Introduce the error detector "warning" to the error_report event and use
the error_report_end tracepoint at the end of a warning report.
This allows in-kernel tests but also userspace to more easily determine
if a warning occurred without polling kernel logs.
Signed-off-by: Marco Elver <[email protected]>
---
include/trace/events/error_report.h | 8 +++++---
kernel/panic.c | 2 ++
2 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/include/trace/events/error_report.h b/include/trace/events/error_report.h
index 96f64bf218b2..ed0164f8e79c 100644
--- a/include/trace/events/error_report.h
+++ b/include/trace/events/error_report.h
@@ -17,14 +17,16 @@
enum error_detector {
ERROR_DETECTOR_KFENCE,
- ERROR_DETECTOR_KASAN
+ ERROR_DETECTOR_KASAN,
+ ERROR_DETECTOR_WARN
};
#endif /* __ERROR_REPORT_DECLARE_TRACE_ENUMS_ONCE_ONLY */
-#define error_detector_list \
+#define error_detector_list \
EM(ERROR_DETECTOR_KFENCE, "kfence") \
- EMe(ERROR_DETECTOR_KASAN, "kasan")
+ EM(ERROR_DETECTOR_KASAN, "kasan") \
+ EMe(ERROR_DETECTOR_WARN, "warning")
/* Always end the list with an EMe. */
#undef EM
diff --git a/kernel/panic.c b/kernel/panic.c
index cefd7d82366f..8e299cae1615 100644
--- a/kernel/panic.c
+++ b/kernel/panic.c
@@ -32,6 +32,7 @@
#include <linux/bug.h>
#include <linux/ratelimit.h>
#include <linux/debugfs.h>
+#include <trace/events/error_report.h>
#include <asm/sections.h>
#define PANIC_TIMER_STEP 100
@@ -609,6 +610,7 @@ void __warn(const char *file, int line, void *caller, unsigned taint,
print_irqtrace_events(current);
print_oops_end_marker();
+ trace_error_report_end(ERROR_DETECTOR_WARN, (unsigned long)caller);
/* Just a warning, don't kill lockdep. */
add_taint(taint, LOCKDEP_STILL_OK);
--
2.34.0.rc1.387.gb447b232ab-goog
On Mon, Nov 15, 2021 at 09:56:30AM +0100, Marco Elver wrote:
> Introduce the error detector "warning" to the error_report event and use
> the error_report_end tracepoint at the end of a warning report.
>
> This allows in-kernel tests but also userspace to more easily determine
> if a warning occurred without polling kernel logs.
...
> enum error_detector {
> ERROR_DETECTOR_KFENCE,
> - ERROR_DETECTOR_KASAN
> + ERROR_DETECTOR_KASAN,
> + ERROR_DETECTOR_WARN
...which exactly shows my point (given many times somewhere else) why comma
is good to have when we are not sure the item is a terminator one in the enum
or array of elements.
> };
--
With Best Regards,
Andy Shevchenko
On Mon, 15 Nov 2021 at 15:38, Andy Shevchenko
<[email protected]> wrote:
>
> On Mon, Nov 15, 2021 at 09:56:30AM +0100, Marco Elver wrote:
> > Introduce the error detector "warning" to the error_report event and use
> > the error_report_end tracepoint at the end of a warning report.
> >
> > This allows in-kernel tests but also userspace to more easily determine
> > if a warning occurred without polling kernel logs.
>
> ...
>
> > enum error_detector {
> > ERROR_DETECTOR_KFENCE,
> > - ERROR_DETECTOR_KASAN
> > + ERROR_DETECTOR_KASAN,
> > + ERROR_DETECTOR_WARN
>
> ...which exactly shows my point (given many times somewhere else) why comma
> is good to have when we are not sure the item is a terminator one in the enum
> or array of elements.
So you want me to add a comma?
(I'm not participating in bikeshedding here, just tell me what to do.)
Thanks,
-- Marco
On Mon, Nov 15, 2021 at 03:40:24PM +0100, Marco Elver wrote:
> On Mon, 15 Nov 2021 at 15:38, Andy Shevchenko
> <[email protected]> wrote:
> > On Mon, Nov 15, 2021 at 09:56:30AM +0100, Marco Elver wrote:
...
> > > ERROR_DETECTOR_KFENCE,
> > > - ERROR_DETECTOR_KASAN
> > > + ERROR_DETECTOR_KASAN,
> > > + ERROR_DETECTOR_WARN
> >
> > ...which exactly shows my point (given many times somewhere else) why comma
> > is good to have when we are not sure the item is a terminator one in the enum
> > or array of elements.
>
> So you want me to add a comma?
Yes. And you see exactly why I'm asking for that.
> (I'm not participating in bikeshedding here, just tell me what to do.)
Done!
--
With Best Regards,
Andy Shevchenko
On Mon, 15 Nov 2021 16:55:17 +0200
Andy Shevchenko <[email protected]> wrote:
> > > > ERROR_DETECTOR_KFENCE,
> > > > - ERROR_DETECTOR_KASAN
> > > > + ERROR_DETECTOR_KASAN,
> > > > + ERROR_DETECTOR_WARN
> > >
> > > ...which exactly shows my point (given many times somewhere else) why comma
> > > is good to have when we are not sure the item is a terminator one in the enum
> > > or array of elements.
> >
> > So you want me to add a comma?
>
> Yes. And you see exactly why I'm asking for that.
>
> > (I'm not participating in bikeshedding here, just tell me what to do.)
>
> Done!
Is there going to be another patch set? Or did I miss it?
-- Steve
On Sat, 11 Dec 2021 at 00:35, Steven Rostedt <[email protected]> wrote:
>
> On Mon, 15 Nov 2021 16:55:17 +0200
> Andy Shevchenko <[email protected]> wrote:
>
> > > > > ERROR_DETECTOR_KFENCE,
> > > > > - ERROR_DETECTOR_KASAN
> > > > > + ERROR_DETECTOR_KASAN,
> > > > > + ERROR_DETECTOR_WARN
> > > >
> > > > ...which exactly shows my point (given many times somewhere else) why comma
> > > > is good to have when we are not sure the item is a terminator one in the enum
> > > > or array of elements.
> > >
> > > So you want me to add a comma?
> >
> > Yes. And you see exactly why I'm asking for that.
> >
> > > (I'm not participating in bikeshedding here, just tell me what to do.)
> >
> > Done!
>
> Is there going to be another patch set? Or did I miss it?
Andrew was kind enough to fix it up for me. :-)
It's already in -mm:
https://www.spinics.net/lists/mm-commits/msg161488.html
https://www.spinics.net/lists/mm-commits/msg161487.html
Thanks,
-- Marco