Commit cb1293e2f594 ("ARM: 8375/1: disable some options on ARMv7-M") causes
build failure on ARMv7-M machines:
CC arch/arm/kernel/asm-offsets.s
In file included from include/linux/sem.h:5:0,
from include/linux/sched.h:35,
from arch/arm/kernel/asm-offsets.c:14:
include/linux/rcupdate.h: In function 'rcu_read_lock_sched_held':
include/linux/rcupdate.h:539:2: error: implicit declaration of function 'arch_irqs_disabled' [-Werror=implicit-function-declaration]
return preempt_count() != 0 || irqs_disabled();
^
Commit cb1293e2f594 was initially introduced to fix build errors with randconfig,
as ARMv7-m selects TRACE_IRQFLAGS_SUPPORT but breaks build when TRACE_IRQFLAGS
is selected.
As there is no proper support yet for TRACE_IRQFLAGS, this patch partially
reverts commit cb1293e2f594, with the downside of build errors with randconfig
when TRACE_IRQFLAGS gets selected.
Cc: Arnd Bergmann <[email protected]>
Cc: Uwe Kleine-König <[email protected]>
Cc: Russell King <[email protected]>
Signed-off-by: Maxime Coquelin <[email protected]>
---
arch/arm/Kconfig | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index b47457d..e4e1694 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -171,7 +171,7 @@ config LOCKDEP_SUPPORT
config TRACE_IRQFLAGS_SUPPORT
bool
- default !CPU_V7M
+ default y
config RWSEM_XCHGADD_ALGORITHM
bool
--
1.9.1
On Tue, Jun 09, 2015 at 12:24:48AM +0200, Maxime Coquelin wrote:
> Commit cb1293e2f594 ("ARM: 8375/1: disable some options on ARMv7-M") causes
> build failure on ARMv7-M machines:
>
> CC arch/arm/kernel/asm-offsets.s
> In file included from include/linux/sem.h:5:0,
> from include/linux/sched.h:35,
> from arch/arm/kernel/asm-offsets.c:14:
> include/linux/rcupdate.h: In function 'rcu_read_lock_sched_held':
> include/linux/rcupdate.h:539:2: error: implicit declaration of function 'arch_irqs_disabled' [-Werror=implicit-function-declaration]
> return preempt_count() != 0 || irqs_disabled();
> ^
The real solution is to provide a definition _in asm-generic_ for
arch_irqs_disabled(), rather than having almost every arch doing:
static inline bool arch_irqs_disabled(void)
{
return arch_irqs_disabled_flags(arch_local_save_flags());
}
I'm personally refusing to take a patch for ARM which adds yet another
copy of the above. This is, after all, exactly the kind of stuff that
should be in asm-generic, or if not, in include/linux but overridable
by arch stuff.
We keep going between the two extremes of "lets push lots of stuff into
arch stuff" and "lets try to extract the common bits out of arch code".
Let's try and settle on one approach, and apply it universally.
In the mean time, I think the right answer is to drop Arnd's patch -
subsituting a randconfig build error for a useful-config build error
is not something we want to do - and even partially reverting the
patch results in randconfig build errors returning, so...
--
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.
2015-06-09 0:47 GMT+02:00 Russell King - ARM Linux <[email protected]>:
> On Tue, Jun 09, 2015 at 12:24:48AM +0200, Maxime Coquelin wrote:
>> Commit cb1293e2f594 ("ARM: 8375/1: disable some options on ARMv7-M") causes
>> build failure on ARMv7-M machines:
>>
>> CC arch/arm/kernel/asm-offsets.s
>> In file included from include/linux/sem.h:5:0,
>> from include/linux/sched.h:35,
>> from arch/arm/kernel/asm-offsets.c:14:
>> include/linux/rcupdate.h: In function 'rcu_read_lock_sched_held':
>> include/linux/rcupdate.h:539:2: error: implicit declaration of function 'arch_irqs_disabled' [-Werror=implicit-function-declaration]
>> return preempt_count() != 0 || irqs_disabled();
>> ^
>
> The real solution is to provide a definition _in asm-generic_ for
> arch_irqs_disabled(), rather than having almost every arch doing:
>
> static inline bool arch_irqs_disabled(void)
> {
> return arch_irqs_disabled_flags(arch_local_save_flags());
> }
>
> I'm personally refusing to take a patch for ARM which adds yet another
> copy of the above. This is, after all, exactly the kind of stuff that
> should be in asm-generic, or if not, in include/linux but overridable
> by arch stuff.
>
> We keep going between the two extremes of "lets push lots of stuff into
> arch stuff" and "lets try to extract the common bits out of arch code".
>
> Let's try and settle on one approach, and apply it universally.
I agree on the idea but I don't measure all the impacts it would have.
>
> In the mean time, I think the right answer is to drop Arnd's patch -
> subsituting a randconfig build error for a useful-config build error
> is not something we want to do - and even partially reverting the
> patch results in randconfig build errors returning, so...
