]> www.infradead.org Git - users/dwmw2/qemu.git/commit
fw_cfg: Don't set callback_opaque NULL in fw_cfg_modify_bytes_read()
authorShameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Tue, 3 Dec 2024 13:18:06 +0000 (13:18 +0000)
committerPhilippe Mathieu-Daudé <philmd@linaro.org>
Tue, 31 Dec 2024 20:21:34 +0000 (21:21 +0100)
commit5a7b6029c1e26c3bb171050938757e048398c576
tree08b29c51df77397c9c96d9069c1acb36d0be11ee
parent0798e09f3badf25a5b490ab0146b0d41ae78a798
fw_cfg: Don't set callback_opaque NULL in fw_cfg_modify_bytes_read()

On arm/virt platform, Chen Xiang reported a Guest crash while
attempting the below steps,

1. Launch the Guest with nvdimm=on
2. Hot-add a NVDIMM dev
3. Reboot
4. Guest boots fine.
5. Reboot again.
6. Guest boot fails.

QEMU_EFI reports the below error:
ProcessCmdAddPointer: invalid pointer value in "etc/acpi/tables"
OnRootBridgesConnected: InstallAcpiTables: Protocol Error

Debugging shows that on first reboot(after hot adding NVDIMM),
Qemu updates the etc/table-loader len,

qemu_ram_resize()
  fw_cfg_modify_file()
     fw_cfg_modify_bytes_read()

And in fw_cfg_modify_bytes_read() we set the "callback_opaque" for
the key entry to NULL. Because of this, on the second reboot,
virt_acpi_build_update() is called with a NULL "build_state" and
returns without updating the ACPI tables. This seems to be
upsetting the firmware.

To fix this, don't change the callback_opaque in fw_cfg_modify_bytes_read().

Fixes: bdbb5b1706d165 ("fw_cfg: add fw_cfg_machine_reset function")
Reported-by: chenxiang <chenxiang66@hisilicon.com>
Acked-by: Igor Mammedov <imammedo@redhat.com>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Message-ID: <20241203131806.37548-1-shameerali.kolothum.thodi@huawei.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
hw/nvram/fw_cfg.c