Patchwork Fix leul_to_cpu on big endian hosts

login
register
mail settings
Submitter Alexander Graf
Date May 11, 2010, 3:38 p.m.
Message ID <1273592297-18370-1-git-send-email-agraf@suse.de>
Download mbox | patch
Permalink /patch/52297/
State New
Headers show

Comments

Alexander Graf - May 11, 2010, 3:38 p.m.
Commit 213acd2e introduced leul_to_cpu with a special code path for big endian
hosts. Unfortunately that code used preprocessor magic that didn't work.

This patch replaces the explicit ##s by glue() which is proven to work reliably,
enabling me to compile qemu on ppc again.

Signed-off-by: Alexander Graf <agraf@suse.de>
---
 bswap.h |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
Alexander Graf - May 31, 2010, 6:38 p.m.
Alexander Graf wrote:
> Commit 213acd2e introduced leul_to_cpu with a special code path for big endian
> hosts. Unfortunately that code used preprocessor magic that didn't work.
>
> This patch replaces the explicit ##s by glue() which is proven to work reliably,
> enabling me to compile qemu on ppc again.
>   

Ping?

Alex
Aurelien Jarno - May 31, 2010, 6:43 p.m.
On Tue, May 11, 2010 at 05:38:17PM +0200, Alexander Graf wrote:
> Commit 213acd2e introduced leul_to_cpu with a special code path for big endian
> hosts. Unfortunately that code used preprocessor magic that didn't work.
> 
> This patch replaces the explicit ##s by glue() which is proven to work reliably,
> enabling me to compile qemu on ppc again.
> 
> Signed-off-by: Alexander Graf <agraf@suse.de>
> ---
>  bswap.h |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/bswap.h b/bswap.h
> index 956f3fa..20caae6 100644
> --- a/bswap.h
> +++ b/bswap.h
> @@ -205,7 +205,7 @@ static inline void cpu_to_be32wu(uint32_t *p, uint32_t v)
>  
>  #ifdef HOST_WORDS_BIGENDIAN
>  #define cpu_to_32wu cpu_to_be32wu
> -#define leul_to_cpu(v) le ## HOST_LONG_BITS ## _to_cpu(v)
> +#define leul_to_cpu(v) glue(glue(le,HOST_LONG_BITS),_to_cpu)(v)
>  #else
>  #define cpu_to_32wu cpu_to_le32wu
>  #define leul_to_cpu(v) (v)

While the patch looks correct, I am not able to reproduce the issue. Do
you have more details about it?
Alexander Graf - May 31, 2010, 6:44 p.m.
Aurelien Jarno wrote:
> On Tue, May 11, 2010 at 05:38:17PM +0200, Alexander Graf wrote:
>   
>> Commit 213acd2e introduced leul_to_cpu with a special code path for big endian
>> hosts. Unfortunately that code used preprocessor magic that didn't work.
>>
>> This patch replaces the explicit ##s by glue() which is proven to work reliably,
>> enabling me to compile qemu on ppc again.
>>
>> Signed-off-by: Alexander Graf <agraf@suse.de>
>> ---
>>  bswap.h |    2 +-
>>  1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/bswap.h b/bswap.h
>> index 956f3fa..20caae6 100644
>> --- a/bswap.h
>> +++ b/bswap.h
>> @@ -205,7 +205,7 @@ static inline void cpu_to_be32wu(uint32_t *p, uint32_t v)
>>  
>>  #ifdef HOST_WORDS_BIGENDIAN
>>  #define cpu_to_32wu cpu_to_be32wu
>> -#define leul_to_cpu(v) le ## HOST_LONG_BITS ## _to_cpu(v)
>> +#define leul_to_cpu(v) glue(glue(le,HOST_LONG_BITS),_to_cpu)(v)
>>  #else
>>  #define cpu_to_32wu cpu_to_le32wu
>>  #define leul_to_cpu(v) (v)
>>     
>
> While the patch looks correct, I am not able to reproduce the issue. Do
> you have more details about it?
>   

