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

Re: [PATCH v7 04/19] xen: introduce generic non-atomic test_*bit()


  • To: Oleksii <oleksii.kurochko@xxxxxxxxx>
  • From: Jan Beulich <jbeulich@xxxxxxxx>
  • Date: Fri, 5 Apr 2024 10:05:05 +0200
  • 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: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>, Ross Lagerwall <ross.lagerwall@xxxxxxxxxx>, Stefano Stabellini <sstabellini@xxxxxxxxxx>, Julien Grall <julien@xxxxxxx>, Bertrand Marquis <bertrand.marquis@xxxxxxx>, Michal Orzel <michal.orzel@xxxxxxx>, Volodymyr Babchuk <Volodymyr_Babchuk@xxxxxxxx>, Andrew Cooper <andrew.cooper3@xxxxxxxxxx>, George Dunlap <george.dunlap@xxxxxxxxxx>, Shawn Anastasio <sanastasio@xxxxxxxxxxxxxxxxxxxxx>, Roger Pau Monné <roger.pau@xxxxxxxxxx>, xen-devel@xxxxxxxxxxxxxxxxxxxx
  • Delivery-date: Fri, 05 Apr 2024 08:05:15 +0000
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

On 05.04.2024 09:56, Oleksii wrote:
> On Fri, 2024-04-05 at 08:11 +0200, Jan Beulich wrote:
>> On 04.04.2024 18:24, Oleksii wrote:
>>> On Thu, 2024-04-04 at 18:12 +0200, Jan Beulich wrote:
>>>> On 04.04.2024 17:45, Oleksii wrote:
>>>>> On Thu, 2024-04-04 at 15:22 +0200, Jan Beulich wrote:
>>>>>> On 03.04.2024 12:19, Oleksii Kurochko wrote:
>>>>>>> --- a/xen/include/xen/bitops.h
>>>>>>> +++ b/xen/include/xen/bitops.h
>>>>>>> @@ -65,10 +65,164 @@ static inline int
>>>>>>> generic_flsl(unsigned
>>>>>>> long
>>>>>>> x)
>>>>>>>   * scope
>>>>>>>   */
>>>>>>>  
>>>>>>> +#define BITOP_BITS_PER_WORD 32
>>>>>>> +/* typedef uint32_t bitop_uint_t; */
>>>>>>> +#define bitop_uint_t uint32_t
>>>>>>
>>>>>> So no arch overrides permitted anymore at all?
>>>>> Not really, I agree that it is ugly, but I expected that arch
>>>>> will
>>>>> use
>>>>> undef to override.
>>>>
>>>> Which would be fine in principle, just that Misra wants us to
>>>> avoid
>>>> #undef-s
>>>> (iirc).
>>> Could you please give me a recommendation how to do that better?
>>>
>>> The reason why I put this defintions before inclusion of
>>> asm/bitops.h
>>> as RISC-V specific code uses these definitions inside it, so they
>>> should be defined before asm/bitops.h; other option is to define
>>> these
>>> definitions inside asm/bitops.h for each architecture.
>>
>> Earlier on you had it that other way already (in a different header,
>> but the principle is the same): Move the generic definitions
>> immediately
>> past inclusion of asm/bitops.h and frame them with #ifndef.
> It can be done in this way:
> xen/bitops.h:
>    ...
>    #include <asm/bitops.h>
>    
>    #ifndef BITOP_TYPE
>    #define BITOP_BITS_PER_WORD 32
>    /* typedef uint32_t bitop_uint_t; */
>    #define bitop_uint_t uint32_t
>    #endif
>    ...
>    
> But then RISC-V will fail as it is using bitop_uint_t inside
> asm/bitops.h.
> So, at least, for RISC-V it will be needed to add asm/bitops.h:
>    #define BITOP_BITS_PER_WORD 32
>    /* typedef uint32_t bitop_uint_t; */
>    #define bitop_uint_t uint32_t
>    
> 
> It seems to me that this breaks the idea of having these macro
> definitions generic, as RISC-V will redefine BITOP_BITS_PER_WORD and
> bitop_uint_t with the same values as the generic ones.

I don't follow. Right now patch 7 has

#undef BITOP_BITS_PER_WORD
#undef bitop_uint_t

#define BITOP_BITS_PER_WORD BITS_PER_LONG
#define bitop_uint_t unsigned long

You'd drop the #undef-s and keep the #define-s. You want to override them
both, after all.

A problem would arise for _another_ arch wanting to use these (default)
types in its asm/bitops.h. Which then could still be solved by having a
types-only header. Recall the discussion on the last summit of us meaning
to switch to such a model anyway (perhaps it being xen/types/bitops.h and
asm/types/bitops.h then), in a broader fashion? IOW for now you could use
the simple approach as long as no other arch needs the types in its
asm/bitops.h. Later we would introduce the types-only headers, thus
catering for possible future uses.

Jan



 


Rackspace

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