... Oh, i fear this patch makes some issues (e.g. changing NOZOMI_MAX_PORTS without adapting all its usages).. as this one (together with [RFC 7/13]) is quite big and i really don't wanna loose the multicard support (which i also cannot test yet here at ...

linux-kernel - Frank Seidel - Nov 10 2007 - 11:42

... ago) so that even the default case is a call. > Dprintk("Asserting INIT.\n"); > > /* > @@ -679,7 +683,8 @@ do_rest: > /* trampoline code not run */ > printk("Not responding.\n"); > #ifdef APIC_DEBUG > - inquire_remote_apic( ...

linux-kernel - Andi Kleen - Mar 25 2008 - 06:25

On Tue, 16 Jan 2007, Paul Menage wrote: > On 1/15/07, Christoph Lameter wrote: > > > > This solution may be a bit hokey. I tried other approaches but this > > one seemed to be the simplest with the least complications. Maybe

linux-kernel - Christoph Lameter - Jan 16 2007 - 16:00

On 1/16/07, Christoph Lameter wrote: > On Tue, 16 Jan 2007, Paul Menage wrote: > > > On 1/15/07, Christoph Lameter wrote: > > > > > > This solution may be a bit hokey. I tried other approaches but this > > > one

linux-kernel - Paul Menage - Jan 16 2007 - 16:06

On Tue, 16 Jan 2007, Paul Menage wrote: > I was thinking runtime, unless MAX_NUMNODES is less than 64 in which > case you can make the decision at compile time. > > > > > If done at compile time then we will end up with a pointer to an unsigned

linux-kernel - Christoph Lameter - Jan 16 2007 - 16:51

On 1/15/07, Christoph Lameter wrote: > > This solution may be a bit hokey. I tried other approaches but this > one seemed to be the simplest with the least complications. Maybe someone > else can come up with a better solution?

linux-kernel - Paul Menage - Jan 16 2007 - 15:52

* Jack Steiner wrote: > Index: linux/arch/x86/kernel/genapic_64.c > @@ -69,7 +73,16 @@ void send_IPI_self(int vector) > > unsigned int get_apic_id(void) > { > - return (apic_read(APIC_ID) >> 24) & 0xFFu; > + unsigned int

linux-kernel - Ingo Molnar - Mar 25 2008 - 10:30

On Tue, Mar 25, 2008 at 03:30:59PM +0100, Ingo Molnar wrote: > > * Jack Steiner wrote: > > > Index: linux/arch/x86/kernel/genapic_64.c > > > @@ -69,7 +73,16 @@ void send_IPI_self(int vector) > > > > unsigned int get_apic_id

linux-kernel - Jack Steiner - Mar 25 2008 - 12:31

On Tue, Mar 25, 2008 at 11:25:34AM +0100, Andi Kleen wrote: > Jack Steiner writes: > > > unsigned int get_apic_id(void) > > { > > - return (apic_read(APIC_ID) >> 24) & 0xFFu; > > + unsigned int id; > > + > > + preempt_disable

linux-kernel - Jack Steiner - Mar 25 2008 - 13:56

> > This should be probably factored properly (didn't Jeremy have smp_ops > > for this some time ago) so that even the default case is a call. > > By factored, do you means something like: > is_uv_legacy_system() > is_us_non_unique_apicid_system

linux-kernel - Andi Kleen - Mar 25 2008 - 14:06

Andi Kleen wrote: > No instead of having lots of if (xyz_system) do_xyz_special() > go through smp_ops for the whole thing so that UV would just have a > special smp_ops that has special implementions or wrappers. > > Oops I see smp_ops are

linux-kernel - Jeremy Fitzhardinge - Mar 25 2008 - 22:23

On Tue, Mar 25, 2008 at 11:23 PM, Jeremy Fitzhardinge wrote: > Andi Kleen wrote: > > No instead of having lots of if (xyz_system) do_xyz_special() > > go through smp_ops for the whole thing so that UV would just have a > > special

linux-kernel - Glauber Costa - Mar 25 2008 - 23:22

On Tue, Mar 25, 2008 at 1:31 PM, Jack Steiner wrote: > On Tue, Mar 25, 2008 at 03:30:59PM +0100, Ingo Molnar wrote: > > > > * Jack Steiner wrote: > > > > > Index: linux/arch/x86/kernel/genapic_64.c > > > >

linux-kernel - Glauber Costa - Mar 25 2008 - 23:24

* Jeremy Fitzhardinge wrote: > Andi Kleen wrote: >> No instead of having lots of if (xyz_system) do_xyz_special() >> go through smp_ops for the whole thing so that UV would just have a >> special smp_ops that has special

linux-kernel - Ingo Molnar - Mar 26 2008 - 03:29

* Jack Steiner wrote: > > > - obj-y += genapic_64.o genapic_flat_64.o > > > + obj-y += genapic_64.o genapic_flat_64.o genx2apic_uv_x.o > > > > Definitely should be a CONFIG > > Not sure that I understand

linux-kernel - Ingo Molnar - Mar 26 2008 - 03:38

On Wed, Mar 26, 2008 at 12:38 AM, Ingo Molnar wrote: > > * Jack Steiner wrote: > > > > > - obj-y += genapic_64.o genapic_flat_64.o > > > > + obj-y

linux-kernel - Yinghai Lu - Mar 30 2008 - 16:23

On Mon, Mar 24, 2008 at 11:21 AM, Jack Steiner wrote: > > UV supports really big systems. So big, in fact, that the APICID register > does not contain enough bits to contain an APICID that is unique across all > cpus. > > The

linux-kernel - Yinghai Lu - Mar 30 2008 - 16:41

On Sun, Mar 30, 2008 at 01:23:12PM -0700, Yinghai Lu wrote: > On Wed, Mar 26, 2008 at 12:38 AM, Ingo Molnar wrote: > > > > * Jack Steiner wrote: > > > > > > > - obj-y += genapic_64

linux-kernel - Jack Steiner - Mar 30 2008 - 17:03

> > unsigned int get_apic_id(void) > > { > > - return (apic_read(APIC_ID) >> 24) & 0xFFu; > > + unsigned int id; > > + > > + preempt_disable(); > > + id = apic_read(APIC_ID); > > + if (uv_system_type >=

linux-kernel - Jack Steiner - Mar 30 2008 - 17:08

> If there was a significant differece between UV and generic kernels > (or hardware), then I would agree. However, the only significant > difference is the APIC model on large systems. Small systems are > exactly compatible. > > The problem with

linux-kernel - Andi Kleen - Mar 30 2008 - 17:18