Re: [Qemu-devel] [PATCH v4 2/2] ACPI/unit-test: Add a new testcase for RAM allocation in numa node

[Igor Mammedov]

On Wed, 23 Aug 2017 21:35:29 +0800
Dou Liyang <address@hidden> wrote:

> Hi Igor,
> 
> At 08/23/2017 08:45 PM, Igor Mammedov wrote:
> > On Wed, 23 Aug 2017 20:12:51 +0800
> > Dou Liyang <address@hidden> wrote:
> >
> >> Hi Igor,
> >>
> >> At 08/23/2017 04:40 PM, Igor Mammedov wrote:
> >>> On Tue, 22 Aug 2017 11:24:10 +0800
> >>> Dou Liyang <address@hidden> wrote:
> >>>
> >>>> As QEMU supports the memory-less node, it is possible that there is
> >>>> no RAM in the first numa node(also be called as node0). eg:
> >>>>   ... \
> >>>>   -m 128,slots=3,maxmem=1G \
> >>>>   -numa node -numa node,mem=128M \
> >>>>
> >>>> But, this makes it hard for QEMU to build a known-to-work ACPI SRAT
> >>>> table. Only fixing it is not enough.
> >>>>
> >>>> Add a testcase for this situation to make sure the ACPI table is
> >>>> correct for guest.
> >>>>
> >>>> Suggested-by: Eduardo Habkost <address@hidden>
> >>>> Signed-off-by: Dou Liyang <address@hidden>
> >>>> ---
> >>>>  tests/acpi-test-data/pc/DSDT.numamem  | Bin 0 -> 6463 bytes
> >>>>  tests/acpi-test-data/pc/SLIT.numamem  | Bin 0 -> 48 bytes
> >>>>  tests/acpi-test-data/pc/SRAT.numamem  | Bin 0 -> 264 bytes
> >>>>  tests/acpi-test-data/q35/DSDT.numamem | Bin 0 -> 9147 bytes
> >>>>  tests/acpi-test-data/q35/SLIT.numamem | Bin 0 -> 48 bytes
> >>>>  tests/acpi-test-data/q35/SRAT.numamem | Bin 0 -> 264 bytes
> >>>>  tests/bios-tables-test.c              |  30 
> >>>> ++++++++++++++++++++++++++++++
> >>>>  7 files changed, 30 insertions(+)
> >>>>  create mode 100644 tests/acpi-test-data/pc/DSDT.numamem
> >>>>  create mode 100644 tests/acpi-test-data/pc/SLIT.numamem
> >>>>  create mode 100644 tests/acpi-test-data/pc/SRAT.numamem
> >>>>  create mode 100644 tests/acpi-test-data/q35/DSDT.numamem
> >>>>  create mode 100644 tests/acpi-test-data/q35/SLIT.numamem
> >>>>  create mode 100644 tests/acpi-test-data/q35/SRAT.numamem
> >>>
> >>>
> >>> considering only SRAT table has been changed and the other
> >>> tables match with default blobs, I'd suggest to keep only
> >>
> >>
> >> Our testcase is：
> >>
> >> +    test_acpi_one(" -m 128,slots=3,maxmem=1G"
> >> +                  " -numa node -numa node,mem=128"
> >> +                  " -numa dist,src=0,dst=1,val=21",
> >> +                  &data);
> >>
> >> The DSDT and SLIT don't match with default blobs.
> > do you actually need SLIT table /i.e. -numa dist/ for test at all?
> > it looks not relevant for the test case at the hand,
> > I'd suggest to drop '-numa dist' option for the test.
> >
> 
> OK, Got it, will drop '-numa dist' option in next version.
> 
> >>
> >> So, they can't be dropped.
> >
> > I wonder what's changed, could you post DSDT diff here?
> >
> 
> Just like memory hot-plug cases, when we use the '-m 128
> 128,slots=3,maxmem=1G' option, As the ACPI spec said, There may be some
> Memory Device in the DSDT table.
for your case '-numa node -numa node,mem=128', there is no need in enabling 
memory hotplug.
If I recall it correctly the default memory for x86 is 128Mb,
hence removing "-m" would probably make DSDT match default one.

[...]