diff mbox series

powerpc/pseries/vas: Pass hw_cpu_id to node associativity HCALL

Message ID 55380253ea0c11341824cd4c0fc6bbcfc5752689.camel@linux.ibm.com (mailing list archive)
State Accepted
Headers show
Series powerpc/pseries/vas: Pass hw_cpu_id to node associativity HCALL | expand

Checks

Context Check Description
snowpatch_ozlabs/github-powerpc_selftests success Successfully ran 10 jobs.
snowpatch_ozlabs/github-powerpc_ppctests success Successfully ran 10 jobs.
snowpatch_ozlabs/github-powerpc_sparse success Successfully ran 4 jobs.
snowpatch_ozlabs/github-powerpc_clang success Successfully ran 6 jobs.
snowpatch_ozlabs/github-powerpc_kernel_qemu success Successfully ran 23 jobs.

Commit Message

Haren Myneni Sept. 29, 2022, 1:57 a.m. UTC
Generally the hypervisor decides to allocate a window on different
VAS instances. But if the user space wishes to allocate on the
current VAS instance where the process is executing, the kernel has
to pass associativity domain IDs to allocate VAS window HCALL. To
determine the associativity domain IDs for the current CPU, passing
smp_processor_id() to node associativity HCALL which may return
H_P2 (-55) error during DLPAR CPU event.

This patch fixes this issue by passing hard_smp_processor_id() with
VPHN_FLAG_VCPU flag (PAPR 14.11.6.1 H_HOME_NODE_ASSOCIATIVITY).

Signed-off-by: Haren Myneni <haren@linux.ibm.com>
---
 arch/powerpc/platforms/pseries/vas.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Nathan Lynch Sept. 29, 2022, 10:16 p.m. UTC | #1
Haren Myneni <haren@linux.ibm.com> writes:
> Generally the hypervisor decides to allocate a window on different
> VAS instances. But if the user space wishes to allocate on the
> current VAS instance where the process is executing, the kernel has
> to pass associativity domain IDs to allocate VAS window HCALL. To
> determine the associativity domain IDs for the current CPU, passing
> smp_processor_id() to node associativity HCALL which may return
> H_P2 (-55) error during DLPAR CPU event.
>
> This patch fixes this issue by passing hard_smp_processor_id() with
> VPHN_FLAG_VCPU flag (PAPR 14.11.6.1 H_HOME_NODE_ASSOCIATIVITY).
>
> Signed-off-by: Haren Myneni <haren@linux.ibm.com>
> ---
>  arch/powerpc/platforms/pseries/vas.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/powerpc/platforms/pseries/vas.c b/arch/powerpc/platforms/pseries/vas.c
> index fe33bdb620d5..533026fd1f40 100644
> --- a/arch/powerpc/platforms/pseries/vas.c
> +++ b/arch/powerpc/platforms/pseries/vas.c
> @@ -348,7 +348,7 @@ static struct vas_window *vas_allocate_window(int vas_id, u64 flags,
>  		 * So no unpacking needs to be done.
>  		 */
>  		rc = plpar_hcall9(H_HOME_NODE_ASSOCIATIVITY, domain,
> -				  VPHN_FLAG_VCPU, smp_processor_id());
> +				  VPHN_FLAG_VCPU, hard_smp_processor_id());
>  		if (rc != H_SUCCESS) {
>  			pr_err("H_HOME_NODE_ASSOCIATIVITY error: %d\n", rc);
>  			goto out;

Yes, it is always wrong to pass Linux CPU numbers to the hypervisor,
which has its own numbering for hardware threads. It usually coincides
with Linux's numbering in practice, which tends to hide bugs like this.

Reviewed-by: Nathan Lynch <nathanl@linux.ibm.com>
Michal Suchanek Sept. 30, 2022, 8:55 a.m. UTC | #2
Hello,

On Thu, Sep 29, 2022 at 05:16:40PM -0500, Nathan Lynch wrote:
> Haren Myneni <haren@linux.ibm.com> writes:
> > Generally the hypervisor decides to allocate a window on different
> > VAS instances. But if the user space wishes to allocate on the
> > current VAS instance where the process is executing, the kernel has
> > to pass associativity domain IDs to allocate VAS window HCALL. To
> > determine the associativity domain IDs for the current CPU, passing
> > smp_processor_id() to node associativity HCALL which may return
> > H_P2 (-55) error during DLPAR CPU event.
> >
> > This patch fixes this issue by passing hard_smp_processor_id() with
> > VPHN_FLAG_VCPU flag (PAPR 14.11.6.1 H_HOME_NODE_ASSOCIATIVITY).
> >
> > Signed-off-by: Haren Myneni <haren@linux.ibm.com>
> > ---
> >  arch/powerpc/platforms/pseries/vas.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/arch/powerpc/platforms/pseries/vas.c b/arch/powerpc/platforms/pseries/vas.c
> > index fe33bdb620d5..533026fd1f40 100644
> > --- a/arch/powerpc/platforms/pseries/vas.c
> > +++ b/arch/powerpc/platforms/pseries/vas.c
> > @@ -348,7 +348,7 @@ static struct vas_window *vas_allocate_window(int vas_id, u64 flags,
> >  		 * So no unpacking needs to be done.
> >  		 */
> >  		rc = plpar_hcall9(H_HOME_NODE_ASSOCIATIVITY, domain,
> > -				  VPHN_FLAG_VCPU, smp_processor_id());
> > +				  VPHN_FLAG_VCPU, hard_smp_processor_id());
> >  		if (rc != H_SUCCESS) {
> >  			pr_err("H_HOME_NODE_ASSOCIATIVITY error: %d\n", rc);
> >  			goto out;
> 
> Yes, it is always wrong to pass Linux CPU numbers to the hypervisor,
> which has its own numbering for hardware threads. It usually coincides
> with Linux's numbering in practice, which tends to hide bugs like this.
> 
> Reviewed-by: Nathan Lynch <nathanl@linux.ibm.com>

This is the code that introduces the problem, right?

