Re: [PATCH v2 2/3] hw/i386/acpi-build: Return a pre-computed _PRT table

["Michael S. Tsirkin" <mst@redhat.com>]

On Fri, Jun 07, 2024 at 08:58:57AM +0000, Ricardo Ribalda wrote:
> When qemu runs without kvm acceleration the ACPI executions take a great
> amount of time. If they take more than the default time (30sec), the
> ACPI calls fail and the system might not behave correctly.

Wow.

> Now the _PRT table is computed on the fly. We can drastically reduce the
> execution of the _PRT method if we return a pre-computed table.
> 
> Without this patch:
> [   51.343484] ACPI Error: Aborting method \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/psparse-529)
> [   51.527032] ACPI Error: Method execution failed \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/uteval-68)
> [   51.530049] virtio-pci 0000:00:02.0: can't derive routing for PCI INT A
> [   51.530797] virtio-pci 0000:00:02.0: PCI INT A: no GSI
> [   81.922901] ACPI Error: Aborting method \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/psparse-529)
> [   82.103534] ACPI Error: Method execution failed \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/uteval-68)
> [   82.106088] virtio-pci 0000:00:04.0: can't derive routing for PCI INT A
> [   82.106761] virtio-pci 0000:00:04.0: PCI INT A: no GSI
> [  112.192568] ACPI Error: Aborting method \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/psparse-529)
> [  112.486687] ACPI Error: Method execution failed \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/uteval-68)
> [  112.489554] virtio-pci 0000:00:05.0: can't derive routing for PCI INT A
> [  112.490027] virtio-pci 0000:00:05.0: PCI INT A: no GSI
> [  142.559448] ACPI Error: Aborting method \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/psparse-529)
> [  142.718596] ACPI Error: Method execution failed \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/uteval-68)
> [  142.722889] virtio-pci 0000:00:06.0: can't derive routing for PCI INT A
> [  142.724578] virtio-pci 0000:00:06.0: PCI INT A: no GSI
> 
> With this patch:
> [   22.938076] ACPI: \_SB_.LNKB: Enabled at IRQ 10
> [   24.214002] ACPI: \_SB_.LNKD: Enabled at IRQ 11
> [   25.465170] ACPI: \_SB_.LNKA: Enabled at IRQ 10
> [   27.944920] ACPI: \_SB_.LNKC: Enabled at IRQ 11
> 
> ACPI disassembly:
>         Scope (PCI0)
>         {
>             Method (_PRT, 0, NotSerialized)  // _PRT: PCI Routing Table
>             {
>                 Return (Package (0x80)
>                 {
>                     Package (0x04)
>                     {
>                         0xFFFF,
>                         Zero,
>                         LNKD,
>                         Zero
>                     },
> 
>                     Package (0x04)
>                     {
>                         0xFFFF,
>                         One,
>                         LNKA,
>                         Zero
>                     },
> 
>                     Package (0x04)
>                     {
>                         0xFFFF,
>                         0x02,
>                         LNKB,
>                         Zero
>                     },
> 
>                     Package (0x04)
>                     {
>                         0xFFFF,
>                         0x03,
>                         LNKC,
>                         Zero
>                     },
> 
>                     Package (0x04)
>                     {
>                         0x0001FFFF,
>                         Zero,
>                         LNKS,
>                         Zero
>                     },
> Context: https://lore.kernel.org/virtualization/20240417145544.38d7b482@imammedo.users.ipa.redhat.com/T/#t
> 
> Signed-off-by: Ricardo Ribalda <ribalda@chromium.org>
> ---
>  hw/i386/acpi-build.c | 118 ++++++++-----------------------------------
>  1 file changed, 21 insertions(+), 97 deletions(-)
> 
> diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c
> index 53f804ac16..4c14d39173 100644
> --- a/hw/i386/acpi-build.c
> +++ b/hw/i386/acpi-build.c
> @@ -725,40 +725,7 @@ static Aml *aml_pci_pdsm(void)
>      return method;
>  }
>  
> -/**
> - * build_prt_entry:
> - * @link_name: link name for PCI route entry
> - *
> - * build AML package containing a PCI route entry for @link_name
> - */
> -static Aml *build_prt_entry(const char *link_name)
> -{
> -    Aml *a_zero = aml_int(0);
> -    Aml *pkg = aml_package(4);
> -    aml_append(pkg, a_zero);
> -    aml_append(pkg, a_zero);
> -    aml_append(pkg, aml_name("%s", link_name));
> -    aml_append(pkg, a_zero);
> -    return pkg;
> -}
> -
> -/*
> - * initialize_route - Initialize the interrupt routing rule
> - * through a specific LINK:
> - *  if (lnk_idx == idx)
> - *      route using link 'link_name'
> - */
> -static Aml *initialize_route(Aml *route, const char *link_name,
> -                             Aml *lnk_idx, int idx)
> -{
> -    Aml *if_ctx = aml_if(aml_equal(lnk_idx, aml_int(idx)));
> -    Aml *pkg = build_prt_entry(link_name);
> -
> -    aml_append(if_ctx, aml_store(pkg, route));
> -
> -    return if_ctx;
> -}
> -


