search for: xs_watch

...e watch fires before the driver sends "OK", which confuses the xenstore-watch client. It seems to me that the client is within its rights to expect the "OK" first. Here''s what is happening: The userspace process xenstore-watch writes to /proc/xen/xenbus with msg_type==XS_WATCH. This is handled by xenbus_write_watch which calls register_xenbus_watch with watch_fired as a callback *before* acquiring the reply_mutex and sending the synthesised "OK" reply. This gives a fast xenstore the opportunity to cause the watch_fired to run (and briefly grab the reply_mutex...

Hello, This patch allows userspace tools on a domU to setup a watch on the xenstore. It does this by intercepting XS_WATCH requests written to /proc/xen/xenbus and then re-submitting the request to the in-kernel xenstore interface, in linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_xs.c. When a callback occurs, an in-kernel function is invoked, which then reconstructs a response in the format expected by userspace, and...

A pair of patches to the xenstore library that: 1) blocks signal delivery before creating the watch wakeup thread 2) reduces the stack size of the watch wakeup thread.

...; --- tools/xenstore/xs.c | 73 +++++++++++++++++++++++++++++++++++++++++++++----- 1 files changed, 65 insertions(+), 8 deletions(-) diff --git a/tools/xenstore/xs.c b/tools/xenstore/xs.c index b951015..df89e37 100644 --- a/tools/xenstore/xs.c +++ b/tools/xenstore/xs.c @@ -753,6 +753,19 @@ bool xs_watch(struct xs_handle *h, const char *path, const char *token) ARRAY_SIZE(iov), NULL)); } + +/* Clear the pipe token if there are no more pending watchs. + * We suppose the watch_mutex is already taken. + */ +static void xs_clear_watch_pipe(struct xs_handle *h) +{ + char c; + + if (list_empty(&a...

...ated message is read. The use case is non-threaded libevent, we have two event A and B: - Event A will destroy something and call xs_unwatch; - Event B is used to notify that a node has changed in XenStore. As the event is called one by one, event A can be handled before event B. So on next xs_watch_read the user could retrieve an unwatch token and a segfault occured if the token store the pointer of the structure (ie: "backend:0xcafe"). To avoid problem with previous application using libXenStore, this behaviour will only be enabled if XS_UNWATCH_SAFE is give to xs_open. Signed-of...

Hello!. How can i use @releaseDomain from dom0 to subscribe for domain destruction? I''m try to add watch for @releaseDomain path with token equal of needed domain id. But when another domain dies, i get domain that i provide via token to xs_watch. Is that possible to get id of domain what released from xenstore? -- Vasiliy Tolstov, Clodo.ru e-mail: v.tolstov@selfip.ru jabber: vase@selfip.ru

...se/drivers/xen/xenbus/xenbus_xs.c --- a/linux-2.6.11-xen-sparse/drivers/xen/xenbus/xenbus_xs.c Thu Aug 4 10:43:03 2005 +++ b/linux-2.6.11-xen-sparse/drivers/xen/xenbus/xenbus_xs.c Thu Aug 4 11:39:03 2005 @@ -321,18 +321,14 @@ return xs_single(XS_GETDOMAINPATH, domid_str, NULL); } -static int xs_watch(const char *path, const char *token, unsigned int priority) -{ - char prio[32]; - struct kvec iov[3]; - - sprintf(prio, "%u", priority); +static int xs_watch(const char *path, const char *token) +{ + struct kvec iov[2]; + iov[0].iov_base = (void *)path; iov[0].iov_len = strlen(path) +...

...} This bug makes the read_buffer not be cleared most of the time. If the xenstore client uses PTHREAD to create a thread to receive reply message, the problem will incur. The new thread can not read what it wants to read, since the list is not empty. I found this problem from the xenstore client xs_watch function. xs_watch creates the new thread on demand. So I recommend that in the function of read_message(xen/tools/xenstore/xs.c), if using thread to receive message, in the case of read fault, it should signal to the listener and print out the error. Jun Zhu Citrix Systems UK ___________________...

Hi experts I use xen-4.0.0 and dom0 is 2.6.33 and domU is 2.6.34 #lspci ............... 01:00.0 Ethernet controller: Intel Corporation Unknown device 10a7 (rev 02) 01:00.1 Ethernet controller: Intel Corporation Unknown device 10a7 (rev 02) 05:00.0 Ethernet controller: Intel Corporation 82571EB Gigabit Ethernet Controller (rev 06) 05:00.1 Ethernet controller: Intel Corporation 82571EB Gigabit

...3:49 2010 +0100 +++ b/tools/libxl/libxl_device.c Thu Sep 02 14:51:46 2010 +0100 @@ -406,6 +406,8 @@ char **l = NULL; xsh = xs_daemon_open(); + if (!xsh) + xsh = xs_domain_open(); path = libxl_sprintf(&gc, "/local/domain/0/device-model/%d/state", domid); xs_watch(xsh, path, path); tv.tv_sec = LIBXL_DEVICE_MODEL_START_TIMEOUT; diff -r eff592364826 tools/libxl/libxl_dom.c --- a/tools/libxl/libxl_dom.c Wed Sep 01 11:23:49 2010 +0100 +++ b/tools/libxl/libxl_dom.c Thu Sep 02 14:51:46 2010 +0100 @@ -284,6 +284,8 @@ snprintf(path, sizeof(path), "/lo...

...a/tools/console/daemon/io.c Fri Sep 01 12:16:21 2006 +0100 +++ b/tools/console/daemon/io.c Fri Sep 01 08:14:07 2006 -0700 @@ -293,12 +293,14 @@ static bool watch_domain(struct domain * bool success; sprintf(domid_str, "dom%u", dom->domid); - if (watch) + if (watch) { success = xs_watch(xs, dom->conspath, domid_str); - else + if (success) + domain_create_ring(dom); + } else { success = xs_unwatch(xs, dom->conspath, domid_str); - if (success) - domain_create_ring(dom); + } + return success; } _______________________________________________ Xen-devel mailing list...

Hello All, I am trying to set up a watch on xentore. According to the documentation xenstore.txt a watch is simply done with the path to the node that the path is going to set, but according to the definitive guide to xen, it is done like a write, path to node and key to be watched. I have tried both and both fail. Probably I have something wrong but it helps to know the correct format. Which

...; --- a/drivers/xen/xenbus/xenbus_dev.c +++ b/drivers/xen/xenbus/xenbus_dev.c @@ -269,18 +269,24 @@ static ssize_t xenbus_dev_write(struct f goto out; } token++; + if (memchr(token, 0, u->u.msg.len - (token - path)) == NULL) { + rc = -EILSEQ; + goto out; + } if (msg_type == XS_WATCH) { watch = kzalloc(sizeof(*watch), GFP_KERNEL); - watch->watch.node = kmalloc(strlen(path)+1, - GFP_KERNEL); - strcpy((char *)watch->watch.node, path); + if (watch == NULL) { + rc = -ENOMEM; + goto out; + } + watch->wat...

..._frontend.c +++ 3.3-rc1-xenbus-dev-write-watch/drivers/xen/xenbus/xenbus_dev_frontend.c @@ -369,6 +369,10 @@ static int xenbus_write_watch(unsigned m goto out; } token++; + if (memchr(token, 0, u->u.msg.len - (token - path)) == NULL) { + rc = -EILSEQ; + goto out; + } if (msg_type == XS_WATCH) { watch = alloc_watch_adapter(path, token);

...do not work from domU userspace (they obviously do for drivers). In some preliminary debugging of this, there are two aspects of the failure: not sending the watch to xenstore and not handling watch events for userspace watches. The former can be easily fixed by adding the missing ''case XS_WATCH:'' to xenbus_dev_write() in xenbus_dev.c. The latter is more difficult to fix. So before I tackle it, I wanted to see if anyone else was already working on a solution or just had some thoughts to share on possible solutions. Joseph Cihula (Linux) Software Security Architect Open Source T...

...__mutex_lock_slowpath+0x90/0x110 Nov 25 15:49:53 k2000 kernel: [ 2408.822722] [<ffffffff8153a536>] ? mutex_lock+0x16/0x30 Nov 25 15:49:53 k2000 kernel: [ 2408.822724] [<ffffffff81397347>] ? xs_talkv+0x67/0x1d0 Nov 25 15:49:53 k2000 kernel: [ 2408.822725] [<ffffffff8139770f>] ? xs_watch+0x4f/0x70 Nov 25 15:49:53 k2000 kernel: [ 2408.822727] [<ffffffff81397bd6>] ? register_xenbus_watch+0x96/0x110 Nov 25 15:49:53 k2000 kernel: [ 2408.822728] [<ffffffff8139a57f>] ? xenbus_file_write+0x48f/0x520 Nov 25 15:49:53 k2000 kernel: [ 2408.822731] [<ffffffff810bab75>] ? c...

The following series of 10 patches is a merge of the xenfb and xenconsoled functionality into the qemu-dm code. The general approach taken is to have qemu-dm provide two machine types - one for xen paravirt, the other for fullyvirt. For compatability the later is the default. The goals overall are to kill LibVNCServer, remove alot of code duplication and/or parallel impls of the same concepts, and

Hello all Xen devels, I''m writing howto''s in brazilian portuguese language about Xen-3.3.0 stubdom, pv-grub and others configurations. Following what is written in xen-3.3.0/stubdom/README, I tried start a HVM but without success. When I start the virtual machine calling ''xm create hvmtest'' command, in a few seconds the virtual machine is dead (see

...ated message is read. The use case is non-threaded libevent, we have two event A and B: - Event A will destroy something and call xs_unwatch; - Event B is used to notify that a node has changed in XenStore. As the event is called one by one, event A can be handled before event B. So on next xs_watch_read the user could retrieve an unwatch token and a segfault occured if the token store the pointer of the structure (ie: "backend:0xcafe"). To avoid problem with previous application using libXenStore, this behaviour will only be enabled if XS_UNWATCH_FILTER is given to xs_open. Signed...

...ated message is read. The use case is non-threaded libevent, we have two event A and B: - Event A will destroy something and call xs_unwatch; - Event B is used to notify that a node has changed in XenStore. As the event is called one by one, event A can be handled before event B. So on next xs_watch_read the user could retrieve an unwatch token and a segfault occured if the token store the pointer of the structure (ie: "backend:0xcafe"). To avoid problem with previous application using libXenStore, this behaviour will only be enabled if XS_UNWATCH_FILTER is given to xs_open. Signed...