| Message ID | 20240226-thermal-fix-fortify-panic-num_trips-v1-1-accc12a341d7@kernel.org |
|---|---|
| State | Accepted |
| Commit | da1983355ccefcfb3f8eb410fff82e250fa87e39 |
| Headers | show |
| Series | thermal: core: Move initial num_trips assignment before memcpy() | expand |
On Wed, Feb 28, 2024 at 09:56:51AM -0700, Nathan Chancellor wrote: > On Wed, Feb 28, 2024 at 08:41:07AM +0000, Lukasz Luba wrote: > > Hi Nathan and Kees, > > > > On 2/27/24 17:00, Kees Cook wrote: > > > On Tue, Feb 27, 2024 at 05:47:44PM +0100, Daniel Lezcano wrote: > > > > Ok my misunderstanding was I thought sizeof() was calling _bdos under the > > > > hood, so when calling sizeof(flex_array), it was returning the computed size > > > > inferring from the __counted_by field. > > > > > > Yeah, sizeof() has a very limited scope. __builtin_object_size() has > > > more flexibility (via the 2nd argument, "type"), but it was still > > > compile-time only. __builtin_dynamic_object_size() was added to bring > > > runtime evaluations into the mix (initially to support the alloc_size > > > attribute, and now includes the counted_by attribute too). > > > > > > > Thanks for your earlier emails explaining these stuff. > > Do you have maybe some presentation about those features > > for the kernel (ideally w/ a video from some conference)? > > I think Kees's 2022 and 2023 talks at LPC are a good place to start: > > https://youtu.be/tQwv79i02ks?si=Nj9hpvmQwPB4K3Y4&t=452 > https://youtu.be/OEFFqhP5sts?si=u6RnOP641S8FkouD&t=614 > > https://outflux.net/slides/2022/lpc/features.pdf > https://outflux.net/slides/2023/lpc/features.pdf I've also got a write-up on the entire topic of array bounds, which ends with some discussion of "the future" (which is now) involving the use of the "counted_by" attribute: https://people.kernel.org/kees/bounded-flexible-arrays-in-c#coming-soon-annotate-bounds-of-flexible-arrays
On 2/28/24 17:48, Kees Cook wrote: > On Wed, Feb 28, 2024 at 09:56:51AM -0700, Nathan Chancellor wrote: >> On Wed, Feb 28, 2024 at 08:41:07AM +0000, Lukasz Luba wrote: >>> Hi Nathan and Kees, >>> >>> On 2/27/24 17:00, Kees Cook wrote: >>>> On Tue, Feb 27, 2024 at 05:47:44PM +0100, Daniel Lezcano wrote: >>>>> Ok my misunderstanding was I thought sizeof() was calling _bdos under the >>>>> hood, so when calling sizeof(flex_array), it was returning the computed size >>>>> inferring from the __counted_by field. >>>> >>>> Yeah, sizeof() has a very limited scope. __builtin_object_size() has >>>> more flexibility (via the 2nd argument, "type"), but it was still >>>> compile-time only. __builtin_dynamic_object_size() was added to bring >>>> runtime evaluations into the mix (initially to support the alloc_size >>>> attribute, and now includes the counted_by attribute too). >>>> >>> >>> Thanks for your earlier emails explaining these stuff. >>> Do you have maybe some presentation about those features >>> for the kernel (ideally w/ a video from some conference)? >> >> I think Kees's 2022 and 2023 talks at LPC are a good place to start: >> >> https://youtu.be/tQwv79i02ks?si=Nj9hpvmQwPB4K3Y4&t=452 >> https://youtu.be/OEFFqhP5sts?si=u6RnOP641S8FkouD&t=614 >> >> https://outflux.net/slides/2022/lpc/features.pdf >> https://outflux.net/slides/2023/lpc/features.pdf > > I've also got a write-up on the entire topic of array bounds, which ends > with some discussion of "the future" (which is now) involving the use of > the "counted_by" attribute: > https://people.kernel.org/kees/bounded-flexible-arrays-in-c#coming-soon-annotate-bounds-of-flexible-arrays > Thank you guys!
diff --git a/drivers/thermal/thermal_core.c b/drivers/thermal/thermal_core.c index bb21f78b4bfa..1eabc8ebe27d 100644 --- a/drivers/thermal/thermal_core.c +++ b/drivers/thermal/thermal_core.c @@ -1354,8 +1354,8 @@ thermal_zone_device_register_with_trips(const char *type, tz->device.class = thermal_class; tz->devdata = devdata; - memcpy(tz->trips, trips, num_trips * sizeof(*trips)); tz->num_trips = num_trips; + memcpy(tz->trips, trips, num_trips * sizeof(*trips)); thermal_set_delay_jiffies(&tz->passive_delay_jiffies, passive_delay); thermal_set_delay_jiffies(&tz->polling_delay_jiffies, polling_delay);
When booting a CONFIG_FORTIFY_SOURCE=y kernel compiled with a toolchain that supports __counted_by() (such as clang-18 and newer), there is a panic on boot: [ 2.913770] memcpy: detected buffer overflow: 72 byte write of buffer size 0 [ 2.920834] WARNING: CPU: 2 PID: 1 at lib/string_helpers.c:1027 __fortify_report+0x5c/0x74 ... [ 3.039208] Call trace: [ 3.041643] __fortify_report+0x5c/0x74 [ 3.045469] __fortify_panic+0x18/0x20 [ 3.049209] thermal_zone_device_register_with_trips+0x4c8/0x4f8 This panic occurs because trips is counted by num_trips but num_trips is assigned after the call to memcpy(), so the fortify checks think the buffer size is zero because tz was allocated with kzalloc(). Move the num_trips assignment before the memcpy() to resolve the panic and ensure that the fortify checks work properly. Fixes: 9b0a62758665 ("thermal: core: Store zone trips table in struct thermal_zone_device") Signed-off-by: Nathan Chancellor <nathan@kernel.org> --- drivers/thermal/thermal_core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) --- base-commit: a85739c8c6894c3b9ff860e79e91db44cb59bd63 change-id: 20240226-thermal-fix-fortify-panic-num_trips-5f94094fb963 Best regards,