[PATCH v2] powerpc/xive: Fix bogus error code returned by OPAL

There's a bug in skiboot that causes the OPAL_XIVE_ALLOCATE_IRQ call
to return the 32-bit value 0xffffffff when OPAL has run out of IRQs.
Unfortunatelty, OPAL return values are signed 64-bit entities and
errors are supposed to be negative. If that happens, the linux code
confusingly treats 0xffffffff as a valid IRQ number and panics at some
point.

A fix was recently merged in skiboot:

e97391ae2bb5 ("xive: fix return value of opal_xive_allocate_irq()")

but we need a workaround anyway to support older skiboots already
in the field.

Internally convert 0xffffffff to OPAL_RESOURCE which is the usual error
returned upon resource exhaustion.

Cc: stable@vger.kernel.org # v4.12+
Signed-off-by: Greg Kurz <redacted>
---
v2: - fix syntax error in changelog
    - Cc stable
    - rename original OPAL wrapper
    - rewrite fixup wrapper (style, use s64 and u32)
---
 arch/powerpc/include/asm/opal.h            |    2 +-
 arch/powerpc/platforms/powernv/opal-call.c |    2 +-
 arch/powerpc/sysdev/xive/native.c          |   11 +++++++++++
 3 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/include/asm/opal.h b/arch/powerpc/include/asm/opal.h
index 57bd029c715e..d5a0807d21db 100644
--- a/arch/powerpc/include/asm/opal.h
+++ b/arch/powerpc/include/asm/opal.h

@@ -272,7 +272,7 @@ int64_t opal_xive_get_vp_info(uint64_t vp,
 int64_t opal_xive_set_vp_info(uint64_t vp,
 			      uint64_t flags,
 			      uint64_t report_cl_pair);
-int64_t opal_xive_allocate_irq(uint32_t chip_id);
+int64_t opal_xive_allocate_irq_raw(uint32_t chip_id);
 int64_t opal_xive_free_irq(uint32_t girq);
 int64_t opal_xive_sync(uint32_t type, uint32_t id);
 int64_t opal_xive_dump(uint32_t type, uint32_t id);

diff --git a/arch/powerpc/platforms/powernv/opal-call.c b/arch/powerpc/platforms/powernv/opal-call.c
index 29ca523c1c79..dccdc9df5213 100644
--- a/arch/powerpc/platforms/powernv/opal-call.c
+++ b/arch/powerpc/platforms/powernv/opal-call.c

@@ -257,7 +257,7 @@ OPAL_CALL(opal_xive_set_queue_info,		OPAL_XIVE_SET_QUEUE_INFO);
 OPAL_CALL(opal_xive_donate_page,		OPAL_XIVE_DONATE_PAGE);
 OPAL_CALL(opal_xive_alloc_vp_block,		OPAL_XIVE_ALLOCATE_VP_BLOCK);
 OPAL_CALL(opal_xive_free_vp_block,		OPAL_XIVE_FREE_VP_BLOCK);
-OPAL_CALL(opal_xive_allocate_irq,		OPAL_XIVE_ALLOCATE_IRQ);
+OPAL_CALL(opal_xive_allocate_irq_raw,		OPAL_XIVE_ALLOCATE_IRQ);
 OPAL_CALL(opal_xive_free_irq,			OPAL_XIVE_FREE_IRQ);
 OPAL_CALL(opal_xive_get_vp_info,		OPAL_XIVE_GET_VP_INFO);
 OPAL_CALL(opal_xive_set_vp_info,		OPAL_XIVE_SET_VP_INFO);

diff --git a/arch/powerpc/sysdev/xive/native.c b/arch/powerpc/sysdev/xive/native.c
index 37987c815913..ad8ee7dd7f57 100644
--- a/arch/powerpc/sysdev/xive/native.c
+++ b/arch/powerpc/sysdev/xive/native.c

@@ -231,6 +231,17 @@ static bool xive_native_match(struct device_node *node)
 	return of_device_is_compatible(node, "ibm,opal-xive-vc");
 }
 
+static s64 opal_xive_allocate_irq(u32 chip_id)
+{
+	s64 irq = opal_xive_allocate_irq_raw(chip_id);
+
+	/*
+	 * Old versions of skiboot can incorrectly return 0xffffffff to
+	 * indicate no space, fix it up here.
+	 */
+	return irq == 0xffffffff ? OPAL_RESOURCE : irq;
+}
+
 #ifdef CONFIG_SMP
 static int xive_native_get_ipi(unsigned int cpu, struct xive_cpu *xc)
 {

On 11/09/2019 17:52, Greg Kurz wrote:

There's a bug in skiboot that causes the OPAL_XIVE_ALLOCATE_IRQ call
to return the 32-bit value 0xffffffff when OPAL has run out of IRQs.
Unfortunatelty, OPAL return values are signed 64-bit entities and
errors are supposed to be negative. If that happens, the linux code
confusingly treats 0xffffffff as a valid IRQ number and panics at some
point.

A fix was recently merged in skiboot:

e97391ae2bb5 ("xive: fix return value of opal_xive_allocate_irq()")

but we need a workaround anyway to support older skiboots already
in the field.

Internally convert 0xffffffff to OPAL_RESOURCE which is the usual error
returned upon resource exhaustion.

Cc: stable@vger.kernel.org # v4.12+
Signed-off-by: Greg Kurz <redacted>

Reviewed-by: Cédric Le Goater <clg@kaod.org>

Thanks,

C.

On Wed, 2019-09-11 at 15:52:18 UTC, Greg Kurz wrote:

There's a bug in skiboot that causes the OPAL_XIVE_ALLOCATE_IRQ call
to return the 32-bit value 0xffffffff when OPAL has run out of IRQs.
Unfortunatelty, OPAL return values are signed 64-bit entities and
errors are supposed to be negative. If that happens, the linux code
confusingly treats 0xffffffff as a valid IRQ number and panics at some
point.

A fix was recently merged in skiboot:

e97391ae2bb5 ("xive: fix return value of opal_xive_allocate_irq()")

but we need a workaround anyway to support older skiboots already
in the field.

Internally convert 0xffffffff to OPAL_RESOURCE which is the usual error
returned upon resource exhaustion.

Cc: stable@vger.kernel.org # v4.12+
Signed-off-by: Greg Kurz <redacted>

Applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/6ccb4ac2bf8a35c694ead92f8ac5530a16e8f2c8

cheers