From mboxrd@z Thu Jan 1 00:00:00 1970 MIME-Version: 1.0 In-Reply-To: References: Date: Fri, 5 Dec 2014 22:39:14 -0700 Message-ID: From: Latchesar Ionkov To: Fans of the OS Plan 9 from Bell Labs <9fans@9fans.net> Content-Type: multipart/alternative; boundary=001a113a5f501fb72c0509859d35 Subject: Re: [9fans] Is the go9p project still maintained? Topicbox-Message-UUID: 34503fa0-ead9-11e9-9d60-3106f5b1d025 --001a113a5f501fb72c0509859d35 Content-Type: text/plain; charset=UTF-8 The 9p server gets valid uid, the Linux kernel sends separate Tattach messages for each user that accesses the server. The issue with remote systems having different uids is no different than any other remote file system in Linux. As somebody already mentioned, your patch isn't very clean, so I am not sure what change of behavior you propose. Thanks, Lucho On Friday, December 5, 2014, Dmitry Golubovsky wrote: > Hi > > Thanks to everybody who answered. I posted another comment to the issue 34 > > https://code.google.com/p/go9p/issues/detail?id=34#c2 > > where explained the use case and provided the diagnostics for the problem. > > Briefly, the problem is: when using 9p kernel mounts in Linux over a > domain socket, a 9p server program never gets a valid numeric user ID, > just -1 regardless of the username. > > Even if there was a way to pass desired numeric UID to a 9p server, > this would make little sense in case of a remote connection because on > the remote system the same user would have different (and often > unknown numeric ID). > > So even in the .u mode, fallback character username recognition is > necessary to use if no valid numeric user ID can be provided. > > Thanks. > > -- > Dmitry Golubovsky > > Anywhere on the Web > > --001a113a5f501fb72c0509859d35 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable The 9p server gets valid uid, the Linux=C2=A0kernel sends separate Tattach = messages for each user that accesses the server. The issue with remote syst= ems having different uids is no different than any other remote file system= in Linux. As somebody already mentioned, your patch isn't very clean, = so I am not sure what change of behavior you propose.

Th= anks,
=C2=A0 =C2=A0 Lucho

On Friday, December 5, 2014, Dmi= try Golubovsky <golubovsky@gmail= .com> wrote:
Hi

Thanks to everybody who answered. I posted another comment to the issue 34<= br>
https://code.google.com/p/go9p/issues/detail?id=3D34#c2

where explained the use case and provided the diagnostics for the problem.<= br>
Briefly, the problem is: when using 9p kernel mounts in Linux over a
domain socket, a 9p server program never gets a valid numeric user ID,
just -1 regardless of the username.

Even if there was a way to pass desired numeric UID to a 9p server,
this would make little sense in case of a remote connection because on
the remote system the same user would have different (and often
unknown numeric ID).

So even in the .u mode, fallback character username recognition is
necessary to use if no valid numeric user ID can be provided.

Thanks.

--
Dmitry Golubovsky

Anywhere on the Web

--001a113a5f501fb72c0509859d35--