hi, I am trying to add xenbus watch support to mini-os, but am getting into a endless loop of watches firing over and over. Below is an excerpt from the xenstore tracefile: IN 0x805acc8 20070103 19:32:18 WATCH (/local/domain/0/backend/vif/26/0/state 00043818 ) CREATE watch 0x805a8b8 OUT 0x805acc8 20070103 19:32:18 WATCH (OK ) OUT 0x805acc8 20070103 19:32:18 WATCH_EVENT (/local/domain/0/backend/vif/26/0/state 00043818 ) IN 0x805acc8 20070103 19:32:18 UNWATCH (/local/domain/0/backend/vif/26/0/state 00043818 ) DESTROY watch 0x805a8b8 OUT 0x805acc8 20070103 19:32:18 UNWATCH (OK ) IN 0x805acc8 20070103 19:32:18 READ (/local/domain/0/backend/vif/26/0/state ) OUT 0x805acc8 20070103 19:32:18 READ (5) IN 0x805acc8 20070103 19:32:18 WATCH (/local/domain/0/backend/vif/26/0/state 00043818 ) CREATE watch 0x806af60 OUT 0x805acc8 20070103 19:32:18 WATCH (OK ) OUT 0x805acc8 20070103 19:32:18 WATCH_EVENT (/local/domain/0/backend/vif/26/0/state 00043818 ) IN 0x805acc8 20070103 19:32:18 UNWATCH (/local/domain/0/backend/vif/26/0/state 00043818 ) DESTROY watch 0x806af60 OUT 0x805acc8 20070103 19:32:18 UNWATCH (OK ) IN 0x805acc8 20070103 19:32:18 READ (/local/domain/0/backend/vif/26/0/state ) OUT 0x805acc8 20070103 19:32:18 READ (5) IN 0x805acc8 20070103 19:32:18 WATCH (/local/domain/0/backend/vif/26/0/state 00043818 ) ... The same watch seems to be firing again and again, even though noone appears to be writing it. Should I some ack the watch, or is this a bug in xenstored? This is with Xen 3.0.3. Regards, Jacob _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On Wed, Jan 03, 2007 at 07:16:03PM +0100, Jacob Gorm Hansen wrote:> hi, > I am trying to add xenbus watch support to mini-os, but am getting into > a endless loop of watches firing over and over. Below is an excerpt from > the xenstore tracefile: > > IN 0x805acc8 20070103 19:32:18 WATCH > (/local/domain/0/backend/vif/26/0/state 00043818 ) > CREATE watch 0x805a8b8 > OUT 0x805acc8 20070103 19:32:18 WATCH (OK ) > OUT 0x805acc8 20070103 19:32:18 WATCH_EVENT > (/local/domain/0/backend/vif/26/0/state 00043818 ) > IN 0x805acc8 20070103 19:32:18 UNWATCH > (/local/domain/0/backend/vif/26/0/state 00043818 ) > DESTROY watch 0x805a8b8 > OUT 0x805acc8 20070103 19:32:18 UNWATCH (OK ) > IN 0x805acc8 20070103 19:32:18 READ > (/local/domain/0/backend/vif/26/0/state ) > OUT 0x805acc8 20070103 19:32:18 READ (5) > IN 0x805acc8 20070103 19:32:18 WATCH > (/local/domain/0/backend/vif/26/0/state 00043818 ) > CREATE watch 0x806af60 > OUT 0x805acc8 20070103 19:32:18 WATCH (OK ) > OUT 0x805acc8 20070103 19:32:18 WATCH_EVENT > (/local/domain/0/backend/vif/26/0/state 00043818 ) > IN 0x805acc8 20070103 19:32:18 UNWATCH > (/local/domain/0/backend/vif/26/0/state 00043818 ) > DESTROY watch 0x806af60 > OUT 0x805acc8 20070103 19:32:18 UNWATCH (OK ) > IN 0x805acc8 20070103 19:32:18 READ > (/local/domain/0/backend/vif/26/0/state ) > OUT 0x805acc8 20070103 19:32:18 READ (5) > IN 0x805acc8 20070103 19:32:18 WATCH > (/local/domain/0/backend/vif/26/0/state 00043818 ) > > ... > > The same watch seems to be firing again and again, even though noone > appears to be writing it. Should I some ack the watch, or is this a bug > in xenstored?As far as xenstored is concerned, someone is explicitly deleting and reregistering that watch -- I don''t think this is a bug in xenstored. You should expect to receive a watch event as soon as a watch is registered -- it looks to me like someone is receiving that event, then unwatching the node, reading its value, and reregistering the watch. Ewan. _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On Wed, 2007-01-03 at 18:25 +0000, Ewan Mellor wrote:> As far as xenstored is