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

Re: [PATCH v7 04/14] xen/arm: add Dom0 cache coloring support


  • To: Carlo Nonato <carlo.nonato@xxxxxxxxxxxxxxx>
  • From: Jan Beulich <jbeulich@xxxxxxxx>
  • Date: Tue, 19 Mar 2024 16:30:52 +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: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>, George Dunlap <george.dunlap@xxxxxxxxxx>, Julien Grall <julien@xxxxxxx>, Stefano Stabellini <sstabellini@xxxxxxxxxx>, Wei Liu <wl@xxxxxxx>, Bertrand Marquis <bertrand.marquis@xxxxxxx>, Michal Orzel <michal.orzel@xxxxxxx>, Volodymyr Babchuk <Volodymyr_Babchuk@xxxxxxxx>, Marco Solieri <marco.solieri@xxxxxxxxxxxxxxx>, xen-devel@xxxxxxxxxxxxxxxxxxxx
  • Delivery-date: Tue, 19 Mar 2024 15:31:02 +0000
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

On 15.03.2024 11:58, Carlo Nonato wrote:
> --- a/docs/misc/xen-command-line.pandoc
> +++ b/docs/misc/xen-command-line.pandoc
> @@ -963,6 +963,15 @@ Controls for the dom0 IOMMU setup.
>  
>  Specify a list of IO ports to be excluded from dom0 access.
>  
> +### dom0-llc-colors
> +> `= List of [ <integer> | <integer>-<integer> ]`
> +
> +> Default: `All available LLC colors`
> +
> +Specify dom0 LLC color configuration. This option is available only when
> +`CONFIG_LLC_COLORING` is enabled. If the parameter is not set, all available
> +colors are used.

My reservation towards this being a top-level option remains.

> --- a/xen/common/llc-coloring.c
> +++ b/xen/common/llc-coloring.c
> @@ -18,6 +18,63 @@ integer_param("llc-nr-ways", llc_nr_ways);
>  /* Number of colors available in the LLC */
>  static unsigned int __ro_after_init max_nr_colors;
>  
> +static unsigned int __initdata dom0_colors[CONFIG_NR_LLC_COLORS];
> +static unsigned int __initdata dom0_num_colors;
> +
> +/*
> + * Parse the coloring configuration given in the buf string, following the
> + * syntax below.
> + *
> + * COLOR_CONFIGURATION ::= COLOR | RANGE,...,COLOR | RANGE
> + * RANGE               ::= COLOR-COLOR
> + *
> + * Example: "0,2-6,15-16" represents the set of colors: 0,2,3,4,5,6,15,16.
> + */
> +static int __init parse_color_config(const char *buf, unsigned int *colors,
> +                                     unsigned int max_num_colors,
> +                                     unsigned int *num_colors)
> +{
> +    const char *s = buf;
> +
> +    *num_colors = 0;
> +
> +    while ( *s != '\0' )
> +    {
> +        unsigned int color, start, end;
> +
> +        start = simple_strtoul(s, &s, 0);
> +
> +        if ( *s == '-' )    /* Range */
> +        {
> +            s++;
> +            end = simple_strtoul(s, &s, 0);
> +        }
> +        else                /* Single value */
> +            end = start;
> +
> +        if ( start > end || (end - start) > (UINT_MAX - *num_colors) ||
> +             (*num_colors + (end - start)) >= max_num_colors )
> +            return -EINVAL;
> +
> +        for ( color = start; color <= end; color++ )
> +            colors[(*num_colors)++] = color;

I can't spot any range check on start/end/color itself. In fact I was first
meaning to ask why the return value of simple_strtoul() is silently clipped
from unsigned long to unsigned int. Don't forget that a range specification
may easily degenerate into a negative number (due to a simple oversight or
typo), which would then be converted to a huge positive one.

> @@ -41,6 +98,22 @@ static void print_colors(const unsigned int *colors, 
> unsigned int num_colors)
>      printk(" }\n");
>  }
>  
> +static bool check_colors(const unsigned int *colors, unsigned int num_colors)
> +{
> +    unsigned int i;
> +
> +    for ( i = 0; i < num_colors; i++ )
> +    {
> +        if ( colors[i] >= max_nr_colors )
> +        {
> +            printk(XENLOG_ERR "LLC color %u >= %u\n", colors[i], 
> max_nr_colors);
> +            return false;
> +        }
> +    }
> +
> +    return true;
> +}

Oh, here's the range checking of the color values themselves. Perhaps
a comment in parse_color_config() would help.

> @@ -91,6 +164,61 @@ void cf_check domain_dump_llc_colors(const struct domain 
> *d)
>      print_colors(d->llc_colors, d->num_llc_colors);
>  }
>  
> +static int domain_set_default_colors(struct domain *d)
> +{
> +    unsigned int *colors = xmalloc_array(unsigned int, max_nr_colors);
> +    unsigned int i;
> +
> +    if ( !colors )
> +        return -ENOMEM;
> +
> +    printk(XENLOG_WARNING
> +           "LLC color config not found for %pd, using all colors\n", d);
> +
> +    for ( i = 0; i < max_nr_colors; i++ )
> +        colors[i] = i;
> +
> +    d->llc_colors = colors;
> +    d->num_llc_colors = max_nr_colors;
> +
> +    return 0;
> +}

If this function is expected to actually come into play, wouldn't it
make sense to set up such an array just once, and re-use it wherever
necessary?

Also right here both this and check_colors() could be __init. I
understand that subsequent patches will also want to use the
functions at runtime, but until then this looks slightly wrong. I'd
like to ask that such aspects be mentioned in the description, to
avoid respective questions.

> +int __init dom0_set_llc_colors(struct domain *d)
> +{
> +    unsigned int *colors;
> +
> +    if ( !dom0_num_colors )
> +        return domain_set_default_colors(d);
> +
> +    if ( !check_colors(dom0_colors, dom0_num_colors) )
> +    {
> +        printk(XENLOG_ERR "Bad LLC color config for %pd\n", d);
> +        return -EINVAL;
> +    }
> +
> +    colors = xmalloc_array(unsigned int, dom0_num_colors);
> +    if ( !colors )
> +        return -ENOMEM;
> +
> +    /* Static type checking */
> +    (void)(colors == dom0_colors);

Btw, a means to avoid this would by to use typeof() in the declaration
of "colors".

Jan



 


Rackspace

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