Cui, Dexuan
2008-Sep-28 06:52 UTC
[Xen-devel] [PATCH] Share the IO_APIC_route_entry with iosapic
The patch moves the struct IO_APIC_route_entry to a common place. This allows us to share the struct with iosapic. Thanks, -- Dexuan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Keir Fraser
2008-Sep-28 09:36 UTC
[Xen-devel] Re: [PATCH] Share the IO_APIC_route_entry with iosapic
If the code using this definition will be shared too, that''s one thing. If you''re actually going to put all the code under arch/ia64 (which you seem to imply), then duplicating this structure in include/asm-ia64 is okay. It''s not like it''s changing, ever. I personally think that striving to share architectural definitions in all cases, even where that may twist the header #includes, is not necessarily a good idea. So it depends what your end goal is. -- Keir On 28/9/08 07:52, "Cui, Dexuan" <dexuan.cui@intel.com> wrote:> The patch moves the struct IO_APIC_route_entry to a common place. > This allows us to share the struct with iosapic. > > Thanks, > -- Dexuan > > >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Cui, Dexuan
2008-Oct-07 07:24 UTC
[Xen-devel] RE: [PATCH] Share the IO_APIC_route_entry with iosapic
Hi Keir, The iosapic continues to stay inside arch/ia64/; on the other hand, I think the struct 'IO_APIC_route_entry' should be placed to a common place so that IPF and x86 can share most of the common codes, like interrupt remapping. To make the most use most use of the current x86 VT-d code, I personally think this movement of thedefinition of the struct IO_APIC_route_entry is necessary here. Could you please comment this more? Thanks, -- Dexuan -----Original Message----- From: Keir Fraser [mailto:keir.fraser@eu.citrix.com] Sent: 2008年9月28日 17:36 To: Cui, Dexuan; 'xen-devel@lists.xensource.com' Subject: Re: [PATCH] Share the IO_APIC_route_entry with iosapic If the code using this definition will be shared too, that's one thing. If you're actually going to put all the code under arch/ia64 (which you seem to imply), then duplicating this structure in include/asm-ia64 is okay. It's not like it's changing, ever. I personally think that striving to share architectural definitions in all cases, even where that may twist the header #includes, is not necessarily a good idea. So it depends what your end goal is. -- Keir On 28/9/08 07:52, "Cui, Dexuan" <dexuan.cui@intel.com> wrote:> The patch moves the struct IO_APIC_route_entry to a common place. > This allows us to share the struct with iosapic. > > Thanks, > -- Dexuan > > >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Keir Fraser
2008-Oct-07 07:34 UTC
[Xen-devel] Re: [PATCH] Share the IO_APIC_route_entry with iosapic
On 7/10/08 08:24, "Cui, Dexuan" <dexuan.cui@intel.com> wrote:> Hi Keir, > The iosapic continues to stay inside arch/ia64/; on the other hand, I think > the struct ''IO_APIC_route_entry'' should be placed to a common place so that > IPF and x86 can share most of the common codes, like interrupt remapping. > To make the most use most use of the current x86 VT-d code, I personally think > this movement of thedefinition of the struct IO_APIC_route_entry is necessary > here. > Could you please comment this more?Okay. Please move all arch-neutral definitions to xen/io_apic.h though -- that will at a minimum be all IO_APIC_reg_xx structures. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Cui, Dexuan
2008-Oct-07 07:49 UTC
[Xen-devel] RE: [PATCH] Share the IO_APIC_route_entry with iosapic
The IO_APIC_reg_xx structures are not used in IPF side. Let me double check if we also need to move them. Thanks, -- Dexuan -----Original Message----- From: Keir Fraser [mailto:keir.fraser@eu.citrix.com] Sent: 2008年10月7日 15:34 To: Cui, Dexuan; 'xen-devel@lists.xensource.com'; Xu, Anthony Subject: Re: [PATCH] Share the IO_APIC_route_entry with iosapic On 7/10/08 08:24, "Cui, Dexuan" <dexuan.cui@intel.com> wrote:> Hi Keir, > The iosapic continues to stay inside arch/ia64/; on the other hand, I think > the struct 'IO_APIC_route_entry' should be placed to a common place so that > IPF and x86 can share most of the common codes, like interrupt remapping. > To make the most use of the current x86 VT-d codes, I personally think > this movement of the definition of the struct IO_APIC_route_entry is necessary > here. > Could you please comment this more?Okay. Please move all arch-neutral definitions to xen/io_apic.h though -- that will at a minimum be all IO_APIC_reg_xx structures. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Keir Fraser
2008-Oct-07 08:07 UTC
[Xen-devel] Re: [PATCH] Share the IO_APIC_route_entry with iosapic
