On Wed 2014-12-03 09:33 -0500, Dave Anderson wrote:
I did a quick check of this patch with my set of sample dumps, and
since
a majority (presumably) of vmcores are generated as the result of an oops,
they will be marked as tainted. So printing the tainted information is
pretty much useless, and actually somewhat misleading.
That's because the kernel will be marked as tainted because TAINT_DIE gets
set by oops_end(), which is called *after* __die(), which is where the
registers and the translation of the tainted mask are dumped:
die()
__die()
show_regs()
show_regs_print_info()
dump_stack_print_info()
print_tainted() --> translates tained_mask
...
oops_end() --> add_taint(TAINT_DIE)
panic()
So the kernel log will (correctly) show "Not tainted" because oops_end() would
not have been called yet, yet this patch would yell out TAINTED.
So I'd like to keep this relegated to "sys -t".
OK - understood.
--
Aaron Tomlin