From: Antoine M. <an...@na...> - 2005-01-19 19:24:44
|
Hi all, I was having some very unusual problems with 2.6.9-bs5 and -bb4 running on an AMD 64 2.6.10-as1 host (2.6.10 + minimal security fixes): some processes would not start with the correct SELinux context (sshd would run in kernel_t instead of sshd_t for no apparent reason) but 2.6.10 vanilla as guest was causing the old process leak problem, so I applied the fix-kill patch manually (originally designed for 2.4.27 - some bits were already merged) and all is working well now! I just thought I'd share this with you. Cheers Antoine |
From: Blaisorblade <bla...@ya...> - 2005-01-19 23:06:25
|
On Wednesday 19 January 2005 21:33, Antoine Martin wrote: > Hi all, > > I was having some very unusual problems with 2.6.9-bs5 and -bb4 running > on an AMD 64 2.6.10-as1 host (2.6.10 + minimal security fixes): Yes, I'm seeing it discussed onto kerneltrap.org... > some > processes would not start with the correct SELinux context (sshd would > run in kernel_t instead of sshd_t for no apparent reason) Please give more details on this - are you speaking of inside UML? Is this in your opinion a bug in the 2.6.9 SELinux code or a bug of UML itself? I don't think it's from UML because there is no SELinux related code in UML, nor any possibly (IMHO) related change in 2.6.10. > but 2.6.10 > vanilla as guest was causing the old process leak problem, so I applied > the fix-kill patch manually (originally designed for 2.4.27 - some bits > were already merged) Yes, the fix for TT mode was not complete originally... the same things you've applied from the 2.4.27 patch could be extracted from the -bb/-bs patchset (in split-out form). The not merged part is likely "uml-jdike-fix-tt-exit-2.6.9-host.patch", inside -bb. > and all is working well now! > I just thought I'd share this with you. Thanks a lot! -- Paolo Giarrusso, aka Blaisorblade Linux registered user n. 292729 http://www.user-mode-linux.org/~blaisorblade |
From: Antoine M. <an...@na...> - 2005-01-20 15:40:28
|
On Thu, 2005-01-20 at 00:18 +0100, Blaisorblade wrote: > On Wednesday 19 January 2005 21:33, Antoine Martin wrote: > > Hi all, > > > > I was having some very unusual problems with 2.6.9-bs5 and -bb4 running > > on an AMD 64 2.6.10-as1 host (2.6.10 + minimal security fixes): > Yes, I'm seeing it discussed onto kerneltrap.org... > > some > > processes would not start with the correct SELinux context (sshd would > > run in kernel_t instead of sshd_t for no apparent reason) > > Please give more details on this - are you speaking of inside UML? Is this in > your opinion a bug in the 2.6.9 SELinux code or a bug of UML itself? I don't > think it's from UML because there is no SELinux related code in UML, nor any > possibly (IMHO) related change in 2.6.10. That's inside the uml, I think you're right, it must be a bug in 2.6.9 and not linked to uml (I couldn't find any info on this on kernel trap), it is just that I couldn't run 2.6.10 until I got the fix-kill sorted out. > > > but 2.6.10 > > vanilla as guest was causing the old process leak problem, so I applied > > the fix-kill patch manually (originally designed for 2.4.27 - some bits > > were already merged) > Yes, the fix for TT mode was not complete originally... the same things you've > applied from the 2.4.27 patch could be extracted from the -bb/-bs patchset > (in split-out form). > > The not merged part is likely "uml-jdike-fix-tt-exit-2.6.9-host.patch", inside > -bb. I'll try that, thanks! > > > and all is working well now! > > I just thought I'd share this with you. > > Thanks a lot! |