On Mon, 1 Mar 2021 11:48:33 +0100 David Hildenbrand <da...@redhat.com> wrote:
> The resizeable memory region that is created for the cmd blob has a maximum > size of ACPI_BUILD_ALIGN_SIZE - 4k. This used to be sufficient, however, > as we try fitting in additional data (e.g., vmgenid, nvdimm, intel-iommu), > we require more than 4k and can crash QEMU when trying to resize the > resizeable memory region beyond its maximum size: > $ build/qemu-system-x86_64 --enable-kvm \ > -machine q35,nvdimm=on \ > -smp 1 \ > -cpu host \ > -m size=2G,slots=8,maxmem=4G \ > -object memory-backend-file,id=mem0,mem-path=/tmp/nvdimm,size=256M \ > -device nvdimm,label-size=131072,memdev=mem0,id=nvdimm0,slot=1 \ > -nodefaults \ > -device vmgenid \ > -device intel-iommu I don't see what's here that would make cmd_blob go above 4k. can you try identify what actually fills it up (perhaps we have a hidden bug elsewhere)? > > Results in: > Unexpected error in qemu_ram_resize() at ../softmmu/physmem.c:1850: > qemu-system-x86_64: Size too large: /rom@etc/table-loader: > 0x2000 > 0x1000: Invalid argument > > We try growing the resizeable memory region (resizeable RAMBlock) beyond > its maximum size. Let's increase the maximum size from 4k to 64k, which > should be good enough for the near future. > > Migration is not concerned with the maximum size of a RAMBlock, only > with the used size - so existing setups are not affected. Of course, we > cannot migrate a VM that would have crash when started on older QEMU from > new QEMU to older QEMU without failing early on the destination when > synchronizing the RAM state: > qemu-system-x86_64: Size too large: /rom@etc/table-loader: 0x2000 > > 0x1000: Invalid argument > qemu-system-x86_64: error while loading state for instance 0x0 of device > 'ram' > qemu-system-x86_64: load of migration failed: Invalid argument > > While at it, replace "etc/table-loader" by ACPI_BUILD_LOADER_FILE in > the microvm. > > Note: we could warn for problematic setups that migration might not > always be possible - similar to how we handle the table blob; or we > could disallow setups that would have crashed until now for compat > machines. But I am not sure if the effort (messing compat machine > properties) is worth it as we fail migration in a safe way early. > > Cc: Alistair Francis <alistair.fran...@xilinx.com> > Cc: Paolo Bonzini <pbonz...@redhat.com> > Cc: "Michael S. Tsirkin" <m...@redhat.com> > Cc: Igor Mammedov <imamm...@redhat.com> > Cc: Peter Maydell <peter.mayd...@linaro.org> > Cc: Shannon Zhao <shannon.zha...@gmail.com> > Cc: Marcel Apfelbaum <marcel.apfelb...@gmail.com> > Cc: Paolo Bonzini <pbonz...@redhat.com> > Cc: Richard Henderson <richard.hender...@linaro.org> > Cc: qemu-...@nongnu.org > Signed-off-by: David Hildenbrand <da...@redhat.com> > --- > hw/arm/virt-acpi-build.c | 3 ++- > hw/i386/acpi-build.c | 3 ++- > hw/i386/acpi-microvm.c | 3 ++- > include/hw/acpi/aml-build.h | 1 + > 4 files changed, 7 insertions(+), 3 deletions(-) > > diff --git a/hw/arm/virt-acpi-build.c b/hw/arm/virt-acpi-build.c > index f9c9df916c..a91550de6f 100644 > --- a/hw/arm/virt-acpi-build.c > +++ b/hw/arm/virt-acpi-build.c > @@ -865,7 +865,8 @@ void virt_acpi_setup(VirtMachineState *vms) > > build_state->linker_mr = > acpi_add_rom_blob(virt_acpi_build_update, build_state, > - tables.linker->cmd_blob, ACPI_BUILD_LOADER_FILE, > 0); > + tables.linker->cmd_blob, ACPI_BUILD_LOADER_FILE, > + ACPI_BUILD_LOADER_MAX_SIZE); > > fw_cfg_add_file(vms->fw_cfg, ACPI_BUILD_TPMLOG_FILE, > tables.tcpalog->data, > acpi_data_len(tables.tcpalog)); > diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c > index 31a5f6f4a5..a75138ea5a 100644 > --- a/hw/i386/acpi-build.c > +++ b/hw/i386/acpi-build.c > @@ -2524,7 +2524,8 @@ void acpi_setup(void) > > build_state->linker_mr = > acpi_add_rom_blob(acpi_build_update, build_state, > - tables.linker->cmd_blob, ACPI_BUILD_LOADER_FILE, > 0); > + tables.linker->cmd_blob, ACPI_BUILD_LOADER_FILE, > + ACPI_BUILD_LOADER_MAX_SIZE); > > fw_cfg_add_file(x86ms->fw_cfg, ACPI_BUILD_TPMLOG_FILE, > tables.tcpalog->data, acpi_data_len(tables.tcpalog)); > diff --git a/hw/i386/acpi-microvm.c b/hw/i386/acpi-microvm.c > index 54b3af478a..fe8a965fe6 100644 > --- a/hw/i386/acpi-microvm.c > +++ b/hw/i386/acpi-microvm.c > @@ -255,7 +255,8 @@ void acpi_setup_microvm(MicrovmMachineState *mms) > ACPI_BUILD_TABLE_MAX_SIZE); > acpi_add_rom_blob(acpi_build_no_update, NULL, > tables.linker->cmd_blob, > - "etc/table-loader", 0); > + ACPI_BUILD_LOADER_FILE, > + ACPI_BUILD_LOADER_MAX_SIZE); > acpi_add_rom_blob(acpi_build_no_update, NULL, > tables.rsdp, > ACPI_BUILD_RSDP_FILE, 0); > diff --git a/include/hw/acpi/aml-build.h b/include/hw/acpi/aml-build.h > index 380d3e3924..93cdfd4006 100644 > --- a/include/hw/acpi/aml-build.h > +++ b/include/hw/acpi/aml-build.h > @@ -6,6 +6,7 @@ > > /* Reserve RAM space for tables: add another order of magnitude. */ > #define ACPI_BUILD_TABLE_MAX_SIZE 0x200000 > +#define ACPI_BUILD_LOADER_MAX_SIZE 0x40000 > > #define ACPI_BUILD_APPNAME6 "BOCHS " > #define ACPI_BUILD_APPNAME8 "BXPC "