[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH] RISCV/bitops: Use Zbb to provide arch-optimised bitops


  • To: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
  • From: Jan Beulich <jbeulich@xxxxxxxx>
  • Date: Fri, 7 Mar 2025 07:45:53 +0100
  • Autocrypt: addr=jbeulich@xxxxxxxx; keydata= xsDiBFk3nEQRBADAEaSw6zC/EJkiwGPXbWtPxl2xCdSoeepS07jW8UgcHNurfHvUzogEq5xk hu507c3BarVjyWCJOylMNR98Yd8VqD9UfmX0Hb8/BrA+Hl6/DB/eqGptrf4BSRwcZQM32aZK 7Pj2XbGWIUrZrd70x1eAP9QE3P79Y2oLrsCgbZJfEwCgvz9JjGmQqQkRiTVzlZVCJYcyGGsD /0tbFCzD2h20ahe8rC1gbb3K3qk+LpBtvjBu1RY9drYk0NymiGbJWZgab6t1jM7sk2vuf0Py O9Hf9XBmK0uE9IgMaiCpc32XV9oASz6UJebwkX+zF2jG5I1BfnO9g7KlotcA/v5ClMjgo6Gl MDY4HxoSRu3i1cqqSDtVlt+AOVBJBACrZcnHAUSuCXBPy0jOlBhxPqRWv6ND4c9PH1xjQ3NP nxJuMBS8rnNg22uyfAgmBKNLpLgAGVRMZGaGoJObGf72s6TeIqKJo/LtggAS9qAUiuKVnygo 3wjfkS9A3DRO+SpU7JqWdsveeIQyeyEJ/8PTowmSQLakF+3fote9ybzd880fSmFuIEJldWxp Y2ggPGpiZXVsaWNoQHN1c2UuY29tPsJgBBMRAgAgBQJZN5xEAhsDBgsJCAcDAgQVAggDBBYC AwECHgECF4AACgkQoDSui/t3IH4J+wCfQ5jHdEjCRHj23O/5ttg9r9OIruwAn3103WUITZee e7Sbg12UgcQ5lv7SzsFNBFk3nEQQCACCuTjCjFOUdi5Nm244F+78kLghRcin/awv+IrTcIWF hUpSs1Y91iQQ7KItirz5uwCPlwejSJDQJLIS+QtJHaXDXeV6NI0Uef1hP20+y8qydDiVkv6l IreXjTb7DvksRgJNvCkWtYnlS3mYvQ9NzS9PhyALWbXnH6sIJd2O9lKS1Mrfq+y0IXCP10eS FFGg+Av3IQeFatkJAyju0PPthyTqxSI4lZYuJVPknzgaeuJv/2NccrPvmeDg6Coe7ZIeQ8Yj t0ARxu2xytAkkLCel1Lz1WLmwLstV30g80nkgZf/wr+/BXJW/oIvRlonUkxv+IbBM3dX2OV8 AmRv1ySWPTP7AAMFB/9PQK/VtlNUJvg8GXj9ootzrteGfVZVVT4XBJkfwBcpC/XcPzldjv+3 HYudvpdNK3lLujXeA5fLOH+Z/G9WBc5pFVSMocI71I8bT8lIAzreg0WvkWg5V2WZsUMlnDL9 mpwIGFhlbM3gfDMs7MPMu8YQRFVdUvtSpaAs8OFfGQ0ia3LGZcjA6Ik2+xcqscEJzNH+qh8V m5jjp28yZgaqTaRbg3M/+MTbMpicpZuqF4rnB0AQD12/3BNWDR6bmh+EkYSMcEIpQmBM51qM EKYTQGybRCjpnKHGOxG0rfFY1085mBDZCH5Kx0cl0HVJuQKC+dV2ZY5AqjcKwAxpE75MLFkr wkkEGBECAAkFAlk3nEQCGwwACgkQoDSui/t3IH7nnwCfcJWUDUFKdCsBH/E5d+0ZnMQi+G0A nAuWpQkjM1ASeQwSHEeAWPgskBQL
  • Cc: Oleksii Kurochko <oleksii.kurochko@xxxxxxxxx>, Xen-devel <xen-devel@xxxxxxxxxxxxxxxxxxxx>
  • Delivery-date: Fri, 07 Mar 2025 06:46:18 +0000
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

On 06.03.2025 21:19, Andrew Cooper wrote:
> On 05/03/2025 7:34 am, Jan Beulich wrote:
>> On 28.02.2025 17:24, Andrew Cooper wrote:
>>> On 27/02/2025 8:11 am, Jan Beulich wrote:
>>>> On 26.02.2025 18:20, Andrew Cooper wrote:
>>>>> --- a/xen/arch/riscv/include/asm/bitops.h
>>>>> +++ b/xen/arch/riscv/include/asm/bitops.h
>>>>> @@ -125,6 +125,13 @@ static inline void clear_bit(int nr, volatile void 
>>>>> *p)
>>>>>  #undef NOT
>>>>>  #undef __AMO
>>>>>  
>>>>> +#define arch_ffs(x)     ((x) ? 1 + __builtin_ctz(x) : 0)
>>>>> +#define arch_ffsl(x)    ((x) ? 1 + __builtin_ctzl(x) : 0)
>>>>> +#define arch_fls(x)     ((x) ? 32 - __builtin_clz(x) : 0)
>>>> I fear you won't like me to say this, but can't we avoid baking in yet
>>>> another assumption on sizeof(int) == 4, by using at least sizeof(int) * 8
>>>> here (yet better might be sizeof(int) * BITS_PER_BYTE)?
>>> Yes and no.
>>>
>>> No, because 32 here is consistent with ARM and PPC when it comes to
>>> arch_fls().  Given the effort it took to get these consistent, I'm not
>>> interested in letting them diverge.
>>>
>>> But, if someone wants to introduce BITS_PER_INT to mirror BITS_PER_LONG
>>> and use it consistently, then that would be ok too.
> 
> Oleksii: I see your patch is committed, but when I said "use it
> consistently", I meant "patch ARM and PPC too".
>> I was actually hoping to eliminate BITS_PER_LONG at some point, in favor
>> of using sizeof(long) * BITS_PER_BYTE. (Surely in common code we could
>> retain a shorthand of that name, if so desired, but I see no reason why
>> each arch would need to provide all three BITS_PER_{BYTE,INT,LONG}.)
> 
> The concern is legibility and clarity.
> 
> This:
> 
>     ((x) ? 32 - __builtin_clz(x) : 0)
> 
> is a clear expression in a way that this:
> 
>     ((x) ? (sizeof(int) * BITS_PER_BYTE) - __builtin_clz(x) : 0)
> 
> is not.  The problem is the extra binary expression, and this:
> 
>     ((x) ? BITS_PER_INT - __builtin_clz(x) : 0)
> 
> is still clear, because the reader doesn't have to perform a multiply to
> just to figure out what's going on.
> 
> 
> It is definitely stupid to have each architecture provide their own
> BITS_PER_*.  The compiler is in a superior position to provide those
> details, and it should be in a common location.
> 
> I don't particularly mind how those constants are derived, but one key
> thing that BITS_PER_* can do which sizeof() can't is be used in #ifdef/etc.

This is a fair point indeed.

Jan



 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.