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

Re: [PATCH v2 7/7] x86/hap: Increase the number of initial mempool_size to 1024 pages


  • To: Petr Beneš <w1benny@xxxxxxxxx>
  • From: Jan Beulich <jbeulich@xxxxxxxx>
  • Date: Tue, 30 Apr 2024 16:47:36 +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: Anthony PERARD <anthony@xxxxxxxxxxxxxx>, Andrew Cooper <andrew.cooper3@xxxxxxxxxx>, George Dunlap <george.dunlap@xxxxxxxxxx>, Roger Pau Monné <roger.pau@xxxxxxxxxx>, xen-devel@xxxxxxxxxxxxxxxxxxxx
  • Delivery-date: Tue, 30 Apr 2024 14:47:47 +0000
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

On 28.04.2024 18:52, Petr Beneš wrote:
> From: Petr Beneš <w1benny@xxxxxxxxx>
> 
> This change anticipates scenarios where `max_altp2m` is set to its maximum
> supported value (i.e., 512), ensuring sufficient memory is allocated upfront
> to accommodate all altp2m tables without initialization failure.

And guests with fewer or even no altp2m-s still need the same bump? You
know the number of altp2m-s upon domain creation, so why bump by any more
than what's strictly needed for that?

> The necessity for this increase arises from the current mechanism where altp2m
> tables are allocated at initialization, requiring one page from the mempool
> for each altp2m view.

So that's the p2m_alloc_table() out of hap_enable()? If you're permitting
up to 512 altp2m-s, I think it needs considering to not waste up to 2Mb
without knowing how many of the altp2m-s are actually going to be used.
How complicate on-demand allocation would be I can't tell though, I have
to admit.

> --- a/tools/tests/paging-mempool/test-paging-mempool.c
> +++ b/tools/tests/paging-mempool/test-paging-mempool.c
> @@ -35,7 +35,7 @@ static struct xen_domctl_createdomain create = {
>  
>  static uint64_t default_mempool_size_bytes =
>  #if defined(__x86_64__) || defined(__i386__)
> -    256 << 12; /* Only x86 HAP for now.  x86 Shadow needs more work. */
> +    1024 << 12; /* Only x86 HAP for now.  x86 Shadow needs more work. */

I also can't derive from the description why we'd need to go from 256 to
1024 here and ...

> --- a/xen/arch/x86/mm/hap/hap.c
> +++ b/xen/arch/x86/mm/hap/hap.c
> @@ -468,7 +468,7 @@ int hap_enable(struct domain *d, u32 mode)
>      if ( old_pages == 0 )
>      {
>          paging_lock(d);
> -        rv = hap_set_allocation(d, 256, NULL);
> +        rv = hap_set_allocation(d, 1024, NULL);

... here. You talk of (up to) 512 pages there only.

Also isn't there at least one more place where the tool stack (libxl I
think) would need changing, where Dom0 ballooning needs are calculated?
And/or doesn't the pool size have a default calculation in the tool
stack, too?

Jan



 


Rackspace

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