[Intel-gfx] [PATCH igt 2/2] igt/debugfs_tests: Record which file is being opened in kmsg

Chris Wilson chris at chris-wilson.co.uk
Wed Dec 6 17:08:15 UTC 2017


Quoting Tvrtko Ursulin (2017-12-06 17:04:40)
> 
> On 06/12/2017 17:02, Chris Wilson wrote:
> > Quoting Tvrtko Ursulin (2017-12-06 16:51:24)
> >>
> >> On 06/12/2017 16:38, Chris Wilson wrote:
> >>> When tracking down the cause of a particular kernel warning, knowing
> >>> which file it is associated with can be a big clue. So write the
> >>> filename into the kernel message log prior to opening it.
> >>>
> >>> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> >>> ---
> >>>    tests/debugfs_test.c | 3 ++-
> >>>    1 file changed, 2 insertions(+), 1 deletion(-)
> >>>
> >>> diff --git a/tests/debugfs_test.c b/tests/debugfs_test.c
> >>> index 268d6e78f..2e87e4420 100644
> >>> --- a/tests/debugfs_test.c
> >>> +++ b/tests/debugfs_test.c
> >>> @@ -61,8 +61,9 @@ static void read_and_discard_sysfs_entries(int path_fd, int indent)
> >>>                        int sub_fd;
> >>>                        ssize_t ret;
> >>>    
> >>> -                     igt_set_timeout(5, "reading sysfs entry");
> >>> +                     igt_kmsg(KMSG_DEBUG "Reading file \"%s\"\n", dirent->d_name);
> >>>                        igt_debug("%sReading file \"%s\"\n", tabs, dirent->d_name);
> >>> +                     igt_set_timeout(5, "reading sysfs entry");
> >>>    
> >>>                        sub_fd = openat(path_fd, dirent->d_name, O_RDONLY);
> >>>                        if (sub_fd == -1) {
> >>>
> >>
> >> If I may suggest an alternative - timestamp igt messages and write a log
> >> interleaver to be used with kernel log and igt log as inputs?
> > 
> > How do you get igt_log over netconsole?
> 
> Hm, ok. But should we then just send igt logs to two places? Stdout/err 
> and kmsg? Could be an environment option or something if required.

There are quite a few igt logs that would be nice in the kernel log,
pretty much anything info+. Maybe debug+ if we push some of the noise
out of debug and down into even lower-priority-debug.
-Chris


More information about the Intel-gfx mailing list