Ok, should I send a revert, or you can still drop Arnd's patch directly?
Thanks,
Maxime
On 09/06/15 10:14, Maxime Coquelin wrote:
>> The real solution is to provide a definition _in asm-generic_ for
>> arch_irqs_disabled(), rather than having almost every arch doing:
>>
>> static inline bool arch_irqs_disabled(void)
>> {
>> return arch_irqs_disabled_flags(arch_local_save_flags());
>> }
>>
>> I'm personally refusing to take a patch for ARM which adds yet another
>> copy of the above. This is, after all, exactly the kind of stuff that
>> should be in asm-generic, or if not, in include/linux but overridable
>> by arch stuff.
>>
>> We keep going between the two extremes of "lets push lots of stuff into
>> arch stuff" and "lets try to extract the common bits out of arch code".
>>
>> Let's try and settle on one approach, and apply it universally.
>
> I agree on the idea but I don't measure all the impacts it would have.
Hey Maxime, just out of interest...
Does the following patch, which makes the arch_irqs_disabled()
implementation from asm-generic available on arm, fix the build for you?
I've only done a real quick 'n dirty check for regression:
multi_v7_defconfig still works ;-)
If the patch is useful I can test it a bit harder...
From d86ef4466cfdbe622a65bcc7e72a8ca0d1dd2879 Mon Sep 17 00:00:00 2001
From: Daniel Thompson <[email protected]>
Date: Tue, 9 Jun 2015 12:35:24 +0100
Subject: [PATCH] ARM: irqflags: Get arch_irqs_disabled from asm-generic
asm-generic/irqflags.h provides an implementation of
arch_irqs_disabled(). Lets grab that implementation.
Suggested-by: Russell King <[email protected]>
Signed-off-by: Daniel Thompson <[email protected]>
---
arch/arm/include/asm/irqflags.h | 11 +++++++++++
1 file changed, 11 insertions(+)
diff --git a/arch/arm/include/asm/irqflags.h
b/arch/arm/include/asm/irqflags.h
index 3b763d6652a0..43908146a5cf 100644
--- a/arch/arm/include/asm/irqflags.h
+++ b/arch/arm/include/asm/irqflags.h
@@ -20,6 +20,7 @@
#if __LINUX_ARM_ARCH__ >= 6
+#define arch_local_irq_save arch_local_irq_save
static inline unsigned long arch_local_irq_save(void)
{
unsigned long flags;
@@ -31,6 +32,7 @@ static inline unsigned long arch_local_irq_save(void)
return flags;
}
+#define arch_local_irq_enable arch_local_irq_enable
static inline void arch_local_irq_enable(void)
{
asm volatile(
@@ -40,6 +42,7 @@ static inline void arch_local_irq_enable(void)
: "memory", "cc");
}
+#define arch_local_irq_disable arch_local_irq_disable
static inline void arch_local_irq_disable(void)
{
asm volatile(
@@ -56,6 +59,7 @@ static inline void arch_local_irq_disable(void)
/*
* Save the current interrupt enable state & disable IRQs
*/
+#define arch_local_irq_save arch_local_irq_save
static inline unsigned long arch_local_irq_save(void)
{
unsigned long flags, temp;
@@ -73,6 +77,7 @@ static inline unsigned long arch_local_irq_save(void)
/*
* Enable IRQs
*/
+#define arch_local_irq_enable arch_local_irq_enable
static inline void arch_local_irq_enable(void)
{
unsigned long temp;
@@ -88,6 +93,7 @@ static inline void arch_local_irq_enable(void)
/*
* Disable IRQs
*/
+#define arch_local_irq_disable arch_local_irq_disable
static inline void arch_local_irq_disable(void)
{
unsigned long temp;
@@ -135,6 +141,7 @@ static inline void arch_local_irq_disable(void)
/*
* Save the current interrupt enable state.
*/
+#define arch_local_save_flags arch_local_save_flags
static inline unsigned long arch_local_save_flags(void)
{
unsigned long flags;
@@ -147,6 +154,7 @@ static inline unsigned long arch_local_save_flags(void)
/*
* restore saved IRQ & FIQ state
*/
+#define arch_local_irq_restore arch_local_irq_restore
static inline void arch_local_irq_restore(unsigned long flags)
{
asm volatile(
@@ -156,10 +164,13 @@ static inline void arch_local_irq_restore(unsigned
long flags)
: "memory", "cc");
}
+#define arch_irqs_disabled_flags arch_irqs_disabled_flags
static inline int arch_irqs_disabled_flags(unsigned long flags)
{
return flags & IRQMASK_I_BIT;
}
+#include <asm-generic/irqflags.h>
+
#endif /* ifdef __KERNEL__ */
#endif /* ifndef __ASM_ARM_IRQFLAGS_H */
--
2.4.2
Hi Daniel,
2015-06-09 13:41 GMT+02:00 Daniel Thompson <[email protected]>:
> On 09/06/15 10:14, Maxime Coquelin wrote:
>>>
>>> The real solution is to provide a definition _in asm-generic_ for
>>> arch_irqs_disabled(), rather than having almost every arch doing:
>>>
>>> static inline bool arch_irqs_disabled(void)
>>> {
>>> return arch_irqs_disabled_flags(arch_local_save_flags());
>>> }
>>>
>>> I'm personally refusing to take a patch for ARM which adds yet another
>>> copy of the above. This is, after all, exactly the kind of stuff that
>>> should be in asm-generic, or if not, in include/linux but overridable
>>> by arch stuff.
