of init process

[Markus Schwarzenberg]

On Wed, 28 Oct 2009 15:30:17 +0000 Ben Dooks <ben-linux@fluff.org> wrote:

> On Wed, Oct 28, 2009 at 04:20:37PM +0100, Markus Schwarzenberg wrote:
> > 
> > booting linux-2.6.32-rc5 on our smdk6400 (S3C6400) fails somewhere in/after 
> > the asm call in arch/arm/kernel/sys_arm.c:kernel_execve, when the init 
> > process is to be run.
> 
> Which is the last kernel version that ran this ok?

None. 2.6.32-rc5 is basically the first we try. (after no success with 
2.6.31-rc6)

 
> Does the kernel have vfp support compiled in?

No.

--
Markus Schwarzenberg

smdk6400/S3C6400 failure on execution of init process

[Markus Schwarzenberg]

On Wed, 28 Oct 2009 21:09:16 +0530 Rabin Vincent <rabin@rab.in> wrote:

> On Wed, Oct 28, 2009 at 04:20:37PM +0100, Markus Schwarzenberg wrote:
> > 
> > booting linux-2.6.32-rc5 on our smdk6400 (S3C6400) fails somewhere in/after 
> > the asm call in arch/arm/kernel/sys_arm.c:kernel_execve, when the init 
> > process is to be run.
> > 
> > The preceding do_execve returned OK, which suggests that the /init file is 
> > found, executable etc. in rootfs. 
> > 
> > The failure happens for both booting from  sd-card as well as from initramfs.
> > I've tried several /init applications, from busybox to a simple wait(1); loop
> > all with the same behavior.
> > 
> > What's the best way to debug this?
> 
> As a first step, enable CONFIG_DEBUG_USER in your kernel and pass
> "user_debug=31" on the kernel command line.  You'll get more info about
> why init's being killed.

thanks, did this, the additional messages are (context see [*]):

  [1] endless-loop-no: obsolete system call 00000000.
  Code: e08f1001 e1a0c000 e3a0702d ef000000 (e3700a01) 

Still have to investigate how to interpret this...

--
Markus Schwarzenberg

[*]


mmc0:     (CMD12): 0: 00000b00 00000000 00000000 00000000
do_execve (/endless-loop-nonstatic-arm) succeeded
kernel_execve calling asm
mmc0: starting CMD18 arg 30868800 flags 000000b5
mmc0:     blksz 512 blocks 72 flags 00000200 tsac 100 ms nsac 0
mmc0:     CMD12 arg 00000000 flags 0000049d
sdhci [sdhci_irq()]: *** mmc0 got interrupt: 0x00000001
sdhci [sdhci_irq()]: *** mmc0 got interrupt: 0x00000002
sdhci [sdhci_irq()]: *** mmc0 got interrupt: 0x00000001
mmc0: req done (CMD18): 0: 00000900 00000000 00000000 00000000
mmc0:     36864 bytes transferred: 0
mmc0:     (CMD12): 0: 00000b00 00000000 00000000 00000000
[1] endless-loop-no: obsolete system call 00000000.
Code: e08f1001 e1a0c000 e3a0702d ef000000 (e3700a01) 
Kernel panic - not syncing: Attempted to kill init!
Backtrace: 
[<c002a9f0>] (dump_backtrace+0x0/0x114) from [<c0275358>] (dump_stack+0x18/0x1c)
 r7:c781ff64 r6:c781d900 r5:c781d900 r4:c034201c
[<c0275340>] (dump_stack+0x0/0x1c) from [<c02753a8>] (panic+0x4c/0x11c)
[<c027535c>] (panic+0x0/0x11c) from [<c0048990>] (do_exit+0x74/0x5b4)
 r3:c032a428 r2:c781d900 r1:c781fc28 r0:c02df366
[<c004891c>] (do_exit+0x0/0x5b4) from [<c0048f54>] (do_group_exit+0x84/0xb8)
[<c0048ed0>] (do_group_exit+0x0/0xb8) from [<c0052cc8>] (get_signal_to_deliver+0x2b8/0x2e8)
 r4:00000004
[<c0052a10>] (get_signal_to_deliver+0x0/0x2e8) from [<c00291c0>] (do_notify_resume+0x64/0x5b8)
[<c002915c>] (do_notify_resume+0x0/0x5b8) from [<c0026f2c>] (work_pending+0x1c/0x20)

smdk6400/S3C6400 failure on execution of init process

[Rabin Vincent]

On Wed, Oct 28, 2009 at 04:59:49PM +0100, Markus Schwarzenberg wrote:
> On Wed, 28 Oct 2009 21:09:16 +0530 Rabin Vincent <rabin@rab.in> wrote:
> 
> > On Wed, Oct 28, 2009 at 04:20:37PM +0100, Markus Schwarzenberg wrote:
> > > 
> > > booting linux-2.6.32-rc5 on our smdk6400 (S3C6400) fails somewhere in/after 
> > > the asm call in arch/arm/kernel/sys_arm.c:kernel_execve, when the init 
> > > process is to be run.
> > 
> > As a first step, enable CONFIG_DEBUG_USER in your kernel and pass
> > "user_debug=31" on the kernel command line.  You'll get more info about
> > why init's being killed.
> 
> thanks, did this, the additional messages are (context see [*]):
> 
>   [1] endless-loop-no: obsolete system call 00000000.
>   Code: e08f1001 e1a0c000 e3a0702d ef000000 (e3700a01) 

EABI-OABI mismatch.  Your userspace is EABI, but your kernel isn't.
Enable "Use the ARM EABI to compile the kernel" in "Kernel Features".

Also, re: your reply to Ben, you have to select one of the floating
point options, otherwise you'll have other such issues.

Rabin.

smdk6400/S3C6400 failure on execution of init process

[Markus Schwarzenberg]

On Wed, 28 Oct 2009 21:40:28 +0530 Rabin Vincent <rabin@rab.in> wrote:

> > thanks, did this, the additional messages are (context see [*]):
> > > >   [1] endless-loop-no: obsolete system call 00000000.
> >   Code: e08f1001 e1a0c000 e3a0702d ef000000 (e3700a01) 
> 
> EABI-OABI mismatch.  Your userspace is EABI, but your kernel isn't.
> Enable "Use the ARM EABI to compile the kernel" in "Kernel Features".

thanks. this was the problem.

--
Markus