On Thu, 31 Aug 2017 20:04:26 +0800
Dou Liyang <douly.f...@cn.fujitsu.com> wrote:
> From: Eduardo Habkost <ehabk...@redhat.com>
>
> Currently, Using the fisrt node without memory on the machine makes
> QEMU unhappy. With this example command line:
> ... \
> -m 1024M,slots=4,maxmem=32G \
> -numa node,nodeid=0 \
> -numa node,mem=1024M,nodeid=1 \
> -numa node,nodeid=2 \
> -numa node,nodeid=3 \
> Guest reports "No NUMA configuration found" and the NUMA topology is
> wrong.
>
> This is because when QEMU builds ACPI SRAT, it regards node 0 as the
> default node to deal with the memory hole(640K-1M). this means the
> node0 must have some memory(>1M), but, actually it can have no
> memory.
>
> Fix this problem by cut out the 640K hole in the same way the PCI
> 4G hole does. Also do some cleanup.
>
> Signed-off-by: Eduardo Habkost <ehabk...@redhat.com>
> Signed-off-by: Dou Liyang <douly.f...@cn.fujitsu.com>
> ---
> hw/i386/acpi-build.c | 30 +++++++++++++++++++++++-------
> 1 file changed, 23 insertions(+), 7 deletions(-)
>
> diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c
> index 98dd424..48525a1 100644
> --- a/hw/i386/acpi-build.c
> +++ b/hw/i386/acpi-build.c
> @@ -2318,6 +2318,9 @@ build_tpm2(GArray *table_data, BIOSLinker *linker)
> (void *)tpm2_ptr, "TPM2", sizeof(*tpm2_ptr), 4, NULL, NULL);
> }
>
> +#define HOLE_640K_START (640 * 1024)
> +#define HOLE_640K_END (1024 * 1024)
> +
> static void
> build_srat(GArray *table_data, BIOSLinker *linker, MachineState *machine)
> {
> @@ -2373,17 +2376,30 @@ build_srat(GArray *table_data, BIOSLinker *linker,
> MachineState *machine)
> next_base = 0;
> numa_start = table_data->len;
>
> - numamem = acpi_data_push(table_data, sizeof *numamem);
> - build_srat_memory(numamem, 0, 640 * 1024, 0, MEM_AFFINITY_ENABLED);
> - next_base = 1024 * 1024;
> for (i = 1; i < pcms->numa_nodes + 1; ++i) {
> mem_base = next_base;
> mem_len = pcms->node_mem[i - 1];
> - if (i == 1) {
> - mem_len -= 1024 * 1024;
> - }
> next_base = mem_base + mem_len;
>
> + /* Cut out the 640K hole */
> + if (mem_base <= HOLE_640K_START &&
> + next_base > HOLE_640K_START) {
> + mem_len -= next_base - HOLE_640K_START;
> + if (mem_len > 0) {
> + numamem = acpi_data_push(table_data, sizeof *numamem);
> + build_srat_memory(numamem, mem_base, mem_len, i - 1,
> + MEM_AFFINITY_ENABLED);
> + }
> +
> + /* Check for the rare case: 640K < RAM < 1M */
> + if (next_base <= HOLE_640K_END) {
> + next_base = HOLE_640K_END;
Is this assignment really necessary?
it seems that next_base will be set at the start of the loop.
> + continue;
> + }
> + mem_base = HOLE_640K_END;
> + mem_len = next_base - HOLE_640K_END;
> + }
> +
> /* Cut out the ACPI_PCI hole */
> if (mem_base <= pcms->below_4g_mem_size &&
> next_base > pcms->below_4g_mem_size) {
> @@ -2395,7 +2411,7 @@ build_srat(GArray *table_data, BIOSLinker *linker,
> MachineState *machine)
> }
> mem_base = 1ULL << 32;
> mem_len = next_base - pcms->below_4g_mem_size;
> - next_base += (1ULL << 32) - pcms->below_4g_mem_size;
> + next_base = mem_base + mem_len;
> }
> numamem = acpi_data_push(table_data, sizeof *numamem);
> build_srat_memory(numamem, mem_base, mem_len, i - 1,
