Hi, Keir, Attached patch renames init_IRQ to xen_init_IRQ, because ia64 version init_IRQ has some other special things to do. Instead of messing arch specific logic into one common init_IRQ, this rename gives different arch possibility to call xen evtchn specific initialization at right time. In the start, I moved init_IRQ to x86 io_apic.c which however doesn''t work if user compiles xen0/xenU by different configuration file because xenU doesn''t compile io_apic by default. I''m not sure whether worthy of changing domU config file and then simply keep it within evtchn.c After this small change, now the whole logic of drivers/xen/core/evtchn.c can be reused by other architectures. I''ve managed to make xen/ia64 working on top of this logic, and related ia64 side changes will be sent out later separately. Signed-off-by Kevin Tian <Kevin.tian@intel.com> Thanks, Kevin _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 17 May 2006, at 05:10, Tian, Kevin wrote:> In the start, I moved init_IRQ to x86 io_apic.c which however > doesn''t > work if user compiles xen0/xenU by different configuration file because > xenU doesn''t compile io_apic by default. I''m not sure whether worthy of > changing domU config file and then simply keep it within evtchn.cHow about we declare it in evtchn.c always, but attribute((weak))? It''s small and is marked __init anyway, so no space concern really. And it''ll avoid putting yet another thing in CONFIG_X86? -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
>From: Keir Fraser [mailto:Keir.Fraser@cl.cam.ac.uk] >Sent: 2006年5月17日 15:02 >On 17 May 2006, at 05:10, Tian, Kevin wrote: > >> In the start, I moved init_IRQ to x86 io_apic.c which however >> doesn''t >> work if user compiles xen0/xenU by different configuration file because >> xenU doesn''t compile io_apic by default. I''m not sure whether worthy >of >> changing domU config file and then simply keep it within evtchn.c > >How about we declare it in evtchn.c always, but attribute((weak))? It''s >small and is marked __init anyway, so no space concern really. And >it''ll avoid putting yet another thing in CONFIG_X86? > > -- KeirThat may work, but I''m not sure whether that style (attribute(weak)) is welcomed in kernel world. Previously I just saw such instances in user level, like libc. :-) Especially, irq_ctx_init is also x86 specific and so it seems reasonable to put it into CONFIG_X86 and only leave real common stuff there... Thanks, Kevin _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 17 May 2006, at 08:55, Tian, Kevin wrote:> That may work, but I''m not sure whether that style (attribute(weak)) is > welcomed in kernel world. Previously I just saw such instances in user > level, like libc. :-)It''s used in various places in Linux. There''s no reason not to use it, if it''s appropriate.> Especially, irq_ctx_init is also x86 specific and so it seems > reasonable > to put it into CONFIG_X86 and only leave real common stuff there...Okay yes, not so appropriate here. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
>From: Keir Fraser [mailto:Keir.Fraser@cl.cam.ac.uk] >Sent: 2006年5月17日 16:11 >On 17 May 2006, at 08:55, Tian, Kevin wrote: > >> That may work, but I''m not sure whether that style (attribute(weak)) is >> welcomed in kernel world. Previously I just saw such instances in user >> level, like libc. :-) > >It''s used in various places in Linux. There''s no reason not to use it, >if it''s appropriate.Ah yes, I just note some in linux kernel source. Thanks, Kevin _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel