[Qemu-devel] [PATCH 0/2] memory: Fix 'Kill old-style I/O dispatch'

[Qemu-devel] [PATCH 0/2] memory: Fix 'Kill old-style I/O dispatch'

[Andreas Färber]

Hello,

This series fixes endianness issues in the memory core.
Apparently the actual byte swapping had not yet been tested.

Together with the BIOS MemoryRegion patch and revert of m48t59 I/O base
this restores the PReP machine to previous behavior.

Regards,
Andreas

Cc: Avi Kivity <avi@redhat.com>
Cc: Anthony Liguori <aliguori@us.ibm.com>
Cc: Alexander Graf <agraf@suse.de>
Cc: Hervé Poussineau <hpoussin@reactos.org>

Andreas Färber (2):
  memory: Fix memory_region_wrong_endianness()
  memory: Fix adjust_endianness()

 memory.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

-- 
1.7.7

[Qemu-devel] [PATCH 1/2] memory: Fix memory_region_wrong_endianness()

[Andreas Färber]

Since commit be675c972088eba210e18dc125613e9f205a6bfb (memory: move
endianness compensation to memory core) it was checking for
TARGET_BIG_ENDIAN instead of TARGET_WORDS_BIGENDIAN, thereby not
swapping correctly for Big Endian targets.

Signed-off-by: Andreas Färber <afaerber@suse.de>
Cc: Avi Kivity <avi@redhat.com>
---
 memory.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/memory.c b/memory.c
index 394cbab..5e55a90 100644
--- a/memory.c
+++ b/memory.c
@@ -843,7 +843,7 @@ static void memory_region_destructor_rom_device(MemoryRegion *mr)
 
 static bool memory_region_wrong_endianness(MemoryRegion *mr)
 {
-#ifdef TARGET_BIG_ENDIAN
+#ifdef TARGET_WORDS_BIGENDIAN
     return mr->ops->endianness == DEVICE_LITTLE_ENDIAN;
 #else
     return mr->ops->endianness == DEVICE_BIG_ENDIAN;
-- 
1.7.7

[Qemu-devel] [PATCH 2/2] memory: Fix adjust_endianness()

[Andreas Färber]

Commit a621f38de85598a13d8d8524d1a94fc6a1818215 (Direct dispatch
through MemoryRegion) moved byte swaps to a central function.

Add a missing break, so that long-sized byte swaps don't abort.

Signed-off-by: Andreas Färber <afaerber@suse.de>
Cc: Avi Kivity <avi@redhat.com>
---
 memory.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/memory.c b/memory.c
index 5e55a90..5ab2112 100644
--- a/memory.c
+++ b/memory.c
@@ -942,6 +942,7 @@ static void adjust_endianness(MemoryRegion *mr, uint64_t *data, unsigned size)
             break;
         case 4:
             *data = bswap32(*data);
+            break;
         default:
             abort();
         }
-- 
1.7.7

Re: [Qemu-devel] [PATCH 0/2] memory: Fix 'Kill old-style I/O dispatch'

[Aurelien Jarno]

On Sat, Jan 07, 2012 at 09:45:13AM +0100, Andreas Färber wrote:
> Hello,
> 
> This series fixes endianness issues in the memory core.
> Apparently the actual byte swapping had not yet been tested.
> 
> Together with the BIOS MemoryRegion patch and revert of m48t59 I/O base
> this restores the PReP machine to previous behavior.

I confirm this totally break the big endian targets. At least MIPS and
PowerPC are affected.

Both patches are:

Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>

-- 
Aurelien Jarno                          GPG: 1024D/F1BCDB73
aurelien@aurel32.net                 http://www.aurel32.net

Re: [Qemu-devel] [PATCH 0/2] memory: Fix 'Kill old-style I/O dispatch'

[Avi Kivity]

On 01/07/2012 10:45 AM, Andreas Färber wrote:
> Hello,
>
> This series fixes endianness issues in the memory core.
> Apparently the actual byte swapping had not yet been tested.

/me hangs head.  But strangely, sparc-test, mips-test, and
ppc-virtex-ml507 (after the idstr fix) all boot for me.

> Together with the BIOS MemoryRegion patch and revert of m48t59 I/O base
> this restores the PReP machine to previous behavior.
>

Thanks, applied.

-- 
error compiling committee.c: too many arguments to function

Re: [Qemu-devel] [PATCH 0/2] memory: Fix 'Kill old-style I/O dispatch'

[Andreas Färber]

Am 08.01.2012 12:11, schrieb Avi Kivity:
> On 01/07/2012 10:45 AM, Andreas Färber wrote:
>> This series fixes endianness issues in the memory core.
>> Apparently the actual byte swapping had not yet been tested.
> 
> /me hangs head.  But strangely, sparc-test, mips-test, and
> ppc-virtex-ml507 (after the idstr fix) all boot for me.

IIUC the byte swaps only have to happen when some MemoryRegion has an
endianness other than the target.

So when CHRP uses BE and native I/O exclusively it may work, whereas
someone once summarized PReP as "a PC with a PowerPC CPU". ;)

Andreas

-- 
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg