[PATCH] powerpc: Fix dynamic relocation

[PATCH] powerpc: Fix dynamic relocation

[Alexander Graf]

Commit 5ac47f7a introduced dynamic relocation of code by manually
relocating TOC entries. However, we need to access the TOC using
the physical address that we have for it, not the virtual address
that we can't even access yet.

Drop the offset from the TOC accessing pointer.

This fixes Linux 3.9 booting on OpenBIOS for me.

Signed-off-by: Alexander Graf <agraf@suse.de>

---

I've also encountered breakage on a G5 Mac, not sure if it's related.
I'll have to test whether this is the culprit on that one too.

Ben, please make sure that this gets into 3.9.
---
 arch/powerpc/kernel/prom_init.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/powerpc/kernel/prom_init.c b/arch/powerpc/kernel/prom_init.c
index 7f7fb7f..2bf7cc3 100644
--- a/arch/powerpc/kernel/prom_init.c
+++ b/arch/powerpc/kernel/prom_init.c
@@ -2851,7 +2851,7 @@ static void reloc_toc(void)
 		(__prom_init_toc_end - __prom_init_toc_start) / sizeof(long);
 
 	/* Need to add offset to get at __prom_init_toc_start */
-	__reloc_toc(__prom_init_toc_start + offset, offset, nr_entries);
+	__reloc_toc(__prom_init_toc_start, offset, nr_entries);
 
 	mb();
 }
-- 
1.6.0.2

Re: [PATCH] powerpc: Fix dynamic relocation

[Benjamin Herrenschmidt]

On Sat, 2013-03-09 at 02:02 +0100, Alexander Graf wrote:
> Commit 5ac47f7a introduced dynamic relocation of code by manually
> relocating TOC entries. However, we need to access the TOC using
> the physical address that we have for it, not the virtual address
> that we can't even access yet.
> 
> Drop the offset from the TOC accessing pointer.
> 
> This fixes Linux 3.9 booting on OpenBIOS for me.
> 
> Signed-off-by: Alexander Graf <agraf@suse.de>
> 
> ---
> 
> I've also encountered breakage on a G5 Mac, not sure if it's related.
> I'll have to test whether this is the culprit on that one too.

That might fix the mac too.. Anton's patch assumed OF ran in real mode
in which case the top bits of the address are ignored, meaning we could
use kernel virtual addresses but that isn't the case with Apple OF which
has MMU enabled and maps memory 1:1. I suppose OpenBIOS has the same
problem ? Or is this a 32-bit issue as well ?

> Ben, please make sure that this gets into 3.9.
> ---
>  arch/powerpc/kernel/prom_init.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/arch/powerpc/kernel/prom_init.c b/arch/powerpc/kernel/prom_init.c
> index 7f7fb7f..2bf7cc3 100644
> --- a/arch/powerpc/kernel/prom_init.c
> +++ b/arch/powerpc/kernel/prom_init.c
> @@ -2851,7 +2851,7 @@ static void reloc_toc(void)
>  		(__prom_init_toc_end - __prom_init_toc_start) / sizeof(long);
>  
>  	/* Need to add offset to get at __prom_init_toc_start */
> -	__reloc_toc(__prom_init_toc_start + offset, offset, nr_entries);
> +	__reloc_toc(__prom_init_toc_start, offset, nr_entries);
>  
>  	mb();
>  }

Re: [PATCH] powerpc: Fix dynamic relocation

[Andreas Schwab]

Benjamin Herrenschmidt <benh@kernel.crashing.org> writes:

> That might fix the mac too..

It does.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: [PATCH] powerpc: Fix dynamic relocation

[Alexander Graf]

Am 09.03.2013 um 08:29 schrieb Benjamin Herrenschmidt <benh@kernel.crashing.=
org>:

> On Sat, 2013-03-09 at 02:02 +0100, Alexander Graf wrote:
>> Commit 5ac47f7a introduced dynamic relocation of code by manually
>> relocating TOC entries. However, we need to access the TOC using
>> the physical address that we have for it, not the virtual address
>> that we can't even access yet.
>>=20
>> Drop the offset from the TOC accessing pointer.
>>=20
>> This fixes Linux 3.9 booting on OpenBIOS for me.
>>=20
>> Signed-off-by: Alexander Graf <agraf@suse.de>
>>=20
>> ---
>>=20
>> I've also encountered breakage on a G5 Mac, not sure if it's related.
>> I'll have to test whether this is the culprit on that one too.
>=20
> That might fix the mac too.. Anton's patch assumed OF ran in real mode
> in which case the top bits of the address are ignored, meaning we could
> use kernel virtual addresses but that isn't the case with Apple OF which
> has MMU enabled and maps memory 1:1. I suppose OpenBIOS has the same
> problem ? Or is this a 32-bit issue as well ?

That's exactly what OpenBIOS does, yes :).

Alex

>=20
>> Ben, please make sure that this gets into 3.9.
>> ---
>> arch/powerpc/kernel/prom_init.c |    2 +-
>> 1 files changed, 1 insertions(+), 1 deletions(-)
>>=20
>> diff --git a/arch/powerpc/kernel/prom_init.c b/arch/powerpc/kernel/prom_i=
nit.c
>> index 7f7fb7f..2bf7cc3 100644
>> --- a/arch/powerpc/kernel/prom_init.c
>> +++ b/arch/powerpc/kernel/prom_init.c
>> @@ -2851,7 +2851,7 @@ static void reloc_toc(void)
>>        (__prom_init_toc_end - __prom_init_toc_start) / sizeof(long);
>>=20
>>    /* Need to add offset to get at __prom_init_toc_start */
>> -    __reloc_toc(__prom_init_toc_start + offset, offset, nr_entries);
>> +    __reloc_toc(__prom_init_toc_start, offset, nr_entries);
>>=20
>>    mb();
>> }
>=20
>=20