| Message ID | 20200603074819.21639-1-kai.heng.feng@canonical.com |
|---|---|
| State | Not Applicable |
| Delegated to: | David Miller |
| Headers | show |
| Series | [v3] libata: Use per port sync for detach | expand |
On 03/06/2020 08:48, Kai-Heng Feng wrote: > Commit 130f4caf145c ("libata: Ensure ata_port probe has completed before > detach") may cause system freeze during suspend. > > Using async_synchronize_full() in PM callbacks is wrong, since async > callbacks that are already scheduled may wait for not-yet-scheduled > callbacks, causes a circular dependency. > > Instead of using big hammer like async_synchronize_full(), use async > cookie to make sure port probe are synced, without affecting other > scheduled PM callbacks. > > Fixes: 130f4caf145c ("libata: Ensure ata_port probe has completed before detach") > BugLink: https://bugs.launchpad.net/bugs/1867983 > Suggested-by: John Garry <john.garry@huawei.com> > Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com> thanks, Tested-by: John Garry <john.garry@huawei.com> > --- > v3: > - Move the comment to properly align with the code. > > v2: > - Sync up to cookie + 1. > - Squash the synchronization into the same loop. > > drivers/ata/libata-core.c | 11 +++++------ > include/linux/libata.h | 3 +++ > 2 files changed, 8 insertions(+), 6 deletions(-) > > diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c > index 69361ec43db5..b1cd4d97bc2a 100644 > --- a/drivers/ata/libata-core.c > +++ b/drivers/ata/libata-core.c > @@ -42,7 +42,6 @@ > #include <linux/workqueue.h> > #include <linux/scatterlist.h> > #include <linux/io.h> > -#include <linux/async.h> > #include <linux/log2.h> > #include <linux/slab.h> > #include <linux/glob.h> > @@ -5778,7 +5777,7 @@ int ata_host_register(struct ata_host *host, struct scsi_host_template *sht) > /* perform each probe asynchronously */ > for (i = 0; i < host->n_ports; i++) { > struct ata_port *ap = host->ports[i]; > - async_schedule(async_port_probe, ap); > + ap->cookie = async_schedule(async_port_probe, ap); > } > > return 0; > @@ -5920,11 +5919,11 @@ void ata_host_detach(struct ata_host *host) > { > int i; > > - /* Ensure ata_port probe has completed */ > - async_synchronize_full(); > - > - for (i = 0; i < host->n_ports; i++) > + for (i = 0; i < host->n_ports; i++) { > + /* Ensure ata_port probe has completed */ > + async_synchronize_cookie(host->ports[i]->cookie + 1); > ata_port_detach(host->ports[i]); > + } > > /* the host is dead now, dissociate ACPI */ > ata_acpi_dissociate(host); > diff --git a/include/linux/libata.h b/include/linux/libata.h > index af832852e620..8a4843704d28 100644 > --- a/include/linux/libata.h > +++ b/include/linux/libata.h > @@ -22,6 +22,7 @@ > #include <linux/acpi.h> > #include <linux/cdrom.h> > #include <linux/sched.h> > +#include <linux/async.h> > > /* > * Define if arch has non-standard setup. This is a _PCI_ standard > @@ -872,6 +873,8 @@ struct ata_port { > struct timer_list fastdrain_timer; > unsigned long fastdrain_cnt; > > + async_cookie_t cookie; > + > int em_message_type; > void *private_data; > >
Hi Jens, > On Jun 3, 2020, at 16:40, John Garry <john.garry@huawei.com> wrote: > > On 03/06/2020 08:48, Kai-Heng Feng wrote: >> Commit 130f4caf145c ("libata: Ensure ata_port probe has completed before >> detach") may cause system freeze during suspend. >> Using async_synchronize_full() in PM callbacks is wrong, since async >> callbacks that are already scheduled may wait for not-yet-scheduled >> callbacks, causes a circular dependency. >> Instead of using big hammer like async_synchronize_full(), use async >> cookie to make sure port probe are synced, without affecting other >> scheduled PM callbacks. >> Fixes: 130f4caf145c ("libata: Ensure ata_port probe has completed before detach") >> BugLink: https://bugs.launchpad.net/bugs/1867983 >> Suggested-by: John Garry <john.garry@huawei.com> >> Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com> > > thanks, > Tested-by: John Garry <john.garry@huawei.com> Can you please review or merge this patch? Kai-Heng > >> --- >> v3: >> - Move the comment to properly align with the code. >> v2: >> - Sync up to cookie + 1. >> - Squash the synchronization into the same loop. >> drivers/ata/libata-core.c | 11 +++++------ >> include/linux/libata.h | 3 +++ >> 2 files changed, 8 insertions(+), 6 deletions(-) >> diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c >> index 69361ec43db5..b1cd4d97bc2a 100644 >> --- a/drivers/ata/libata-core.c >> +++ b/drivers/ata/libata-core.c >> @@ -42,7 +42,6 @@ >> #include <linux/workqueue.h> >> #include <linux/scatterlist.h> >> #include <linux/io.h> >> -#include <linux/async.h> >> #include <linux/log2.h> >> #include <linux/slab.h> >> #include <linux/glob.h> >> @@ -5778,7 +5777,7 @@ int ata_host_register(struct ata_host *host, struct scsi_host_template *sht) >> /* perform each probe asynchronously */ >> for (i = 0; i < host->n_ports; i++) { >> struct ata_port *ap = host->ports[i]; >> - async_schedule(async_port_probe, ap); >> + ap->cookie = async_schedule(async_port_probe, ap); >> } >> return 0; >> @@ -5920,11 +5919,11 @@ void ata_host_detach(struct ata_host *host) >> { >> int i; >> - /* Ensure ata_port probe has completed */ >> - async_synchronize_full(); >> - >> - for (i = 0; i < host->n_ports; i++) >> + for (i = 0; i < host->n_ports; i++) { >> + /* Ensure ata_port probe has completed */ >> + async_synchronize_cookie(host->ports[i]->cookie + 1); >> ata_port_detach(host->ports[i]); >> + } >> /* the host is dead now, dissociate ACPI */ >> ata_acpi_dissociate(host); >> diff --git a/include/linux/libata.h b/include/linux/libata.h >> index af832852e620..8a4843704d28 100644 >> --- a/include/linux/libata.h >> +++ b/include/linux/libata.h >> @@ -22,6 +22,7 @@ >> #include <linux/acpi.h> >> #include <linux/cdrom.h> >> #include <linux/sched.h> >> +#include <linux/async.h> >> /* >> * Define if arch has non-standard setup. This is a _PCI_ standard >> @@ -872,6 +873,8 @@ struct ata_port { >> struct timer_list fastdrain_timer; >> unsigned long fastdrain_cnt; >> + async_cookie_t cookie; >> + >> int em_message_type; >> void *private_data; >> >
On 6/3/20 1:48 AM, Kai-Heng Feng wrote: > Commit 130f4caf145c ("libata: Ensure ata_port probe has completed before > detach") may cause system freeze during suspend. > > Using async_synchronize_full() in PM callbacks is wrong, since async > callbacks that are already scheduled may wait for not-yet-scheduled > callbacks, causes a circular dependency. > > Instead of using big hammer like async_synchronize_full(), use async > cookie to make sure port probe are synced, without affecting other > scheduled PM callbacks. Queued up for 5.8, thanks.
diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c index 69361ec43db5..b1cd4d97bc2a 100644 --- a/drivers/ata/libata-core.c +++ b/drivers/ata/libata-core.c @@ -42,7 +42,6 @@ #include <linux/workqueue.h> #include <linux/scatterlist.h> #include <linux/io.h> -#include <linux/async.h> #include <linux/log2.h> #include <linux/slab.h> #include <linux/glob.h> @@ -5778,7 +5777,7 @@ int ata_host_register(struct ata_host *host, struct scsi_host_template *sht) /* perform each probe asynchronously */ for (i = 0; i < host->n_ports; i++) { struct ata_port *ap = host->ports[i]; - async_schedule(async_port_probe, ap); + ap->cookie = async_schedule(async_port_probe, ap); } return 0; @@ -5920,11 +5919,11 @@ void ata_host_detach(struct ata_host *host) { int i; - /* Ensure ata_port probe has completed */ - async_synchronize_full(); - - for (i = 0; i < host->n_ports; i++) + for (i = 0; i < host->n_ports; i++) { + /* Ensure ata_port probe has completed */ + async_synchronize_cookie(host->ports[i]->cookie + 1); ata_port_detach(host->ports[i]); + } /* the host is dead now, dissociate ACPI */ ata_acpi_dissociate(host); diff --git a/include/linux/libata.h b/include/linux/libata.h index af832852e620..8a4843704d28 100644 --- a/include/linux/libata.h +++ b/include/linux/libata.h @@ -22,6 +22,7 @@ #include <linux/acpi.h> #include <linux/cdrom.h> #include <linux/sched.h> +#include <linux/async.h> /* * Define if arch has non-standard setup. This is a _PCI_ standard @@ -872,6 +873,8 @@ struct ata_port { struct timer_list fastdrain_timer; unsigned long fastdrain_cnt; + async_cookie_t cookie; + int em_message_type; void *private_data;
Commit 130f4caf145c ("libata: Ensure ata_port probe has completed before detach") may cause system freeze during suspend. Using async_synchronize_full() in PM callbacks is wrong, since async callbacks that are already scheduled may wait for not-yet-scheduled callbacks, causes a circular dependency. Instead of using big hammer like async_synchronize_full(), use async cookie to make sure port probe are synced, without affecting other scheduled PM callbacks. Fixes: 130f4caf145c ("libata: Ensure ata_port probe has completed before detach") BugLink: https://bugs.launchpad.net/bugs/1867983 Suggested-by: John Garry <john.garry@huawei.com> Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com> --- v3: - Move the comment to properly align with the code. v2: - Sync up to cookie + 1. - Squash the synchronization into the same loop. drivers/ata/libata-core.c | 11 +++++------ include/linux/libata.h | 3 +++ 2 files changed, 8 insertions(+), 6 deletions(-)