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

Re: [PATCH 06/10] x86/ucode: Enforce invariant about module selection


  • To: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
  • From: Jan Beulich <jbeulich@xxxxxxxx>
  • Date: Mon, 28 Oct 2024 14:56:20 +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: Roger Pau Monné <roger.pau@xxxxxxxxxx>, "Daniel P . Smith" <dpsmith@xxxxxxxxxxxxxxxxxxxx>, Xen-devel <xen-devel@xxxxxxxxxxxxxxxxxxxx>
  • Delivery-date: Mon, 28 Oct 2024 13:56:31 +0000
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

On 28.10.2024 10:18, Andrew Cooper wrote:
> The work to add the `ucode=nmi` cmdline option left a subtle corner case.
> Both scan and an explicit index could be selected, and we could really find a
> CPIO archive and explicit microcode file.
> 
> Worse, because the if/else chains for processing ucode_{blob,mod} are opposite
> ways around in early_microcode_load() and microcode_init_cache(), we can
> genuinely perform early microcode loading from the CPIO archive, then cache
> from the explicit file.
> 
> Therefore, enforce that only one selection method can be active.

Question is - is this really the best of all possible behaviors? One may want
to use one approach as the fallback for the other, e.g. preferably use what
the CPIO has, but fall back to something pre-installed on the boot or EFI
partition.

> @@ -139,12 +148,16 @@ static int __init cf_check parse_ucode(const char *s)
>          else if ( !ucode_mod_forced ) /* Not forced by EFI */
>          {
>              if ( (val = parse_boolean("scan", s, ss)) >= 0 )
> -                ucode_scan = val;
> +            {
> +                opt_scan = val;
> +                opt_mod_idx = 0;
> +            }
>              else
>              {
>                  const char *q;
>  
> -                ucode_mod_idx = simple_strtol(s, &q, 0);
> +                opt_scan = false;
> +                opt_mod_idx = simple_strtol(s, &q, 0);
>                  if ( q != ss )
>                      rc = -EINVAL;
>              }

I think this latter part rather wants to be

                opt_mod_idx = simple_strtol(s, &q, 0);
                if ( q != ss )
                {
                    opt_mod_idx = 0;
                    rc = -EINVAL;
                }
                else
                    opt_scan = false;

to prevent a malformed ucode= to clobber an earlier wellformed ucode=scan.
(There are limits to this of course, as an out-of-range value would still
invalidate the "scan" request.)

> @@ -817,17 +830,42 @@ static int __init early_microcode_load(struct boot_info 
> *bi)
>      const void *data = NULL;
>      size_t size;
>      struct microcode_patch *patch;
> +    int idx = opt_mod_idx;
> +
> +    /*
> +     * Cmdline parsing ensures this invariant holds, so that we don't end up
> +     * trying to mix multiple ways of finding the microcode.
> +     */
> +    ASSERT(idx == 0 || !opt_scan);
>  
> -    if ( ucode_mod_idx < 0 )
> -        ucode_mod_idx += bi->nr_modules;
> -    if ( ucode_mod_idx <= 0 || ucode_mod_idx >= bi->nr_modules ||
> -         !__test_and_clear_bit(ucode_mod_idx, bi->module_map) )
> -        goto scan;
> -    ucode_mod = *bi->mods[ucode_mod_idx].mod;
> - scan:

Oh, the goto and label are going away here anyway. Never mind the comment on
the earlier patch then.

Jan



 


Rackspace

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