Re: [Qemu-devel] [PATCH 4/8] nvdimm acpi: implement Read FIT function

[Xiao Guangrong <guangrong.xiao@linux.intel.com>]

On 07/14/2016 08:17 PM, Stefan Hajnoczi wrote:

>> +/* Read FIT data, defined in docs/specs/acpi_nvdimm.txt. */
>> +static void nvdimm_dsm_func_read_fit(NvdimmDsmIn *in, hwaddr dsm_mem_addr)
>> +{
>> +    NvdimmFuncReadFITIn *read_fit;
>> +    NvdimmFuncReadFITOut *read_fit_out;
>> +    GSList *device_list = nvdimm_get_plugged_device_list();
>> +    GArray *fit = nvdimm_build_device_structure(device_list);
>> +    uint32_t read_len = 0, func_ret_status;
>> +    int left, size;
>> +
>> +    read_fit = (NvdimmFuncReadFITIn *)in->arg3;
>> +    le32_to_cpus(&read_fit->offset);
>> +
>> +    nvdimm_debug("Read FIT: offset %#x FIT size %#x.\n", read_fit->offset,
>> +                 fit->len);
>> +
>> +    left = fit->len - read_fit->offset;
>> +    if (left < 0) {
>
> Signed integer overflow leads to memory disclosure in memcpy() below.
> The problem occurs when (guint)fit->len - (uint32_t)read_fit->offset >
> INT_MAX.
>
> Please perform the check like this:
>
>    if (fit->offset >= fit->len) {
>

Ah, yes, you are right, thank you for pointing it out. Will fix it.