| Message ID | 20230810053958.14992-2-yangx.jy@fujitsu.com |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [NDCTL,1/2] daxctl: Don't check param.no_movable when param.no_online is set | expand |
Xiao Yang wrote: > Try to make daxctl reconfigure-device with system-ram mode > offline memory when both param.no_online and param.force > are set but daxctl_dev_will_auto_online_memory returns true. So is the goal here to try to save some steps in the case where the kernel already onlined the device? It should probably emit a warning that the memory was onlined automatically so the admin can consider changing the default kernel policy. Otherwise, it may be too late to undo the onlining at this point. > > Signed-off-by: Xiao Yang <yangx.jy@fujitsu.com> > --- > daxctl/device.c | 9 +++++++-- > 1 file changed, 7 insertions(+), 2 deletions(-) > > diff --git a/daxctl/device.c b/daxctl/device.c > index ba31eb6..dfa7f79 100644 > --- a/daxctl/device.c > +++ b/daxctl/device.c > @@ -83,7 +83,7 @@ OPT_STRING('m', "mode", ¶m.mode, "mode", "mode to switch the device to"), \ > OPT_BOOLEAN('N', "no-online", ¶m.no_online, \ > "don't auto-online memory sections"), \ > OPT_BOOLEAN('f', "force", ¶m.force, \ > - "attempt to offline memory sections before reconfiguration"), \ > + "attempt to offline memory sections for reconfiguration"), \ > OPT_BOOLEAN('C', "check-config", ¶m.check_config, \ > "use config files to determine parameters for the operation") > > @@ -734,8 +734,13 @@ static int reconfig_mode_system_ram(struct daxctl_dev *dev) > return rc; > } > > - if (param.no_online) > + if (param.no_online) { > + if (param.force && daxctl_dev_will_auto_online_memory(dev)) { > + rc = dev_offline_memory(dev); > + return rc; It is not clear that this is an error that should fail the reconfigure-device, because the reconfiguration succeeded. The fact that the kernel policy forced the memory online is the administrators fault for setting conflicting policy. This is why I think a warning is appropriate because the administrator is confused if they are letting kernel an daxctl policy conflict.
diff --git a/daxctl/device.c b/daxctl/device.c index ba31eb6..dfa7f79 100644 --- a/daxctl/device.c +++ b/daxctl/device.c @@ -83,7 +83,7 @@ OPT_STRING('m', "mode", ¶m.mode, "mode", "mode to switch the device to"), \ OPT_BOOLEAN('N', "no-online", ¶m.no_online, \ "don't auto-online memory sections"), \ OPT_BOOLEAN('f', "force", ¶m.force, \ - "attempt to offline memory sections before reconfiguration"), \ + "attempt to offline memory sections for reconfiguration"), \ OPT_BOOLEAN('C', "check-config", ¶m.check_config, \ "use config files to determine parameters for the operation") @@ -734,8 +734,13 @@ static int reconfig_mode_system_ram(struct daxctl_dev *dev) return rc; } - if (param.no_online) + if (param.no_online) { + if (param.force && daxctl_dev_will_auto_online_memory(dev)) { + rc = dev_offline_memory(dev); + return rc; + } return 0; + } return dev_online_memory(dev); }
Try to make daxctl reconfigure-device with system-ram mode offline memory when both param.no_online and param.force are set but daxctl_dev_will_auto_online_memory returns true. Signed-off-by: Xiao Yang <yangx.jy@fujitsu.com> --- daxctl/device.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-)