[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH for-4.21 1/2] x86/AMD: avoid REP MOVSB for Zen3/4
- To: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
- From: Jan Beulich <jbeulich@xxxxxxxx>
- Date: Thu, 9 Oct 2025 09:27:13 +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: Roger Pau Monné <roger.pau@xxxxxxxxxx>, Oleksii Kurochko <oleksii.kurochko@xxxxxxxxx>, "xen-devel@xxxxxxxxxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxxxxxxxxx>
- Delivery-date: Thu, 09 Oct 2025 07:27:22 +0000
- List-id: Xen developer discussion <xen-devel.lists.xenproject.org>
On 08.10.2025 18:33, Andrew Cooper wrote:
> On 25/09/2025 11:46 am, Jan Beulich wrote:
>> --- a/xen/arch/x86/copy_page.S
>> +++ b/xen/arch/x86/copy_page.S
>> @@ -57,6 +57,6 @@ END(copy_page_cold)
>> .endm
>>
>> FUNC(copy_page_hot)
>> - ALTERNATIVE copy_page_movsq, copy_page_movsb, X86_FEATURE_ERMS
>> + ALTERNATIVE copy_page_movsq, copy_page_movsb,
>> X86_FEATURE_XEN_REP_MOVSB
>> RET
>> END(copy_page_hot)
>
> Hmm.
>
> Overall I think this patch is an improvement.
>
> But, for any copy_page variants, we know both pointers are 4k aligned,
> so will not tickle the problem case.
Then I fear I still haven't understood the bad "may overlap" condition.
I thought that with the low 12 bits all identical in a page-copy, this
case would _specifically_ trigger the bad behavior.
> This does mess with the naming of the synthetic feature.
Short of better naming suggestions, I would keep it as is.
>> --- a/xen/arch/x86/cpu/amd.c
>> +++ b/xen/arch/x86/cpu/amd.c
>> @@ -1386,6 +1386,10 @@ static void cf_check init_amd(struct cpu
>>
>> check_syscfg_dram_mod_en();
>>
>> + if (c == &boot_cpu_data && cpu_has(c, X86_FEATURE_ERMS)
>> + && c->family != 0x19 /* Zen3/4 */)
>
> Even if this is Linux style, && on the previous line please.
No, precisely because it is Linux style. If and when we change the file
to Xen style (which probably we should), such operators would move.
Jan
|
