[PATCH] drm/amdkfd: add ACPI SRAT parsing for topology

Eric Huang jinhuieric.huang at amd.com
Tue May 4 14:35:38 UTC 2021


Good point. Thanks.

Eric

On 2021-05-04 10:30 a.m., Lazar, Lijo wrote:
>
> [AMD Official Use Only - Internal Distribution Only]
>
>
> Converting using pxm_to_node and then comparing against pxm value 
> looks a bit odd. Shouldn't the comparsion be between equals - node to 
> node or pxm to pxm?
>
> Thanks,
> Lijo
> ------------------------------------------------------------------------
> *From:* Huang, JinHuiEric <JinHuiEric.Huang at amd.com>
> *Sent:* Tuesday, May 4, 2021 7:30:44 PM
> *To:* Lazar, Lijo <Lijo.Lazar at amd.com>; amd-gfx at lists.freedesktop.org 
> <amd-gfx at lists.freedesktop.org>
> *Subject:* Re: [PATCH] drm/amdkfd: add ACPI SRAT parsing for topology
> Like I answer Oak's question,
> "For GCD parsing, the relation of GCD to CCD is defined by AMD, generic
> parsing in srat.c is considering a GCD as a new numa node which is not
> suitable for our need."
>
> GCD's pxm domain will get a wrong numa node which may be bigger than CCD
> domains, so I have to do a sanity check to correct it.
>
> Regards,
> Eric
>
> On 2021-05-04 3:46 a.m., Lazar, Lijo wrote:
> > [AMD Public Use]
> >
> >> *numa_node > max_pxm
> > Why numa node number is compared to a proximity domain? Since you 
> are already using pxm_to_node() API, assume that should take care.
> >
> > That also will avoid parsing ACPI_SRAT_TYPE_CPU_AFFINITY structs.
> >
> > Thanks,
> > Lijo
> >
> >
> > -----Original Message-----
> > From: amd-gfx <amd-gfx-bounces at lists.freedesktop.org> On Behalf Of 
> Eric Huang
> > Sent: Wednesday, April 28, 2021 8:42 PM
> > To: amd-gfx at lists.freedesktop.org
> > Cc: Huang, JinHuiEric <JinHuiEric.Huang at amd.com>
> > Subject: [PATCH] drm/amdkfd: add ACPI SRAT parsing for topology
> >
> > In NPS4 BIOS we need to find the closest numa node when creating 
> topology io link between cpu and gpu, if PCI driver doesn't set it.
> >
> > Signed-off-by: Eric Huang <jinhuieric.huang at amd.com>
> > ---
> >   drivers/gpu/drm/amd/amdkfd/kfd_crat.c | 94 ++++++++++++++++++++++++++-
> >   1 file changed, 91 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c 
> b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
> > index 38d45711675f..57518136c7d7 100644
> > --- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
> > +++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
> > @@ -1759,6 +1759,87 @@ static int kfd_fill_gpu_memory_affinity(int 
> *avail_size,
> >        return 0;
> >   }
> >
> > +#ifdef CONFIG_ACPI
> > +static void kfd_find_numa_node_in_srat(struct kfd_dev *kdev,
> > +             int *numa_node)
> > +{
> > +     struct acpi_table_header *table_header = NULL;
> > +     struct acpi_subtable_header *sub_header = NULL;
> > +     unsigned long table_end, subtable_len;
> > +     u32 pci_id = pci_domain_nr(kdev->pdev->bus) << 16 |
> > +                     pci_dev_id(kdev->pdev);
> > +     u32 bdf;
> > +     acpi_status status;
> > +     struct acpi_srat_cpu_affinity *cpu;
> > +     struct acpi_srat_generic_affinity *gpu;
> > +     int pxm = 0, max_pxm = 0;
> > +     bool found = false;
> > +
> > +     /* Fetch the SRAT table from ACPI */
> > +     status = acpi_get_table(ACPI_SIG_SRAT, 0, &table_header);
> > +     if (status == AE_NOT_FOUND) {
> > +             pr_warn("SRAT table not found\n");
> > +             return;
> > +     } else if (ACPI_FAILURE(status)) {
> > +             const char *err = acpi_format_exception(status);
> > +             pr_err("SRAT table error: %s\n", err);
> > +             return;
> > +     }
> > +
> > +     table_end = (unsigned long)table_header + table_header->length;
> > +
> > +     /* Parse all entries looking for a match. */
> > +
> > +     sub_header = (struct acpi_subtable_header *)
> > +                     ((unsigned long)table_header +
> > +                     sizeof(struct acpi_table_srat));
> > +     subtable_len = sub_header->length;
> > +
> > +     while (((unsigned long)sub_header) + subtable_len  < table_end) {
