On 23.03.2025 15:57, Jürgen Groß wrote:
> On 23.03.25 01:01, Samuel Thibault wrote:
>> Juergen Gross, le ven. 21 mars 2025 10:31:44 +0100, a ecrit:
>>> Add a file operations fstat hook to the 9pfs frontend.
>>>
>>> Signed-off-by: Juergen Gross <jgr...@suse.com>
>>> Reviewed-by: Jason Andryuk <jason.andr...@amd.com>
>>> ---
>>> V2:
>>> - or file access mode into st_mode (Jason Andryuk)
>>> ---
>>> 9pfront.c | 29 +++++++++++++++++++++++++++++
>>> 1 file changed, 29 insertions(+)
>>>
>>> diff --git a/9pfront.c b/9pfront.c
>>> index 1741d600..7257a07e 100644
>>> --- a/9pfront.c
>>> +++ b/9pfront.c
>>> @@ -85,6 +85,8 @@ struct file_9pfs {
>>>
>>> #define P9_QID_SIZE 13
>>>
>>> +#define QID_TYPE_DIR 0x80 /* Applies to qid[0]. */
>>> +
>>> struct p9_header {
>>> uint32_t size;
>>> uint8_t cmd;
>>> @@ -950,6 +952,32 @@ static int write_9pfs(struct file *file, const void
>>> *buf, size_t nbytes)
>>> return ret;
>>> }
>>>
>>> +static int fstat_9pfs(struct file *file, struct stat *buf)
>>> +{
>>> + struct file_9pfs *f9pfs = file->filedata;
>>> + struct p9_stat stat;
>>> + int ret;
>>> +
>>> + ret = p9_stat(f9pfs->dev, f9pfs->fid, &stat);
>>> + if ( ret )
>>> + {
>>> + errno = EIO;
>>> + return -1;
>>> + }
>>> +
>>> + buf->st_mode = (stat.qid[0] == QID_TYPE_DIR) ? S_IFDIR : S_IFREG;
>>> + buf->st_mode |= stat.mode & 0777;
>>> + buf->st_atime = stat.atime;
>>> + buf->st_mtime = stat.mtime;
>>
>> Should we perhaps also fill st_ctime? Leaving it at 0 could surprise
>> other software layers.
>
> I can set it to the same value as st_mtime.
Maybe the smaller of atime and mtime?
Jan
> I don't think there is a way
> to get the "real" ctime value via the 9p protocol.
>
>
> Juergen
