Re: [PATCH] kunit: Taint kernel if any tests run

From: Greg KH
Date: Fri Apr 29 2022 - 08:07:41 EST


On Fri, Apr 29, 2022 at 02:54:25PM +0300, Jani Nikula wrote:
> On Fri, 29 Apr 2022, Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> > On Fri, Apr 29, 2022 at 02:21:26PM +0300, Jani Nikula wrote:
> >> On Fri, 29 Apr 2022, Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> >> > On Fri, Apr 29, 2022 at 12:39:14PM +0800, David Gow wrote:
> >> >> KUnit tests are not supposed to run on production systems: they may do
> >> >> deliberately illegal things to trigger errors, and have security
> >> >> implications (assertions will often deliberately leak kernel addresses).
> >> >>
> >> >> Add a new taint type, TAINT_KUNIT to signal that a KUnit test has been
> >> >> run. This will be printed as 'N' (for kuNit, as K, U and T were already
> >> >> taken).
> >> >>
> >> >> This should discourage people from running KUnit tests on production
> >> >> systems, and to make it easier to tell if tests have been run
> >> >> accidentally (by loading the wrong configuration, etc.)
> >> >>
> >> >> Signed-off-by: David Gow <davidgow@xxxxxxxxxx>
> >> >> ---
> >> >>
> >> >> This is something I'd been thinking about for a while, and it came up
> >> >> again, so I'm finally giving it a go.
> >> >>
> >> >> Two notes:
> >> >> - I decided to add a new type of taint, as none of the existing ones
> >> >> really seemed to fit. We could live with considering KUnit tests as
> >> >> TAINT_WARN or TAINT_CRAP or something otherwise, but neither are quite
> >> >> right.
> >> >> - The taint_flags table gives a couple of checkpatch.pl errors around
> >> >> bracket placement. I've kept the new entry consistent with what's
> >> >> there rather than reformatting the whole table, but be prepared for
> >> >> complaints about spaces.
> >> >>
> >> >> Thoughts?
> >> >> -- David
> >> >>
> >> >> ---
> >> >> Documentation/admin-guide/tainted-kernels.rst | 1 +
> >> >> include/linux/panic.h | 3 ++-
> >> >> kernel/panic.c | 1 +
> >> >> lib/kunit/test.c | 4 ++++
> >> >> 4 files changed, 8 insertions(+), 1 deletion(-)
> >> >>
> >> >> diff --git a/Documentation/admin-guide/tainted-kernels.rst b/Documentation/admin-guide/tainted-kernels.rst
> >> >> index ceeed7b0798d..8f18fc4659d4 100644
> >> >> --- a/Documentation/admin-guide/tainted-kernels.rst
> >> >> +++ b/Documentation/admin-guide/tainted-kernels.rst
> >> >> @@ -100,6 +100,7 @@ Bit Log Number Reason that got the kernel tainted
> >> >> 15 _/K 32768 kernel has been live patched
> >> >> 16 _/X 65536 auxiliary taint, defined for and used by distros
> >> >> 17 _/T 131072 kernel was built with the struct randomization plugin
> >> >> + 18 _/N 262144 a KUnit test has been run
> >> >> === === ====== ========================================================
> >> >>
> >> >> Note: The character ``_`` is representing a blank in this table to make reading
> >> >> diff --git a/include/linux/panic.h b/include/linux/panic.h
> >> >> index f5844908a089..1d316c26bf27 100644
> >> >> --- a/include/linux/panic.h
> >> >> +++ b/include/linux/panic.h
> >> >> @@ -74,7 +74,8 @@ static inline void set_arch_panic_timeout(int timeout, int arch_default_timeout)
> >> >> #define TAINT_LIVEPATCH 15
> >> >> #define TAINT_AUX 16
> >> >> #define TAINT_RANDSTRUCT 17
> >> >> -#define TAINT_FLAGS_COUNT 18
> >> >> +#define TAINT_KUNIT 18
> >> >> +#define TAINT_FLAGS_COUNT 19
> >> >> #define TAINT_FLAGS_MAX ((1UL << TAINT_FLAGS_COUNT) - 1)
> >> >>
> >> >> struct taint_flag {
> >> >> diff --git a/kernel/panic.c b/kernel/panic.c
> >> >> index eb4dfb932c85..b24ca63ed738 100644
> >> >> --- a/kernel/panic.c
> >> >> +++ b/kernel/panic.c
> >> >> @@ -404,6 +404,7 @@ const struct taint_flag taint_flags[TAINT_FLAGS_COUNT] = {
> >> >> [ TAINT_LIVEPATCH ] = { 'K', ' ', true },
> >> >> [ TAINT_AUX ] = { 'X', ' ', true },
> >> >> [ TAINT_RANDSTRUCT ] = { 'T', ' ', true },
> >> >> + [ TAINT_KUNIT ] = { 'N', ' ', false },
> >> >
> >> > As kunit tests can be in modules, shouldn't this be "true" here?
> >> >
> >> > Overall, I like it, makes sense to me. The "N" will take some getting
> >> > used to, and I have no idea why "T" was for "struct randomization", that
> >> > would have allowed you to use "T" instead. Oh well.
> >>
> >> Would you consider a patch adding more self-explanatory taint flag
> >> strings to the output?
> >
> > Where would those strings go? In the oops report? Or somewhere else?
>
> I was thinking the oops report. Basically most times I look at an oops
> with taint, I need to double check what the flags mean. There are soon
> 19 of them, you need to look at a lot of oops to remember them all.

I agree, it isn't easy to remember.

> Currently we also print ' ' (or 'G' in case of non-properietary module)
> for every unset taint flag. If we stopped doing that we wouldn't even
> need that much more horizontal space for the strings, unless several
> flags were set. (I assume people who do remember all the flags by heart
> would still want to keep them too.)

I recommend keeping the current layout, but maybe adding a new line that
gives the "key" for what the current taint flags mean?

For example, the oops report here:
https://lore.kernel.org/r/20220413033425.GM16799@magnolia
Has the lines:
kernel BUG at mm/filemap.c:1653!
invalid opcode: 0000 [#1] PREEMPT SMP
CPU: 0 PID: 1349866 Comm: 0:116 Tainted: G W 5.18.0-rc2-djwx #rc2 19cc48221d47ada6c8e5859639b6a0946c9a3777
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS ?-20171121_152543-x86-ol7-builder-01.us.oracle.com-4.el7.1 04/01/2014
Workqueue: xfs-conv/sda4 xfs_end_io [xfs]
RIP: 0010:folio_end_writeback+0x79/0x80

Perhaps we add another line right before or after "Hardware name:" that
lists the flags that are set at the moment and what they mean:

Taint flags: [G]=PROPRIETARY_MODULE, [W]=WARN

Or something like that (format was a first guess only).

Anyway, might be helpful?

thanks,

greg k-h