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

Re: [PATCH] AMD/IOMMU: Improve register_iommu_exclusion_range()


  • To: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
  • From: Jan Beulich <jbeulich@xxxxxxxx>
  • Date: Wed, 19 Jun 2024 09:45:52 +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>, Xen-devel <xen-devel@xxxxxxxxxxxxxxxxxxxx>
  • Delivery-date: Wed, 19 Jun 2024 07:46:01 +0000
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

On 18.06.2024 20:31, Andrew Cooper wrote:
>  * Use 64bit accesses instead of 32bit accesses
>  * Simplify the constant names
>  * Pull base into a local variable to avoid it being reloaded because of the
>    memory clobber in writeq().
> 
> No functional change.
> 
> Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
> ---
> CC: Jan Beulich <JBeulich@xxxxxxxx>
> CC: Roger Pau Monné <roger.pau@xxxxxxxxxx>
> 
> RFC.  This is my proposed way of cleaning up the whole IOMMU file.  The
> diffstat speaks for itself.

Absolutely.

> I've finally found the bit in the AMD IOMMU spec which says 64bit accesses are
> permitted:
> 
>   3.4 IOMMU MMIO Registers:
> 
>   Software access to IOMMU registers may not be larger than 64 bits. Accesses
>   must be aligned to the size of the access and the size in bytes must be a
>   power of two. Software may use accesses as small as one byte.

I take it that the use of 32-bit writes was because of the past need
also work in a 32-bit hypervisor, not because of perceived restrictions
by the spec.

> --- a/xen/drivers/passthrough/amd/iommu-defs.h
> +++ b/xen/drivers/passthrough/amd/iommu-defs.h
> @@ -338,22 +338,10 @@ union amd_iommu_control {
>  };
>  
>  /* Exclusion Register */
> -#define IOMMU_EXCLUSION_BASE_LOW_OFFSET              0x20
> -#define IOMMU_EXCLUSION_BASE_HIGH_OFFSET     0x24
> -#define IOMMU_EXCLUSION_LIMIT_LOW_OFFSET     0x28
> -#define IOMMU_EXCLUSION_LIMIT_HIGH_OFFSET    0x2C
> -#define IOMMU_EXCLUSION_BASE_LOW_MASK                0xFFFFF000U
> -#define IOMMU_EXCLUSION_BASE_LOW_SHIFT               12
> -#define IOMMU_EXCLUSION_BASE_HIGH_MASK               0xFFFFFFFFU
> -#define IOMMU_EXCLUSION_BASE_HIGH_SHIFT              0
> -#define IOMMU_EXCLUSION_RANGE_ENABLE_MASK    0x00000001U
> -#define IOMMU_EXCLUSION_RANGE_ENABLE_SHIFT   0
> -#define IOMMU_EXCLUSION_ALLOW_ALL_MASK               0x00000002U
> -#define IOMMU_EXCLUSION_ALLOW_ALL_SHIFT              1
> -#define IOMMU_EXCLUSION_LIMIT_LOW_MASK               0xFFFFF000U
> -#define IOMMU_EXCLUSION_LIMIT_LOW_SHIFT              12
> -#define IOMMU_EXCLUSION_LIMIT_HIGH_MASK              0xFFFFFFFFU
> -#define IOMMU_EXCLUSION_LIMIT_HIGH_SHIFT     0
> +#define IOMMU_MMIO_EXCLUSION_BASE           0x20
> +#define   EXCLUSION_RANGE_ENABLE            (1 << 0)
> +#define   EXCLUSION_ALLOW_ALL               (1 << 1)
> +#define IOMMU_MMIO_EXCLUSION_LIMIT          0x28

Just one question here: Previously you suggested we switch to bitfields
for anything like this, and we've already done so with e.g.
union amd_iommu_control and union amd_iommu_ext_features. IOW I wonder
if we wouldn't better strive to be consistent in this regard. Or if not,
what the (written or unwritten) guidelines are when to use which
approach.

Jan



 


Rackspace

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