>>>
>>> We keep going between the two extremes of "lets push lots of stuff into
>>> arch stuff" and "lets try to extract the common bits out of arch code".
>>>
>>> Let's try and settle on one approach, and apply it universally.
>
> Does the following patch, which makes the arch_irqs_disabled()
> implementation from asm-generic available on arm, fix the build for you?
I confirm it fixes the build on Russell's for-next branch with efm32_defconfig.
I have no efm32 HW to test it though.
>
> I've only done a real quick 'n dirty check for regression:
> multi_v7_defconfig still works ;-)
>
> If the patch is useful I can test it a bit harder...
I can also test it this evening with my stm32 config.
Thanks,
Maxime
On Tue, Jun 09, 2015 at 12:41:50PM +0100, Daniel Thompson wrote:
> Does the following patch, which makes the arch_irqs_disabled()
> implementation from asm-generic available on arm, fix the build for you?
Yes, this is exactly the kind of fix for this I'm looking for. Once
everyone's happy with it, it can find it's way to the patch system.
Thanks.
--
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.
On 09/06/15 16:01, Russell King - ARM Linux wrote:
> On Tue, Jun 09, 2015 at 12:41:50PM +0100, Daniel Thompson wrote:
>> Does the following patch, which makes the arch_irqs_disabled()
>> implementation from asm-generic available on arm, fix the build for you?
>
> Yes, this is exactly the kind of fix for this I'm looking for. Once
> everyone's happy with it, it can find it's way to the patch system.
From my side, using v4.1-rc6 plus the patch, I can build all defconfigs
and both versatile_defconfig and multi_v7_defconfig remain bootable (so
both ways through the arch #ifdef in irqflag.h).
Similarly working on linux-next-20150608 plus the patch I am able to
build and boot Maxime's stm32 code (and show that without the patch it
doesn't build).
From my side I think that makes it good to go.
So... in the absense of objections I will send it to the patch tracker
tomorrow.
Daniel.
2015-06-09 19:37 GMT+02:00 Daniel Thompson <[email protected]>:
> On 09/06/15 16:01, Russell King - ARM Linux wrote:
>>
>> On Tue, Jun 09, 2015 at 12:41:50PM +0100, Daniel Thompson wrote:
>>>
>>> Does the following patch, which makes the arch_irqs_disabled()
>>> implementation from asm-generic available on arm, fix the build for you?
>>
>>
>> Yes, this is exactly the kind of fix for this I'm looking for. Once
>> everyone's happy with it, it can find it's way to the patch system.
>
>
> From my side, using v4.1-rc6 plus the patch, I can build all defconfigs and
> both versatile_defconfig and multi_v7_defconfig remain bootable (so both
> ways through the arch #ifdef in irqflag.h).
>
> Similarly working on linux-next-20150608 plus the patch I am able to build
> and boot Maxime's stm32 code (and show that without the patch it doesn't
> build).
Ok, thanks for having tested on STM32.
>
> From my side I think that makes it good to go.
>
> So... in the absense of objections I will send it to the patch tracker
> tomorrow.
For what it is worth, you can add:
Acked-by: Maxime Coquelin <[email protected]>
Regards,
Maxime
Hi Daniel,
On 9 June 2015 at 19:37, Daniel Thompson <[email protected]> wrote:
> On 09/06/15 16:01, Russell King - ARM Linux wrote:
>>
>> On Tue, Jun 09, 2015 at 12:41:50PM +0100, Daniel Thompson wrote:
>>>
>>> Does the following patch, which makes the arch_irqs_disabled()
>>> implementation from asm-generic available on arm, fix the build for you?
>>
>>
>> Yes, this is exactly the kind of fix for this I'm looking for. Once
>> everyone's happy with it, it can find it's way to the patch system.
>
>
> From my side, using v4.1-rc6 plus the patch, I can build all defconfigs and
> both versatile_defconfig and multi_v7_defconfig remain bootable (so both
> ways through the arch #ifdef in irqflag.h).
>
> Similarly working on linux-next-20150608 plus the patch I am able to build
> and boot Maxime's stm32 code (and show that without the patch it doesn't
> build).
>
> From my side I think that makes it good to go.
Since next didn't build for me either I applied this and it works for
the LPC18xx (Cortex-M4) platform.
Tested-by: Joachim Eastwood <[email protected]>
regards,
Joachim Eastwood