Skip to contentSkip to conversation

SKAS3: prerelease of V8 (against 2.6.9) uploaded

uml-devel Blaisorblade 5 participants 8 messages Jan 28, 2005 – Feb 3, 2005 User-mode Linux (UML)

lore
Share & tools
lore
#0 lore raw

Ok, I've uploaded the first prerelease of SKAS3 V8... It's uploaded on the web-site, you can start playing for it if you want:

http://www.user-mode-linux.org/~blaisorblade/patches/skas3-2.6/skas-2.6.9-v8-rc1/

if you want, impressions are welcome.... I've not yet tested it thoroughly, but it should work well enough.

I'm particularly interested in seeing it tested on SMP host systems, since I've added some locking and I want to be sure there is no deadlock possibility.

However, *not* on production systems for now... I want to check it more carefully.

Signature · 14 lines
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729
http://www.user-mode-linux.org/~blaisorblade


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
User-mode-linux-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
#1 lore raw
Quoted 9 lines · Blaisorblade

Ok, I've uploaded the first prerelease of SKAS3 V8... It's uploaded on the web-site, you can start playing for it if you want:

http://www.user-mode-linux.org/~blaisorblade/patches/skas3-2.6/skas-2.6.9-v8-rc1/

if you want, impressions are welcome.... I've not yet tested it thoroughly, but it should work well enough.

I'm particularly interested in seeing it tested on SMP host systems, since I've added some locking and I want to be sure there is no deadlock possibility.

However, *not* on production systems for now... I want to check it more carefully.

I just tested this release on a vanilla single-proc AMD system. I see one major problem (at least with how we run UML here).

With 2.6.7 with skas3 and earlier releases it is possible to kill a virtual server from the host with a standard kill command. All of the sub-threads die and you can unmount/unmap the various stuff that the virtual used. This is an unclean shutdown but at least lets you kill virtuals that have misbehaved while leaving the host up.

With 2.6.9 skas3 v8 a plain kill does nothing. With a kill -9, 3 of the four thread die, but one remains and the only way to get rid of it is with a server reboot.

Not sure what is going on here, but I have not been able to get anything past 2.6.7 to work while killing virtual rudely.

--------------------------------------------------------------------

Doug Dumitru     800-470-2756     (610-237-2000)
EasyCo LLC       [email protected]  http://easyco.com

--------------------------------------------------------------------

-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

#2 lore raw

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Doug Dumitru wrote:
<snip>

| With 2.6.9 skas3 v8 a plain kill does nothing.  With a kill -9, 3 of the
| four thread die, but one remains and the only way to get rid of it is
| with a server reboot.
|
| Not sure what is going on here, but I have not been able to get anything
| past 2.6.7 to work while killing virtual rudely.

Will they die if you send a "kill -CONT" followed by the "kill -9"?

Frank
- --
Frank Sorenson - KD7TZK
Systems Manager, Computer Science Department
Brigham Young University
[email protected]
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFB/yYaaI0dwg4A47wRAqayAJ9SG2inXL5ts32AaHp4wFmThEBxvwCdG/zP IlYmpS0xqBIG9HoS+MgAkUk=
=4jiQ
-----END PGP SIGNATURE-----

-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

#3 lore raw
Quoted 13 lines · Frank Sorenson

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Doug Dumitru wrote:
<snip>

| With 2.6.9 skas3 v8 a plain kill does nothing.  With a kill -9, 3 of the
| four thread die, but one remains and the only way to get rid of it is
| with a server reboot.
|
| Not sure what is going on here, but I have not been able to get anything
| past 2.6.7 to work while killing virtual rudely.
Quoted 3 lines

Will they die if you send a "kill -CONT" followed by the "kill -9"?

Frank

If a kill -9 needs to be followed by a CONT signal to actually take effect on a stopped process, why is this not a kernel bug? (Kill -9 is supposed to _mean_ it. I've always thought of it as the "with shotgun" option...)

Rob

-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

#4 lore raw

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Rob Landley wrote:
| If a kill -9 needs to be followed by a CONT signal to actually take effect on

| a stopped process, why is this not a kernel bug?  (Kill -9 is supposed to
| _mean_ it.  I've always thought of it as the "with shotgun" option...)
|
| Rob

I believe there were PTRACE changes in 2.6.9 that caused this:

http://marc.theaimsgroup.com/?l=linux-kernel&m=109877438509709&w=2

Frank
- --
Frank Sorenson - KD7TZK
Systems Manager, Computer Science Department
Brigham Young University
[email protected]
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFB/5+naI0dwg4A47wRApyhAJwKVrAZiH8bcxwGlx/6GSSyPdNr7gCglcfo i0d67JyivfwJSiZv+yk7Wp8=
=W8eb
-----END PGP SIGNATURE-----

-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

#5 lore raw

With the problems we had with 2.6.9 as a host, I was wondering why Blaisorblade chose it when he released skas-v8-rc1.

Here is skas-v8-rc1 diffed against 2.6.10:

http://www.theshore.net/~caker/patches/skas-v8-rc1-2.6.10.patch

-Chris

-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

#6 lore raw
Quoted 8 lines · Christopher S. Aker

With the problems we had with 2.6.9 as a host, I was wondering why Blaisorblade chose it when he released skas-v8-rc1.

Here is skas-v8-rc1 diffed against 2.6.10:

http://www.theshore.net/~caker/patches/skas-v8-rc1-2.6.10.patch

-Chris

Thank you for the patch.

It compiles cleanly and does run UML here, but still has the kill behaviour. To recap:

kill -TERM pidof linux

does nothing

kill -KILL pidof linux

kills all but one thread, after which you can ...

kill -CONT pidof linux

which finally kills it off completely.

Having a kill -9 not work really sounds like a bug to me. It can be worked around, but is "non obvious".

--------------------------------------------------------------------

Doug Dumitru     800-470-2756     (610-237-2000)
EasyCo LLC       [email protected]  http://easyco.com

--------------------------------------------------------------------

-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

#7 lore raw
In reply to Christopher S. Aker
Quoted 2 lines · you

With the problems we had with 2.6.9 as a host, I was wondering why Blaisorblade chose it when he released skas-v8-rc1.

I wondered too :-), however the reason was just that I wanted to get it out for early review... I've then been busy and now I'm going to port it to 2.6.10.

Quoted 3 lines

Here is skas-v8-rc1 diffed against 2.6.10:

http://www.theshore.net/~caker/patches/skas-v8-rc1-2.6.10.patch

I'm having a look at it, however IIRC there weren't too many subtle ptrace change in 2.6.10, so I guess you've done it well...

Actually I have some doubt about ptrace.c, for this part:

                  child->exit_code = data;
  -               /* make sure the single step bit is not set. */
  -               clear_singlestep(child);
  +               put_stack_long(child, EFL_OFFSET,tmp);
                  wake_up_process(child);
                  ret = 0;
                  break;

I understand the purpose, and it makes sense... but I'll have to find a way to clean it up...

Signature · 16 lines
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729
http://www.user-mode-linux.org/~blaisorblade




-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
User-mode-linux-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel