xenbus: Fix S3 frontend resume when xenstored is not running
v2 cover xenbus: Fix S3 frontend resume when xenstored is not running
Hi,
This patch series fixes the S3 resume of a domain running xenstored running a frontend over xenbus (xen-netfront in my use case).
As device resume is happening before process resume, the xenbus frontend resume is hanging if xenstored is not running, thus causing a deadlock. This patch series is fixing that issue by deferring the xenbus frontend resume when we are running xenstored in that same domain.
Aurelien Chartier (2):
xenbus: save xenstore local status for later use
xenbus: defer xenbus frontend resume if xenstored is not running
drivers/xen/xenbus/xenbus_comms.h | 1 + drivers/xen/xenbus/xenbus_probe.c | 27 ++++++++++------------ drivers/xen/xenbus/xenbus_probe.h | 12 ++++++++++ drivers/xen/xenbus/xenbus_probe_frontend.c | 34 +++++++++++++++++++++++++++- 4 files changed, 58 insertions(+), 16 deletions(-)
Signature · 1 line
1.7.10.4
PATCH v2 1/2 xenbus: save xenstore local status for later use
Save the xenstore local status computed in xenbus_init. It can then be used later to check if xenstored is running in that domain.
--- drivers/xen/xenbus/xenbus_comms.h | 1 + drivers/xen/xenbus/xenbus_probe.c | 27 ++++++++++++--------------- drivers/xen/xenbus/xenbus_probe.h | 7 +++++++ 3 files changed, 20 insertions(+), 15 deletions(-)
drivers/xen/xenbus/xenbus_comms.h +1 -0
@@ -45,6 +45,7 @@ int xb_wait_for_data_to_read(void);int xs_input_avail (void );extern struct xenstore_domain_interface *xen_store_interface;extern int xen_store_evtchn; +extern enum xenstore_init xen_store_domain;extern const struct file_operations xen_xenbus_fops;
drivers/xen/xenbus/xenbus_probe.c +12 -15
@@ -69,6 +69,9 @@ EXPORT_SYMBOL_GPL(xen_store_evtchn);struct xenstore_domain_interface *xen_store_interface;EXPORT_SYMBOL_GPL (xen_store_interface); +enum xenstore_init xen_store_domain; +EXPORT_SYMBOL_GPL (xen_store_domain); +static unsigned long xen_store_mfn;static BLOCKING_NOTIFIER_HEAD (xenstore_chain);
@@ -719,17 +722,11 @@ static int __init xenstored_local_init(void)return err; } -enum xenstore_init { -UNKNOWN , -PV , -HVM , -LOCAL , -};static int __init xenbus_init (void ) {int err =0 ; -enum xenstore_init usage =UNKNOWN ;uint64_t v =0 ; + xen_store_domain =XS_UNKNOWN ;if (!xen_domain ())return -ENODEV ;
@@ -737,29 +734,29 @@ static int __init xenbus_init(void)xenbus_ring_ops_init ();if (xen_pv_domain ()) - usage =PV ; + xen_store_domain =XS_PV ;if (xen_hvm_domain ()) - usage =HVM ; + xen_store_domain =XS_HVM ;if (xen_hvm_domain () &&xen_initial_domain ()) - usage =LOCAL ; + xen_store_domain =XS_LOCAL ;if (xen_pv_domain () && !xen_start_info->store_evtchn) - usage =LOCAL ; + xen_store_domain =XS_LOCAL ;if (xen_pv_domain () && xen_start_info->store_evtchn) xenstored_ready =1 ; -switch (usage) { -case LOCAL : +switch (xen_store_domain) { +case XS_LOCAL : err =xenstored_local_init ();if (err) goto out_error; xen_store_interface =mfn_to_virt (xen_store_mfn);break ; -case PV : +case XS_PV : xen_store_evtchn = xen_start_info->store_evtchn; xen_store_mfn = xen_start_info->store_mfn; xen_store_interface =mfn_to_virt (xen_store_mfn);break ; -case HVM : +case XS_HVM : err =hvm_get_parameter (HVM_PARAM_STORE_EVTCHN , &v);if (err) goto out_error;
drivers/xen/xenbus/xenbus_probe.h +7 -0
@@ -47,6 +47,13 @@ struct xen_bus_type {struct bus_type bus; }; +enum xenstore_init { +XS_UNKNOWN , +XS_PV , +XS_HVM , +XS_LOCAL , +}; +extern struct device_attribute xenbus_dev_attrs[];extern int xenbus_match (struct device *_dev,struct device_driver *_drv);
Signature · 1 line
1.7.10.4
PATCH v2 2/2 xenbus: defer xenbus frontend resume if xenstored is not running
In reply to Aurelien ChartierIf the xenbus frontend is running in a domain running xenstored, the device resume is hanging because it is happening before the process resume. This patch adds extra logic to the resume code to check if we are the domain running xenstored and delay the resume if needed.
Changes in v2:
- Instead of bypassing the resume, process it in a workqueue
--- drivers/xen/xenbus/xenbus_probe.h | 5 ++++ drivers/xen/xenbus/xenbus_probe_frontend.c | 34 +++++++++++++++++++++++++++- 2 files changed, 38 insertions(+), 1 deletion(-)
drivers/xen/xenbus/xenbus_probe.h +5 -0
@@ -47,6 +47,11 @@ struct xen_bus_type {struct bus_type bus; }; +struct xenbus_resume_work { +struct work_struct w; +struct device *dev; +}; +enum xenstore_init {XS_UNKNOWN ,XS_PV ,
drivers/xen/xenbus/xenbus_probe_frontend.c +33 -1
@@ -28,6 +28,7 @@#include "xenbus_comms.h" #include "xenbus_probe.h" +static struct workqueue_struct *xenbus_frontend_resume_wq;/* device/<type>/<id> => <type>-<id> */ static int frontend_bus_id (char bus_id[XEN_BUS_ID_SIZE ],const char *nodename)
@@ -89,9 +90,38 @@ static void backend_changed(struct xenbus_watch *watch,xenbus_otherend_changed (watch, vec, len,1 ); } +static void xenbus_frontend_delayed_resume (struct work_struct *w) +{ +struct xenbus_resume_work *resume_work = (struct xenbus_resume_work *) w; + +xenbus_dev_resume (dev); + +kfree (w); +} + +static int xenbus_frontend_dev_resume (struct device *dev) +{ +/* +* If xenstored is running in that domain, we cannot access the backend +* state at the moment, so we need to defer xenbus_dev_resume +*/ +if (xen_store_domain ==XS_LOCAL ) { +struct xenbus_resume_work *work = +kmalloc (sizeof (struct xenbus_resume ),GFP_KERNEL ); + +INIT_WORK ((struct work_struct *) work, xenbus_frontend_delayed_resume); + resume_work->dev = dev; + queue_work(xenbus_frontend_resume_wq, (struct work_struct *) work) + +return 0 ; + } + +return xenbus_dev_resume (dev); +} +static const struct dev_pm_ops xenbus_pm_ops = { .suspend = xenbus_dev_suspend, - .resume = xenbus_dev_resume, + .resume = xenbus_frontend_dev_resume, .freeze = xenbus_dev_suspend, .thaw = xenbus_dev_cancel, .restore = xenbus_dev_resume,
@@ -440,6 +470,8 @@ static int __init xenbus_probe_frontend_init(void)register_xenstore_notifier (&xenstore_notifier); + xenbus_frontend_resume_wq =create_workqueue ("xenbus_frontend_resume" ); +return 0 ; }subsys_initcall (xenbus_probe_frontend_init);
Signature · 1 line
1.7.10.4
Re: [PATCH V2 2/2] xenbus: defer xenbus frontend resume if xenstored is not running
Quoted 58 lines · Aurelien Chartier
If the xenbus frontend is running in a domain running xenstored, the device resume is hanging because it is happening before the process resume. This patch adds extra logic to the resume code to check if we are the domain running xenstored and delay the resume if needed.
Signed-off-by Aurelien Chartier <[email protected]>Changes in v2:
- Instead of bypassing the resume, process it in a workqueue--- drivers/xen/xenbus/xenbus_probe.h | 5 ++++ drivers/xen/xenbus/xenbus_probe_frontend.c | 34 +++++++++++++++++++++++++++- 2 files changed, 38 insertions(+), 1 deletion(-)drivers/xen/xenbus/xenbus_probe.h +5
-0@@ -47,6 +47,11 @@ struct xen_bus_type {struct bus_type bus; }; +struct xenbus_resume_work { +struct work_struct w; +struct device *dev; +}; +enum xenstore_init {XS_UNKNOWN ,XS_PV ,drivers/xen/xenbus/xenbus_probe_frontend.c +13
-0@@ -28,6 +28,7 @@#include "xenbus_comms.h" #include "xenbus_probe.h" +static struct workqueue_struct *xenbus_frontend_resume_wq;/* device/<type>/<id> => <type>-<id> */ static int frontend_bus_id (char bus_id[XEN_BUS_ID_SIZE ],const char *nodename)@@ -89,9 +90,38 @@ static void backend_changed(struct xenbus_watch *watch,xenbus_otherend_changed (watch, vec, len,1 ); } +static void xenbus_frontend_delayed_resume (struct work_struct *w) +{ +struct xenbus_resume_work *resume_work = (struct xenbus_resume_work *) w; + +xenbus_dev_resume (dev); + +kfree (w); +} + +static int xenbus_frontend_dev_resume (struct device *dev) +{ +/*
drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 114
Quoted 10 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +10
-0+* If xenstored is running in that domain, we cannot access the backend +* state at the moment, so we need to defer xenbus_dev_resume +*/ +if (xen_store_domain ==XS_LOCAL ) { +struct xenbus_resume_work *work = +kmalloc (sizeof (struct xenbus_resume ),GFP_KERNEL ); + +INIT_WORK ((struct work_struct *) work, xenbus_frontend_delayed_resume); + resume_work->dev = dev; +queue_work (xenbus_frontend_resume_wq, (struct work_struct *) work)
Missing a semicolon here.
I should have read that patch once more, apologies for that.
Quoted 23 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +10
-1+ +return 0 ; + } + +return xenbus_dev_resume (dev); +} +static const struct dev_pm_ops xenbus_pm_ops = { .suspend = xenbus_dev_suspend, - .resume = xenbus_dev_resume, + .resume = xenbus_frontend_dev_resume, .freeze = xenbus_dev_suspend, .thaw = xenbus_dev_cancel, .restore = xenbus_dev_resume,@@ -440,6 +470,8 @@ static int __init xenbus_probe_frontend_init(void)register_xenstore_notifier (&xenstore_notifier); + xenbus_frontend_resume_wq =create_workqueue ("xenbus_frontend_resume" ); +return 0 ; }subsys_initcall (xenbus_probe_frontend_init);
Re: [PATCH V2 2/2] xenbus: defer xenbus frontend resume if xenstored is not running
In reply to Aurelien Chartierdrivers/xen/xenbus/xenbus_probe.h struct xen_bus_type { · line 54
Quoted 12 lines
Quoted 1 line
Quoted 1 line · Aurelien Chartier
drivers/xen/xenbus/xenbus_probe.h +5
-0@@ -47,6 +47,11 @@ struct xen_bus_type {struct bus_type bus; }; +struct xenbus_resume_work { +struct work_struct w; +struct device *dev; +}; +
I don't think this structure needs to be in a header - it's being used in a single source file only.
Quoted 11 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +1
-0@@ -28,6 +28,7 @@#include "xenbus_comms.h" #include "xenbus_probe.h" +static struct workqueue_struct *xenbus_frontend_resume_wq;/* device/<type>/<id> => <type>-<id> */ static int frontend_bus_id (char bus_id[XEN_BUS_ID_SIZE ],const char *nodename)
drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 95
Quoted 6 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +3
-0@@ -89,9 +90,38 @@ static void backend_changed(struct xenbus_watch *watch,xenbus_otherend_changed (watch, vec, len,1 ); } +static void xenbus_frontend_delayed_resume (struct work_struct *w) +{ +struct xenbus_resume_work *resume_work = (struct xenbus_resume_work *) w;
container_of()
drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 97
Quoted 2 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +2
-0+ +xenbus_dev_resume (dev);
Does this build at all? I don't see where "dev" is being declared/ initialized?
drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 99
Quoted 2 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +2
-0+ +kfree (w);
kfree(resume_work) - otherwise you have a hidden dependency on "w" being the first member of struct xenbus_resume_work.
drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 110
Quoted 11 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +11
-0+} + +static int xenbus_frontend_dev_resume (struct device *dev) +{ +/* +* If xenstored is running in that domain, we cannot access the backend +* state at the moment, so we need to defer xenbus_dev_resume +*/ +if (xen_store_domain ==XS_LOCAL ) { +struct xenbus_resume_work *work = +kmalloc (sizeof (struct xenbus_resume ),GFP_KERNEL );
Missing NULL return check (don't know what you should do in that case). Perhaps dynamic allocation is the wrong approach here, and you want to rather add a new field to struct xenbus_device (which gets populated only for frontend devices, and - if possible - only when xen_store_domain == XS_LOCAL.
Also - GFP_KERNEL really suitable here?
drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 112
Quoted 2 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +2
-0+ +INIT_WORK ((struct work_struct *) work, xenbus_frontend_delayed_resume);
&work->w (without any cast).
drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 114
Quoted 2 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +2
-0+ resume_work->dev = dev; +queue_work (xenbus_frontend_resume_wq, (struct work_struct *) work)
Again.
Jan
Re: [PATCH V2 2/2] xenbus: defer xenbus frontend resume if xenstored is not running
Quoted 34 lines · Jan Beulich
drivers/xen/xenbus/xenbus_probe.h struct xen_bus_type { · line 54
Quoted 12 lines
Quoted 1 line
Quoted 1 line · Aurelien Chartier
drivers/xen/xenbus/xenbus_probe.h +5
-0@@ -47,6 +47,11 @@ struct xen_bus_type {struct bus_type bus; }; +struct xenbus_resume_work { +struct work_struct w; +struct device *dev; +}; +I don't think this structure needs to be in a header - it's being used in a single source file only.
Quoted 11 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +1
-0@@ -28,6 +28,7 @@#include "xenbus_comms.h" #include "xenbus_probe.h" +static struct workqueue_struct *xenbus_frontend_resume_wq;/* device/<type>/<id> => <type>-<id> */ static int frontend_bus_id (char bus_id[XEN_BUS_ID_SIZE ],const char *nodename)drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 95
Quoted 6 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +3
-0@@ -89,9 +90,38 @@ static void backend_changed(struct xenbus_watch *watch,xenbus_otherend_changed (watch, vec, len,1 ); } +static void xenbus_frontend_delayed_resume (struct work_struct *w) +{ +struct xenbus_resume_work *resume_work = (struct xenbus_resume_work *) w;
container_of()
Quoted 4 lines
drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 97
Quoted 2 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +2
-0+ +xenbus_dev_resume (dev);Does this build at all? I don't see where "dev" is being declared/ initialized?
I messed up my config file and ended up with CONFIG_XEN_XENBUS_FRONTEND wiped out, so I did not catch this one. Sorry for that, it was meant to be resume_work->dev.
Quoted 23 lines
drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 99
Quoted 3 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +2
-0+ +kfree (w);
kfree(resume_work)- otherwise you have a hidden dependency on "w" being the first member of structxenbus_resume_work.drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 110
Quoted 11 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +11
-0+} + +static int xenbus_frontend_dev_resume (struct device *dev) +{ +/* +* If xenstored is running in that domain, we cannot access the backend +* state at the moment, so we need to defer xenbus_dev_resume +*/ +if (xen_store_domain ==XS_LOCAL ) { +struct xenbus_resume_work *work = +kmalloc (sizeof (struct xenbus_resume ),GFP_KERNEL );Missing NULL return check (don't know what you should do in that case). Perhaps dynamic allocation is the wrong approach here, and you want to rather add a new field to struct
xenbus_device(which gets populated only for frontend devices, and - if possible - only whenxen_store_domain==XS_LOCAL.
Quoted 1 line
Also -
GFP_KERNELreally suitable here?
Adding a new field for xenbus_device for that purpose seemed too much for me, but it is probably a cleaner solution in the end.
I don't really see a way to avoid having that field for backend devices. We can have a #ifdef CONFIG_XEN_XENBUS_FRONTEND, but that won't prevent domains with both frontend and backend to have that field set in all xenbus devices.
Quoted 8 lines
drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 112
Quoted 3 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +2
-0+ +INIT_WORK ((struct work_struct *) work, xenbus_frontend_delayed_resume);
&work->w(without any cast).drivers/xen/xenbus/xenbus_probe_frontend.c static void backend_changed(struct xenbus_watch *watch, · line 114
Quoted 2 lines
drivers/xen/xenbus/xenbus_probe_frontend.c +2
-0+ resume_work->dev = dev; +queue_work (xenbus_frontend_resume_wq, (struct work_struct *) work)Again.
Thanks for the review. I will fix the errors in my next patch series.
Aurelien.