[Qemu-devel] [PATCH] Always swap endianness in DBDMA

[Qemu-devel] [PATCH] Always swap endianness in DBDMA

[Alexander Graf]

When we get an MMIO request, we always get variables in host endianness. The
only time we need to actually reverse byte order is when we read bytes from
guest memory.

Apparently the DBDMA implementation is different there. A lot of the logic
in there depends on values being big endian. Now, qemu does all the conversion
in the MMIO handlers for us already though, so it turns out that we're in
the same byte order from a C point of view, but cpu_to_be32 and be32_to_cpu
end up being nops.

This makes the code work differently on x86 (little endian) than on ppc (big
endian). On x86 it works, on ppc it doesn't.

This patch (while being seriously hacky and ugly) makes dbdma emulation work
on ppc hosts. I'll leave the real fixing to someone else.

Signed-off-by: Alexander Graf <agraf@suse.de>
CC: Laurent Vivier <Laurent@vivier.eu>
---
 hw/mac_dbdma.c |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)

diff --git a/hw/mac_dbdma.c b/hw/mac_dbdma.c
index 98dccfd..4dbfc16 100644
--- a/hw/mac_dbdma.c
+++ b/hw/mac_dbdma.c
@@ -40,6 +40,14 @@
 #include "isa.h"
 #include "mac_dbdma.h"
 
+/*
+ * XXX This is just plain wrong. Apparently we don't want to have big endian
+ *     values, but reversed endian ones. The code as is doesn't work on big
+ *     endian hosts. With these defines it does.
+ */
+#define cpu_to_be32 bswap32
+#define be32_to_cpu bswap32
+
 /* debug DBDMA */
 //#define DEBUG_DBDMA
 
-- 
1.6.0.2

[Qemu-devel] Re: [PATCH] Always swap endianness in DBDMA

[Michael S. Tsirkin]

On Tue, Dec 22, 2009 at 11:24:18AM +0100, Alexander Graf wrote:
> When we get an MMIO request, we always get variables in host endianness. The
> only time we need to actually reverse byte order is when we read bytes from
> guest memory.
> 
> Apparently the DBDMA implementation is different there. A lot of the logic
> in there depends on values being big endian. Now, qemu does all the conversion
> in the MMIO handlers for us already though, so it turns out that we're in
> the same byte order from a C point of view, but cpu_to_be32 and be32_to_cpu
> end up being nops.
> 
> This makes the code work differently on x86 (little endian) than on ppc (big
> endian). On x86 it works, on ppc it doesn't.
> 
> This patch (while being seriously hacky and ugly) makes dbdma emulation work
> on ppc hosts. I'll leave the real fixing to someone else.

Come on, 

#define cpu_to_dbdma32 bswap32
#define dbdma_to_cpu32 bswap32

and then

s/cpu_to_be32/cpu_to_dbdma32/g
s/be32_to_cpu/dbdma32_to_cpu/g

is not too hard, is it?

> Signed-off-by: Alexander Graf <agraf@suse.de>
> CC: Laurent Vivier <Laurent@vivier.eu>
> ---
>  hw/mac_dbdma.c |    8 ++++++++
>  1 files changed, 8 insertions(+), 0 deletions(-)
> 
> diff --git a/hw/mac_dbdma.c b/hw/mac_dbdma.c
> index 98dccfd..4dbfc16 100644
> --- a/hw/mac_dbdma.c
> +++ b/hw/mac_dbdma.c
> @@ -40,6 +40,14 @@
>  #include "isa.h"
>  #include "mac_dbdma.h"
>  
> +/*
> + * XXX This is just plain wrong. Apparently we don't want to have big endian
> + *     values, but reversed endian ones. The code as is doesn't work on big
> + *     endian hosts. With these defines it does.
> + */
> +#define cpu_to_be32 bswap32
> +#define be32_to_cpu bswap32
> +
>  /* debug DBDMA */
>  //#define DEBUG_DBDMA
>  
> -- 
> 1.6.0.2
> 
> 

