- User Since
- Jun 4 2014, 10:38 AM (172 w, 2 d)
Sun, Sep 10
- u_int v_free_count; /* (f) pages free */ u_int v_inactive_target; /* (c) pages desired inactive */ u_int v_pageout_free_min; /* (c) min pages reserved for kernel */ u_int v_interrupt_free_min; /* (c) reserved pages for int code */ u_int v_free_severe; /* (c) severe page depletion point */ + u_int v_free_count VMMETER_ALIGNED; /* (f) pages free */ u_int v_wire_count VMMETER_ALIGNED; /* (a) pages wired down */
Updated the diff to only move counters to the end + pad them.
Sat, Sep 9
Annotating mid-struct has 2 problems:
- creates unnecessary holes between stuff prior to and after the counter, i.e. the cacheline preceeding the padded field can have free space which is now unused. extreme case in point, v_inactive_target is getting sandwiched between 2 annotated fields (see below)
- the annotation sets the alignment of the field itself, but does not prevent following fields from placed immediately afterwards. once more case in point with v_inactive_target
Fri, Sep 8
Wed, Sep 6
Mon, Sep 4
Mon, Aug 28
since this is a spin mutex even failed trylock adds a trip through disabling/enabling preemption + interrupts. add a cacheline fetch. "fortunately" it so happens that spin trylock does not dirty it if it sees a taken lock. this is slow.
Aug 2 2017
Jul 31 2017
Jul 19 2017
There is definitely something messed up. Now there is no diff at all, but just a patched file. I don't know this tooling so can't you with it. Regardless, I can still provide comments.
Jul 15 2017
This does not look like a diff against head, for instance it includes the freeing of the old table.
Jul 14 2017
This looks much better overall.
I don't see what's the benefit of having the callers of fdgrowtable decide whether the old table can be freed. If you are worried about the case where a multithreaded process forks, that's covered by the first comment in the function about fd_lastfile == -1. This remark aside, the patch checks for correct things but does it wrong. :)
I have no basis to comment on this patch. I suggest you contact jeffr@.
Jul 11 2017
Jul 6 2017
Jul 2 2017
Jun 1 2017
May 30 2017
May 16 2017
May 4 2017
Mar 24 2017
Mar 23 2017
Mar 16 2017
Mar 14 2017
Mar 1 2017
I think the jail_attach interface is fundamentally unsuitable for this purpose. The problem is the process appears in the jail.