concerned, someone is explicitly deleting and > reregistering that watch -- I don''t think this is a bug in xenstored. > You > should expect to receive a watch event as soon as a watch is > registered -- it > looks to me like someone is receiving that event, then unwatching the > node, > reading its value, and reregistering the watch.That explains it, I was trying a simple single-threaded setup where the watch is unregistered once it has fired. What is the rationale for triggering the watch on registration? Jacob _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On Thu, Jan 04, 2007 at 01:18:32PM +0100, Jacob Gorm Hansen wrote:> On Wed, 2007-01-03 at 18:25 +0000, Ewan Mellor wrote: > > > As far as xenstored is concerned, someone is explicitly deleting and > > reregistering that watch -- I don''t think this is a bug in xenstored. > > You > > should expect to receive a watch event as soon as a watch is > > registered -- it > > looks to me like someone is receiving that event, then unwatching the > > node, > > reading its value, and reregistering the watch. > > That explains it, I was trying a simple single-threaded setup where the > watch is unregistered once it has fired. What is the rationale for > triggering the watch on registration?It was believed (not by me) that it made it easier to write client code, because your initial reads from the store would be triggered off the first watch, and so you didn''t need initialisation code separate from your state-change-handling code. We''re stuck with it now. Ewan. _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Could you send mini-os diff? Cheers Gregor Jacob Gorm Hansen wrote:> hi, > I am trying to add xenbus watch support to mini-os, but am getting into > a endless loop of watches firing over and over. Below is an excerpt from > the xenstore tracefile: > > IN 0x805acc8 20070103 19:32:18 WATCH > (/local/domain/0/backend/vif/26/0/state 00043818 ) > CREATE watch 0x805a8b8 > OUT 0x805acc8 20070103 19:32:18 WATCH (OK ) > OUT 0x805acc8 20070103 19:32:18 WATCH_EVENT > (/local/domain/0/backend/vif/26/0/state 00043818 ) > IN 0x805acc8 20070103 19:32:18 UNWATCH > (/local/domain/0/backend/vif/26/0/state 00043818 ) > DESTROY watch 0x805a8b8 > OUT 0x805acc8 20070103 19:32:18 UNWATCH (OK ) > IN 0x805acc8 20070103 19:32:18 READ > (/local/domain/0/backend/vif/26/0/state ) > OUT 0x805acc8 20070103 19:32:18 READ (5) > IN 0x805acc8 20070103 19:32:18 WATCH > (/local/domain/0/backend/vif/26/0/state 00043818 ) > CREATE watch 0x806af60 > OUT 0x805acc8 20070103 19:32:18 WATCH (OK ) > OUT 0x805acc8 20070103 19:32:18 WATCH_EVENT > (/local/domain/0/backend/vif/26/0/state 00043818 ) > IN 0x805acc8 20070103 19:32:18 UNWATCH > (/local/domain/0/backend/vif/26/0/state 00043818 ) > DESTROY watch 0x806af60 > OUT 0x805acc8 20070103 19:32:18 UNWATCH (OK ) > IN 0x805acc8 20070103 19:32:18 READ > (/local/domain/0/backend/vif/26/0/state ) > OUT 0x805acc8 20070103 19:32:18 READ (5) > IN 0x805acc8 20070103 19:32:18 WATCH > (/local/domain/0/backend/vif/26/0/state 00043818 ) > > ... > > The same watch seems to be firing again and again, even though noone > appears to be writing it. Should I some ack the watch, or is this a bug > in xenstored? > > This is with Xen 3.0.3. > > Regards, > Jacob > > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Maybe Matching Threads
- [PATCH V2] libxenstore: filter watch events in libxenstore when we unwatch
- [PATCH 6/11] Xenstore watch rework
- [PATCH V3] libxenstore: filter watch events in libxenstore when we unwatch
- [PATCH V5] libxenstore: filter watch events in libxenstore when we unwatch
- [PATCH V4] libxenstore: filter watch events in libxenstore when we unwatch