Currently there is a small window where a badly timed migration could
cause in_dbg_master() to spuriously return true. Specifically if we
migrate to a new core after reading the processor id and the previous
core takes a breakpoint then we will evaluate true if we read
kgdb_active before we get the IPI to bring us to halt.
Fix this by checking irqs_disabled() first. Interrupts are always
disabled when we are executing the kgdb trap so this is an acceptable
prerequisite. This also allows us to replace raw_smp_processor_id()
with smp_processor_id() since the short circuit logic will prevent
warnings from PREEMPT_DEBUG.
Fixes: dcc7871128e9 ("kgdb: core changes to support kdb")
Suggested-by: Will Deacon <[email protected]>
Signed-off-by: Daniel Thompson <[email protected]>
---
include/linux/kgdb.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/include/linux/kgdb.h b/include/linux/kgdb.h
index b072aeb1fd78..4d6fe87fd38f 100644
--- a/include/linux/kgdb.h
+++ b/include/linux/kgdb.h
@@ -323,7 +323,7 @@ extern void gdbstub_exit(int status);
extern int kgdb_single_step;
extern atomic_t kgdb_active;
#define in_dbg_master() \
- (raw_smp_processor_id() == atomic_read(&kgdb_active))
+ (irqs_disabled() && (smp_processor_id() == atomic_read(&kgdb_active)))
extern bool dbg_is_early;
extern void __init dbg_late_init(void);
extern void kgdb_panic(const char *msg);
base-commit: 6a8b55ed4056ea5559ebe4f6a4b247f627870d4c
--
2.25.1
Hi,
On Wed, May 6, 2020 at 9:42 AM Daniel Thompson
<[email protected]> wrote:
>
> Currently there is a small window where a badly timed migration could
> cause in_dbg_master() to spuriously return true. Specifically if we
> migrate to a new core after reading the processor id and the previous
> core takes a breakpoint then we will evaluate true if we read
> kgdb_active before we get the IPI to bring us to halt.
>
> Fix this by checking irqs_disabled() first. Interrupts are always
> disabled when we are executing the kgdb trap so this is an acceptable
> prerequisite. This also allows us to replace raw_smp_processor_id()
> with smp_processor_id() since the short circuit logic will prevent
> warnings from PREEMPT_DEBUG.
>
> Fixes: dcc7871128e9 ("kgdb: core changes to support kdb")
> Suggested-by: Will Deacon <[email protected]>
> Signed-off-by: Daniel Thompson <[email protected]>
> ---
> include/linux/kgdb.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/include/linux/kgdb.h b/include/linux/kgdb.h
> index b072aeb1fd78..4d6fe87fd38f 100644
> --- a/include/linux/kgdb.h
> +++ b/include/linux/kgdb.h
> @@ -323,7 +323,7 @@ extern void gdbstub_exit(int status);
> extern int kgdb_single_step;
> extern atomic_t kgdb_active;
> #define in_dbg_master() \
> - (raw_smp_processor_id() == atomic_read(&kgdb_active))
> + (irqs_disabled() && (smp_processor_id() == atomic_read(&kgdb_active)))
> extern bool dbg_is_early;
> extern void __init dbg_late_init(void);
> extern void kgdb_panic(const char *msg);
Reviewed-by: Douglas Anderson <[email protected]>
On Wed, May 06, 2020 at 05:42:23PM +0100, Daniel Thompson wrote:
> Currently there is a small window where a badly timed migration could
> cause in_dbg_master() to spuriously return true. Specifically if we
> migrate to a new core after reading the processor id and the previous
> core takes a breakpoint then we will evaluate true if we read
> kgdb_active before we get the IPI to bring us to halt.
>
> Fix this by checking irqs_disabled() first. Interrupts are always
> disabled when we are executing the kgdb trap so this is an acceptable
> prerequisite. This also allows us to replace raw_smp_processor_id()
> with smp_processor_id() since the short circuit logic will prevent
> warnings from PREEMPT_DEBUG.
>
> Fixes: dcc7871128e9 ("kgdb: core changes to support kdb")
> Suggested-by: Will Deacon <[email protected]>
> Signed-off-by: Daniel Thompson <[email protected]>
> ---
> include/linux/kgdb.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/include/linux/kgdb.h b/include/linux/kgdb.h
> index b072aeb1fd78..4d6fe87fd38f 100644
> --- a/include/linux/kgdb.h
> +++ b/include/linux/kgdb.h
> @@ -323,7 +323,7 @@ extern void gdbstub_exit(int status);
> extern int kgdb_single_step;
> extern atomic_t kgdb_active;
> #define in_dbg_master() \
> - (raw_smp_processor_id() == atomic_read(&kgdb_active))
> + (irqs_disabled() && (smp_processor_id() == atomic_read(&kgdb_active)))
> extern bool dbg_is_early;
> extern void __init dbg_late_init(void);
> extern void kgdb_panic(const char *msg);
Cheers, Daniel. I assume you'll route this one via the kgdb tree? I can
live with the "small window" in the arm64 for-next/core branch ;)
Will
On Thu, May 07, 2020 at 09:39:30AM +0100, Will Deacon wrote:
> On Wed, May 06, 2020 at 05:42:23PM +0100, Daniel Thompson wrote:
> > Currently there is a small window where a badly timed migration could
> > cause in_dbg_master() to spuriously return true. Specifically if we
> > migrate to a new core after reading the processor id and the previous
> > core takes a breakpoint then we will evaluate true if we read
> > kgdb_active before we get the IPI to bring us to halt.
> >
> > Fix this by checking irqs_disabled() first. Interrupts are always
> > disabled when we are executing the kgdb trap so this is an acceptable
> > prerequisite. This also allows us to replace raw_smp_processor_id()
> > with smp_processor_id() since the short circuit logic will prevent
> > warnings from PREEMPT_DEBUG.
> >
> > Fixes: dcc7871128e9 ("kgdb: core changes to support kdb")
> > Suggested-by: Will Deacon <[email protected]>
> > Signed-off-by: Daniel Thompson <[email protected]>
> > ---
> > include/linux/kgdb.h | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/include/linux/kgdb.h b/include/linux/kgdb.h
> > index b072aeb1fd78..4d6fe87fd38f 100644
> > --- a/include/linux/kgdb.h
> > +++ b/include/linux/kgdb.h
> > @@ -323,7 +323,7 @@ extern void gdbstub_exit(int status);
> > extern int kgdb_single_step;
> > extern atomic_t kgdb_active;
> > #define in_dbg_master() \
> > - (raw_smp_processor_id() == atomic_read(&kgdb_active))
> > + (irqs_disabled() && (smp_processor_id() == atomic_read(&kgdb_active)))
> > extern bool dbg_is_early;
> > extern void __init dbg_late_init(void);
> > extern void kgdb_panic(const char *msg);
>
> Cheers, Daniel. I assume you'll route this one via the kgdb tree? I can
> live with the "small window" in the arm64 for-next/core branch ;)
Yes. I'll get this one applied very soon (thanks for Doug for the quick
review).
Daniel.