Re: [Qemu-devel] [RESEND PATCH 1/2] nvdimm: warn if the backend is not a DAX device

[Dan Williams]

[ adding linux-fsdevel ]

On Thu, Jun 1, 2017 at 5:00 AM, Xiao Guangrong <address@hidden> wrote:
>
>
> On 05/26/2017 10:32 AM, Haozhong Zhang wrote:
>
>> +static void nvdimm_check_dax(HostMemoryBackend *hostmem)
>> +{
>> +    char *mem_path =
>> +        object_property_get_str(OBJECT(hostmem), "mem-path", NULL);
>> +    char *dev_name = NULL, *sysfs_path = NULL;
>> +    bool is_dax = false;
>> +
>> +    if (!mem_path) {
>> +        goto out;
>> +    }
>> +
>> +    if (!g_str_has_prefix(mem_path, "/dev/dax")) {
>> +        goto out;
>> +    }
>> +
>> +    dev_name = mem_path + strlen("/dev/");
>> +    sysfs_path = g_strdup_printf("/sys/class/dax/%s", dev_name);
>> +    if (access(sysfs_path, F_OK)) {
>> +        goto out;
>> +    }
>> +
>> +    is_dax = true;
>> +
>
>
> So only dax raw disk has write-persistence guaranty， a pre-allocated
> file which locates on a DAX-enabled filesystem can not?

Correct, it is not guaranteed by any existing filesystem. It may work
by accident today on ext4 with a pre-allocated file, but a filesystem
is otherwise permitted to drop any writes that have not been synced.
Until we get an explicit MMAP_SYNC or METADATA_IMMUTABLE feature into
a filesystem the only interface that we can use to reliably pass
persistent memory through to a guest is device-dax.