I have been asked whether it is safe to call vkdb_printf() from vprintk_nmi() in NMI context. It seems that it is not safe. Well, is not a big deal.
But I have noticed suspicious patterns when looking at the vkdb_printf() locking code and prepared two patches that should avoid some possible races. Please, note that I am not familiar with the kdb implementation. I hope that I have got the requested behavior right. I did some basic testing and did not found any problem. But I am sure that I did not test all usecases. Petr Mladek (2): kdb: Properly synchronize vkdb_printf() calls with other CPUs kdb: Call vkdb_printf() from vprintk_default() only when wanted include/linux/kdb.h | 1 + kernel/debug/kdb/kdb_io.c | 41 +++++++++++++++++------------------------ kernel/debug/kdb/kdb_private.h | 1 - kernel/printk/printk.c | 4 +++- 4 files changed, 21 insertions(+), 26 deletions(-) -- 1.8.5.6
