diff mbox

[v4,2/3] VT-d: Reduce spin timeout to 1ms, which can be boot-time changed.

Message ID 945CA011AD5F084CBEA3E851C0AB28894B818444@SHSMSX101.ccr.corp.intel.com (mailing list archive)
State New, archived
Headers show

Commit Message

Quan Xu Jan. 14, 2016, 1:59 a.m. UTC
On 14.01.2016 at 12:58am, <JBeulich@suse.com> wrote:
> >>> On 23.12.15 at 09:25, <quan.xu@intel.com> wrote:
> > --- a/xen/drivers/passthrough/vtd/qinval.c
> > +++ b/xen/drivers/passthrough/vtd/qinval.c
> > @@ -28,6 +28,11 @@
> >  #include "vtd.h"
> >  #include "extern.h"
> >
> > +static int __read_mostly iommu_qi_timeout_ms = 1;
> > +integer_param("iommu_qi_timeout_ms", iommu_qi_timeout_ms);
> 
> But wait - this needs to be accompanied by an entry in
> docs/misc/xen-command-line.markdown.
> 

Jan, Is the following right?



-Quan

Comments

Jan Beulich Jan. 14, 2016, 9:03 a.m. UTC | #1
>>> On 14.01.16 at 02:59, <quan.xu@intel.com> wrote:
> On 14.01.2016 at 12:58am, <JBeulich@suse.com> wrote:
>> >>> On 23.12.15 at 09:25, <quan.xu@intel.com> wrote:
>> > --- a/xen/drivers/passthrough/vtd/qinval.c
>> > +++ b/xen/drivers/passthrough/vtd/qinval.c
>> > @@ -28,6 +28,11 @@
>> >  #include "vtd.h"
>> >  #include "extern.h"
>> >
>> > +static int __read_mostly iommu_qi_timeout_ms = 1;
>> > +integer_param("iommu_qi_timeout_ms", iommu_qi_timeout_ms);
>> 
>> But wait - this needs to be accompanied by an entry in
>> docs/misc/xen-command-line.markdown.
>> 
> 
> Jan, Is the following right?

No - you appear to add to the "iommu" option description. Either you
need to change the implementation to match that (which probably
would be a good idea, as that's what "iommu" is being used for in
other cases), or you need to add a new standalone entry with the
correct option name.

Jan

> --- a/docs/misc/xen-command-line.markdown
> +++ b/docs/misc/xen-command-line.markdown
> @@ -901,6 +901,14 @@ debug hypervisor only).
> 
>  >> Control the use of Queued Invalidation.
> 
> +> `qi_timeout_ms` (VT-d)
> +
> +> Default: `1`
> +
> +> `= <integer>`
> +
> +>> Specify the timeout of the VT-d Queued Invalidation in ms.
> +
>  > `snoop` (Intel)
> 
>  > Default: `true`
> 
> 
> -Quan
Quan Xu Jan. 14, 2016, 10:29 a.m. UTC | #2
On 14.01.2016 at 5:04pm, <JBeulich@suse.com> wrote:
> >>> On 14.01.16 at 02:59, <quan.xu@intel.com> wrote:
> > On 14.01.2016 at 12:58am, <JBeulich@suse.com> wrote:
> >> >>> On 23.12.15 at 09:25, <quan.xu@intel.com> wrote:
> >> > --- a/xen/drivers/passthrough/vtd/qinval.c
> >> > +++ b/xen/drivers/passthrough/vtd/qinval.c
> >> > @@ -28,6 +28,11 @@
> >> >  #include "vtd.h"
> >> >  #include "extern.h"
> >> >
> >> > +static int __read_mostly iommu_qi_timeout_ms = 1;
> >> > +integer_param("iommu_qi_timeout_ms", iommu_qi_timeout_ms);
> >>
> >> But wait - this needs to be accompanied by an entry in
> >> docs/misc/xen-command-line.markdown.
> >>
> >
> > Jan, Is the following right?
> 
> No - you appear to add to the "iommu" option description. Either you need to
> change the implementation to match that (which probably would be a good
> idea, as that's what "iommu" is being used for in other cases), or you need to
> add a new standalone entry with the correct option name.
> 

I prefer to add a new standalone entry.
Maybe I can modify the option name from "iommu_qi_timeout_ms" to "qi_timeout_ms". 


-Quan



> 
> > --- a/docs/misc/xen-command-line.markdown
> > +++ b/docs/misc/xen-command-line.markdown
> > @@ -901,6 +901,14 @@ debug hypervisor only).
> >
> >  >> Control the use of Queued Invalidation.
> >
> > +> `qi_timeout_ms` (VT-d)
> > +
> > +> Default: `1`
> > +
> > +> `= <integer>`
> > +
> > +>> Specify the timeout of the VT-d Queued Invalidation in ms.
> > +
> >  > `snoop` (Intel)
> >
> >  > Default: `true`
> >
> >
> > -Quan
> 
>
Jan Beulich Jan. 14, 2016, 12:17 p.m. UTC | #3
>>> On 14.01.16 at 11:29, <quan.xu@intel.com> wrote:
> On 14.01.2016 at 5:04pm, <JBeulich@suse.com> wrote:
>> >>> On 14.01.16 at 02:59, <quan.xu@intel.com> wrote:
>> > On 14.01.2016 at 12:58am, <JBeulich@suse.com> wrote:
>> >> >>> On 23.12.15 at 09:25, <quan.xu@intel.com> wrote:
>> >> > --- a/xen/drivers/passthrough/vtd/qinval.c
>> >> > +++ b/xen/drivers/passthrough/vtd/qinval.c
>> >> > @@ -28,6 +28,11 @@
>> >> >  #include "vtd.h"
>> >> >  #include "extern.h"
>> >> >
>> >> > +static int __read_mostly iommu_qi_timeout_ms = 1;
>> >> > +integer_param("iommu_qi_timeout_ms", iommu_qi_timeout_ms);
>> >>
>> >> But wait - this needs to be accompanied by an entry in
>> >> docs/misc/xen-command-line.markdown.
>> >>
>> >
>> > Jan, Is the following right?
>> 
>> No - you appear to add to the "iommu" option description. Either you need to
>> change the implementation to match that (which probably would be a good
>> idea, as that's what "iommu" is being used for in other cases), or you need 
> to
>> add a new standalone entry with the correct option name.
>> 
> 
> I prefer to add a new standalone entry.
> Maybe I can modify the option name from "iommu_qi_timeout_ms" to 
> "qi_timeout_ms". 

vtd-qi-timeout-ms might be okay, but qi as the prefix is bad. The
trailing ms may also not be needed.

Jan
diff mbox

Patch

--- a/docs/misc/xen-command-line.markdown
+++ b/docs/misc/xen-command-line.markdown
@@ -901,6 +901,14 @@  debug hypervisor only).

 >> Control the use of Queued Invalidation.

+> `qi_timeout_ms` (VT-d)
+
+> Default: `1`
+
+> `= <integer>`
+
+>> Specify the timeout of the VT-d Queued Invalidation in ms.
+
 > `snoop` (Intel)

 > Default: `true`