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

Re: [PATCH v1 1/4] xen/riscv: introduce preinit_xen_time()


  • To: Oleksii Kurochko <oleksii.kurochko@xxxxxxxxx>
  • From: Jan Beulich <jbeulich@xxxxxxxx>
  • Date: Mon, 24 Mar 2025 16:56:10 +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: Alistair Francis <alistair.francis@xxxxxxx>, Bob Eshleman <bobbyeshleman@xxxxxxxxx>, Connor Davis <connojdavis@xxxxxxxxx>, Andrew Cooper <andrew.cooper3@xxxxxxxxxx>, Anthony PERARD <anthony.perard@xxxxxxxxxx>, Michal Orzel <michal.orzel@xxxxxxx>, Julien Grall <julien@xxxxxxx>, Roger Pau Monné <roger.pau@xxxxxxxxxx>, Stefano Stabellini <sstabellini@xxxxxxxxxx>, xen-devel@xxxxxxxxxxxxxxxxxxxx
  • Delivery-date: Mon, 24 Mar 2025 15:56:19 +0000
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

On 24.03.2025 16:29, Oleksii Kurochko wrote:
> 
> On 3/20/25 8:36 AM, Jan Beulich wrote:
>> On 19.03.2025 18:29, Oleksii Kurochko wrote:
>>> On 3/17/25 4:24 PM, Jan Beulich wrote:
>>>> On 11.03.2025 17:19, Oleksii Kurochko wrote:
>>>>> --- /dev/null
>>>>> +++ b/xen/arch/riscv/time.c
>>>>> @@ -0,0 +1,38 @@
>>>>> +#include <xen/device_tree.h>
>>>>> +#include <xen/init.h>
>>>>> +#include <xen/lib.h>
>>>>> +#include <xen/sections.h>
>>>>> +
>>>>> +unsigned long __ro_after_init cpu_khz; /* CPU clock frequency in kHz. */
>>>>> +unsigned long __read_mostly boot_count;
>>>> Why not also __ro_after_init? And what is this variable actually needed
>>>> for? Common code doesn't use it, so a better name (describing what it
>>>> really holds) might be desirable, even if this then means not being in
>>>> sync with Arm code.
>>> To calculate more accurate amount of time since boot.
>> Okay. But how does the name of the variable reflect that? I.e. what it
>> is that the count of is being stored? The only meaning I could associate
>> to a variable of this name is the number of boot cycles a system went
>> through. I.e. nothing that an OS (or hypervisor) would normally count.
> 
> But an OS (or hypervisor) doesn't count it, they initialize a variable
> once (in my case, it was named as boot_count) and then just subtract it from
> get_cycles() to get time relative to this variable (so since Xen boot) and not
> since power on as nothing guarantee (at least, I can't find that in the 
> RISC-V spec)
> that after power on the value in CSR_TIME will start from 0 what could lead to
> some issues, if my understanding is correct, such as if on SoC A timer starts 
> from
> let it be 1000 and on SoC B timer value starts from 5000 then all 
> measurements will be
> incorrect as Xen will think that for SoC B it was spent more time then for 
> SoC A.
> 
> What do you think if boot_count will be renamed to xen_start_time or 
> {initial_}boot_time?

Something like that, yes. Whether "time" in there is unambiguous enough I'm not
sure. "cycles" or "clock_cycles" or some such may help. I don't really want to
restrict you in what name you choose, just so long as the name reflects the
purpose in good enough a way.

Jan



 


Rackspace

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