Legacy pm ops to dev_pm_ops patches
Hi Rafael,
I checked on the status of the legacy pm ops to dev_pm_ops conversion patches. Several went in. I have about 13 patches I would like to send. These patches are similar in nature. One of the issues I am having is being able to test these patches. Except for drivers/pcmcia/ds.c patch, others I am limited to compile testing only. Would it be okay to I send these as RFT patches cc'ing you and linux-pm and in addition to the maintainers for each of these patches?
Patch details below:
5 mmc host driver patches:
drivers/mmc/host/au1xmmc.cdrivers/mmc/host/bfin_sdh.cdrivers/mmc/host/cb710-mmc.cdrivers/mmc/host/msm_sdcc.cdrivers/mmc/host/tmio_mmc.c
3 arm platform driver patches:
arch/arm/common/locomo.carch/arm/common/sa1111.carch/arm/common/scoop.c
1 macintosh driver patch:drivers/macintosh/adb.c
3 bus drivers patches:drivers/base/isa.cdrivers/pcmcia/ds.cdrivers/s390/crypto/ap_bus.c
The nature if changes is very similar:
1. Define dev_pm_ops as in the example below:
+static const struct dev_pm_ops pcmcia_bus_dev_pm_ops = {
+ .suspend = pcmcia_dev_suspend, + .resume = pcmcia_dev_resume, + /* Hibernate callbacks */ + .freeze = pcmcia_dev_freeze, + .thaw = pcmcia_dev_resume, + .poweroff = pcmcia_dev_poweroff, + .restore = pcmcia_dev_resume,
+};
struct bus_type pcmcia_bus_type = {
.name = "pcmcia",@@ -1405,8 +1432,7 @@ struct bus_type pcmcia_bus_type = {
.dev_groups = pcmcia_dev_groups,
.probe = pcmcia_device_probe,
.remove = pcmcia_device_remove,
- .suspend = pcmcia_dev_suspend, - .resume = pcmcia_dev_resume, + .pm = &pcmcia_bus_dev_pm_ops,
};
2. Define new interfaces for poweroff, freeze which in turn call existing suspend interface, as in the example below. thaw, resume, and restore are straight forward and just point to existing resume() hook
+static int pcmcia_dev_suspend(struct device *dev)
+{
+ __pcmcia_dev_suspend(dev, PMSG_SUSPEND);
+}
+
+static int pcmcia_dev_poweroff(struct device *dev)
+{
+ __pcmcia_dev_suspend(dev, PMSG_HIBERNATE);
+}
+
+static int pcmcia_dev_freeze(struct device *dev)
+{
+ __pcmcia_dev_suspend(dev, PMSG_FREEZE);
+}
+
+static int __pcmcia_dev_suspend(struct device *dev, pm_message_t state)
{
struct pcmcia_device *p_dev = to_pcmcia_dev(dev);
struct pcmcia_driver *p_drv = NULL;@@ -1397,6 +1415,15 @@ static struct class_interface pcmcia_bus_interface __refdata = {
.remove_dev = &pcmcia_bus_remove_socket,
};
thanks,
-- Shuah
Signature · 4 lines
Shuah Khan Senior Linux Kernel Developer - Open Source Group Samsung Research America(Silicon Valley) [email protected] | (970) 672-0658
Quoted 1 line · Shuah Khan
Hi Rafael,
Hi,
Quoted 80 lines
I checked on the status of the legacy pm ops to
dev_pm_opsconversion patches. Several went in. I have about 13 patches I would like to send. These patches are similar in nature. One of the issues I am having is being able to test these patches. Except fordrivers/pcmcia/ds.cpatch, others I am limited to compile testing only. Would it be okay to I send these as RFT patches cc'ing you and linux-pm and in addition to the maintainers for each of these patches?Patch details below:
5 mmc host driver patches:
drivers/mmc/host/au1xmmc.cdrivers/mmc/host/bfin_sdh.cdrivers/mmc/host/cb710-mmc.cdrivers/mmc/host/msm_sdcc.cdrivers/mmc/host/tmio_mmc.c3 arm platform driver patches:
arch/arm/common/locomo.carch/arm/common/sa1111.carch/arm/common/scoop.c1 macintosh driver patch:
drivers/macintosh/adb.c3 bus drivers patches:
drivers/base/isa.cdrivers/pcmcia/ds.cdrivers/s390/crypto/ap_bus.cThe nature if changes is very similar:
1. Definedev_pm_opsas in the example below:(quoted diff) +10
-2+static const struct dev_pm_ops pcmcia_bus_dev_pm_ops = { + .suspend = pcmcia_dev_suspend, + .resume = pcmcia_dev_resume, + /* Hibernate callbacks */ + .freeze = pcmcia_dev_freeze, + .thaw = pcmcia_dev_resume, + .poweroff = pcmcia_dev_poweroff, + .restore = pcmcia_dev_resume, +}; struct bus_type pcmcia_bus_type = { .name = "pcmcia",@@ -1405,8 +1432,7 @@ struct bus_type pcmcia_bus_type = { .dev_groups = pcmcia_dev_groups, .probe = pcmcia_device_probe, .remove = pcmcia_device_remove, - .suspend = pcmcia_dev_suspend, - .resume = pcmcia_dev_resume, + .pm = &pcmcia_bus_dev_pm_ops, };2. Define new interfaces for poweroff, freeze which in turn call existing suspend interface, as in the example below. thaw, resume, and restore are straight forward and just point to existing
resume()hook(quoted diff) +16
-0+static int pcmcia_dev_suspend(struct device *dev) +{ + __pcmcia_dev_suspend(dev, PMSG_SUSPEND); +} + +static int pcmcia_dev_poweroff(struct device *dev) +{ + __pcmcia_dev_suspend(dev, PMSG_HIBERNATE); +} + +static int pcmcia_dev_freeze(struct device *dev) +{ + __pcmcia_dev_suspend(dev, PMSG_FREEZE); +} + +static int __pcmcia_dev_suspend(struct device *dev, pm_message_t state) { struct pcmcia_device *p_dev = to_pcmcia_dev(dev); struct pcmcia_driver *p_drv = NULL;
Review static struct class_interface · line 1418
Quoted 4 lines
(quoted diff) +0
-0@@ -1397,6 +1415,15 @@ static struct class_interface pcmcia_bus_interface __refdata = { .remove_dev = &pcmcia_bus_remove_socket, };
OK
Can you please send all of these patches in one series to me and linux-pm and CC the LKML and the maintainers of the subsystems in question (only the patches concerning them directly)?
Rafael