Can't say I like defines like this. Explains almost nothing
and is far divorced from code.
Better to do it near the 1st use:

 int nroutes = 128 /* Explain why it's 128 here */;

 rt_pkg = aml_varpackage(nroutes);



>  /*
>   * build_prt - Define interrupt rounting rules
>   *
> @@ -771,74 +738,31 @@ static Aml *initialize_route(Aml *route, const char *link_name,
>   */
>  static Aml *build_prt(bool is_pci0_prt)
>  {
> -    Aml *method, *while_ctx, *pin, *res;
> +    Aml *rt_pkg, *method;
> +    const char link_name[][2] = {"D", "A", "B", "C"};

Wouldn't it be clearer if we just made it LNKA, LNKB etc?

> +    int i;
>  
>      method = aml_method("_PRT", 0, AML_NOTSERIALIZED);
> -    res = aml_local(0);
> -    pin = aml_local(1);
> -    aml_append(method, aml_store(aml_package(128), res));
> -    aml_append(method, aml_store(aml_int(0), pin));
> +    rt_pkg = aml_varpackage(N_ROUTES);


Again, initialize where defined.

>  
> -    /* while (pin < 128) */
> -    while_ctx = aml_while(aml_lless(pin, aml_int(128)));
> -    {
> -        Aml *slot = aml_local(2);
> -        Aml *lnk_idx = aml_local(3);
> -        Aml *route = aml_local(4);
> -
> -        /* slot = pin >> 2 */
> -        aml_append(while_ctx,
> -                   aml_store(aml_shiftright(pin, aml_int(2), NULL), slot));
> -        /* lnk_idx = (slot + pin) & 3 */
> -        aml_append(while_ctx,
> -            aml_store(aml_and(aml_add(pin, slot, NULL), aml_int(3), NULL),
> -                      lnk_idx));
> -
> -        /* route[2] = "LNK[D|A|B|C]", selection based on pin % 3  */
> -        aml_append(while_ctx, initialize_route(route, "LNKD", lnk_idx, 0));
> -        if (is_pci0_prt) {
> -            Aml *if_device_1, *if_pin_4, *else_pin_4;
> -
> -            /* device 1 is the power-management device, needs SCI */
> -            if_device_1 = aml_if(aml_equal(lnk_idx, aml_int(1)));
> -            {
> -                if_pin_4 = aml_if(aml_equal(pin, aml_int(4)));
> -                {
> -                    aml_append(if_pin_4,
> -                        aml_store(build_prt_entry("LNKS"), route));
> -                }
> -                aml_append(if_device_1, if_pin_4);
> -                else_pin_4 = aml_else();
> -                {
> -                    aml_append(else_pin_4,
> -                        aml_store(build_prt_entry("LNKA"), route));
> -                }
> -                aml_append(if_device_1, else_pin_4);
> -            }
> -            aml_append(while_ctx, if_device_1);
> -        } else {
> -            aml_append(while_ctx, initialize_route(route, "LNKA", lnk_idx, 1));
> +    for (i = 0; i < N_ROUTES; i++) {
> +        Aml *pkg = aml_package(4);
> +        const char *name;
> +
> +        name = link_name[((i >> 2) + i) & 3];

better to combine definition and initialization of name.

And can we get some explanation of what is going on here?


> +
> +        if (is_pci0_prt && i == 4) {
> +            name = "S";
>          }

So this discards the previous calculation.
if {} else {} would be better.


> -        aml_append(while_ctx, initialize_route(route, "LNKB", lnk_idx, 2));
> -        aml_append(while_ctx, initialize_route(route, "LNKC", lnk_idx, 3));
> -
> -        /* route[0] = 0x[slot]FFFF */
> -        aml_append(while_ctx,
> -            aml_store(aml_or(aml_shiftleft(slot, aml_int(16)), aml_int(0xFFFF),
> -                             NULL),
> -                      aml_index(route, aml_int(0))));
> -        /* route[1] = pin & 3 */
> -        aml_append(while_ctx,
> -            aml_store(aml_and(pin, aml_int(3), NULL),
> -                      aml_index(route, aml_int(1))));
> -        /* res[pin] = route */
> -        aml_append(while_ctx, aml_store(route, aml_index(res, pin)));
> -        /* pin++ */
> -        aml_append(while_ctx, aml_increment(pin));
> +
> +        aml_append(pkg, aml_int((i << 14) | 0xFFFF));
> +        aml_append(pkg, aml_int(i & 3));

document and explain above &/|/<< logic too, please.

> +        aml_append(pkg, aml_name("LNK%s", name));
> +        aml_append(pkg, aml_int(0));
> +        aml_append(rt_pkg, pkg);
>      }
> -    aml_append(method, while_ctx);
> -    /* return res*/
> -    aml_append(method, aml_return(res));
> +
> +    aml_append(method, aml_return(rt_pkg));
>  
>      return method;
>  }
> -- 
> 2.45.2.505.gda0bf45e8d-goog