panic: move panic_print before kmsg dumpers
authorGuilherme G. Piccoli <gpiccoli@igalia.com>
Mon, 28 Feb 2022 23:02:31 +0000 (10:02 +1100)
committerStephen Rothwell <sfr@canb.auug.org.au>
Mon, 28 Feb 2022 23:02:31 +0000 (10:02 +1100)
commit4b1b7021af559abae63f42a21431b6bf1458cb4b
tree79b5cd0a1f98e2de0715a0ba6b2c5403339c6d0f
parent24bb124d1aaf76acb98c2f3a1ced6b9e1a52744c
panic: move panic_print before kmsg dumpers

The panic_print s\10etting allows users to collect more information in a
panic event, like memory stats, tasks, CPUs backtraces, etc.
This is an interesting debug mechanism, but currently the print event
happens *after* kmsg_dump(), meaning that pstore, for example, cannot
collect a dmesg with the panic_print extra information.

This patch changes that in 2 steps:

(a) The panic_print setting allows to replay the existing kernel log
    buffer to the console (bit 5), besides the extra information dump.
    This functionality makes sense only at the end of the panic()
    function.  So, we hereby allow to distinguish the two situations by a
    new boolean parameter in the function panic_print_sys_info().

(b) With the above change, we can safely call panic_print_sys_info()
    before kmsg_dump(), allowing to dump the extra information when using
    pstore or other kmsg dumpers.

The additional messages from panic_print could overwrite the oldest
messages when the buffer is full.  The only reasonable solution is to use
a large enough log buffer, hence we added an advice into the kernel
parameters documentation about that.

Link: https://lkml.kernel.org/r/20220214141308.841525-1-gpiccoli@igalia.com
Signed-off-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
Acked-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Petr Mladek <pmladek@suse.com>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Feng Tang <feng.tang@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Documentation/admin-guide/kernel-parameters.txt
kernel/panic.c