2021-03-08 12:54:05

by Tiezhu Yang

Subject: [PATCH] MIPS: Check __clang__ to avoid performance influence with GCC in csum_tcpudp_nofold()

The asm code in csum_tcpudp_nofold() is performance-critical, I am sorry
for the poorly considered implementation about the performance influence
with GCC in the commit 198688edbf77 ("MIPS: Fix inline asm input/output
type mismatch in checksum.h used with Clang").

With this patch, we can build successfully by both GCC and Clang,
at the same time, we can avoid the potential performance influence
with GCC.

Signed-off-by: Tiezhu Yang <[email protected]>
---
arch/mips/include/asm/checksum.h | 10 ++++++++++
1 file changed, 10 insertions(+)

diff --git a/arch/mips/include/asm/checksum.h b/arch/mips/include/asm/checksum.h
index 1e6c135..64d353e 100644
--- a/arch/mips/include/asm/checksum.h
+++ b/arch/mips/include/asm/checksum.h
@@ -130,7 +130,9 @@ static inline __wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr,
__u32 len, __u8 proto,
__wsum sum)
{
+#ifdef __clang__
unsigned long tmp = (__force unsigned long)sum;
+#endif

__asm__(
" .set push # csum_tcpudp_nofold\n"
@@ -159,7 +161,11 @@ static inline __wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr,
" addu %0, $1 \n"
#endif
" .set pop"
+#ifdef __clang__
: "=r" (tmp)
+#else
+ : "=r" (sum)
+#endif
: "0" ((__force unsigned long)daddr),
"r" ((__force unsigned long)saddr),
#ifdef __MIPSEL__
@@ -169,7 +175,11 @@ static inline __wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr,
#endif
"r" ((__force unsigned long)sum));

+#ifdef __clang__
return (__force __wsum)tmp;
+#else
+ return sum;
+#endif
}
#define csum_tcpudp_nofold csum_tcpudp_nofold

--
2.1.0

2021-03-08 16:54:25

by David Laight

Subject: RE: [PATCH] MIPS: Check __clang__ to avoid performance influence with GCC in csum_tcpudp_nofold()

From: Tiezhu Yang
> Sent: 08 March 2021 12:50
>
> The asm code in csum_tcpudp_nofold() is performance-critical, I am sorry
> for the poorly considered implementation about the performance influence
> with GCC in the commit 198688edbf77 ("MIPS: Fix inline asm input/output
> type mismatch in checksum.h used with Clang").
>
> With this patch, we can build successfully by both GCC and Clang,
> at the same time, we can avoid the potential performance influence
> with GCC.
>
> Signed-off-by: Tiezhu Yang <[email protected]>
> ---
> arch/mips/include/asm/checksum.h | 10 ++++++++++
> 1 file changed, 10 insertions(+)
>
> diff --git a/arch/mips/include/asm/checksum.h b/arch/mips/include/asm/checksum.h
> index 1e6c135..64d353e 100644
> --- a/arch/mips/include/asm/checksum.h
> +++ b/arch/mips/include/asm/checksum.h
> @@ -130,7 +130,9 @@ static inline __wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr,
> __u32 len, __u8 proto,
> __wsum sum)
> {
> +#ifdef __clang__
> unsigned long tmp = (__force unsigned long)sum;
> +#endif

What happens if you make the above:
#ifdef __clang__
unsigned long tmp = (__force unsigned long)sum;
#else
__wsum tmp = sum;
#endif

and then leave the rest of the function the same for both compilers.
Maybe do s/sum/sum_in/,s/tmp/sum/ to reduce the changes.

David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)

2021-03-09 02:25:21

by Tiezhu Yang

Subject: Re: [PATCH] MIPS: Check __clang__ to avoid performance influence with GCC in csum_tcpudp_nofold()

On 03/09/2021 12:52 AM, David Laight wrote:
> From: Tiezhu Yang
>> Sent: 08 March 2021 12:50
>>
>> The asm code in csum_tcpudp_nofold() is performance-critical, I am sorry
>> for the poorly considered implementation about the performance influence
>> with GCC in the commit 198688edbf77 ("MIPS: Fix inline asm input/output
>> type mismatch in checksum.h used with Clang").
>>
>> With this patch, we can build successfully by both GCC and Clang,
>> at the same time, we can avoid the potential performance influence
>> with GCC.
>>
>> Signed-off-by: Tiezhu Yang <[email protected]>
>> ---
>> arch/mips/include/asm/checksum.h | 10 ++++++++++
>> 1 file changed, 10 insertions(+)
>>
>> diff --git a/arch/mips/include/asm/checksum.h b/arch/mips/include/asm/checksum.h
>> index 1e6c135..64d353e 100644
>> --- a/arch/mips/include/asm/checksum.h
>> +++ b/arch/mips/include/asm/checksum.h
>> @@ -130,7 +130,9 @@ static inline __wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr,
>> __u32 len, __u8 proto,
>> __wsum sum)
>> {
>> +#ifdef __clang__
>> unsigned long tmp = (__force unsigned long)sum;
>> +#endif
> What happens if you make the above:
> #ifdef __clang__
> unsigned long tmp = (__force unsigned long)sum;
> #else
> __wsum tmp = sum;
> #endif
>
> and then leave the rest of the function the same for both compilers.
> Maybe do s/sum/sum_in/,s/tmp/sum/ to reduce the changes.

Hi David,

Thank you very much.

As you suggested, the following changes looks much better,
I will test it and then send v2 later.

diff --git a/arch/mips/include/asm/checksum.h
b/arch/mips/include/asm/checksum.h
index 1e6c135..80eddd4 100644
--- a/arch/mips/include/asm/checksum.h
+++ b/arch/mips/include/asm/checksum.h
@@ -128,9 +128,13 @@ static inline __sum16 ip_fast_csum(const void *iph,
unsigned int ihl)

static inline __wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr,
__u32 len, __u8 proto,
- __wsum sum)
+ __wsum sum_in)
{
- unsigned long tmp = (__force unsigned long)sum;
+#ifdef __clang__
+ unsigned long sum = (__force unsigned long)sum_in;
+#else
+ __wsum sum = sum_in;
+#endif

__asm__(
" .set push # csum_tcpudp_nofold\n"
@@ -159,7 +163,7 @@ static inline __wsum csum_tcpudp_nofold(__be32
saddr, __be32 daddr,
" addu %0, $1 \n"
#endif
" .set pop"
- : "=r" (tmp)
+ : "=r" (sum)
: "0" ((__force unsigned long)daddr),
"r" ((__force unsigned long)saddr),
#ifdef __MIPSEL__
@@ -169,7 +173,7 @@ static inline __wsum csum_tcpudp_nofold(__be32
saddr, __be32 daddr,
#endif
"r" ((__force unsigned long)sum));

- return (__force __wsum)tmp;
+ return (__force __wsum)sum;
}
#define csum_tcpudp_nofold csum_tcpudp_nofold

Thanks,
Tiezhu

>
> David
>
> -
> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
> Registration No: 1397386 (Wales)