IO_APIC broken by 45edfd1db02f818b3dc7e4743ee8585af6b78f78

IO_APIC broken by 45edfd1db02f818b3dc7e4743ee8585af6b78f78

[Tuncer Ayaz]

I've bisected the non-working'ness of HD-Audio and USB Mouse on one of
my x86_64 boxes back to the following commit.

The machine is an HP xw4400 Core 2 Duo E6600 with the Intel 975X chipset.
Please let me know if you need any debug info.

45edfd1db02f818b3dc7e4743ee8585af6b78f78 is first bad commit
commit 45edfd1db02f818b3dc7e4743ee8585af6b78f78
Author: Yinghai Lu <yinghai.lu@amd.com>
Date:   Sat Oct 21 18:37:01 2006 +0200

    [PATCH] x86-64: typo in __assign_irq_vector when updating pos for
vector and offset

    typo with cpu instead of new_cpu

    Signed-off-by: Yinghai Lu <yinghai.lu@amd.com>
    Signed-off-by: Andi Kleen <ak@suse.de>

:040000 040000 1d64801d89bfb23ef4d63d1625f47122d01ded6c
03bb862d9bfdfb05fb5382d56d52ffe1c5f8aba2 M	arch

Accordingly ther revert-diff is this
diff --git a/arch/x86_64/kernel/io_apic.c b/arch/x86_64/kernel/io_apic.c
index b000017..1fa95d5 100644
--- a/arch/x86_64/kernel/io_apic.c
+++ b/arch/x86_64/kernel/io_apic.c
@@ -651,12 +651,12 @@ next:
                if (vector == IA32_SYSCALL_VECTOR)
                        goto next;
                for_each_cpu_mask(new_cpu, domain)
-                       if (per_cpu(vector_irq, new_cpu)[vector] != -1)
+                       if (per_cpu(vector_irq, cpu)[vector] != -1)
                                goto next;
                /* Found one! */
                for_each_cpu_mask(new_cpu, domain) {
-                       pos[new_cpu].vector = vector;
-                       pos[new_cpu].offset = offset;
+                       pos[cpu].vector = vector;
+                       pos[cpu].offset = offset;
                }
                if (old_vector >= 0) {
                        int old_cpu;

Re: IO_APIC broken by 45edfd1db02f818b3dc7e4743ee8585af6b78f78

[Muli Ben-Yehuda]

On Wed, Oct 25, 2006 at 02:37:57PM +0200, Tuncer Ayaz wrote:
> I've bisected the non-working'ness of HD-Audio and USB Mouse on one of
> my x86_64 boxes back to the following commit.
> 
> The machine is an HP xw4400 Core 2 Duo E6600 with the Intel 975X chipset.
> Please let me know if you need any debug info.

These two patches should fix it:

http://marc.theaimsgroup.com/?l=linux-kernel&m=116157813623508&w=2
http://marc.theaimsgroup.com/?l=linux-kernel&m=116157837104613&w=2

Cheers,
Muli

Re: IO_APIC broken by 45edfd1db02f818b3dc7e4743ee8585af6b78f78

[Tuncer Ayaz]

On 10/25/06, Muli Ben-Yehuda <muli@il.ibm.com> wrote:
> On Wed, Oct 25, 2006 at 02:37:57PM +0200, Tuncer Ayaz wrote:
> > I've bisected the non-working'ness of HD-Audio and USB Mouse on one of
> > my x86_64 boxes back to the following commit.
> >
> > The machine is an HP xw4400 Core 2 Duo E6600 with the Intel 975X chipset.
> > Please let me know if you need any debug info.
>
> These two patches should fix it:
>
> http://marc.theaimsgroup.com/?l=linux-kernel&m=116157813623508&w=2
> http://marc.theaimsgroup.com/?l=linux-kernel&m=116157837104613&w=2

Thanks, a clean v2.6.19-rc3 plus those two patches solves the issues.
I assume that the fixes will be pulled into Linus' tree before 2.6.19
is released.

Re: IO_APIC broken by 45edfd1db02f818b3dc7e4743ee8585af6b78f78

[Tuncer Ayaz]

On 10/25/06, Tuncer Ayaz <tuncer.ayaz@gmail.com> wrote:
> On 10/25/06, Muli Ben-Yehuda <muli@il.ibm.com> wrote:
> > On Wed, Oct 25, 2006 at 02:37:57PM +0200, Tuncer Ayaz wrote:
> > > I've bisected the non-working'ness of HD-Audio and USB Mouse on one of
> > > my x86_64 boxes back to the following commit.
> > >
> > > The machine is an HP xw4400 Core 2 Duo E6600 with the Intel 975X chipset.
> > > Please let me know if you need any debug info.
> >
> > These two patches should fix it:
> >
> > http://marc.theaimsgroup.com/?l=linux-kernel&m=116157813623508&w=2
> > http://marc.theaimsgroup.com/?l=linux-kernel&m=116157837104613&w=2
>
> Thanks, a clean v2.6.19-rc3 plus those two patches solves the issues.
> I assume that the fixes will be pulled into Linus' tree before 2.6.19
> is released.

Linus' tree as of this morning CEST time has the two patches included and
everything's fine. Case closed.