Re: [bug] very high non-preempt latency in context_struct_compute_av()

2007-06-07 Thread James Morris
On Thu, 7 Jun 2007, Ingo Molnar wrote: > i have just tried your patch and it completely solves the issue! Without > the patch, a simple script that keeps logging in on a box: > > while :; do ssh testbox true; done > > would cause glxgears to get into a very jerky motion due to the > late

Re: [bug] very high non-preempt latency in context_struct_compute_av()

2007-06-07 Thread Ingo Molnar
* Stephen Smalley <[EMAIL PROTECTED]> wrote: > Can you try the patch below to see whether it helps? > > In security_get_user_sids, move the transition permission checks > outside of the section holding the policy rdlock, and use the AVC to > perform the checks, calling cond_resched after each

Re: [bug] very high non-preempt latency in context_struct_compute_av()

2007-06-07 Thread Stephen Smalley
On Mon, 2007-06-04 at 13:27 +0200, Ingo Molnar wrote: > a simple ssh login triggers a ~130 msecs non-preemptible latency even > with CONFIG_PREEMPT enabled, on a fast Core2Duo CPU (!). > > the latency is caused by a _very_ long loop in the SELinux code: > > sshd-4828 0.N.. 465894us : avtab_

Re: [bug] very high non-preempt latency in context_struct_compute_av()

2007-06-04 Thread Paul Moore
On Monday 04 June 2007 5:39:00 pm Stephen Smalley wrote: > On Mon, 2007-06-04 at 17:11 -0400, Paul Moore wrote: > > I'm not an expert on the SELinux security server guts like the other > > people on the To/CC line of this thread, but here are my two cents on the > > issue above. > > > > From what I

Re: [bug] very high non-preempt latency in context_struct_compute_av()

2007-06-04 Thread James Morris
On Mon, 4 Jun 2007, Paul Moore wrote: > Okay, for some reason I thought someone had found a way to make > RCU "preemptable" through the real-time work, maybe I'm just confused > again :) It is preemptible in the RT kernel, but as Ingo points out, nothing should be even trying to do something f

Re: [bug] very high non-preempt latency in context_struct_compute_av()

2007-06-04 Thread Stephen Smalley
On Mon, 2007-06-04 at 17:11 -0400, Paul Moore wrote: > On Monday, June 4 2007 7:27:45 am Ingo Molnar wrote: > > a simple ssh login triggers a ~130 msecs non-preemptible latency even > > with CONFIG_PREEMPT enabled, on a fast Core2Duo CPU (!). > > > > the latency is caused by a _very_ long loop in t

Re: [bug] very high non-preempt latency in context_struct_compute_av()

2007-06-04 Thread Paul Moore
On Monday, June 4 2007 7:27:45 am Ingo Molnar wrote: > a simple ssh login triggers a ~130 msecs non-preemptible latency even > with CONFIG_PREEMPT enabled, on a fast Core2Duo CPU (!). > > the latency is caused by a _very_ long loop in the SELinux code: > > sshd-4828 0.N.. 465894us : avtab_sear

Re: [bug] very high non-preempt latency in context_struct_compute_av()

2007-06-04 Thread Ingo Molnar
* James Morris <[EMAIL PROTECTED]> wrote: > > the latency is caused by a _very_ long loop in the SELinux code: > > > > sshd-4828 0.N.. 465894us : avtab_search_node > > (context_struct_compute_av) > > What do the 0DNs fields mean and what did you use to create this > trace? i used the la

Re: [bug] very high non-preempt latency in context_struct_compute_av()

2007-06-04 Thread James Morris
On Mon, 4 Jun 2007, Ingo Molnar wrote: > > a simple ssh login triggers a ~130 msecs non-preemptible latency even > with CONFIG_PREEMPT enabled, on a fast Core2Duo CPU (!). Thanks for the report. > > the latency is caused by a _very_ long loop in the SELinux code: > > sshd-4828 0.N.. 465

[bug] very high non-preempt latency in context_struct_compute_av()

2007-06-04 Thread Ingo Molnar
a simple ssh login triggers a ~130 msecs non-preemptible latency even with CONFIG_PREEMPT enabled, on a fast Core2Duo CPU (!). the latency is caused by a _very_ long loop in the SELinux code: sshd-4828 0.N.. 465894us : avtab_search_node (context_struct_compute_av) sshd-4828 0.N.. 4658