If you''re making an arch-neutral io-apic header, you should move them. If you just want a convenient stash for vt-d struct definitions, duplicate the route_entry structure in a header under drivers/passthrough. -- Keir On 7/10/08 08:49, "Cui, Dexuan" <dexuan.cui@intel.com> wrote:> The IO_APIC_reg_xx structures are not used in IPF side. > Let me double check if we also need to move them. > > Thanks, > -- Dexuan > > > -----Original Message----- > From: Keir Fraser [mailto:keir.fraser@eu.citrix.com] > Sent: 2008年10月7日 15:34 > To: Cui, Dexuan; ''xen-devel@lists.xensource.com''; Xu, Anthony > Subject: Re: [PATCH] Share the IO_APIC_route_entry with iosapic > > On 7/10/08 08:24, "Cui, Dexuan" <dexuan.cui@intel.com> wrote: > >> Hi Keir, >> The iosapic continues to stay inside arch/ia64/; on the other hand, I think >> the struct ''IO_APIC_route_entry'' should be placed to a common place so that >> IPF and x86 can share most of the common codes, like interrupt remapping. >> To make the most use of the current x86 VT-d codes, I personally think >> this movement of the definition of the struct IO_APIC_route_entry is >> necessary >> here. >> Could you please comment this more? > > Okay. Please move all arch-neutral definitions to xen/io_apic.h though -- > that will at a minimum be all IO_APIC_reg_xx structures. > > -- Keir > >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Cui, Dexuan
2008-Oct-09 08:27 UTC
[Xen-devel] RE: [PATCH] Share the IO_APIC_route_entry with iosapic
Hi Keir, I think the idea "making an arch-neutral io-apic header" may need too many lines of code. I tried the idea "duplicating the IO_APIC_route_entry structure in a header under drivers/passthrough" and find it's not ok: xen/drivers/passthrough/vtd/intremap.c includes xen/sched.h which has the IO_APIC_route_entry defined in include/asm-x86/io_apic.h. To use the newly-duplicated definition in intremap.c, I have to remove the inclusion of xen/sched.h and I find doing this causes many header files compilation issue which seems not easy to resolve. Another drawback of this idea is: we have 2 different definitions for the same struct name -- this can cause great confusion, I think. A better idea may be: we duplicate the IO_APIC_route_entry structure in a header under drivers/passthrough/ and rename it to IO_xAPIC_route_entry -- this new name will only be used by files in drivers/passthrough/. This requires the least changes and avoids confusion. How do you like this? Thanks, -- Dexuan -----Original Message----- From: xen-devel-bounces@lists.xensource.com [mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of Keir Fraser Sent: 2008年10月7日 16:07 To: Cui, Dexuan; 'xen-devel@lists.xensource.com'; Xu, Anthony Subject: [Xen-devel] Re: [PATCH] Share the IO_APIC_route_entry with iosapic If you're making an arch-neutral io-apic header, you should move them. If you just want a convenient stash for vt-d struct definitions, duplicate the route_entry structure in a header under drivers/passthrough. -- Keir On 7/10/08 08:49, "Cui, Dexuan" <dexuan.cui@intel.com> wrote:> The IO_APIC_reg_xx structures are not used in IPF side. > Let me double check if we also need to move them. > > Thanks, > -- Dexuan > > > -----Original Message----- > From: Keir Fraser [mailto:keir.fraser@eu.citrix.com] > Sent: 2008年10月7日 15:34 > To: Cui, Dexuan; 'xen-devel@lists.xensource.com'; Xu, Anthony > Subject: Re: [PATCH] Share the IO_APIC_route_entry with iosapic > > On 7/10/08 08:24, "Cui, Dexuan" <dexuan.cui@intel.com> wrote: > >> Hi Keir, >> The iosapic continues to stay inside arch/ia64/; on the other hand, I think >> the struct 'IO_APIC_route_entry' should be placed to a common place so that >> IPF and x86 can share most of the common codes, like interrupt remapping. >> To make the most use of the current x86 VT-d codes, I personally think >> this movement of the definition of the struct IO_APIC_route_entry is >> necessary >> here. >> Could you please comment this more? > > Okay. Please move all arch-neutral definitions to xen/io_apic.h though -- > that will at a minimum be all IO_APIC_reg_xx structures. > > -- Keir > >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Keir Fraser
2008-Oct-09 09:00 UTC
Re: [Xen-devel] RE: [PATCH] Share the IO_APIC_route_entry with iosapic
On 9/10/08 09:27, "Cui, Dexuan" <dexuan.cui@intel.com> wrote:> A better idea may be: > we duplicate the IO_APIC_route_entry structure in a header under > drivers/passthrough/ and rename it to IO_xAPIC_route_entry -- this new name > will only be used by files in drivers/passthrough/. > This requires the least changes and avoids confusion. > How do you like this?Sounds okay. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Maybe Matching Threads
- [PATCH] vtd: Define a struct IO_xAPIC_route_entry to accommodate both ioapic and iosapic
- A race condition introduced by changeset 15175: Re-init hypercall stubs page after HVM save/restore
- Weekly VMX status report. Xen: #18560 & Xen0: #682
- Anyone succeeds HVM on latest x86-64 xen
- Can we disable secondary_bus_reset in runtime?