When building on a KVM capable PPC host (or S390 for that matter), I get
the following:

cc1: warnings being treated as errors
/home/agraf/release/qemu/kvm-all.c: In function
‘kvm_get_dirty_pages_log_range’:
/home/agraf/release/qemu/kvm-all.c:305: error: implicit declaration of
function ‘leHOST_LONG_BITS_to_cpu’
make[1]: *** [kvm-all.o] Error 1
make[1]: *** Waiting for unfinished jobs....
make: *** [subdir-ppc64-softmmu] Error 2


Alex
Aurelien Jarno - May 31, 2010, 6:58 p.m.
On Mon, May 31, 2010 at 08:44:55PM +0200, Alexander Graf wrote:
> Aurelien Jarno wrote:
> > On Tue, May 11, 2010 at 05:38:17PM +0200, Alexander Graf wrote:
> >   
> >> Commit 213acd2e introduced leul_to_cpu with a special code path for big endian
> >> hosts. Unfortunately that code used preprocessor magic that didn't work.
> >>
> >> This patch replaces the explicit ##s by glue() which is proven to work reliably,
> >> enabling me to compile qemu on ppc again.
> >>
> >> Signed-off-by: Alexander Graf <agraf@suse.de>
> >> ---
> >>  bswap.h |    2 +-
> >>  1 files changed, 1 insertions(+), 1 deletions(-)
> >>
> >> diff --git a/bswap.h b/bswap.h
> >> index 956f3fa..20caae6 100644
> >> --- a/bswap.h
> >> +++ b/bswap.h
> >> @@ -205,7 +205,7 @@ static inline void cpu_to_be32wu(uint32_t *p, uint32_t v)
> >>  
> >>  #ifdef HOST_WORDS_BIGENDIAN
> >>  #define cpu_to_32wu cpu_to_be32wu
> >> -#define leul_to_cpu(v) le ## HOST_LONG_BITS ## _to_cpu(v)
> >> +#define leul_to_cpu(v) glue(glue(le,HOST_LONG_BITS),_to_cpu)(v)
> >>  #else
> >>  #define cpu_to_32wu cpu_to_le32wu
> >>  #define leul_to_cpu(v) (v)
> >>     
> >
> > While the patch looks correct, I am not able to reproduce the issue. Do
> > you have more details about it?
> >   
> 
> When building on a KVM capable PPC host (or S390 for that matter), I get
> the following:
> 
> cc1: warnings being treated as errors
> /home/agraf/release/qemu/kvm-all.c: In function
> ‘kvm_get_dirty_pages_log_range’:
> /home/agraf/release/qemu/kvm-all.c:305: error: implicit declaration of
> function ‘leHOST_LONG_BITS_to_cpu’
> make[1]: *** [kvm-all.o] Error 1
> make[1]: *** Waiting for unfinished jobs....
> make: *** [subdir-ppc64-softmmu] Error 2
> 

Ok, that make sense now. I have tried to build on a big endian host, but
KVM was not enabled. Strangely this function is also used at other
places, so I was expecting the build to fail.

Applied now.

Patch

diff --git a/bswap.h b/bswap.h
index 956f3fa..20caae6 100644
--- a/bswap.h
+++ b/bswap.h
@@ -205,7 +205,7 @@  static inline void cpu_to_be32wu(uint32_t *p, uint32_t v)
 
 #ifdef HOST_WORDS_BIGENDIAN
 #define cpu_to_32wu cpu_to_be32wu
-#define leul_to_cpu(v) le ## HOST_LONG_BITS ## _to_cpu(v)
+#define leul_to_cpu(v) glue(glue(le,HOST_LONG_BITS),_to_cpu)(v)
 #else
 #define cpu_to_32wu cpu_to_le32wu
 #define leul_to_cpu(v) (v)