Fixes: b22f2d88e435 ("powerpc/pseries/vas: Integrate API with open/close windows")

Thanks

Michal
Michael Ellerman Sept. 30, 2022, 10:35 a.m. UTC | #3
Michal Suchánek <msuchanek@suse.de> writes:
> On Thu, Sep 29, 2022 at 05:16:40PM -0500, Nathan Lynch wrote:
>> Haren Myneni <haren@linux.ibm.com> writes:
>> > Generally the hypervisor decides to allocate a window on different
>> > VAS instances. But if the user space wishes to allocate on the
>> > current VAS instance where the process is executing, the kernel has
>> > to pass associativity domain IDs to allocate VAS window HCALL. To
>> > determine the associativity domain IDs for the current CPU, passing
>> > smp_processor_id() to node associativity HCALL which may return
>> > H_P2 (-55) error during DLPAR CPU event.
>> >
>> > This patch fixes this issue by passing hard_smp_processor_id() with
>> > VPHN_FLAG_VCPU flag (PAPR 14.11.6.1 H_HOME_NODE_ASSOCIATIVITY).
>> >
>> > Signed-off-by: Haren Myneni <haren@linux.ibm.com>
>> > ---
>> >  arch/powerpc/platforms/pseries/vas.c | 2 +-
>> >  1 file changed, 1 insertion(+), 1 deletion(-)
>> >
>> > diff --git a/arch/powerpc/platforms/pseries/vas.c b/arch/powerpc/platforms/pseries/vas.c
>> > index fe33bdb620d5..533026fd1f40 100644
>> > --- a/arch/powerpc/platforms/pseries/vas.c
>> > +++ b/arch/powerpc/platforms/pseries/vas.c
>> > @@ -348,7 +348,7 @@ static struct vas_window *vas_allocate_window(int vas_id, u64 flags,
>> >  		 * So no unpacking needs to be done.
>> >  		 */
>> >  		rc = plpar_hcall9(H_HOME_NODE_ASSOCIATIVITY, domain,
>> > -				  VPHN_FLAG_VCPU, smp_processor_id());
>> > +				  VPHN_FLAG_VCPU, hard_smp_processor_id());
>> >  		if (rc != H_SUCCESS) {
>> >  			pr_err("H_HOME_NODE_ASSOCIATIVITY error: %d\n", rc);
>> >  			goto out;
>> 
>> Yes, it is always wrong to pass Linux CPU numbers to the hypervisor,
>> which has its own numbering for hardware threads. It usually coincides
>> with Linux's numbering in practice, which tends to hide bugs like this.
>> 
>> Reviewed-by: Nathan Lynch <nathanl@linux.ibm.com>
>
> This is the code that introduces the problem, right?
>
> Fixes: b22f2d88e435 ("powerpc/pseries/vas: Integrate API with open/close windows")

Yeah, I tagged it when applying. Thanks.

cheers
Michael Ellerman Oct. 4, 2022, 1:26 p.m. UTC | #4
On Wed, 28 Sep 2022 18:57:33 -0700, Haren Myneni wrote:
> Generally the hypervisor decides to allocate a window on different
> VAS instances. But if the user space wishes to allocate on the
> current VAS instance where the process is executing, the kernel has
> to pass associativity domain IDs to allocate VAS window HCALL. To
> determine the associativity domain IDs for the current CPU, passing
> smp_processor_id() to node associativity HCALL which may return
> H_P2 (-55) error during DLPAR CPU event.
> 
> [...]

Applied to powerpc/next.

[1/1] powerpc/pseries/vas: Pass hw_cpu_id to node associativity HCALL
      https://git.kernel.org/powerpc/c/f3e5d9e53e74d77e711a2c90a91a8b0836a9e0b3

cheers
diff mbox series

Patch

diff --git a/arch/powerpc/platforms/pseries/vas.c b/arch/powerpc/platforms/pseries/vas.c
index fe33bdb620d5..533026fd1f40 100644
--- a/arch/powerpc/platforms/pseries/vas.c
+++ b/arch/powerpc/platforms/pseries/vas.c
@@ -348,7 +348,7 @@  static struct vas_window *vas_allocate_window(int vas_id, u64 flags,
 		 * So no unpacking needs to be done.
 		 */
 		rc = plpar_hcall9(H_HOME_NODE_ASSOCIATIVITY, domain,
-				  VPHN_FLAG_VCPU, smp_processor_id());
+				  VPHN_FLAG_VCPU, hard_smp_processor_id());
 		if (rc != H_SUCCESS) {
 			pr_err("H_HOME_NODE_ASSOCIATIVITY error: %d\n", rc);
 			goto out;