[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <201912171607.73EE8133@keescook>
Date: Tue, 17 Dec 2019 16:08:02 -0800
From: Kees Cook <keescook@...omium.org>
To: Will Deacon <will@...nel.org>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Andrey Ryabinin <aryabinin@...tuozzo.com>,
Elena Petrova <lenaptr@...gle.com>,
Alexander Potapenko <glider@...gle.com>,
Dmitry Vyukov <dvyukov@...gle.com>,
Linus Torvalds <torvalds@...ux-foundation.org>,
Dan Carpenter <dan.carpenter@...cle.com>,
"Gustavo A. R. Silva" <gustavo@...eddedor.com>,
Arnd Bergmann <arnd@...db.de>,
Ard Biesheuvel <ard.biesheuvel@...aro.org>,
kasan-dev@...glegroups.com, linux-kernel@...r.kernel.org,
kernel-hardening@...ts.openwall.com
Subject: Re: [PATCH v2 1/3] ubsan: Add trap instrumentation option
On Mon, Dec 16, 2019 at 10:26:56AM +0000, Will Deacon wrote:
> Hi Kees,
>
> On Thu, Nov 21, 2019 at 10:15:17AM -0800, Kees Cook wrote:
> > The Undefined Behavior Sanitizer can operate in two modes: warning
> > reporting mode via lib/ubsan.c handler calls, or trap mode, which uses
> > __builtin_trap() as the handler. Using lib/ubsan.c means the kernel
> > image is about 5% larger (due to all the debugging text and reporting
> > structures to capture details about the warning conditions). Using the
> > trap mode, the image size changes are much smaller, though at the loss
> > of the "warning only" mode.
> >
> > In order to give greater flexibility to system builders that want
> > minimal changes to image size and are prepared to deal with kernel code
> > being aborted and potentially destabilizing the system, this introduces
> > CONFIG_UBSAN_TRAP. The resulting image sizes comparison:
> >
> > text data bss dec hex filename
> > 19533663 6183037 18554956 44271656 2a38828 vmlinux.stock
> > 19991849 7618513 18874448 46484810 2c54d4a vmlinux.ubsan
> > 19712181 6284181 18366540 44362902 2a4ec96 vmlinux.ubsan-trap
> >
> > CONFIG_UBSAN=y: image +4.8% (text +2.3%, data +18.9%)
> > CONFIG_UBSAN_TRAP=y: image +0.2% (text +0.9%, data +1.6%)
> >
> > Additionally adjusts the CONFIG_UBSAN Kconfig help for clarity and
> > removes the mention of non-existing boot param "ubsan_handle".
> >
> > Suggested-by: Elena Petrova <lenaptr@...gle.com>
> > Signed-off-by: Kees Cook <keescook@...omium.org>
> > ---
> > lib/Kconfig.ubsan | 22 ++++++++++++++++++----
> > lib/Makefile | 2 ++
> > scripts/Makefile.ubsan | 9 +++++++--
> > 3 files changed, 27 insertions(+), 6 deletions(-)
> >
> > diff --git a/lib/Kconfig.ubsan b/lib/Kconfig.ubsan
> > index 0e04fcb3ab3d..9deb655838b0 100644
> > --- a/lib/Kconfig.ubsan
> > +++ b/lib/Kconfig.ubsan
> > @@ -5,11 +5,25 @@ config ARCH_HAS_UBSAN_SANITIZE_ALL
> > config UBSAN
> > bool "Undefined behaviour sanity checker"
> > help
> > - This option enables undefined behaviour sanity checker
> > + This option enables the Undefined Behaviour sanity checker.
> > Compile-time instrumentation is used to detect various undefined
> > - behaviours in runtime. Various types of checks may be enabled
> > - via boot parameter ubsan_handle
> > - (see: Documentation/dev-tools/ubsan.rst).
> > + behaviours at runtime. For more details, see:
> > + Documentation/dev-tools/ubsan.rst
> > +
> > +config UBSAN_TRAP
> > + bool "On Sanitizer warnings, abort the running kernel code"
> > + depends on UBSAN
> > + depends on $(cc-option, -fsanitize-undefined-trap-on-error)
> > + help
> > + Building kernels with Sanitizer features enabled tends to grow
> > + the kernel size by around 5%, due to adding all the debugging
> > + text on failure paths. To avoid this, Sanitizer instrumentation
> > + can just issue a trap. This reduces the kernel size overhead but
> > + turns all warnings (including potentially harmless conditions)
> > + into full exceptions that abort the running kernel code
> > + (regardless of context, locks held, etc), which may destabilize
> > + the system. For some system builders this is an acceptable
> > + trade-off.
>
> Slight nit, but I wonder if it would make sense to move all this under a
> 'menuconfig UBSAN' entry, so the dependencies can be dropped? Then you could
> have all of the suboptions default to on and basically choose which
> individual compiler options to disable based on your own preferences.
Sure; I can do that. I'll respin the series.
--
Kees Cook
Powered by blists - more mailing lists