[Qemu-devel] Re: [PATCH] Always swap endianness in DBDMA

[Alexander Graf]

Michael S. Tsirkin wrote:
> On Tue, Dec 22, 2009 at 11:24:18AM +0100, Alexander Graf wrote:
>   
>> When we get an MMIO request, we always get variables in host endianness. The
>> only time we need to actually reverse byte order is when we read bytes from
>> guest memory.
>>
>> Apparently the DBDMA implementation is different there. A lot of the logic
>> in there depends on values being big endian. Now, qemu does all the conversion
>> in the MMIO handlers for us already though, so it turns out that we're in
>> the same byte order from a C point of view, but cpu_to_be32 and be32_to_cpu
>> end up being nops.
>>
>> This makes the code work differently on x86 (little endian) than on ppc (big
>> endian). On x86 it works, on ppc it doesn't.
>>
>> This patch (while being seriously hacky and ugly) makes dbdma emulation work
>> on ppc hosts. I'll leave the real fixing to someone else.
>>     
>
> Come on, 
>
> #define cpu_to_dbdma32 bswap32
> #define dbdma_to_cpu32 bswap32
>
> and then
>
> s/cpu_to_be32/cpu_to_dbdma32/g
> s/be32_to_cpu/dbdma32_to_cpu/g
>
> is not too hard, is it?
>   

Well, if I'd want to do a sed'ish approach I'd just

s/cpu_to_be32/bswap32/g
s/be32_to_cpu/bswap32/g

The problem is that the whole define is just plain wrong which tells me
that the code is using the bswap functions incorrectly. This really
needs to be fixed by someone who knows the dbdma device. I don't see how
calling incorrect calls even more incorrect makes any difference.

Alex

[Qemu-devel] Re: [PATCH] Always swap endianness in DBDMA

[Michael S. Tsirkin]

On Tue, Dec 22, 2009 at 01:07:20PM +0100, Alexander Graf wrote:
> Michael S. Tsirkin wrote:
> > On Tue, Dec 22, 2009 at 11:24:18AM +0100, Alexander Graf wrote:
> >   
> >> When we get an MMIO request, we always get variables in host endianness. The
> >> only time we need to actually reverse byte order is when we read bytes from
> >> guest memory.
> >>
> >> Apparently the DBDMA implementation is different there. A lot of the logic
> >> in there depends on values being big endian. Now, qemu does all the conversion
> >> in the MMIO handlers for us already though, so it turns out that we're in
> >> the same byte order from a C point of view, but cpu_to_be32 and be32_to_cpu
> >> end up being nops.
> >>
> >> This makes the code work differently on x86 (little endian) than on ppc (big
> >> endian). On x86 it works, on ppc it doesn't.
> >>
> >> This patch (while being seriously hacky and ugly) makes dbdma emulation work
> >> on ppc hosts. I'll leave the real fixing to someone else.
> >>     
> >
> > Come on, 
> >
> > #define cpu_to_dbdma32 bswap32
> > #define dbdma_to_cpu32 bswap32
> >
> > and then
> >
> > s/cpu_to_be32/cpu_to_dbdma32/g
> > s/be32_to_cpu/dbdma32_to_cpu/g
> >
> > is not too hard, is it?
> >   
> 
> Well, if I'd want to do a sed'ish approach I'd just
> 
> s/cpu_to_be32/bswap32/g
> s/be32_to_cpu/bswap32/g

This would throw away some information: it is better to know whether you
are converting from or to cpu mode.  Hopefully at some point we will add
sparce annotations which will make this even more important.

> The problem is that the whole define is just plain wrong which tells me
> that the code is using the bswap functions incorrectly. This really
> needs to be fixed by someone who knows the dbdma device. I don't see how
> calling incorrect calls even more incorrect makes any difference.
> 
> Alex

At least build will not break in strange ways e.g. when someone changes
cpu_to_be32 to a macro.  I don't really know about this hardware either,
but why make it more fragile than it already is?

-- 
MST