| Message ID | 20250318204841.373116-2-jeremy.linton@arm.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | arm64: Enable UPROBES with GCS | expand |
On Tue, Mar 18, 2025 at 03:48:35PM -0500, Jeremy Linton wrote: > Mark Rutland noticed that the task parameter is ignored and > 'current' is being used instead. Since this is usually > what its passed, it hasn't yet been causing problems but likely > will as the code gets more testing. Reviewed-by: Mark Brown <broonie@kernel.org>
On Tue, Mar 18, 2025 at 03:48:35PM -0500, Jeremy Linton wrote: > Mark Rutland noticed that the task parameter is ignored and > 'current' is being used instead. Since this is usually > what its passed, it hasn't yet been causing problems but likely > will as the code gets more testing. Are we sure nothing is relying upon the bug? For example, in copy_thread_gcs(): copy_thread_gcs(p, ...) { ... gcs = gcs_alloc_thread_stack(p, ...) { ... if (!task_gcs_el0_enabled(p)) return 0; ... < actually allocate here > } ... p->thread.gcs_el0_mode = current->thread.gcs_el0_mode; ... } Either that later assignment is redundant, or copy_thread_gcs() was accidentally relying upon task_gcs_el0_enabled() reading from 'current' rather than 'p', and this change opens up another bug... Mark. > > Fixes: fc84bc5378a8 ("arm64/gcs: Context switch GCS state for EL0") > Signed-off-by: Jeremy Linton <jeremy.linton@arm.com> > --- > arch/arm64/include/asm/gcs.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/arch/arm64/include/asm/gcs.h b/arch/arm64/include/asm/gcs.h > index f50660603ecf..5bc432234d3a 100644 > --- a/arch/arm64/include/asm/gcs.h > +++ b/arch/arm64/include/asm/gcs.h > @@ -58,7 +58,7 @@ static inline u64 gcsss2(void) > > static inline bool task_gcs_el0_enabled(struct task_struct *task) > { > - return current->thread.gcs_el0_mode & PR_SHADOW_STACK_ENABLE; > + return task->thread.gcs_el0_mode & PR_SHADOW_STACK_ENABLE; > } > > void gcs_set_el0_mode(struct task_struct *task); > -- > 2.48.1 >
On Wed, Mar 19, 2025 at 02:26:26PM +0000, Mark Rutland wrote: > On Tue, Mar 18, 2025 at 03:48:35PM -0500, Jeremy Linton wrote: > if (!task_gcs_el0_enabled(p)) > return 0; > p->thread.gcs_el0_mode = current->thread.gcs_el0_mode; > Either that later assignment is redundant, or copy_thread_gcs() was > accidentally relying upon task_gcs_el0_enabled() reading from 'current' > rather than 'p', and this change opens up another bug... copy_thread_gcs() looks buggy here - we should move the allocation of the new stack to the bottom of the function after the assignments. Like you say it does currently work due to the check of the source thread. The other users are the prctl and signal code which always work with current and the check to see if we should do a GCSB DSYNC in gcs_thread_switch() which currently misses a sync if switching from a non-GCS to GCS task.
diff --git a/arch/arm64/include/asm/gcs.h b/arch/arm64/include/asm/gcs.h index f50660603ecf..5bc432234d3a 100644 --- a/arch/arm64/include/asm/gcs.h +++ b/arch/arm64/include/asm/gcs.h @@ -58,7 +58,7 @@ static inline u64 gcsss2(void) static inline bool task_gcs_el0_enabled(struct task_struct *task) { - return current->thread.gcs_el0_mode & PR_SHADOW_STACK_ENABLE; + return task->thread.gcs_el0_mode & PR_SHADOW_STACK_ENABLE; } void gcs_set_el0_mode(struct task_struct *task);
Mark Rutland noticed that the task parameter is ignored and 'current' is being used instead. Since this is usually what its passed, it hasn't yet been causing problems but likely will as the code gets more testing. Fixes: fc84bc5378a8 ("arm64/gcs: Context switch GCS state for EL0") Signed-off-by: Jeremy Linton <jeremy.linton@arm.com> --- arch/arm64/include/asm/gcs.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)