> > +             /*
> > +              * If length is 0, break from this loop to avoid
> > +              * infinite loop.
> > +              */
> > +             if (subtable_len == 0) {
> > +                     pr_err("SRAT invalid zero length\n");
> > +                     break;
> > +             }
> > +
> > +             switch (sub_header->type) {
> > +             case ACPI_SRAT_TYPE_CPU_AFFINITY:
> > +                     cpu = (struct acpi_srat_cpu_affinity *)sub_header;
> > +                     pxm = *((u32 *)cpu->proximity_domain_hi) << 8 |
> > + cpu->proximity_domain_lo;
> > +                     if (pxm > max_pxm)
> > +                             max_pxm = pxm;
> > +                     break;
> > +             case ACPI_SRAT_TYPE_GENERIC_AFFINITY:
> > +                     gpu = (struct acpi_srat_generic_affinity 
> *)sub_header;
> > +                     bdf = *((u16 *)(&gpu->device_handle[0])) << 16 |
> > +                                     *((u16 
> *)(&gpu->device_handle[2]));
> > +                     if (bdf == pci_id) {
> > +                             found = true;
> > +                             *numa_node = 
> pxm_to_node(gpu->proximity_domain);
> > +                     }
> > +                     break;
> > +             default:
> > +                     break;
> > +             }
> > +
> > +             if (found)
> > +                     break;
> > +
> > +             sub_header = (struct acpi_subtable_header *)
> > +                             ((unsigned long)sub_header + 
> subtable_len);
> > +             subtable_len = sub_header->length;
> > +     }
> > +
> > +     /* workaround bad cpu-gpu binding case */
> > +     if (found && (*numa_node < 0 || *numa_node > max_pxm))
> > +             *numa_node = 0;
> > +}
> > +#endif
> > +
> >   /* kfd_fill_gpu_direct_io_link - Fill in direct io link from GPU
> >    * to its NUMA node
> >    *  @avail_size: Available size in the memory
> > @@ -1774,6 +1855,9 @@ static int 
> kfd_fill_gpu_direct_io_link_to_cpu(int *avail_size,
> >                        uint32_t proximity_domain)
> >   {
> >        struct amdgpu_device *adev = (struct amdgpu_device *)kdev->kgd;
> > +#ifdef CONFIG_NUMA
> > +     int numa_node = 0;
> > +#endif
> >
> >        *avail_size -= sizeof(struct crat_subtype_iolink);
> >        if (*avail_size < 0)
> > @@ -1805,9 +1889,13 @@ static int 
> kfd_fill_gpu_direct_io_link_to_cpu(int *avail_size,
> >
> >        sub_type_hdr->proximity_domain_from = proximity_domain;  
> #ifdef CONFIG_NUMA
> > -     if (kdev->pdev->dev.numa_node == NUMA_NO_NODE)
> > -             sub_type_hdr->proximity_domain_to = 0;
> > -     else
> > +     if (kdev->pdev->dev.numa_node == NUMA_NO_NODE) { #ifdef 
> CONFIG_ACPI
> > +             kfd_find_numa_node_in_srat(kdev, &numa_node); #endif
> > +             sub_type_hdr->proximity_domain_to = numa_node;
> > + set_dev_node(&kdev->pdev->dev, numa_node);
> > +     } else
> >                sub_type_hdr->proximity_domain_to = 
> kdev->pdev->dev.numa_node;  #else
> >        sub_type_hdr->proximity_domain_to = 0;
> > --
> > 2.17.1
> >
> > _______________________________________________
> > amd-gfx mailing list
> > amd-gfx at lists.freedesktop.org
> > 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Famd-gfx&data=04%7C01%7Clijo.lazar%40amd.com%7C96808a6aab7b40861eeb08d90a580524%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637552195438132467%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=ipBmGTX%2Fokto1zRuQ8jlDA8p%2B8BOjHZa5WGGKNJszEY%3D&reserved=0 
> <https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Famd-gfx&data=04%7C01%7Clijo.lazar%40amd.com%7C96808a6aab7b40861eeb08d90a580524%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637552195438132467%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=ipBmGTX%2Fokto1zRuQ8jlDA8p%2B8BOjHZa5WGGKNJszEY%3D&reserved=0>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/amd-gfx/attachments/20210504/1539e8b0/attachment-0001.htm>


More information about the amd-gfx mailing list